help-with-hw-design

[pale umbra]

Also first time I used it I was used as an example of what not to do because I didnt change my default 😦

like it dimensioned my parts with 10.0000 cm instead of 10.0cm

thankfully it was caught before the university send to the workshop to have it made for the next class to get visual feedback

otherwise it would have costs like 6000$ instead of 6$ 🤣

[idle python]

idk what to do now

do i get a pi zero or a radxa zero

[pale umbra]

why not get a pi 4b since they are as rare or a video processing specific MCU ?

[idle python]

because i need an sbc thats small

and i cant just get a video processing unit because i also need something to manage the lte module

right now im stuck wondering if a radxa zero on software codec is faster than a raspberry pi zero on hardware codec

[pale umbra]

I wasnt able to acquire a zero 😦 so I cant help much and I have no idea what a radxa is

[idle python]

radxa zero is like the raspberry pi zero but its much faster than the zero, and faster than the zero2

the only downside about the radxa zero is that it doesnt get enough software support

thats why you can still buy it

and you cant buy a pi zero nowadays

[pale umbra]

oh I could if I really wanted to. My local shop is an official reseller and doesnt accept non-local orders or internet orders

but Id have to walk to it

since Im in a small city and it's kinda niche vs the other products they sell (computer parts / car audio systems) the Pis often collect dust for a long time

[idle python]

lucky

i live in romania theres probably only like 15 people who own a pi zero in this country

[pale umbra]

Like they have 8x PI4B 8GB atm at 134 CAD$ atm vs 110$ on canadian pishop

but they have an habit of getting confused in product description and sometimes you walk to the store to find out it was a case or a bigger product with a pi 4 not included in it or the wrong model number etc

[idle python]

still worth it

pi 4bs are pretty rare

[pale umbra]

mine was 135$ in basic starter kit package last june because I needed the cables / adapter / sdcard / sdcard reader flob etc

the flob also works on a computer so it's really handy

and support 3 sizes of sd-card

[idle python]

i wish i knew someone who had a pi zero atm

[pale umbra]

my makerspace put their 12 or so of each pi model in a big safe since a couple of years 🤣

they get them throught an educational channel / corporate sponsorship so Im not sure how long they have to wait

so is it better to buy the pi zero W or normal pi zero ?

meh nevermind they dont have a pi zero 2W without a kit...

[idle python]

i dont see why you would buy a normal pi zero without wifi unless the normal pi zero is a lot cheaper than a pi zero w

[idle python]

cant they just grab a pi zero 2w from the kits and sell you it seperately?

[pale umbra]

no, usually the agreement with raspberry pi prevent them from doing so

it doesnt even show as out of stock

[supple pollen]

Half the price, last I checked

[violet portal]

g'day! hope everyone is well in Adafruit land

(this could equally be in #help-with-linux-sbcs I guess, but let's tryi here)

I am hooking up a Micro USB breakout I picked up yesterday to a Device Under Test; the terminals are labeled 1, 2, 3, 4, 5, SHL and the DUT just has D+/D+/Vcc/Gnd

DUT -> Breakout

Vcc -> Pin 1
D- -> Pin 2
D+ -> Pin 3
Gnd -> Pin 5
(Pin 5 / SHL are NC)

on the host side I am seeing

[  550.953227] usb 1-1-port3: Cannot enable. Maybe the USB cable is bad?
[  550.953891] usb 1-1-port3: attempt power cycle
[  552.121075] usb 1-1-port3: Cannot enable. Maybe the USB cable is bad?
[  552.973084] usb 1-1-port3: Cannot enable. Maybe the USB cable is bad?
[  552.973762] usb 1-1-port3: unable to enumerate USB device```

according to this, my pinout is ok... (and if it wasn't then the host wouldn't show anything)

on the DUT side, I am using a set of PCBite probes, qhich are designed for USB, SPI, CAN etc

but, thinking a signal integrity issue?

i have a short length of "ordinary" jumper/dupont cable connect the probes to the breakout; then a short USB cable (which I know is good as I use it for cellular modems) to the host

the USB port on the DUT should appear on the host as a 115200 Serial port, so not high speed at all

(but yes! this is horrible probing technique!)

[violet portal]

Other info:

• the Host is a Raspberry Pi 400, running Bullseye lite 32bit
• have swapped the micro usb breakout for IC clips, same problem

I have a USB Type A Plug Breakout Cable with Premium Female Jumpers - 30cm long (https://www.adafruit.com/product/4448) on the way which can then go direct to the probe tips

the probes are PCBite SQ10 from Sensepeek (https://sensepeek.com/4x-sq10-probes-with-test-wires)

[supple pollen]

Could be the pins are numbered differently on that breakout, could be the wiring just isn't clean enough for USB. Even though you're taking to a USB peripheral that's not doing something data intensive, it might still be trying to negotiate at high/full speed. You could also have a ground loop problem. You probably want to hook up the shield to minimize interference.

[violet portal]

yeah, I am pretty sure its a signal integrity issue... i just don't want to admit it 😆

and agreed on all your points... though as these are Test Points, i think my usage of probes is ok - though (and I have never probed USB before) I wonder what the best practice for this is? after all a USB cable is twisted pair and I just have long test leads flappin' in the breeze... the USB Breakout cable should be a good help

[supple pollen]

I dunno, "cannot enable" seems like it might be a power issue. Some useful discussion here https://askubuntu.com/questions/206614/usb-slots-stop-working-suddenly-from-time-to-time/833569#833569

[violet portal]

that's true; i've been stung by power issues before on Pi's with USB... I have a log on it somewhere...

hrmmm here

[  212.557335] usb 1-1.4: new high-speed USB device number 18 using dwc_otg
[  212.997344] usb 1-1.4: device not accepting address 18, error -71
[  213.097327] usb 1-1.4: new high-speed USB device number 19 using dwc_otg
[  213.537334] usb 1-1.4: device not accepting address 19, error -71
[  213.537467] usb 1-1-port4: unable to enumerate USB device```

that turned out to be an undervoltage issue

this is interesting... https://forums.adafruit.com/viewtopic.php?f=20&t=122375; this was a bad batch of pre-programmed microcontrollers

[supple pollen]

I considered asking what the target device actually is

[violet portal]

the target device is a cellular modem, and this is the debug USB port

after some more reading (and mea culpa, I have never dug into USB, and so I enjoy when bugs like this happen - good chance to learn) - the ID Pin on the micro USB should also be grounded? (https://electronics.stackexchange.com/questions/35462/why-does-micro-usb-2-0-have-5-pins-when-the-a-type-only-has-4)

[supple pollen]

Wiring integrity is extra important with a cellular modem, the cellular signal is a high powered transmitter than can interfere with nearby wiring. Maybe adding some shielding between the cellular antenna and your wiring could help. Additionally, cellular modems draw a significant amount of power. It might be worth hooking up the power lead to a separate supply to see if that helps.

[violet portal]

absolutley; unfortunately I can't provide more info than i have; but suffice to say the board itself is all good; and these test points are debug and go direct to the USB of the modem - so; 99.9% the issue is my crappy wiring effort; if i have to, ill solder a shielded USB cable direct to the TP and bypass everything else - but if I can avoid that, it would be best for all involved (i have limited test devices) - so, fingers crossed if i can put a USB cable direct to the test probes (the adafruit cable i mentioned) that will resolve the issue 🙂

a few things to try anyway 🙂

[vast flume]

Hey guys, still trying to figure out this dang i2c sensor :(. I was looking through the datasheet again and found this:

And this is my setup:

Is it possible that not having those 10 ohm resistors could be causing this?

Its being powered from a regulator so I didn't think it would be an issue

[cursive sentinel]

It has a warning about limiting inrush current. 10us is pretty slow.

[vast flume]

Is it possible it could come back to life if I add that resitor now or no?

[cursive sentinel]

to avoid damages from repeated power cycles
If this happened it is irreversible.

[vast flume]

True

The confusing part is that the sensor worked fine on a previous version

[cursive sentinel]

repeated power cycles
It is not immediate

[vast flume]

Ik but I've gone through 3 sensors on this version of the board so I would've expected at least one to be discoverable before dying

Literally worked for hours on the last version of the board

[rustic linden]

Can/should you connect VDD and VDDIO?

[vast flume]

What do you mean?

[surreal tangle]

It looks like you need to tie CSB to VDDIO... "For using I2C and I3C, it is recommended to hard-wire the CSB line to VDDIO"

[supple pollen]

I'll ask again: are other I2C devices on the bus working?

[vast flume]

Its the only i2c device on this pcb

I have it on another pcb which does have 2 other devices on it as well but that one is beyond screwed at the moment so I cant test

[vast flume]

In the datasheet it also says that you can keep csb floating bc it has an internal pullup

[supple pollen]

Presumably you do have I2C pull-up resistors

[vast flume]

Yeah 4.7k

Is it possible they could be too small

[supple pollen]

Then I'm guessing it's an orientation problem. That orientation illustration on the data sheet is confusing at best.

[vast flume]

Its so confusing

[supple pollen]

4.7k seems fine.

[vast flume]

But I was able to confirm the orientation is correct

[supple pollen]

How did you confirm that?

[vast flume]

Apparently my iphone 14 has a macro lens that I never knew about

Zoomed way in and was able to see the pin 1 marker on the underside of the sensor

Flipped it over and realized the dot on the top of the package isn't the vent hole, it is a pin 1 marker

The vent hole is actually 2 holes in a different corner

[surreal tangle]

Are you sure it is enabled by default... This makes it sound like you have to enable the pullup

[vast flume]

Bruh thats so stupid

So it has an internal pullup but you need to communicate with it over i2c to enable it 🤦‍♂️

Why does that even exist

[surreal tangle]

Doah

[supple pollen]

This does not appear to match

[pale umbra]

well they says to uses 71000 to 130000 pull-up externally

but I cant find the value of the internal pull-up

[vast flume]

Yeah it was in the wrong orientation in that picture 😅

That was before i realized those holes in the bottom right are actually the port holes

[supple pollen]

The internal pull-up isn't a resistor, it's some weird thing like a MOSFET current source

[vast flume]

Is it possible to fix this in software?

By communicating over a different protocol to enable the pullup?

[supple pollen]

I'd just connect the external pullup.

[vast flume]

How?

[supple pollen]

Connect a resistor from the CSB pin to 3V3

[vast flume]

So I have to order a new pcb 😦

[supple pollen]

You didn't break out that pin?

[vast flume]

Im a dummy

So nope 😅

[supple pollen]

Do you have any more unassembled PCBs?

[vast flume]

Yea

In theory can't I just remove the sensor and bridge the two pads

(Ignore the unrouted nets, making some other changes)

[supple pollen]

That should work, as long as nothing shorts to the bottom of the housing

[vast flume]

Ok

[pale umbra]

eh that internal pull-up thing is kinda ridiculous . Kinda like an ISP that tells you to check the knowledge base at some URL to fix the internet not working when you cant connect to the internet 🤣

[vast flume]

Ikr 🤣

I literally can't think of when someone would use that

[supple pollen]

Yeah, that's some Verizon level dopiness.

[vast flume]

lol

[vast flume]

Well this is my attempt

I tested and confirmed that the little wire is connected to 3v3

Theres no way for me to confirm if it is/isn't connected to csb

But it doesn't seem to work

After removing the sensor this is how it looks

So it def was connecting the two pins together

[supple pollen]

I'm worried that the wire was shorting to the metal case of the sensor

[vast flume]

hmm

Is there another way that I should do this?

[supple pollen]

I think you have the right idea, but a) the wire may be thick enough it keeps the other pads from making good contact, and b) the possibility of a short circuit. Unfortunately, I don't have a good suggestion. Insulated wire or polyimide tape might help with the short circuit issue, but it will make it even thicker, which is worse for the "good contact" issue.

[vast flume]

I ended up somehow shortening the wire:

Still wont work 😦

[supple pollen]

If I were you, I'd tack wires to the I2C pins and run them to some other I2C device off-board to see if the I2C bus itself is working.

[vast flume]

True

I'll give that a shot

[vast flume]

Just tested with an i2c display

It found it

So I guess that rules out an i2c bus issue

[distant raven]

The signal integrity comment only applies to signals over 10GHz, IBM disproved the right angle signal integrity myth a while ago with empirical evidence.

If right angle traces were bad for integrity, no one would use inverted F, or square meandered Wi-Fi antenna or use via for that matter

https://www.nwengineeringllc.com/article/right-angle-pcb-traces-its-time-to-kill-the-myths.php

[rigid plume]

Yeap I agree, manufacturing processes for PCBs can sometimes struggle with right-angle bends.

Although I still think it is best practice to not do 90* angles. But it just depends

[distant raven]

Sure, manufacturing is the biggest concern here. Acid traps, adhesion..

For the average design though, 90 vs 45 is going to be an aesthetic choice

[rigid plume]

Yes definitely, many prototype board houses like JLCPCB should be able to handle 90* perfectly.

[distant raven]

Or if you’re making like.. a 40G board, you’re going to use smoothed meandering for traces, not even rough 45

[rigid plume]

More arc like in that case

[meager coral]

what is the best practice for attaching gears to the end of servos? is there a cad file for the pre-installed end piece?

[rigid plume]

What's the application?

