help-with-hw-design

[unreal flax]

Just as an aside, the ESP32 tends to use the flow-control pins for reset and boot-mode control in its serial debugger hookup, which is convenient for USB adapters.

[distant raven]

I believe they use them with mosfets?

[unreal flax]

Yeah, they're typically hooked up as an XOR operation, to distinguish manual control of the pins versus just having them idle.

[supple pollen]

Arduinos often use RTS or DTR for reset too.

[vast flume]

Is connecting VBUS to my boards 5v bus a bad idea?

In the ds for the CP2102n it says that vbus is used to detect when a usb device is connected, so I guess with this configuration vbus would be high even when no usb device is connected. I don't know if that will cause issues though

Or would it just be safer to not try to power my board off of usb

[supple pollen]

That depends, does your board have another source of 5V besides USB?

[distant raven]

I should note, be sure to read the errata for the cp2102n

There’s an issue that’s pretty common on them where the com port doesn’t come up immediately and basically you need like 3 extra components, one of which is a mosfet

Op just one component extra

[vast flume]

Battery power, but if I go with usb power then it wouldn't power itself off of battery at the same time

[distant raven]

https://www.silabs.com/documents/public/errata/cp2102n-errata.pdf

[vast flume]

Thanks, I did not know about that

[distant raven]

I personally haven’t encountered the issue but others have so 🤷‍♂️

[vast flume]

OK actually this seems to be correct I think

[onyx kernel]

How stupid is my latest idea? Would it actually work? Its a ring, with an accelerometer and a button. When you press the button you can turn your hand in the air to change a variable, this value will be sent via bluetooth to lets say a nother arduino, which then dimms a light or something.

[distant raven]

Seems like a fun idea

The tricky part will be determining orientation

So maybe an IMU would help with accuracy

[onyx kernel]

good point. OOHh wit a compass (magnetometer) you could actually point at the lights in your room and turn them on and off. Thats some harry potter type s*** xD

[vast flume]

That wouldn't really work bc that would also depend on position which is next to impossible to accurately determine with a hobby-grade IMU. What if you had an IR beam coming out of the front and a reciever on each target device?

[jade wedge]

That would be neat too, but I can tell you from experience, IR communication from remote controls is a lot more sensitive than one would expect. It's not at all uncommon to be able to take a standard TV remote and bounce the signal off two walls. You would need some serious beam control to target a single device in a room.

[vast flume]

Oh wow thats true

[unique patio]

you can use a diode to provide a separate +5v that is not VBUS. For instance, here is part of the schematic for the Adafruit QT Py RP2040. JP2 is usually open (it's closed if you want to backpower the USB jack, which you normally would NOT want to do).

[distant raven]

A little sensor fusion could make this possible. It’s just a lot of math

Essentially using a Kalman filter

Plus some other maths

I wouldn’t say it’s an easy solution, but there’s a lot of literature on how to do it

[unique patio]

similar here on QT Py ESP32. https://cdn-learn.adafruit.com/assets/assets/000/111/951/original/adafruit_products_QTP_sch.png?1653427567

with CP2102N

[distant raven]

Anyway, if you’re interested in kalman filters and sensor fusion: https://medium.com/@satya15july_11937/sensor-fusion-with-kalman-filter-c648d6ec2ec2

It’s a lot of math but super cool in application 🙂

[onyx kernel]

my room isnt that big. i think roughly knowing the direction would work. if not i could still say if i point north turn that light on and so on...

[distant raven]

Add a microphone and voice recognition so you can cast spells

😬

[vast flume]

Been there done that, doesn't work :(. You'd need something to fuse the accelerometer data with, and usually use would use GPS but that isn't going to work on such a small device, especially indoors

[vast flume]

Thanks for the schematic! Does it look like this woudl work?

[distant raven]

Look up inertial sense

I nearly got a job with them

[vast flume]

Oh wow nice!