[meager coral]

two applications: 1. is a rack and pinion to dispense starburst. 2. is spinning a spool of 3mm rope to lift and lower it.

both with a continuous servo

[cursive sentinel]

It's really a matter of what you are trying to accomplish. Any time your trace length matches up with an integer multiple of the wavelength of your signal (or a harmonic of it) you can get reflections. Digital signals with fast edges can have significant harmonics high above the switching frequency. In the case of antennas, you often want that reflection.

[cursive sentinel]

Acid traps are not nearly as much of a concern anymore, and if they were you'd have trouble using surface mount packages smaller than SOIC.

[rigid plume]

you could attach a pinion gear directly to the servo shaft.

or a hub or horn

[meager coral]

the servo shaft is the white nubby guy at the end?

do I just make a circular hole that size and use loctite on it?

[vast flume]

Thats worked for me if you're FDM printing it. Just make sure to not get any glue inside of the servo itself. In my experience, hot glue has worked well since its too thick to get inside

[distant raven]

Right, modern processes circulate the etchant which helps prevent trapping

[elder peak]

Acid traps are such a "Are we talking about the D&D adventure or PCB" moment.

[pale umbra]

nice to have acid traps in PCB in case you encounter a troll

[graceful sparrow]

Oh gosh, when I was in middle school shop class.... "hot glue and duct tape fixes everything" was the word... bringing back memories.

[vast flume]

Hey guys have a quick question about the esp01

It turns out that that is what has been shorting out my other pcb (I think, with 99% certainty)

The wifi antenna on it gets extremely hot

But the thing is, I have no idea what could be the issue

This is how I connected it, super simple

Nvm turns out its a footprint issue 😅

[vast flume]

Actually never mind it seems that that wasn't the issue

I removed the esp-01 but I'm still unable to program the board and the inductor is still making tons of noise (and heating up)

I replaced the inductor and stm32 but still no luck

[tough matrix]

for linear motion:
you can buy a ready kit: https://www.servocity.com/servo-driven-gear-rack-kit-single-gear-rack/
or just a pinion gear: https://www.servocity.com/all-servo-gears/

[tough matrix]

for spool: use a plastic spool like this: https://www.servocity.com/3410-series-servo-mount-winch-pulley-25t-spline-dual-spool-112mm-circumference/
or a combination of hub: https://www.servocity.com/servo-hubs/
and matching spool

[supple pollen]

I doubt the inductor is the WiFi antenna

[vast flume]

I originally thought there was a short with how I connected the esp01 but if that was the short the inductor would've stopped being so noisy & hot

I'm using the same inductor on a different pcb where it works fine

To be fair in this board it is stepping down from a higher voltage (~16v down to 5v vs stepping 3.7v up to 5v)

[supple pollen]

It can do that if the inductor saturates

[vast flume]

What could cause it to saturate?

[cursive sentinel]

All inductors have a finite amount of energy they can store in their magnetic field. That's fundamentally what inductance is: it relates energy to current.

In SI base units 1H works out to be 1J/A².

1 H = 1 kg⋅m²/(s²⋅A²)
    = 1 kg⋅m²/s² ⋅ 1/A²
    = 1 J/A²

[vast flume]

Just looked it up and its also inducance * current, and typically saturation is ~0.3

Also it looks like mine is rated up to 4 amps

Idk how i would figure out the current in the buck converter circuit though 😅

[cursive sentinel]

On average it will be whatever your load is drawing, but the peak current can be considerably higher.

This is called ripple current and how you would calculate it depends on the design of the converter.

[vast flume]

I see

The average current should be essentially 0 though unless there is another short which is why I'm confused

All thats in the circuit is an stm32, 3 memory modules, a pressure sensor, and an imu

I've tested without the imu and pressure sensor and it still had the same issue

Also tried measuring the current into the circuit and it was much lower than 1a

[cursive sentinel]

How so?

[vast flume]

Just bc theres not really anything on this board

Besides the stm32 and sensors

[cursive sentinel]

That doesn't make the average current 0.

The first thing you'll have to determine is if your regulator is running in CCM (continuous conduction mode) or DCM (discontinuous conduction mode). In DCM the inductor is fully discharged between switching cycles.

[vast flume]

I thought you said the average current is whatever the load is drawing?

https://www.monolithicpower.com/en/documentview/productdocument/index/version/2/document_type/Datasheet/lang/en/sku/MP2307/document_id/503

This is the datasheet for the buck converter I am using

[cursive sentinel]

Yes, but you implied that the load is drawing nothing.

I think the first thing you want to do is determine your idle and peak load.

Also, what is your input voltage?

[vast flume]

15v

[cursive sentinel]

Output of 3.3?

[vast flume]

5v

Then goes into a 3.3v linear regulator

[cursive sentinel]

What inductor are you using?

[vast flume]

https://www.digikey.com/en/products/detail/monolithic-power-systems-inc/MP2307DN-LF-Z/5292007

[cursive sentinel]

Inductor, not regulator

[vast flume]

Oops sorry copied the wrong link

https://www.digikey.com/en/products/detail/abracon-llc/ASPI-0630LR-100M-T15/3059602?s=N4IgTCBcDaIIYGcAOBLAtABgGwGYMBsAnNARgwwFsQBdAXyA

[cursive sentinel]

But you don't have any idea what your board draws while idle?

[vast flume]

Barely any

If you mean the current pulled by the regulation circuit I'm not too sure

But everything downstream of that is pulling an extremely minimal amount of current

I can show you the schematic if you want to see

Or the layout

[cursive sentinel]

Schematic would be fine.

[vast flume]

Ok one min

Heres the voltage regulation circuit

Heres the microcontroller

Heres the onboard storage

Heres the sensors

And then heres the display

[cursive sentinel]

22uF 6.3V ceramic caps is pretty sketchy with a 5V output. Ceramic caps—especially ones with a large capacitance in a small package—have a significant DC bias characteristic. You need to derate for the reduced capacitance.
https://community.infineon.com/t5/Knowledge-Base-Articles/DC-Bias-Characteristic-of-Multilayer-Ceramic-Capacitor-MLCC-KBA223633/ta-p/250035

[vast flume]

I have some other caps should I try swapping it out

[cursive sentinel]

Yeah might be a good idea. You could also probably get away with putting more of them in parallel.

What caps are you using now?

[vast flume]

I actually think I already did that

I have 2 22uF caps in parallel

So 4 total

Should I try throwing some more in parallel?

[cursive sentinel]

Also are you able to hook a scope to the output?

[vast flume]

Sadly not 😦

I don't have one

[cursive sentinel]

Bummer

[vast flume]

Yeah for real

[cursive sentinel]

Do you have a datasheet for those caps?

[vast flume]

Yeah

https://www.digikey.com/en/products/detail/samsung-electro-mechanics/CL21A226MOCLRNC/5961264

[cursive sentinel]

Ok they don't have a graph for your specific cap in the datasheet, but here's one from the same family. You can try putting more caps on, but I can't guarantee you it will help. As a general rule the X5R dielectrics are crap.

Gotta go for now.

It wouldn't surprise me if your actual capacitance is 30 or 40% lower than you were expecting.

[vast flume]

Ah I see

Thanks for the help! I'll just keep adding them in parallel and see what happens then

[distant raven]

Also a good point along this, larger packages derate at a slower pace of voltage increases

0805 and 1206 will have better derating than 0603 and much better than 0402 and smaller

[rigid fiber]

O'm trying to make a schematic for this circuit from an Ali Express item page forna tiny OLED display. Q1 and Q2 are confusing me. The symbols looklike depeletion type mosfets, but I can only find FDN338P & FDN335N as enhancement type. I thought that you needed to wire depletion vs enhancement differently?

[rigid fiber]

But I think what it's wanting to do is Q1 is ESD protection a sthat seems to be a pull-up resistor to keep the gate closed and let current flow Vbat (which powers the tiny display). And R2 looks like a pull-down resistor which keeps Q2's gate opne, preventing current to VDD unless GPIO is asserted high.

Does that seem correct? Or did Is just show myself to be a dummy? 🙂

[vast flume]

@cursive sentinel @distant raven Well I swapped out the caps that were being used with some 16v 22uf ones, and that seems to have maybe brought the noise down a little bit but I'm not 100% sure. I'm still not able to program the stm32 for some reason though 😦

[distant raven]

Do you have a reset button somewhere?

I’m curious why you didn’t connect reset to the SWD headers

[vast flume]

Nope 😦

Haven't needed one before with this chip

Albiet I'm a complete noob

But my other stm board that I made is able to be programed with no issues

[distant raven]

Idk, usually reset is a signal you need for stable swd, or it’s recommended anyway by the developer docs from Arm

[vast flume]

I see

Is there a way I can "manually" trigger this to test?

Oh actually I guess not bc the pin isn't connected

[distant raven]

You can use a bodge wire

[dry pelican]

The MOSFETs are probably just for switching power to VBAT

[echo zephyr]

i have 3 methods of splitting routes for buttons but i need to know which gives the least voltage loss, on the first one i think that one wire could have less voltage because some electrons could be on a single side, on the second if the first button was pressed then the second button would get less or no voltage, on the third one the result may be the same as the first one. which is the best method or is there some other method i haven't used?

the pcb is designed to handle multiple buttons being pressed without like more than 1V loss

[rustic linden]

i think voltage loss in any of those will be negligible by themselves, because it depends on the loads connected to the buttons

[echo zephyr]

alright then, I’ll use the first one and hopefully the electricity from the buttons won’t come back and create arcing from 90 to 135 degree angles

oh the maximum is 90 degree angles and 0.3mm routing width shouldn’t produce arcing with 3.3V

[rustic linden]

You'd need so much more to arc

[echo zephyr]

makes sense

the voltage and current is too low to cause even small arcing

why did I just realise that

[echo zephyr]

also another question, is a via with 0.915mm diameter and 0.458mm drill diameter enough for 5V with 1A?

[supple pollen]

"Probably". It depends on the copper weight, via fill weight, and hole accuracy. Normally if I'm designing something to carry non-trivial current, I'll use several vias ("via stitching")

[distant raven]

I like to do via stitching for large current with smaller via because I make increase the overall surface area for current to travel

[supple pollen]

Good point: a single via will funnel the current down to the area of the trace that's nearest the via, wasting part of the current capability of the trace and increasing voltage drop, heating, etc.

[idle python]

how many mbps could a raspberry pi zero handle on download and upload

deciding whether i should get a7670e or sim7600g-h

[supple pollen]

I'm guessing the Pi won't be the bottleneck, although if you're sending data to the SD card, that could be.

[idle python]

no im trying to stream video from csi camera with a 4g m odule

average upload speed is 11mbps where i live but a7670e has a max upload rate of 5mbps

[supple pollen]

1.2MBps seems to me like something the Pi could handle, depending how much much processing is involved. However, I really don't know. It would be easy enough to test just having it stream from the camera through one of the USB ports. It may also depend on what the interface to your cellular module is. PCIe is plenty fast, but I don't think the Pi supports that.

[idle python]

the module i found is usb only

[supple pollen]

Ah, then a streaming-to-USB test will be pretty representative of the Pi's capabilities.

[idle python]

representative?

[idle python]

i did a quick test on obs and a h264 stream at 1920x1080 60fps with 4000kbps bitrate is only 3000kbps on bandwidth

baseline profile btw

[woven bluff]

pi4B can do 54Mbps on 5G at best

[idle python]

i bought the a767e

[vast flume]

How possible is it that I've killed my Segger J-Link programming adapter? Neither of my 2 pcbs is able to be programmed now so I'm super confused

I guess I'll try ordering some more stm32s to rule out having killed both microcontrollers

[supple pollen]

It might be worth using some other piece of gear (depending on what you have available) to see if the J-link is producing the expected signals

Or emulating the protocol with something else to see if your PCBs answer up

[dusky raptor]

Is there anything I should add between the rasberry pi pico and the rotary encoder???

[supple pollen]

You probably want to hook up the ground

Ah, I see it now, the dark blue on black didn't show up very well on my screen

[vast flume]

If I've burnt out an stm32 is it possible that the power and ground pins could short?

[woven bluff]

if you power it up and the chip gets hot immediately then it's wasted.

[vast flume]

Ah yup that explains it 💀

Ig my programmer isn't dead its just the chips

[jade wedge]

I'm trying to work out the usage of a shift register, and I want to make sure I'm reading this correctly. The goal is to read the state of 8 dipswitches one at a time, and I think the 4021B will do this for me.
https://www.ti.com/lit/ds/symlink/cd4021b-q1.pdf
Now, if I'm correct, I can latch the 8 inputs by bringing pin 9 high, then pulsing pin 10 high. After that I can bring 9 low, and get the data out on pin 3. Shifting with positive pulses on pin 10. Pin 11 can be tied to either ground or VCC depending if I want the registers to fill with highs or lows.

That all sound about right? I don't want to use an I2C input expander as I don't need anything fancy from this, and I'm also hoping to limit my use of external libraries.

[supple pollen]

A 4021? That's old school. I usually reach for a 74HC595 for such duties.

[jade wedge]

Mostly going by what's available on Digikey in thru hole format.
Plus they have the 597 which seems to be more complex, and has a less helpful datasheet.

[supple pollen]

Ah, that makes sense. I like the '595 because its popularity yields helpful pages like https://docs.arduino.cc/tutorials/communication/guide-to-shift-out. These look to be in stock if you care to go that way. https://www.digikey.com/en/products/detail/texas-instruments/SN74AHC595N/375877

[jade wedge]

Oh, also the 595 is the opposite function. Unless I'm reading it wrong. The 595 looks like it takes a serial input and makes it parallel. I have parallel input I need to make serial.

[supple pollen]

Whoops, sorry about that. I had it backwards.

[jade wedge]

Well, assuming I have the usage of the 4021 correct, I have at least built myself a schematic, and board. Ok, they're probably not that great, and I'm definitely putting the cart before the horse, but I've got little else to do for this project while waiting for parts, and learning how to program an ATTiny directly. Still, I don't think they're too bad for a first try. The idea is a simple NeoPixel style controller board. I don't think I'll need a cap for smoothing on the LEDs as I'm only going to use two of them. Mildly annoyed that I couldn't find a footprint for a through hole USB Micro connector.

[supple pollen]

Good use of a resistor network.

[jade wedge]

Figured better than 8 discrete pulldown resistors. Also took a bit of finagling to get the routing to not require a huge amount of over/under.

I realize now though that I'll likely need at least one to pull reset high on the ATTiny. Probably not good to leave that floating.

But hey, like I said, I've got time to work on this.

Breadboarding is first anyway.

Ahh... Phooey. I can't use the 4021. "Quantity entered must be a multiple of 451"

Guess I'm learning the 74H597.

[supple pollen]

Ah, marketplace part? That's annoying. In this design, I took advantage of the ability to remap pins to make routing neater.

[jade wedge]

Clever. In my case there is an advantage of swapping chips. 7 of the 8 inputs are on one side, and in a row. Super convenient for me.

[supple pollen]

Oh, I appreciate that sort of thing!

[jade wedge]

Indeed. Convenient, and visually pleasing. A proper flow of data from one end to the other.

[vast flume]

Hey guys, would somebody be able to help me fix the inductor heating issue on my buck circuit?

[supple pollen]

Remote debugging of switching power supplies is tricky, but I can try...

[vast flume]

What would be the first place to start at? 😅 I've tried adding more output caps in parallel & replacing the inductor

This is the current circuit for regulation

And this is the buck ic I'm using https://www.digikey.com/en/products/detail/monolithic-power-systems-inc/MP2307DN-LF-Z/5292007

[woven bluff]

SMPS is not trivial, you need at least a scope to find out what's wrong

[supple pollen]

It looks like it's hooked up wrong. As in the inductor does not go between the supply voltage and the SW pin

Oh wait, +5V is the output, not the input. Sorry, I'm used to seeing the input on the left and the output on the right.

What's the output current you're drawing? Which inductor are you using?

[woven bluff]

output filter cap voltage rating is too low

[supple pollen]

Also, layout is important in these designs. You have high currents and fast edges in the power side of the circuitry, which can easily couple to the feedback node and cause problems.

[woven bluff]

the schematic on the first page is for 3v3 output

[supple pollen]

If the capacitors are getting warm, you have an overvoltage, ripple current, or leakage issue (monolithic ceramic capacitors with high capacitance in a small volume are particularly susceptible to this)

[woven bluff]

did you follow recommended layout on page 10?

also why did you chose a chip that is NRND?

[vast flume]

Shouldn't draw much current at all, since I don't have much downstream currently. Just have an stm32, and some sensors. I've tested with those all removed, and even with the 3.3v linear regulator removed to remove the 3.3v circuit, but it still occurs

[vast flume]

I've since switched to a higher rated cap (16v) https://www.digikey.com/en/products/detail/samsung-electro-mechanics/CL21A226MOCLRNC/5961264

[vast flume]

Someone here had recommended it to me as one of the easier chips to get working I think

[woven bluff]

x5r is ok, but it's better to use x7r

[supple pollen]

I may have mentioned that chip, because it's the one used in scads of inexpensive but solid UBECs including the one AdaFruit sells.

[vast flume]

Actually I didn't 😅 . Here is what my layout looks like:

[woven bluff]

traces too thin, inductor too far

[vast flume]

Ah jeez 😦

Just curious, how does trace length correspond to the heating?

[supple pollen]

Feedback loop too big, it'll pick up noise inductively and capacitively

The long traces in the switching loop broadcast signals like an antenna, transformer, and capacitor all at once. The long traces in the feedback loop pick up those signals as noise, confusing the chip and causing it to operate improperly.

[vast flume]

I see

So thicker traces and shorten the feedback loop?

[supple pollen]

Shorten both the feedback and power loops.

From the datasheet, the PCB Layout Guide, paragraph 1 says "Keep the path of switching current short
and minimize the loop area formed by Input cap., high-side MOSFET and low-side MOSFET.", paragraph 3 says "Ensure all feedback connections are short and direct. Place the feedback resistors and compensation components as close to the chip as possible", and paragraph 4 points out "ROUT SW away from sensitive analog areas such as FB." All of these details are important for a switchmode power supply to operate smoothly.

[vast flume]

Ah ok perfect thanks

[pale umbra]

"2 V Sealed Lead Acid (SLA, VRLA) Battery Rechargeable (Secondary) 200Ah" - Why?!

[elder peak]

Are you feeling lead astray?

[supple pollen]

I've seen cells like that in phone company offices for power backup.

[pale umbra]

No I just fail to comprehend why youd need such a long duration power source and ruins it with such a low non-standard voltage

[supple pollen]

Often there will be several such cells in series to provide whatever voltage the equipment needs. Having it in separate cells means several heavy items instead of a single massively heavy one. It also makes cell balancing easier, as the cell-to-cell connections are available. In the phone company office, there were 24 such cells in series to provide the 48V the switching equipment used.

[dry pelican]

Longer increases resistance, which increases power loss

I use ~1mm trace width per amp when I can

That's my rule of thumb

There are also calcs

[vast flume]

Thanks I'll keep that in mind!

I've just been using the same trace width for everything which obviously only gets you so far lmao

So ig I'd size the traces off of 4 amps? So 4mm?

[distant raven]

Do 2mm and 2oz copper, you should get the same carrying capacity

Less board space occupied. You should use 2oz anyway when possible with high current applications

[supple pollen]

Since it's the inductor that's heating instead of the traces, I suspect trace resistance isn't your current [pun intended] problem. But once you get that sorted, you may find you have some warm traces.

[distant raven]

I suspect there’s probably some strange induction loops in the board, non-optimal return paths increasing ESR

It’s amazing how you can introduce noise by having not great return paths for signals

Which is a different discussion

But something I’ve been thinking about recently

[supple pollen]

The larger the loop area on the switched side, the more noise it transmits. The larger the loop area on the feedback side, the more noise it receives.

[distant raven]

Yup yup

[dry pelican]

Why do you need that much current?

[vast flume]

Its a buck converter so the datasheet has the parts spec'd out to 4a

[dry pelican]

you only need traces rated for the max current the converter will be supplying

Here's my layout for an LMR51420 buck

probably not the best

but it works with no inductor overheating

[dusky raptor]

is this schematic okay ? I am really new to this project and I am trying to build a macropad of my own

[rustic linden]

Here's a tip! You can use net labels so you don't have to individually wire up things, you can just place a label at the points you want to connect

[supple pollen]

Your LEDs won't be individually controllable, and they need current limiting resistors.

[pale umbra]

Is there a way to filter all available chip by connector types on digi-key (or another site) by anything that works in throught-hole or that is very easy to use an adapter for (like SOC) /TOS/those that have cheap socket you put them into ?

[dusky raptor]

should I add current limiting resistor to each individual LED or just add a big one on parallel with them, also how much kohm/ohm should I add with them?

[dusky raptor]

also about controlling individual led, I was hoping for a matrix configuration , but as this is my first board, I don't know how am I supposed to do it

[supple pollen]

Yes, DigiKey offers a general "through hole" filter, as well as more specific package filters

[supple pollen]

You generally need a resistor per LED. For matrixing, you normally connect the anodes to various different GPIOs and the cathodes as well to various different GPIOs, and then scan through them by activation combinations of GPIOs to select the LED you want to light, and cycle through them.

[jade wedge]

I've been mulling over this 74HC597 truth table, and I think I can "cheat" my pin usage a little bit. I don't care about clearing the registers, so MR can just go straight to VCC. The clock pins can be combined as there's no case where both ST and SH can both accept a pulse. Which just leaves PL.

[pale umbra]

I meant site side, without having to select hundreds of individual categories or at least being able to filter all ICs. Like from this page I dont see any filter and I would like to select all possible DIP-8/in stock with datasheet

[jade wedge]

If you filtered at the site level, you'd still end up with tens of thousands of results. Lots of components come in through hole variety. Even some rather large DIP configurations.

[pale umbra]

That is fine. I have to keep redesigning circuits because I cant find the right IC in stock with the right package. Would be easier to start from a virtual bin of available chips and work from there

ie: rather than have my project pick a chip it would be the opposite, available chips determines the projects I can make

I think so far only API calls would be able to do this but I wanted to validate here first

Seems doing an API will require me to pay for hosting though 😦

Basically the X=>Y problem Im trying to solve it that finding the correct chip in stock when I need it is the major limiting factor for my projects

So I want to check if the opposite approach will work. getting a vault of components and that will determine what I can do / being able to enter what I need into some sort of assistant and it's smart enough to find things speced for 3.3V/5V boards along with a socket/adapter if needed so at the end I just have a BOM to enter in the site and I know everything is in stock and I can see the price

[supple pollen]

Those filters show up after you've chosen a category.

[pale umbra]

I know but there are thousands of category I cant possibly spend a week manually filtering everything 😦

[supple pollen]

You don't want things from thousands of categories. Generally when you're looking for a chip, you know what sort of chip you want. Having a jumble of every 8-pin DIP would not be useful. But every 8-pin DIP packaged op-amp could be useful.

[pale umbra]

ok 😦 that's the thing sometimes I want something but I dont know what all the subcategories means . Like if I want a RF receiver. I dont know if anything from those other categories is required or if it will be provided on the chip if I pay more.

ie: those things might be included in an adafruit breakout but I wouldn't know if I get chips by themselves without a breakout

[dusky raptor]

I just made my research and My smd 3528 LED RGB has a vF(3-3.2v) and my board supply voltage is vS(3.3v) Imax = 20mA ... if my math is correct I need 5 ohm of current limiting resistor ... is it worth it though ?

[pale umbra]

I wish they had like a definition pdf where they explain what each category mean ,,,

[supple pollen]

Let's see, 3.3V (supply) - 3V (minimum forward voltage) yields a voltage drop of 0.3 volts. 0.3 volts divided by 0.020 amps is 15 ohms. That is pretty low. However, you'd probably damage LEDs without it. Do you need the full 20mA?

[supple pollen]

The actual receivers would be included in finished units, receivers, and transceiver ICs and modules. The problem of knowing what you want is more of an electronics problem than a DigiKey problem. When you say "I want an RF receiver", that's a very general term. Normally when you're buying a part, you're trying to solve a specific problem like "receiving the signal from a garage door opener transmitter" instead of "receive any kind of RF at any frequency, power level, and modulation"

[pale umbra]

Im just trying to make my own kit of generic IC to enable me to do various projects without it needing to be perfect. Like if I want to listen to some audio an lm386 would do until I have something better. In the same thinking if I want to identify the frequency of a transmitter for an electronic a rf receiver X might do to enable this project until I get something specific etc

ie: I have this part X in my ICs vault but it's not perfect but at least I can continue my project until the part I want is back in stock and it get shipped in 8 months/2 years

[supple pollen]

That's a rather large range of capability. Things like "listen to some audio" or "blink an LED" are fairly basic tasks for which there are popular ICs available. Things like "identify the frequency of a transmitter" are fairly advanced tasks that call for things like frequency counters or spectrum analyzers: more of a test equipement task than a "drawer of parts" task.

[pale umbra]

eh I thought that tunning manually to a frequency close to what the transmitter is would be sufficient 😄

[supple pollen]

That said, there are parts kits available with a useful range of chips and associated components, both generic ones, as well as more specific ones (like "analog" or "digital" or "RF"). Additionally there are kits offered by manufacturers to show off their range of available products. For a pretty comprehensive list, the "Learning the Art of Electronics" site https://learningtheartofelectronics.com/parts-lists/parts-lists/ has spreadsheets (digital and analog) you can download of the parts used in their book of lab exercises, as well as buy lists from various suppliers.

The "tuning manually" approach is a valid one, you just need a tunable receiver and some sort of signal strength indicator. However, you still need to decide what range of frequencies you need to tune through. You can cover a lot of ground with an inexpensive SDR module, but you need software and an antenna to make it work. Physically tuned receivers generally have a limited range (for example, an ordinary AM radio tuner covers around 540kHz to 1.7MHz, which is not useful if you're looking for a signal in the microwave band).

[pale umbra]

nah it would mostly be stuff in the 900mhz and 2400mhz range. Someone suggested this approach if I buy a toy/r/c stuff and I want to control it myself. They said they look at the schematic and figure out the frequency multiplier to give a range says 2401.56mhz to 2401.70mhz then they manually tune with a SDR usb key until they get something using the remote right next to it. I though this would be nice to have. This might also helps me to figure out if there are still ham sin my area as well

For the other need it would be nice to have a generic/chip IC for each category on digikey as part of a component vault. I saw a lot of kits but they often have very cheap ICs that have a lot of tradeoffs like a DHT11

seems like a waste to get something very cheap/bad like a DHT11 that Id immediately replace for most project and spend 1$+5$ for a decent sensor vs just getting the 5$ one to start with

the lm386 is another example. Could get a stereo version instead and pre-assemble an interface with it on an adafruit perfboard with sockets. That way if I quickly need to listen to something I just have to plug it

[supple pollen]

R/C modules are generally 2.4GHz (with a variety of spread spectrum modulations) or 27 or 49 MHz (with simple PPM modulation). The schematics don't give a lot of detail, but looking up their FCC IDs will give the frequency range.