[vast flume]

Their IMU is $200 tho 😦

[distant raven]

They do some crazy gps constrained position sensing

It’s not an IMU, but a module that uses IMUs, Magnetometers, accelerometers, and whatnot to generate a fused 10DoF sensor

[vast flume]

ohhhh I see

Thats actually really cool, I might see if it has a use case on one of our rockets at the rocket team

Rn I'm leading a team for active control but we're just working on 1d stuff at the moment. Once we go for 6d control it seems that module will make things a lot simpler

[distant raven]

Oh definitely

[unique patio]

I'm confused -- what other way is there to power the board?

[vast flume]

The board normally gets power from a battery. Thats on a different sheet. But the battery goes through a 5v boost converter and then a 3.3v linear converter

[unique patio]

the idea of the diode is to prevent an external 5v supply from backpowering the USB port, which could damage a host computer that is plugged in. In your example, if VUSB powered the internal regulator on the CP2102N, it would backpower the 3.3V line, which would try to power other things on the board. I think a diode is an easy solution

[vast flume]

Ah ok makes sense

[onyx kernel]

Litle update. Ive addet a MPU-9250 and got rid of the accelerometer. Are there any obvious flaws?

[rigid kraken]

Im designing a gaming controller and im having trouble finding a good microcontroller to use.
Requirements: USB/HID compatible
Easy to program
cheap and accessible
Preferences:
5v
able to run code without needing to launch a program everythime/ plug and play

[supple pollen]

5V really narrows it down, I'm thinking 32U4

[rigid kraken]

im ok with 3.3 volts but that adds more components since it is essentually just a couple buttons, leds, and a switch. Right now im using the RP2040 for the prototype but I want to eventually mass produce and sell them so I need something a little more robust.

[supple pollen]

I have a real fondness for the SAMD21/SAMD51 chips, but they're still low availability these days, I think

[distant raven]

https://www.digikey.com/en/products/detail/ATSAMD21G18A-MFT/5057241

Samd21 not so much

Samd51/54.. yeah

[rigid kraken]

I do like that the 21 has usb host and device built in

[distant raven]

SAME51 surprisingly more available than Samd51

SAME51 https://www.digikey.com/en/products/detail/ATSAME51J19A-AU/7390364

[unique patio]

it has automotive applications, maybe that's why?

[distant raven]

Yeah, very likely

Downside of most common m0/m4 is lack of trustzone support which probably would be better for devices that can control other devices

[rigid kraken]

i took apart an old USB off brand xbox controller and tried to find what chip they used there but the top had been lasered off

[distant raven]

The stm32f103 is also a good option https://www.digikey.com/en/products/detail/STM32F103RCT6TR/2035381

It has decent availability for being an stm32

[rigid kraken]

price is a little steep for that one.

[distant raven]

Sure, but the analog peripherals are generally better than the samd21

Which are important for joysticks and analog triggers

[onyx kernel]

i actually am planning something simmelar and i think ill use a esp32 with the Bluetooth HID library. idk if it works tho

[rigid kraken]

i looked into those but for the controller im building, wifi and bluetooth is overkill

i plan to eventually make a UI program that would allow users to reassign the key functions and adjust settings without needing to go into the code

[onyx kernel]

sound cool

[rigid kraken]

its pretty ambicious but ive gotten a lot of interest in it

[onyx kernel]

thats what i always wanted to do for my shortcut keyboard. never got to it tho...

[supple pollen]

There's some useful recent discussion on #general-chat about which ESP chips have full USB support (with HID), or partial (serial only), or none at all.

[inland jungle]

are there any rules of thumb for when to use bipolar transistors or mosfets when it comes to things like switching loads, e.g. LEDs, relays, solenoids, etc... especially when using GPIO pins for control

[distant raven]

I tend to like BJT for LED or current dependent circuits

But mosfets can in many cases also fit that bill

[supple pollen]

I like BJTs for 3.3V logic, as that's not much gate voltage for MOSFETs.

[distant raven]

Mosfets are definitely more sensitive to gate voltage, that’s for sure

There are some mosfets with low threshold voltages though

[inland jungle]

do BJTs dissipate more power in general, or is it really component/circuit dependent?

[distant raven]

They can

But mosfets can as well

It is going to depend on your circuit and what you’re doing

Mosfets used in switching circuits can have a lot of switching loss

[inland jungle]

I can find low Rdson Mosfets pretty easily, but is there anything similar on the BJT side? Low saturation voltage?