It's true, a lot of kits out there are mostly cheap stuff that's not very useful.

[pale umbra]

gates is another huge class of ICs that Id like to have two of every kind. So whenever I need to do logic circuits I already have something

memory too I guess... 64mb...

[supple pollen]

There are only a few basic gates (NAND, NOR, AND, XOR, inverter) with varying numbers of inputs. Getting the 2-input versions of each plus a hex inverter is 5 chips, no big deal.

[cursive sentinel]

You can also build every other gate with NAND or NOR gates, although it's less convenient.

[supple pollen]

Eh, 6 chips (I forgot OR, but you can use NAND + an inverter to get AND and similarly for OR)

Some people like to add flip-flops, counters, maybe a 7-segment encoder/driver, but it's still a manageable number of chips.

[pale umbra]

a cheap gate with a couple of port should be enough for a basic need I guess

It's really just to have a vault of parts so that I "have something" I can sorta work with until I get something better. Like my arduinos kits dont have any speakers or anything like an lm386 or a stereo version. It's mostly build my own kit based on the types of IC in the list madbodger gave and the one in the ARRL handbook, eevblog list etc

[cursive sentinel]

I wouldn't invest too heavily in individual logic gates. They are occasionally useful, but in general I find it best to accumulate chips as you do different projects.

[supple pollen]

Even the "Learning the Art of Electronics" list seems a little excessive with its 16 chips. Do you really need six different kinds of counters?

[pale umbra]

Im glad you understand what Im trying to do here. Been browsing the whole adafruit catalog to get some idea and getting some stuff to enable me to do some projects. But digikey obviously has way too much stuff to do this 🤣

[supple pollen]

DigiKey does have a category for integrated circuit kits https://www.digikey.com/en/products/filter/integrated-circuits-ics-kits/658

[pale umbra]

On the hardware side I just need a generic cheap camera and a generic cheap H-motor shield. I have a variety of displays to be able to debug several things. Also when possible I pick digital sensors with built-in accessories like pull-up resistors or various functional capacitors so ideally I just feed it power and it does everything for me and works with my raspberry pi too 🙂

[unique patio]

it's good to start with buying stuff with some projects in mind, and buy parts for those, rather than just accumulating inventory. Inevitably I find that I am missing one particular thing, even though I may have a stock of other stuff.

[pale umbra]

Im trying to break the "create project from scratch" to create actually useful stuff complexity barrier but Im finding that when you dont have a tutorial all sort of problems prop up

[cursive sentinel]

If you do want general purpose logic gates, I recommend the 74LVC1G99.

[supple pollen]

That's sort of what Adam Savage gets at in his book "Every Tool is a Hammer". He tries to minimize things that use up his motivation, like waiting for parts or trying to find the right thing.

[pale umbra]

like LCD keep staying open so I have to use a button/proximity sensor. Have to worry about battery life when I cant leave something always connected to USB etc

[cursive sentinel]

It's only available in surface mount unfortunately, but depending on the pins you use it can be turned into pretty much any gate you want, plus a 2:1 multiplexer.

[pale umbra]

Dont you need to have a clue for that ? Have a good general knowledge of tools/materials/3d printing/folding stuff and glue etc ?

[supple pollen]

Proto-Advantage and other services will procure it and mount it on a DIP carrier for you if you like

[cursive sentinel]

I just mounted mine on a bunch of Adafruit breakouts.

[pale umbra]

It's kinda like when you start C/C++ and you dont know the std library. You dont have a clue / overview of what is possible / what libraries to use. That is kinda where I am at with electronics. I worry about "syntax" without having a general overview/clue of knowing what type of parts to get for a specific end result

[unique patio]

but also "the perfect is the enemy of the good" (Voltaire); sometimes i don't use something because it is slightly not exactly what I want. Like "Oh, don't use that project box because it's a little too big, but this other one that I don't have woul be better." Could just use what I have.

[cursive sentinel]

That's where you figure out what project to do, and then research what you need in order to do it.

[supple pollen]

My main takeaway from the Savage book is that I spend too much time rummaging around in drawers for tools I commonly use, so I swiped his idea of a "tool ladder" on wheels with the commonly used tools all in easy reach. As for the part supply problem, I took a different approach.

[supple pollen]

That's what MulhorandEmperor is getting at: which bits are needed to have a "close enough" part on hand for a variety of projects.

[pale umbra]

yeah even with construction systems I basically have nothing but low-force cardboard boxes

it's ok to hold very light parts but not much else. Im looking at lego technic clone kits to solve that problem currently

[cursive sentinel]

FPGA board.

[pale umbra]

Since hard plastic can have more weight than cardboard

[supple pollen]

I use cardboard a lot for quick construction, but it's not always up to the task (as you can see, I made my tool ladder out of chipboard, which was cheap and good enough for the job)

[pale umbra]

Im ok on the tools though. Like my landlord electrician came here the other day to repair my range hood and I had everything he was trying to find in his kit 🤣 Jokingly said he should hire me as his electrician assistant

[cursive sentinel]

3d printer go

[pale umbra]

no I dont have the space / time to wait/supervise it

enclosures not really the problem atm anyway. I have a ton of small cardboard and metal boxes anyway that I can sorta work with if precision/weight capacity isnt required

[supple pollen]

One thing I did was buy some trays of drawers full of random electronics bits from a seller on Craigslist. It wasn't very expensive, and while a lot of it was of limited use, I do like having a handful each of 40 different 7400 series TTL chips just available for when I need something.

[cursive sentinel]

Time isn't really a huge factor, as building the things you want will invariably take longer than waiting for the printer to do it for you.

[supple pollen]

For 3D printing, I generally just light off a print before I go to bed, and the part (or random bird nest of filament) is waiting for me the next morning, like a "Santa Clause" machine.

[pale umbra]

I mean I cant leave a 3d printer without supervision for 18 hours since it heats a lot. Kinda like you dont leave a dryer running for 16 hours while you are not at home. And it would probably be too noisy to sleep

[unique patio]

yes, that's why to start with some specific projects. Things like resistor assortments are obviously general purpose. But otherwise I started knowing what to accumulate by looking at a LOT of projects and seeing which parts were common. This was 35 years ago, so it was all on paper (magazines, books). I think it was actually easier than trying to find out interesting things on the Internet, which suffers from way too much stuff that's not already curated for you. Your start with the ARRL handbook is good, but is short on projects. Make or Hackspace magazines, and the Adafruit Learn system, are also good sources of curated material. It's a pity Popular Electronics and the like are defunct now. The books "Electronics from the Ground Up" and "Make:Electronics" are good starts.

[pale umbra]

And as I said I dont it it yet. Lot of videos around on using lego technic with electronics and I used it a lot before. Same for tamiya plate carriers

[pale umbra]

yeah you are so right about popular electronics 😦 I saw it but Im not sure tutorials from 1978 will be of much uses. Also not sure a 555 is of much uses when there is a lot of upgrades for it

[supple pollen]

Gates are gates, but these days I'd suggest CMOS instead of TTL. I still use 555 chips, they're robust and handy.

[cursive sentinel]

They really aren't that noisy unless you go for a fast one. As far as leaving it without supervision is concerned, I wouldn't leave mine unattended for 18 hours, but you don't have to constantly watch it either. Also only really large parts will take 18+ hours. My typical print times are 2-4 hours.

[pale umbra]

I dont have specific projects in mind at the moment. My plan is to take a semi-complicated one and extend it but for this I need generic parts. ie: kinda like when you have a box of generic lego bricks and start from the pirate ship then it ends up being a flying sci-fi ships with cops and turrets and leds and sensors. ie: just want to use creativity

[supple pollen]

I wouldn't leave house with a dryer or 3D printer running, but I don't sit around and watch them either. Mine's in the basement, so I can hear it sometimes from my bedroom, but it's not intrusive.

[cursive sentinel]

I think I've said this before, but electronics are not legos. You can't just plug things in like that. More often than not you will need something specific to do what you are describing.

[unique patio]

i have not tried to just pause a 3D print, but my printer software does have that. I'd think it would start up again fine after waiting for the temperature to go up again.

[cursive sentinel]

The bed still needs to stay hot in order to keep the part from detaching. There can also be visible artifacts since it's hard to stop it in a specific place.

[unique patio]

i use PLA on an unheated bed 🙂

[cursive sentinel]

I don't use PLA.

PETG and ASA mostly

[supple pollen]

Some electronics are like Legos, but that's more already-integrated stuff like LittleBits. Although there are some basic snap-together electronics like Snap Circuits and Lectron.

[pale umbra]

No but you can sure have a lot of modular kits that add a capability in any circuit. Like I have a dice sized OLED with the pullups already on it on a perfboard. All I have to do is to connect it with others to give display capability to anything. Idea would be the same with a stereo version of LM386. Same with an LCD with a proximity sensor to save power only keep it on for a few seconds when a human show up with a button to turn it on/off manually

[supple pollen]

STEMMA/Grove lean in that direction as well

[pale umbra]

so if I took these two for instance I could make some basic project and easily add audio and oled feedback to it in a couple of seconds

[cursive sentinel]

It's often more complicated than that. You may be able to power off the LCD, but you'd also have to disable the logic outputs on whatever is driving it or it may backpower.

[pale umbra]

same with a datalogging modular perfboard or a air 9$ adafruit thing to add wireless to any project

[cursive sentinel]

I'm not saying you can't modify things, but having general purpose logic gates won't really help you since you generally need something more specific.

[pale umbra]

My point is that I can do these modular things and make them fit in any project to do modular thing once they are perfect. Instead of trying to fit in all the components for that part of the circuit I can just plug in with a JST/qwiik connector. As I make more and modular things a custom project will get more and more simple

but yeah kinda like seed/grove/m5/littlebits/bbc microbit but without the 300% markup 😄

[supple pollen]

A lot of my prototyping takes a similar approach, sticking breakout boards, chips, components, etc. into a solderless breadboard to assemble whatever functionality I'm looking for at the moment.

[cursive sentinel]

Aside from LittleBits, none of those have a 300% markup. Adafruit's boards are more expensive because they are based in New York and go to great lengths to get genuine parts (something the hundreds of AmazonExpress distributors don't do) and provide actual support for their products.

Stemma QT/Qwiic are plug-n-play because they were designed to be.

[supple pollen]

A lot of the breakouts in that picture are AdaFruit, for exactly that reason: I'm trying to build stuff, not waste time debugging bizarre issues caused by cheap knockoff components.

[cursive sentinel]

Grove claims to be plug-n-play but they really only did the plug part, and they use weird proprietary plugs.

[pale umbra]

Also some ICs are very good and they are a whole project by themselves you just have to find them

[pale umbra]

yeah Im trying to do something like that except remove as much wires as possible to replace them with stemma qt connectors or JST or female/male headers

I dont like seeing wires or accidently deconnecting one and I have to start the whole thing over

In an ideal world any project I do would looks like a motherboard

ie: no wires at all, uses ribbon cables and other connectors all the way

[cursive sentinel]

Then you don't want bare ICs.

[supple pollen]

There are some products like that, with a motherboard that has the interconnections, and snap-in modules that do various things. You just put in the modules you want for the function you're building. However, a general solution is impractical, so all those products are limited to a fairly specific set of kinds of functionality.

[cursive sentinel]

And sometimes you still need an odd bodge wire or two... or 10.

[supple pollen]

Since I build all kinds of things, I just accept the wires for the ability to connect anything to anything in any configuration I like.

[pale umbra]

also RE: the 3d printer. Ill when I know solidworks enough and have enough things to print. Until then I can cardboard/lego technic

[supple pollen]

Like if I suddenly want to play with Nixie tubes (hey, it happens), I just get some out, along with a power supply and the appropriate driver chips and wire them up. There aren't a lot of kits out there that offer nixie tubes as a component.

[pale umbra]

would probably be a creality cr-x for the same reason I got an hakko fx-888d . Easy of order of spare parts / avoid the customs 😄

[pale umbra]

I understand that/ If I take the stereo amplifier I mentionned in #projects I can make a perfboard for it and instead of trying to figure out with it has a 20 uF cap each time and wasting time finding the datasheet etc I can trust that I did my research when I did the perfboard and I can just connect to it with my 4 audio sources, 5V/3.3V and GND

[cursive sentinel]

Unlike Hakko, Creality makes cheap crap.

Case in point: this is a multi-thousand dollar characterization board at work. In order to perform the measurements I needed I had to bodge a signal through a spare level shifter channel.

[supple pollen]

Or just buy this, with the engineering baked in https://www.adafruit.com/product/1752

[cursive sentinel]

(reuploading because there was a name on it)

[pale umbra]

but where's the fun if I dont make my own breakouts ? 🤣

how am I ever supposed to understand transistors/filter caps if I dont make my own op-amps one day besides the very basic theory in a tutorial?

[supple pollen]

Heh, that looks familiar. I'm also fond of using little MCUs as daughterboards.

Yes, there is real value in making your own breakouts.

[cursive sentinel]

Yeah we use them mostly as an I2C bridge.

[pale umbra]

For my next experiment Id like to try to get that stereo thing running from a chip, make the sound as nice as possible etc

The problem with kits is they often have transistors and I never had to uses them yet because of break-out etc

Also I want to get more familiers with buck/step-up/step-down

[elder peak]

Also, making your own adafruit-esque breakouts is pretty darn great.

[pale umbra]

with a fire extinguisher nearby 🤣

my 10Ah / 5V usb-c battery bank cames in a lipoly fire bag. Not sure what to makes of it. Could be A) here is a bonus or B) You'll need this for normal operation of it

[jade wedge]

Probably a mailing requirement.

[supple pollen]

Learning basic transistor operation is useful

[jade wedge]

Also just a sanity check question about pull up/down resistors. I don't need one if I'm tying an input one way or the other with absolutely no intention of ever toggling that input, right? At least as long as my source voltage doesn't exceed the pin's maximum rating.

[pale umbra]

Is there a point to having a ADC or DAC > 12 bits ? Also if you connect it to a GPIO youd have to translate the signal before it gets back to the GPIO right ?

[supple pollen]

Yes, pull up/down resistors are used when an input might be not connected to something, to provide a default level.

12 bits is generally enough DAC/ADC performance for general circuitry. For precision or high fidelity, you might need more bits, but you also need lots of other considerations.

I'm not sure what you mean by "translate the signal before it gets back to the GPIO".

[jade wedge]

That much I get, but in this case the pin will never be used for anything except the default value. From what I understand the resistor is so you don't short your power supply when you try to override the default.

[supple pollen]

Right.

[jade wedge]

This specific case, I have the reset pin on an ATTiny, but I never plan to use it. So is it safe to forego the resistor, and go straight to VCC?

[supple pollen]

Many times I'll just connect inputs to Vcc or ground if I'm not planning on switching them.

[jade wedge]

Ok, that's what I figured, but I just wanted to do a sanity check.

[supple pollen]

I think if you never plan to reset it, you can just tie that pin to Vcc.

Or configure it as an output, and get an extra GPIO

[jade wedge]

The only thing I worry about doing by that is losing the ability to reprogram the chip since I think without the reset pin you need to use "high voltage" programming.

At least, I think that's a side effect

[supple pollen]

Correct.

If you put the chip in a socket, and the socket pin is tied to Vcc, you can always remove the chip for programming.

[jade wedge]

Oh I meant setting the pin to an output loses me the ability to reprogram. Obviously connecting it to Vcc without the ability to disconnect would do the same.

[pale umbra]

like if I do something like this signal - GPIO 14 - GPIO 12 - 24 bits DAC - GPIO 4 - read input it's useless because the gpio have a built-in DAC that will ruins the 24 bits precision one.

ie: I have to translate it to a 24 bits digital stream before it comes out of the gpio or the arduino/whatever built-in ADC will convert it to a lower precision digital signal

and after that it's too late

[supple pollen]

DAC is digital to analog, digital on one side, analog on the other. ADC is analog to digital, analog on one side, digital on the other.

You wouldn't read a digital signal with an ADC anyway.

[pale umbra]

yeah I know that but my point is that the arduino have one

if I buy the same with better precision and Im not careful where I put it the arduino will ruin the measure

kinda like having a gigabit ethernet card is useless if you have 10 mbits network cables

[supple pollen]

An Arduino has some ADC pins (not a DAC). However, if you have an external higher precision ADC, it's a digital signal after the ADC. The Ardino's built-in ADC wouldn't be part of the signal chain.

Although you can use an Arduino's ADC pins as additional digital inputs.

It's like if you have a computer with a low-end video output on the motherboard, and a fancy graphics card. You can ignore the built-in one if you want and just use the fancy one.

[dry pelican]

Yeah the arduino shouldn't be a bottleneck (at least not on accuracy or precision) with a high precision ADC

Maybe speed

But that's only if you're moving data really fast

[pale umbra]

It's just that I read somewhere that if you uses a 24 bit DAC and ever uses the data back to analog / store it in an analog play etc everything has to follow from it it's the hidden cost of it

[dry pelican]

Nonono

That sounds like an audiophile myth

[pale umbra]

like if you play it on a speaker now you need a 24 bit ADC. If you read some signals like RF/video you may need to convert it several time between analog and digital and all the components now need to have the same bitness

[dry pelican]

Digital is digital

Doesn't matter if your SSD is gold plated

But digital is quantized

And there can be signal errors

But that's only if things get really bad

[pale umbra]

but if you uses says an amp and the signal is now digital

wont you need to convert it to analog, same bits, then convert it back same bits ?

[dry pelican]

???

[pale umbra]

same if you need to condition it, filter it etc ?

[dry pelican]

?

[pale umbra]

Im not sure what I said wrong

but surely a 14 microvolts microphone signal needs to be amplified as usual

[dry pelican]

Yes

[pale umbra]

if it's digital you need to convert it to analog before amplifying it ?

so that eventually you can play it on a speakers that uses 40W and needs the signal amplified in the 24V range analog ?

[dry pelican]

No

You can do amplification in digita

From your raw (mic) signal, you need to amplify it, then you can convert it to digital, it can be passed through DSP and other things, and then it obly has to be converted to analog when it goes to speakers

[pale umbra]

so when it goes to speakers. If I uses a 12 bits DAC at this point wont I lose fidelity on the sound ?

if it was converted to 24 bits digital initially ?

[dry pelican]

If you used 24 bit adc then 12 bit dac, you will lose resolution

Not necessarily fidelity though

The signal values will just be rounded up or down

[cursive sentinel]

It's... more complicated than that. There's an entire field dedicated to A-to-D and D-to-A, and getting accurate reproduction of a signal through a DAC often involves a combination of digital and analog filtering.

[pale umbra]

ok Ill try to find the context and asks later. So book was proposing a circuit and said that of ofc if I uses more than the ADC/DAC that comes with the arduino I have to the same precision on both sides otherise it's pointless as information will be lost. Seemed very obvious but some nuances must have been lost on me

[cursive sentinel]

First, you have to determine whether you care about information being lost. Human hearing is a bit... odd.

[pale umbra]

I dont know. Never been much swayed by audiophile arguments. I guess I let my guard down. I just know that it is a field that is well-known for selling pseudo-science based 1200$ power cords

[cursive sentinel]

That's pretty accurate.

That being said, good audio reproduction does exist, and doesn't need a $1200 placebo.

[pale umbra]

DAC are basically 10$+ so I wanted to check here if it was generic enough. Where as a 30$ ultrasonic 40 feet on A band 12Hz/ <4 feet on B band 48Hz is a no brainer because it can do both commonly needed features of ultrasonic

[cursive sentinel]

All of the frequencies you listed there are subsonic. 48kHz is ultrasonic.

[pale umbra]

? it's not a frequency

it's how often it measure per second

48 times per second at close range / 12 times per second at long range

[supple pollen]

Yes, if you convert a signal to digital at 24 bits, then back to analog, you've lost a little. If you then convert that analog signal back to digital with a 12-bit ADC, then you've lost more (now you have 12-bit resolution). However, in most cases you can avoid that back and forth conversion. If you start with a microphone signal, you'd amplify that in analog. Then you could either route that analog signal to an amplifier, or you could convert it to digital. HOWEVER, if you convert that amplified analog signal to digital with a 24-bit converter, then do some processing, then convert it back to analog with a 12-bit converter, yes, you will again have lost some accuracy.

So you're sampling an ultrasonic ... something? rangefinder? at 12 samples per second? That has nothing to do with audio.

[pale umbra]

the ultrasonic was another question. "DAC are basically 10$+ so I wanted to check here if it was generic enough. Where as a 30$ ultrasonic 40 feet" . ie: Not sure I should get a 24 bit DAC especially if I may need more than one to convert back and forth in a typical project. Where as rotating an ultrasonic around can save a lot of money so it's a no brainer. ie: as much as Im certain about the ultrasonic Im not about the much cheaper DAC

but they are not related

[supple pollen]

What does a DAC have to do with ultrasonic? What are you doing here?

[dry pelican]

What is the ultrasonic? Like a ping sensor?

What exactly are you trying to do?

[pale umbra]

Im trying to assemble a kit with various ICs like I said this morning

[dry pelican]

Well for most arduino purposes, you don't need 24 bits

[supple pollen]

You mention some specific things like "40 feet" and "12Hz", which sounds like a particular project.

[dry pelican]

12 or 16 bits would work fine and be virtually indistiguishable probably

[pale umbra]

Eh Im busted I guess. My current project is that I got this which is no longer made along with a nice powerful servo so Im trying to make a rotating sensored thing

[supple pollen]

I still don't see where a DAC comes into the equation.

[pale umbra]

Im a bit confused. Is having several things going at once against the rules ?

I just have a bit more cash than usual and Im just trying to fill a digikey order with useful generic ICs

where as I already have that ultrasonic. I just didnt need to ask about it because it was obvious I need it. In comparaison to a DAC which Im not sure I need in a generic ICs kit

[supple pollen]

I'm a bit confused too. Having several things going on is normal, but it can make it difficult to figure out just what you're asking.

In particular, this: "DAC are basically 10$+ so I wanted to check here if it was generic enough. Where as a 30$ ultrasonic 40 feet" . ie: Not sure I should get a 24 bit DAC especially if I may need more than one to convert back and forth in a typical project.

So, yes, you can get a $10 DAC. I'm not sure what "generic enough" means here, so I ignored that bit. Then you mention "$30 ultrasonic 40 feet". Which seems to me like a different subject (as in, I don't know what a DAC has to do with an ultrasonic sensor). Then you say "Not sure I should get a 24 bit DAC". Again, I'm not sure what a DAC has to do with an ultrasonic sensor. THEN you say " especially if I may need more than one to convert back and forth in a typical project." which seems to harken back to a different thread? So yes, I'm confused.

[pale umbra]

Ok forget about the ultrasonic and let's reset. Is there a use for a 24 bit DAC in many electronic projects and if so do I need more than one as they are 10$ each ? Or the arduino DAC will be sufficient for most general uses

And 2- Should I get a matching spec ADC to convert it back

sorry Ill try to focus on one question at a time in the future

My gut feeling is that if I need something like that it will be better to get a breakout at my level of knowledge which will already include the correct components of these type on the PCB for the signal the breakout is for

[supple pollen]

No. A 24-bit DAC is way more resolution than you'll normally need, and to get all that resolution, all the supporting circuitry has to be able to provide enough signal quality for all those bits to mean anything. To illustrate, if you had a 0-3.3V signal going into a 24-bit DAC, that means the least significant bit would be 197 NANOVOLTS. That's getting to the amount of noise a small resistor produces by itself just sitting there (due to Johnson noise).

As for matching specs, it makes sense to do so if the intervening circuitry doesn't degrade the signal more, and it's appropriate to the signal you have. Most of the time, I use 8 and 10 bit DACs and ADCs, they're sufficient for many purposes. Sometimes I'll use a 12-bit unit. For more bits than that, you rapidly get into the range of fancy low noise circuitry, shielding, special power supplies, etc.

There are some DSP boxes that have a 16-bit ADC on the input, and a 10 or 12 bit DAC on the output, because all the math inside introduces enough inaccuracy that the digital stream only has that many valid bits remaining.

[pale umbra]

I assume the caps and resistors would be more expensive as well (lower tolerance such as 0.5% instead of 2%)

because using 12 ohm instead of 10.5 ohm accidently might matter

[supple pollen]

Value tolerance isn't usually critical to circuit linearity, except in certain realms like frequency determining components, voltage dividers, etc.

These days, even 1% and 2% resistors are cheap enough I just use those, even when I don't need that kind of precision.

[pale umbra]

Ok. Question about another component because the DAC thing have been answered throughly. A suggested flip-flio from the kit this morning is out of stock/not made anymore: https://www.ti.com/lit/ds/scls177g/scls177g.pdf?ts=1686163048884 . So I tried to pick a new one matching the specs. I went with under 33mhz speed, quiescent under 1mA, operating amp under 6mA, voltage supply with a range included in 3 to 6V, D-type, throught-hole or easily soldered surface mounts (tssoc/ssoc/dip/soic). I have 20 left after excluding anything without either a datasheet/picture or not in stock. Original one is a tri-state, do I still want that or I could upgrade this? I noticed tri-state all have 1 bit/element 3 elements/several bits for the others. Should I even care about the clock speed / output type and reset function ? Original doesn't specify reset type

My understanding is that I would want D/T/J-K/S-R flip flops but not necessarily also one of each output type

And Im not sure what happens if I pick a 20mhz one and I uses it on a 16mhz arduino. As long as I dont send over 16M bits per second to it should be fine ?!

not seeing any mention of crystals in flip-flop datasheets so I guess the clock frequency dont have to sync with each other

[unique patio]

it depends on what you are doing. Flip-flops just take a clock signal. Where that comes from is up to you. You don't attach a crystal -- you drive it from something that is producing a clock. The clock could be a regular signal periodic signal, or it could be aperiodic. The flip-flop is a pretty low-level thing.