[supple pollen]

It's trickier to find MOSFETs that achieve low Rdson with low gate voltage (threshold voltage is where it just begins to conduct)

[distant raven]

2n2222, 2n3906/3904 tend to have low saturation current required

[distant raven]

Dmg3415 have pretty low RdsOn given the threshold voltage

[supple pollen]

I'm fond of Darlingtons like the TIP120 for switching high current, and 3.3V logic still has enough voltage to forward bias both B-E junctions in series.

[inland jungle]

yeah, low Rdson + low gate voltage tends seems to correlate with high gate capacitance, as far as I can tell

[distant raven]

But p-channel

[distant raven]

Yeah, that fits

There definitely are trade offs

What are you looking to drive?

[distant raven]

Darlington are great

[inland jungle]

nothing specific in mind, just gleaning knowledge/experience from you guys

[distant raven]

Gotcha

[inland jungle]

was triggered by the nifty 8 channel darlignton IC

actually, one specific question I've had for a while, Mosfet data sheets, Safe Operation Area: https://www.mouser.com/datasheet/2/115/DIOD_S_A0003383573_1-2542431.pdf
Figure 12.
Why does it appear that the continuous drain current spec. is always outside the DC SOA? I notice this in every Mosfet datasheet.

[latent jungle]

Until you get until the hundreds (and more) of watts, modern BJTs and MOSFETs are interchangeable. But. Finding modern devices is difficult because most tutorials are written around decades old devices.

[cursive sentinel]

At very high power levels (hundreds of watts) they are actually more efficient. They act more like current sources/sinks (PNP/NPN) than FETs, using both majority and minority carriers.

[latent jungle]

because you shouldn’t operate a device at its “Maximum Ratings.” That’s where you need to do testing.

(if you decide to operate at or exceeding them)

[inland jungle]

Sure, but it looks like the DC current is derated by ~x6 in this particular datasheet

[latent jungle]

Welcome to writing reading datasheets. It can handle 60A of current, in specific circumstances

[inland jungle]

ok, just wanted to make sure I wasn't making an error in reading the charts

it's still a bit confusing to me though, how can you even have a VDS of 10V with VG=10V. According to Fig. 1, VDS should be <<1V

[cursive sentinel]

"Absolute maximum ratings" literally means just that. Often it is based on lab tests with the die temperature forced low enough to prevent smoke from escaping. Part of my job involves pushing these limits with the help of a massive air compressor.

[latent jungle]

Vds is the drop from drain to source.

[inland jungle]

shouldn't that essentially be Rdson * ID?

[cursive sentinel]

Also, keep in mind that the temperature ratings are for die temperature, and that is what all of the derating curves are defined against. In terms of ambient temperature, this means you can run quite a bit below the minimum, but not all the way up to the maximum.

[latent jungle]

It doesn’t take into account the voltage difference of the gate and drain.

[latent jungle]

sure. that’s a different graph though

[cursive sentinel]

Or temperature variations.

[distant raven]

Also RdsOn varies based on gate voltage too

[inland jungle]

yeah, but for the SOA fig, they use VGS=10, so hard on

[latent jungle]

If you take the time to correlate all the graphs, you’ll realize they make sense. But they are all drawn around different circumstances.

[distant raven]

Most datasheets will specify RdsOn at a specific gate voltage and current test.

[latent jungle]

So one assumption doesn’t apply to all of them

[inland jungle]

yeah, I just don't understand the condition where you're dropping 10V across VDS while it's on

[latent jungle]

and (worse) the same assumptions do not apply to some of them.

[inland jungle]

all of the other figures make sense to me

[distant raven]

Picking mosfets ends up being a “what roughly meets my needs” game

[latent jungle]

if you’re talking about SOA, then it’s showing you the Pd for how long it takes it turn on.

that’s why each line is based on pulse width

[distant raven]

For fast switching and high power situations, often find something the roughly fits and use a gate driver. I discovered the magic of gate drivers when doing my senior design project

Which was working on making a visible light communications device

[inland jungle]

yeah, I've come to appreciate gate drivers after burning my finger on a PWMing mosfet that wasn't being driven hard enough 🙂

[latent jungle]

because of Miller (the smart one, not the crappy beer one), FETs aren’t perfect. They have two dynamic capacitances that vary their parameters as conditions change. Which also change because of their dynamic characteristics.

it’s also why you can’t look at a single graph and understand the full behavior

[elder peak]

Well, if you get it right, you will be FETed for sure.