[supple pollen]

The frequency is "up to" so it will work at any lower frequency

You can even clock one with a (debounced) pushbutton, which is a good way to get a feel for how one really works

[pale umbra]

Im not doing anything yet. I want a part of type X when I need one without having to put my project on hold for 8 months like it keeps happening

so that if I need a flip-flop I have one, for major types of parts

Im doing this to have enough parts in reserve to fill up 2 digi-key orders the to min 100$ order

[unique patio]

i have not used bare flip flops or gates since the 1990's

op amps, yes

[supple pollen]

You don't need a T F/F if you have a D F/F available

A couple of the more versatile ones will fill any of the usual rôles

[unique patio]

i always liked D flip-flops better than JK

[silver marsh]

You can make anything if you have enough NAND gates.

[dry pelican]

Has anyone done anything with PVDF piezoelectrics? I'm working on a project for an internship, and I need a way to attach electrodes to some PVDF for sensing. I just have a sheet of PVDF. Would copper tape work? (I don't think the adhesive is strong enough though)

[pale umbra]

I have to admit there is a bit of curiosity to it since I learned logic circuits at university a couple of years ago and always wanted to try it for real. Also isnt it faster to do boolean logic operations with hardware vs in arduino C or circuitpython ?

Isnt that why a lot of peoples like fpga here because it's kinda logic gates exponent n ?

[supple pollen]

Yes, hardware implementations are generally much faster than software ones

[pale umbra]

eh you know what I though about it and I think a 3d printer will enable me to do more. Especially with all the time I spend trying to find a construction system

ie: getting 200$ of parts in case I need it vs getting something Ill actually a lot and which mean freedom

[vast flume]

Ah I see, makes sense

[dusky raptor]

interesting, so this is my per led configuration, is this correct ? the 2 pin is my 3.3v input and the 1,3,4 represents red, blue, green GND ... should I use this schematic with my board??

[spice turtle]

If you want them on all at the same time, yea. Other wise you can common ground them and turn on the anodes individually

[rustic linden]

I think the anodes are internally tied together

[worldly schooner]

Are you sure forward voltage for all three channels is 3-3.2v? There does exist a red LED with similar voltage drop, but many red LEDs used in RGB applications are 1.8-2.2v. If that’s the case, your current configuration is likely to overdraw current through the red channel and break it.

[dusky raptor]

I understand what you're talking about

my reds are infact 1.8-2v

but it has a common cathode/ 3.3v input

[worldly schooner]

Also, if the goal is just white, white LEDs will be far more power efficient if you have access to them.

[dusky raptor]

I just simplified it, but I am not going to lit it up as white

Is the diagram correct though?

[worldly schooner]

You’re going to need separate resistors for each anode.

If you’re not aiming for white, you’ll have to adjust the current limits of each channel to get the color you want.

I’d attach pin 2 directly to power, and put per-channel limiting resistors to ground on 134.

[dusky raptor]

so I need 3 separate resistor per led ? bruh

[supple pollen]

Either that, or an LED controller that does current limiting for you.

[dusky raptor]

interesting, what is that ?

any examples or any showcases?

[supple pollen]

https://www.adafruit.com/product/453 and the HT16K33 are examples I've used in the past. There's also the IS31Fl3741 but it's "Charlieplexed" so it's harder to work with (AdaFruit's LED glasses are a tour de force of using that chip with 116 separate RGB LEDs and some difficult PCB routing)

[worldly schooner]

https://www.adafruit.com/product/1378 is another example of an LED controller, though these are typically used for saving GPIO pins with a microcontroller.

You may have heard of its sibling, the WS2812. Same idea, but the chip is built into the LED.

[rustic linden]

I'm planning on using the WS2811s for a project, would recommend

[dry pelican]

you could also have 3 resistors for all the LEDs, but that would require some more tuning, and brightness would be uneven

just 3 resistors. Not 3 per

but dedicated while LEDs also have better color accuracy and look better

most RGB LED strips use 3 resistors per 3 LEDs

[dry pelican]

I'm working on this matrix of piezoelectric sensors. I'm using a low leakage current multiplexer (max336) to switch a ground reference to one of the rows of piezo sensors and keep the others at high impedance (floating). Each column is attached to a high input impedance op amp buffer. Those signals are then passed into a multiplexer that feeds a selected signal into a 60hz notch filter (maybe an amplifier) and then into an ADC. The leakage current of the first multiplexer is 3pA, and the leakage current (is this also known as input bias current) of the op amps may be even lower at like 1pA or 10fA . Since the op amps have an even higher impedance than the multiplexer, would signals from deactivated rows (those kept at high impedance by the 3pA leakage current mux) be able to get into the op amps?

[supple pollen]

It's not going to be an all or nothing sort of situation. If you need to damp down the signals from the unselected ones, you might need some additional switching. It really depends on how good is good enough for your application.

[dry pelican]

Actually, thinking about it more, it probably won't matter

This is based off a group that did the same thing, so if theirs worked, this should work

[supple pollen]

One thing I would do is use the "spare" op-amp as a buffer for the voltage divider driving the COM pin.

[dry pelican]

Also, the schematic just shows a smaller matrix than it will actually be. The final product will be 5x5 or bigger

Also it was very quickly put together

Just to show the idea

Is input bias current the spec to look for?

[supple pollen]

Not exactly, that's often referring to the amount of current away from "zero", like a DC offset.

Although since you have a high impedance source, the input bias voltage can be the amount of voltage that current induces across the impedance of your source, which can become significant.

[cursive sentinel]

Keep in mind that high impedance amplifiers can also be very, very sensitive to any kind of DC offset.

[dry pelican]

It's just going to be acting as a follower or a simple charge amplifier because 60Hz noise is probably going to be an issue

[cursive sentinel]

One thing to keep in mind: the capacitance on the '4067 switches will likely be significant.

This could cause issues when switching between piezos.

[dry pelican]

That first '4067 will be replaced by a MAX336 for ultra-low leakage current

[cursive sentinel]

That's good

[dry pelican]

They advertise low capacitance as well

[cursive sentinel]

If you do have stability issues you can try putting some 100k resistors on the opamp outputs.

[dry pelican]

Well for clarification, the first one (vref switcher) is a max336 while the second will be the standard '4076

Both max336 would be expensive overkill

$24 for 2xMAX336

[coral rampart]

i want to make something to stick over my washing machine's LED display, something clear so it doesnt block what it says. the purpose is to make a script that alerts me when the washing machine is done. so far this has been the best option i can think of as the housemates arent comfortable with the idea of a camera in the house. just wondering now what parts would make this work?

so it would likely need to not just sense light but sense patterns

[unique patio]

how about sensing vibration instead? If no vibration/movement for several minutes, then it's done.

[coral rampart]

wouldnt work cause the machine goes through like 10 minutes of doing nothing but gently spinning to sense weight

[unique patio]

an accelerometer would check this. You would want to do some data collection to see what looks like washing and what doesn't

[coral rampart]

and as far as i can tell its kinda random in when it does that

[unique patio]

gentle spinning could still probably be detected.

are there single LEDs that are on or off when done, but not when washing? Ours turns off its display completely when done, and also some "in progress" lights go off.

what's the model of the washing machine?

would a timer work as well? We just set a timer for the cycle time + a few minutes

[coral rampart]

thats a good point. i was not basing it off any single LED since there's no like "done" light or anything. the machine just totally turns off

but maybe i can use that

not sure of the model. will need to check but prolly not til later in the day

fyi if anyone is thinking i could just splice into the wire that sounds the chime when its done, that was my first idea but then that would void the warranty

wait a sec

👂

eh listening to the chime could still give the same paranoia my housemates are concerned about having with a camera

[unique patio]

you can have a light detector. But don't you know the approximate cycle time? Ours has a countdown. It's off a little bit (sometimes the spin takes longer than it says), but it's mostly accurate

[dry pelican]

What does the ohm||pF unit mean? Is it saying that there is an equivalent 10^13 ohm resistance going to ground and an equivalent 6pF capacitance in parallel?

for common mode

differential mode impedance would be capacitor and resistor between inverting and noninverting inputs right?

and in that case, there's a 2pF equivalent capacitance

[cursive sentinel]

It's giving you the value in terms of capacitance and impedance (the unit is ohms, but it is not resistance per se).

Impedance is more or less "equivalent resistance with respect to frequency"

So that impedance value will hold true at whatever their test frequency was.

If you look further down in the datasheet there are probably bode plots of both Zcm and Zdiff.

Also, keep in mind that with FET input opamps that the input bias and offset currents may increase exponentially with temperature.

[quartz shadow]

Question, if you wanted to try to connect a custom PCB to a sheet of ITO, would conductive adhesve, or a pressure mount (or both), be the most reliable connection?

[supple pollen]

Like edge lighting in reverse...

[pale umbra]

Id use an accelemeter here too and do a few test runs to get a sense of the range. A working washing machine should be vibrating constantly. Of course the very best if you are qualified would be to hook it up to the washing machine

[coral rampart]

not a newer machine. they have cycles where they just sit and do nothing for a random amount of time while sensing or draining

[dry pelican]

I think pressure mount rubber things are used in cheap LCDs. Don't really know what they're called

[unreal flax]

Zebra connectors?

[pale umbra]

I wish I had the training to connect directly to my appliances instead but no I gotta try to cook something too. Except in my case it's mostly emergency (dryer getting too hot, water under the washing machine)

[cursive sentinel]

Nearly every engineer I've met is suspicious of 'smart' appliances and actively avoids them.

[pale umbra]

Im just trying to make the point that the easiest way is to connect to the wire that actually tell the water valve to turn on/off / electrically switch the belt on/off etc

since electricity is far more simple than electronics cuz it's kinda brute force

[quartz shadow]

I'll look into those, thanks! My roommate said zebra stripes, but I wasn't able to get him to actually clarify what that was. Makes perfect sense now

[pale umbra]

this might give you some ideas that is how I started doing this: https://www.youtube.com/watch?v=_UlkMRrUIlI for non-invasive appliances monitoring

[stoic smelt]

Is anyone aware of a device (IC) that is similar to an i2c I/O expander, except it the output ports are pulled "high" and "low" based on separately provided voltage references. i.e., not pulling to the i2c High or GND. Almost describing an i2C opto-isolator bank, but don't necessarily need opto isolation.

[unreal flax]

You should be able to find GPIO expanders that can do level-shifting with a separate I/O supply voltage on the high side, but finding something that has a separate non-ground low voltage would be much more unusual.

[stoic smelt]

yea, not sure it exists, or if I just don't know the correct search terms. Been looking through datasheets for the past couple hours and mostly just finding GPIO expanders.

[unreal flax]

One option might be to use a regular GPIO expander with pulldown resistors to your low voltage, and switch the pin between output high and high-Z input mode. Kind of the opposite of the usual open-drain scheme.

[supple pollen]

There are some gate driver ICs that let you provide different supplies for the high and low sides, but you'd need another I2C-to-GPIO type IC driving them

[stoic smelt]

Thanks @unreal flax and @supple pollen , These are some ideas to explore, I'll take a look.

[tough matrix]

RIP Eagle
https://www.hackster.io/news/autodesk-announces-the-end-of-eagle-707864d95d7e.amp

[supple pollen]

I was a fan of Eagle right up until they invalidated my expensive license and moved to a subscription model. Same with PhotoShop. In some software, a subscription model works for me, but for those two, where I'd already sunk a substantial amount of money into them, and could imagine having a bad month where I couldn't pay the subscription and needed the income from using them, it was a non-starter.

[tough matrix]

Licensing apart, I liked Eagle.
Somehow I like electronic design part of F360 much less.
Anyway, I switched to KiKad 2 years ago and am quite happy with it ..

[supple pollen]

When my 9 year old computer finally stops working, I'll have to switch.

KiCAD seems like an obvious choice for PCB design, not sure what I'll use for photo editing. Probably a combination of Preview, The Gimp, and Inkscape.

[elder peak]

I just carefully protect my CS4 media.

[stoic smelt]

Would be nice if they released the source so that the open source software community could take it over, but I am sure they won't. Just another example of a mega company buying up something and then running it into the ground :/

[tough matrix]

to their defence, I am sure they actually used that as basis for F360 electronic design, so it is not so much "ran into ground" as "used and discarded"

[stoic smelt]

Fair

[distant raven]

Hey, so I appreciate the caution. Im not entirely sure creating a launch system for fireworks is a particularly safe idea given that fireworks have a common likelihood of going wrong. This kind of walks a gray area if the Code of Conduct so the help most could provide is going to be with a heavy leaning towards, “sure, but this can be dangerous”

I’ll let Andon explain more though

[grand fossil]

I'll delete it. Sorry.

[distant raven]

Thank you for being cautious though and seeking help

[marsh nest]

Hello! I am happy that you're trying to be safe, but I'm going to have to disagree with skerr. Fireworks are something that can be incredibly dangerous even if you do everything right, which makes it outside the scope of this server.

[grand fossil]

Gotcha, I was inspired by the Adafruit project using the same circuit in a smaller sense but I understand.

[scarlet jay]

Hey guys, I need help with a few things on here

• Bluetooth button programming

• DAC controlled via app

• POTENTIOMETER controlled via app

That is everything 🙂

[rustic linden]

For the third one, would a digital potentiometer work?

[broken zenith]

Hi, I'm designing an RP2040-based board and can't figure out what a pin does on my flash chip. Is EP the same as WP (write protect)? If so, should I leave it floating or pull it high? The datasheet says WP is active low. Thanks!

(not done connecting everything yet)

[unreal flax]

EP is likely the Exposed Pad underneath the chip, which would usually be grounded.

[broken zenith]

Ah, I'm glad I asked! Where could I confirm that's the case?

Although now that you mention that, it does seem pretty likely.

I thought it was "Edit protect" 🙃

[cursive sentinel]

EP pretty much always stands for exposed pad.

[hushed smelt]

I didn't know that, good info.

[broken zenith]

I guess I'll roll with that then. Thanks!

[unreal flax]

I'd check the chip datasheet to be sure. If it's an 8-pin package, i.e. there's no pin 9 except for the exposed pad, then that'd be a good sign.

[hushed smelt]

There should be a data sheet for the W25Q32JVZP, look it up, see what the datasheet has to say about it.

[broken zenith]

No pin 9, so it's probably an exposed pad?

[cursive sentinel]

Check the package dimensions (should be near the end of the datasheet).

[broken zenith]

Good idea, looking now

"Note:
The metal pad area on the bottom center of the package is not connected to any internal electrical signals. It can be
left floating or connected to the device ground (GND pin). Avoid placement of exposed PCB vias under the pad."

Thanks for the clarification and help!

[jade wedge]

10k pullup resistor should generally be fine for a 5v design, right?

[cursive sentinel]

Yeah

But the Pi is not a 5V device.

[jade wedge]

I'm not the one doing Pi stuff.

I'm back in Atmel land.

[cursive sentinel]

Ah

One thing to keep in mind: 10k is on the upper end for I2C pullup values. It's fine at 100kHz and should generally be ok at 400kHz. Anything beyond that and you'll likely run into issues.

[jade wedge]

Fortunately in this case it's not even data. It's a set of dip switches that'll be read via shift register.

I just know that I can't leave them floating.

[cursive sentinel]

Yeah 10k is fine then

[jade wedge]

Cool.

[broken zenith]

I'm looking for an analog hall effect sensor that can sense along the axis in the diagram I've attached. I've poked around in Digi-Key but I don't understand what the datasheets say about the axis (or, there's a handy diagram but it's the wrong axis) Any ideas would be appreciated!

[jade wedge]

Supposedly the MLX90333 line of chips will do what you want, but...
Good luck figuring out how to configure them. It specifically calls out being able to do the lateral measurements that you're asking for, and it says that it has an analog output mode. I just can't seem to figure out how you actually configure any of the user configurable parameters.

[broken zenith]

Thanks, I'm reading over the datasheet. It looks like it can be configured over SPI, but I'm not entirely sure.

[jade wedge]

Probably. Though judging by the pinouts, once it's in analog mode it might be stuck there until reset. Could be wrong though.

[broken zenith]

I don't mind that, I just want a super simple way to detect a magnet along that axis. Most of the sub-$1 sensors I looked at were only sensitive on the vertical axis 😦

This is promising: https://www.digikey.com/en/products/detail/allegro-microsystems/ALS31313KLEATR-JOY/9674752

I2C is OK, I just thought analog would be cheaper/have more options

[jade wedge]

Oh if I2C is an option you've got loads of chips to pick from.

[broken zenith]

Huh, I assumed it was the other way around. The more you know 🤷

Something being rated for "automotive" just means it's more reliable, right?

[jade wedge]

Yeah, I'm pretty sure the ratings are just levels of reliability. Not something hobbyists will normally worry about.

[broken zenith]

Good to know! In that case, I think I found one: https://www.digikey.com/en/products/detail/infineon-technologies/TLE493DW2B6A0HTSA1/8627120

[jade wedge]

Glad to hear it.

[unreal flax]

Yeah, typically a wider temperature range, for example.

[broken zenith]

Awesome, thanks for the help!

[jade wedge]

Don't go for medical, military, or aviation ratings. You don't need that level of reliability, and you'll pay through the nose for it.

[broken zenith]

Gotcha! Thanks again!

[supple pollen]

Automotive spec is one of the more demanding ones out there, but automotive parts are made in such volume that such specs aren't nearly as cost prohibitive as the others. Generally automotive spec parts are a bargain. The problem is the market is fairly uneven and fluid, so if you design in a part available one quarter, you may well have to switch to some other part if you want more next quarter.

[distant raven]

Which is why it’s always nice when chip manufacturers make the automotive part in compatible footprints with other products in that line.

[cursive sentinel]

I work on automotive chips.

[woven bluff]

hall sensor needs shielding and calibration

are automotive parts RHoS?

I though lead is allowed in automotive

[cursive sentinel]

You'll have to read the datasheets.

[jade wedge]

RoHS is per component, so yeah, check the datasheets. If you're using Digikey, you can filter by RoHS compliance.

[limpid nest]

Is there a way to create a DXF from a PNG or similar that is filled in and not just wires?

[cursive sentinel]

Oh hi oats.

[latent jungle]

How are you creating the PNG?

[spice zenith]

Just confirming, I can use the AdaFruit 4Mb FRAM with 3.3v supply and 3.3v signalling. https://www.adafruit.com/product/4719 I'm attaching to a modern Teensy chip if it makes a difference.
Reason being I have everything hooked up correctly (I think) and the chip isn't responding. So I'm thinking it's either not connected right, or the chip is fried.
I could easily attach +5 to VIN, and I know it'd regulate the supply down to 3.3v, I'm just afraid of what voltage I'd see on MISO. I do not want to start feeding 5v signals to the Teensy.

[jade wedge]

Well, according to the page, it should work on 3.3v
Just to double check, you're connecting your supply to the VIN pin, not the 3V3 pin, right?

[limpid nest]

I figured it out! Turns out kicad is way less arcane than fusion.

[limpid nest]

Oh hi mark

[spice zenith]

Yes, confirming I have the 3.3v supply connected to VIN.

I checked the voltage on the supply pins on the FRAM breakout with a test meter, 3.29v, so that's good.
Now it's down to either a bad hookup, or the chip is fried. I hope it's the former. I've got some new hookup wires in flight from Adafruit, I'll see what happens when I try those.

Aha!!! I get the "idiot of the day" award today. 🙂 I had MOSI and MISO swapped. Doh!
Hooking it up properly made it work I'm only doing a ReadID command on the chip, but that sends back a very recognizable response, which is coming through exactly as expected.

[latent jungle]

Standards committees can only go so far with their signal names.

multiple times over decades

[cursive sentinel]

Some have also started using the names POCI (Peripheral Out, Controller In) and PICO (Peripheral In, Controller Out) to get rid of the master/slave terminology. At some point I want to make a PCB where the signal names are 👉 (pokie) and 🤏 (pico).

[supple pollen]

I've also seen host/target

[jade wedge]

Everyone earns the idiot of the day award at least once.

Most of us earn it multiple times.
"I've been debugging this code for HOURS why doesn't it work???"
"Oh, yeah. I put the wrong character here. Oops."

[supple pollen]

We were discussing that very thing recently https://discordapp.com/channels/327254708534116352/336571109430263816/1117788078212395078

[devout stag]

greetings

i got myself an RP2040 feather, but i need to find out which resistor is R3

i want to replace the internal pull up with some more elaborate circuitry so i can optimize power consumption while unused

the only thing i can download are PDFs and pictures of schematic and board view (silk screen and some routing) but no info on parts placvement anywhere

point is, the schematic is incomplete, it shows the EN pin on the JP3 but EN is nowhere as a net visible only on U2 and this deignates a connection between VHI and EN via 10k and the only resistor i found seems to not have any effect at all

i measure a voltage of less than 10mV on En pin but the board still is powred

an i mean the pad and the EN pin on JP3 , no idea whree U2 is located

[unreal flax]

Following the EN trace from the header pin, it looks like this resistor is probably R3.

[devout stag]

i am boggled then, i have removed it, made a 22k from the En pin on JP3 to the GND pin but when i power it via battery it still turns on

i wish those boards had an option to properly use the En pin at all. the pull up means there will ALWAYS be power flowing though the pull up und the connection to ground

now i invested so much time and money into thosee boards and they do not work like the schematic suggests this is very annoying

and since i flipped that resistor off i can not refund it

neithercan i refund the other coponents i already used

[unreal flax]

That's weird... I'm not sure what's going on.

[jade wedge]

I don't know the exact composition of SMD resistors, but unless you removed it entirely might the original resistor still be making a complete circuit? In that case it would still override a 22k pulldown resistor.

[devout stag]

nah i took it off completely and i double and triple checked all connections, point is, i measure voltages BELOW 0.4V which is the LOW threshold of the chip by datasheet

but the board is happily running

i think there might be a side path somewhere that is powering the board but i do not have the full schematic

what i want to do is, i put three sources of enable signals, the USB pin, D13 and a capacitor of 47uF, that is charged from the BAT pin when i push a external button. Each enable source vial a diode to enable. When i push the button, the capacitor is charged and enables the circuit, and then drains over around 2-3 seconds until either the application puts high on the D13 or somebody plugs in a USB source and powersd the board from there .... if the app has timed out or did not measure enough battery voltage it will clear D13 and power down the board

then i put that in a box and do my stuff with a push of the button until i am finished and select to quit or the app times out ... all feather boards lack that feature to truly power them down, the enable pin will always slowly drain the power

if somebody has an idea or solid solution how i can power the RP2040 feather from battery with a push of a button without slowly draining the battery while offline i would welcome it ... the device might be outside even in winter and the old but crappy to maintain ESP board with a Power1000 worked fine with that solution but i want a single board solution this time .. i alos have no other solution than push button due to the sealed box

[unique patio]

do you have I2C pullups powered from somewhere else? Could be backpowering

[jade wedge]

A mosfet controlled by an OR gate maybe? Stick the mosfet in line with the battery, each of your inputs goes to one input of a 3 input OR gate.

[unique patio]

to find a part you can use the Eagle files from the repo: https://github.com/adafruit/Adafruit-Feather-RP2040-PCB

[unique patio]

maybe of interest: https://www.adafruit.com/product/1400

[jade wedge]

Actually I suppose since you're already blocking with diodes, you could forego the gate and just common the cathodes of your diodes into the mosfet.

[earnest folio]

Quick question: does rp2040 support RX/TX pin swap without PIO?

[surreal tangle]

Not for hardware serial. In the datasheet, you can see what functions (section 1.4.3) each pin can perform: https://datasheets.raspberrypi.com/rp2040/rp2040-datasheet.pdf . There is an implementation of software serial, but I've not used it...

[earnest folio]

I see...

[split obsidian]

Is it possible to fake a potentiometer with GPIO. As in, one MCU's ADC is reading a faked potentiometer with a GPIO? Like is it possible to vary resistance with a PWM signal? What kind of electronics might complete that task?

[cursive sentinel]

Resistance no, but it is possible to filter a PWM signal and read a voltage.

Digital potentiometers do exist. Internally they use a switch network.

[split obsidian]

like a switched resistor network?

[cursive sentinel]

Yeah

[split obsidian]

oh interesting

I wonder how hard that would be to do manually

[cursive sentinel]

Not hard, but it would be very tedious and expensive (lots of resistors and analog switch ICs).

That being said some test equipment actually does this, since digital pots aren't very accurate.

[split obsidian]

how much is a digital pot?

I'm just trying to fake out a joystick potentiometer

basically to control an xbox adaptive controller input with gpio

[cursive sentinel]

You can get them for like $2 in single quantity. They'll definitely be accurate enough for that.

[split obsidian]

are they fast?

[cursive sentinel]

Microchip and Analog Devices make a lot of them.

Fast enough to drive game controller inputs.

[split obsidian]

https://learn.adafruit.com/ds3502-i2c-potentiometer << Think this one on i2c would be fast enough?

[cursive sentinel]

Should be alright as long as you aren't writing to the EEPROM every time.

[split obsidian]

What eeprom?

[jade wedge]

Likely the EEPROM to set the default startup value.

[split obsidian]

ohhh

yeah idk why I would ever use that (for this project)

[cursive sentinel]

Yeah it has onboard EEPROM to store its value across power cycles. A common use case for these chips is power-on configuration for something more complex.

[split obsidian]

yeah, I'm looking to hook up a magnetic encoder joystick to a MCU with one of these basically providing output from said joystick to an xbox adaptive controller

the goal here is to make a forza game car drive and I quote the guy I'm building this for 'drive like my wheelchair'

but the thing is there are so many forza controls that involve like...braking and steering that's not really super easy to do

as far as I've looked into it that is...it's been a minute

[jade wedge]

Well, not knowing much about Forza itself, you could potentially get away with doing this with as few as two digital pots.
While the default input is undoubtedly triggers and L analog X axis, I know a number of games allow R analog Y to be used as throttle/brake

[cursive sentinel]

You may want to use higher resolution digitpots.

[split obsidian]

yeah, I think forza splits throttle and brake to two separate analog inputs

haha, so 128 steps isn't enough?

that'

[jade wedge]

64 in each direction. It's probably doable.

[split obsidian]

is that what you mean by resolution?

I think that's fair

I mean, we're not really going for game performance as much as quick to delivery and consistency in product performance

he's only going to be so good of a driver

(at least at first, it might be worth an upgrade down the road)

[cursive sentinel]

I'm mostly thinking to try to avoid jitter. Some pots can produce small transients when changing values.

[split obsidian]

aah

some digipots you mean?

[cursive sentinel]

Yeah

[split obsidian]

can you define what a transient is?

[cursive sentinel]

In layman's terms, it's a little blip. In more technical terms you may see the resistance suddenly spike as it is changing from one value to another.

I can do some datasheet digging tonight and come up with a few part options if I remember.

[jade wedge]

Wow. You can get all the way up to 1024 taps on Digikey.

[cursive sentinel]

Yeah, that's overkill here. What you want to look for is something with low DNL/INL (i.e. good output linearity) and possibly some sort of make-before-break switching mechanism to make sure transitions are smooth.

[supple pollen]

You can do it with an analog switch and a series resistor, in some circumstances

[split obsidian]

How would one search for make before break?

[split obsidian]

Do you have any examples of this?

[dry pelican]

Maybe you can use a lowpass filter and a digipot

[supple pollen]

There's some discussion in this old StackExchange thread https://electronics.stackexchange.com/questions/129803/pwm-controlled-variable-resistance

[cursive sentinel]

It probably won't be in parametric search. You'd have to look through datasheets.

[cursive sentinel]

The trouble is that in order to get a reasonably smooth signal you'd need a low pass filter with a giant cap, which will result in a very slow response time.

[rigid heart]

Has anyone implemented usbc/usb3.2? just looking for a good starting point

[cursive sentinel]

Which are you referring to? Because USB-C is just a connector. It can support a number of different protocols, including USB3.2. But in order to get USB3.2 speeds you are going to need to understand the signal integrity implications.

That being said Arya has made an entire series of hackaday articles about USB-C.
https://hackaday.com/series_of_posts/all-about-usb-c/

[rigid heart]

Yeah specifically 3.2. Ive seen a few applications that use the slower d+ and d- lines but nothing that does the higher speeds. Working on understanding signal integrity for a few projects i want to do. Ill read through this. Thanks for the resource!

[devout stag]

nice to know, thank you

[devout stag]

oh wow ... i found the path ... it was my battery voltage sensing line ... i totally forgot that most port pins have revers polarity- or zener-diodes ... and i think i powered the board over those from the A0 pin on which i wanted to check my battery voltage ....

[cursive sentinel]

You mean the ESD diodes?

[devout stag]

yeah

now i need to find out how i can meaasure my battery voltage without bringing in that bypass

P-FET ?

[cursive sentinel]

You could probably just run it through a resistor divider.

[devout stag]

yeah that is something i am already doing

but that means i will still have battery drain when not enabled

i need to uncouple the battery pin from the A0 pin

[cursive sentinel]

How much of a drain?

[devout stag]

the targeted device is outside every day an night and more or less rarely used

so no drain at all would be prefereable so the lipo doesnt die

[cursive sentinel]

Yeah ok you don't want leakage on that.

[devout stag]

i fear going in the high 3digit kOhm range might make the measurement quite noisy

[cursive sentinel]

Yeah that's a possibility.

[devout stag]

can't i use a FET but which type? N or P

and connect it to 3.3.V

[cursive sentinel]

Honestly I'd be tempted to go with a mechanical relay.

[devout stag]

the gate i mean

or will there be a voltage drop like with J transistors ?

[cursive sentinel]

I'm assuming you use this battery to power the microcontroller?

[devout stag]

it's a RP2040 feather , removed the R3 pull up of the enable pin

the battery is connected to the board and i wanted to use a capacitor delay, USB or GPIO pin to keep the enable pin high

the capacitor deleay is charged via push button from the bat pin

then drains within 2 or 3 seconds

until the app pulls up the hold pin

[cursive sentinel]

I'll have to give it some thought in the morning. Do you have a schematic (or a least some kind of block diagram?

[devout stag]

can i link falstad.com links?

[cursive sentinel]

Sure

[devout stag]

or i DM you the circuit

just give me a minute toput it together 🙂

[cursive sentinel]

Nah just go ahead and post here unless you're worried about privacy or something.

Like I said I'll have to look in the morning.

[devout stag]

just get some rest 🙂

https://www.falstad.com/circuit/circuitjs.html?ctz=CQAgjCAMB0l3BWcBmBsBsyCcyBMAWLSXXSAdhHSxCX2RoFMBaMMAKACVL8QSfJeBKMPwD+wmAk6CerXDO4SQomksnTkydAuQAOHnyUrk0egPPQpAeR2QDQrbolsATrdlltyO8NLw2AObgniDeHl4I8uau7gqGAmBwbAAmCmAhhpkgyQwAZgCGAK4ANgAuKbxk8plCWTkFJeWp6do19gbZeUVlbADOldUOWgoJIAXFvQx9aWCD4cIQ45NsAMbBXj4tI1Cw8PgUTDBYVAToZAgI6Vg42jBw7P3o7eC4TvHgY-kTU26zb0Jbd6JSBsfKhHx4AR6HiQkDUMBIFhIYHwCz8aBURIIfAEVhEZD4MBOFFJIL4J6hUjKCmoKJQGLQyliMBeKkJVEaCHDMLKOBGNFqSxsADu4JhPkMPJBLPkhNZYhpVPk9W6pSYxQYzUFkHYAHtQooxJAsE4kDAIAk2EA 4V input is BAT, 5V is USB, 3.3V is the application GPIO and the "out" is the Enable pin .. now i need to know how to measure by BAT input when enabled only ... PS: i added a 0.4V comparator for a more complete simulation of the power off threshold 😄

i think i might add some parallel resistor tot he capaitor to keep it draining when the diode is backpowered

[tidal elbow]

Hi, I need a help. I want to drive a 12V relay using MCU 3.3V GPIO. The circuit is working ok but it's survival rate is only about 50% after it continously run 6h+. The goal is to make it able to drive AC load fot 8h. The failure point is the Mosfet. I've tried swaping the 2N7000 mosfet with BC547 BJT, but the result is about the same. Bellow is attached the schematic. Please kindly give me an advice to make this circuit work.

[unique patio]

Glad you figured it out!

[worldly schooner]

How much current is your mosfet pulling? Generally the peak current and continuous current ratings are different, so if you don’t have enough resistance on that line, your mosfet will have a much shorter lifespan than expected.

Also, do you have a voltage source off-board? Relying on the mosfet to drive current without voltage seems tough on the semiconductor junctions.

[tidal elbow]

The Mosfet is used to drive SPDT relay (12V at 30mA). To drive the Mosfet an optocoupler is used with pull up resistor (letting 6mA) voltage from on board regulator (ams1117). If my calculations is correct, the power dissipation is about 360mW (30mA x 12v), and the continuous gate drain current is well below its maximum rating (200mA). Attached below is the Mosfet absolute max rating. Please correct me if you spot a flaws in my calculations or understanding.

[supple pollen]

I suspect the problem is a voltage spike. While you do have a TVS diode, I'm guessing either that diode has failed or it's too slow or somesuch. I normally just use a reverse diode for spike protection when driving inductive loads. Note that the 2N7000 does have a parasitic body diode, but that may not be sufficient.

[violet portal]

Does anyone have any tips for soldering to find pitch test points? I have a board for testing and there are 4 TPs for USB; despite everything I’ve tried with probes, it just refuses to enumerate - the Pi sees the usb port, but won’t enumerate sadly… so I was hoping to not have to solder but that seems to be my last hope; it’s a bit finer pitch than I’m usually comfortable with and I’m most concerned with lifting a pad

I’m thinking to tin the usb cable and then with a fine tip and some flux attach to the TP and hope for the best!

[dry pelican]

Agreed. Freewheeling diode and/or snubber

[inland jungle]

Is the mosfet getting hot? You're voltage dividing your gate input, and mosfets typically don't have a very good Rdson at such voltages unless they are specifically tailored for it

plus, if the failure is while it's constantly on (you're not switching) then it's unlikely to be voltage spikes or lack of a diode/snubber

can you replace the 3.3V driving the gate with 12V?

[supple pollen]

I think the bipolar transistor was a better bet, but it failed as well.

[inland jungle]

yeah, that should have been driven will enough with his circuit

[worldly schooner]

Gate drain current should be negligible, as a MOSFET gate is voltage controlled. The drain current is usually drain-source current, so the voltage at your drain and resistance of your load define how much continuous current you run through your MOSFET when it’s on.

[surreal tangle]

So what is the voltage drop across the optocoupler. I grabbed one and it was 1.1v. So that could be lowering your GVS too much and putting it into linear mode. You could just replace R107 with a 10 ohm resistor and see if that solves the issue.

[tidal elbow]

I checked the inductive Voltage spike with oscilloscope and the tvs diode able to clamp it well.

[tidal elbow]

Its not get hot. The room environment itself also keep around 28°C.

[tidal elbow]

The Mosfet drain voltage is around 1.2V when its driven by the optocoupler. I supposed its adequate to sufficiently drive the Mosfet.

[supple pollen]

Not even close for the 2N7000. It doesn't really conduct until the gate voltage gets to 5-6 volts.

[jade wedge]

Good thing I looked this up first. Pretty sure I can't run NeoPixels unless I set the ATTiny85 to 4MHz or above. Might be able to get away with 2MHz, but not 1. I could go for broke and just run it at the full 20MHz that the internal clock supports. It's not like I'm that worried about power consumption.

Oh. The internal only goes to 16. Still plenty fast

[tidal elbow]

Thank you for the spec check. The funny thing with my circcuit is the 2N700 is able to turn on the relay (30 mA draw) properly at 2.5V base voltage (I previously mentioned 1.2 V, but that was an error on my part). However, I am still struggling because it cannot satisfy my 8-hour continuous on-state requirement. Right now, I am thinking of swapping the 2N700 with a power MOSFET. Could you guys recommend a power MOSFET that can be driven with 5 V and drive a 12 V inductive load?

[supple pollen]

I'd suggest a BJT solution myself, but that's mostly because I've had better luck with BJTs with low voltage drive and with surviving voltage spikes. You did point out you'd tried a BC547 as well and it had failed too, which seems curious to me. Did you just use the same driving arrangement?

[tidal elbow]

yes, i only swapping the 2n700 with BC547 but not changing the driving circuit.

[supple pollen]

Hmm, that would dump maybe 3mA into the base, which should be plenty for switching a 30mA relay coil. I'm unsure why that would have failed. You've checked the things I would have looked at (seen if the transistor heated up and looked for spikes with an oscilloscope). Sometimes the simplest circuits are the headscratchers.

I would expect the LED not to light with the BJT installed, as it would drag the voltage at its base down to 0.7V or so.

[tidal elbow]

@supple pollen do you think placing a resistor directly before the MOSFET drain would do anything different? (Please see R107 of my previous schematic.)

[supple pollen]

I'm guessing you mean gate (pin 2), not drain (pin 3) in this instance. Since it's a series circuit, R107 should limit the gate charge current as effectively as a resistor directly before the gate.

If you want to stick with a MOSFET, you could try a beefier one such as STP160N3LL or DMT6009LJ3

Those two tend to conduct well with the low gate drive available.

[tidal elbow]

Thank you. Could you point me to SMD package equifalent?

[devout stag]

i want to measure the voltage of my battery on my rp2040 but i did an oopsie on my first setup and powered the unpowered board through the ESD diodes ... now my knowledge of specific operation parameters around transistors and diodes is bad and i have no idea if a simple voltage divider is enough to prevent any unwanted flow of power even when the battery is not powering the boaRD

[supple pollen]

Maybe DMS1019USN, which is available in a tiny SOT-23 package, among others. Since the other parts you had mentioned were through hole, I had guessed that was what you were looking for.

I'm still boggled when I see parts in tiny packages like that rated for several amperes

Other possibilities are AO3400, BSS138, FDV303N, SI2302, or the grunty IRFR3708 in a D-Pak package

[tidal elbow]

Thank you. I'm trying to make a board adapter to make the power MOSFET work with my current PCB. I've been trying to switch from local IC marketplaces to overseas marketplaces, as I've seen some quality improvements lately. Maybe I just got unlucky with the MOSFET I got (as the 2n700 are an old batch)

[supple pollen]

That last one is designed for driving inductive loads in switching power supplies, so may be robust as a relay driver.

I just noticed it's obsolete now. DigiKey suggests DMN3016LK3 as a replacement, which is also designed for switching inductive loads

[tidal elbow]

What's your oppinon on IRLL024NPbF as replacment? it is small that it could easly fit on my current board arrangment.

[supple pollen]

It is avalanche rated (which is good for your application), but seems to have a higher gate voltage than several of the others.

Among the small ones, the BSS138 and FDV303N seem robust for SOT-23 packaged devices

[torn plover]

Not so much a design issue more of a decision issue im stuck with. Im building a power bank for longer fishint trips. Do i go 3s liion and fully charge them or 4s and not charge to full. All the devices powered from this bank are going to be either car devices (head unit and usb chargers) or 5v and all devices can cope with 10-15v. I wont be overcharging the 3s idea. I cant make my mind up. Going 3s means ill be needing to charge before ive "fully discharged" but 4s means ill never be fully charging and can run to the safe discharge voltage

Ive got 42 18650s for the project and because its only for me bms is going to be just a low pack voltage cut out

Its either going to be 4s10p or 3s14p

[supple pollen]

Your cells will last much longer if not fully charged. Charging lithium cells all the way to 4.2V gets the maximum capacity out of them, but those last few tenths of a volt cause physical damage to the cathodes as the last lithium ions get crammed into the matrix. Charging cells to just 3.9V or 4V gives slightly lower capacity but greatly increased lifetime.

[torn plover]

Thats why i was torn id lose 7.5 Ah capacity from fully charged voltages

But i was also aware they will cope better with less full charge cycles.

Ill be going 4s i think