But, yeah, I did some tests with a Arduino PWMing a load with a MOSFET and that was a real educimiation about the benefits of a gate driver.

[inland jungle]

going full circle, can you use BJT or darlingtons as effective mosfet gate drivers?

[supple pollen]

Yes, done it many times

[elder peak]

@latent jungle wrote a tutorial on doing it.

[latent jungle]

Sure. You can also use MOSFETs as gate drivers. The real questions become “at what voltage/frequency”

[elder peak]

Oh, also, they make pin-compatible MOSFET versions of the ULN2003A / ULN2803A darlingtons.

[latent jungle]

@elder peak thanks 😉

(also, if you rarely need to use a darlington as a gate driver.)

[latent jungle]

do you have a part number? asking for a friend …

[elder peak]

https://www.digikey.com/en/products/detail/toshiba-semiconductor-and-storage/TBD62084APG/5514090

Toshiba makes an assortment, some of which lack the flyback diodes.

through-hole, even.

And then TI and Diodes Inc make some as well. Diodes Inc just announced.

I'd posted a second one, but it's the same performance as the other one. switching voltage is down at 2.5v so it's happier with a 3.3v GPIO line than a ULN2003A.

[inland jungle]

yeah, common datasheet for the 2

[elder peak]

One nice thing about a gate driver is that if you are driving a 3.3v GPIO line to a higher voltage to make the MOSFET happy, might as well drive it at a much higher voltage so that there's less resistance.

(restricted by the "at what voltage/amperage/frequency" question, of course)

[supple pollen]

Oh yeah, I generally run my gate drivers at 10-12V

[elder peak]

If you want not just average performance, but gate performance.

[inland jungle]

do you voltage divide down from the mosfet drain voltage?

if you're running them > 20V for example

[supple pollen]

Normally I'll have a separate supply for the gate drivers. They'll cheerfully dump 2 amperes into quickly charging the gate capacitance.

[elder peak]

There are gate drivers that are happy at 24V.

Not all of them are, ofc.

[supple pollen]

I've also been known to (mis)use gate drivers as general purpose drivers.

[elder peak]

Same.

[inland jungle]

yeah, I get tempted by that as well

[elder peak]

So, there's an appendix in Art Of Electronics that explains how to use a gate driver and some resistors to shove digital signals down a 50/75 ohm coax cable.

It's shoving, not really coaxing.

I've wanted to see exactly how far I can make that go but have been otherwise distracted.

[supple pollen]

I've used them as VFD power supplies, AC coupling them to the filament, and a voltage multiplier for the grid/anode supply, back when the LM9022 stopped being made (until I found out the LM9022 was just a marketing rebadge for another chip which is still available)

[elder peak]

I feel like somewhere in history there's got to have been a MOSFET gate driver datasheet that was completely absurd with a "Typical Applications" setting that makes Desert Topping and Floor Polish look normal with all of the bizarre applications that you can make a gate driver do.

[supple pollen]

Some of Jim Williams' app notes when he got really into LCD backlighting went into a lot of odd corners, including one where he referred to a helium-neon laser as an "evil" load, and modified a CCFL supply into a current regulated laser power supply with like 12000V of compliance.

[elder peak]

Thankfully, he was able to replace some ROM chips on the B-1 bomber containing his evil laser and fill some university dean's house with popcorn.

(I was at a wedding where one of the other guests did a PhD on laser stuff of some sort, I forget, and they absolutely had seen Real Genius so my references to the movie landed as I'd intended)

[supple pollen]

I remember seeing that in the theatre! I also have a copy on laserdisc, which seems apropos.

[elder peak]

That probably raised my desire to actually study the old Jim Williams datasheets and appnotes.

I feel like my life came out the way it did because I saw Real Genius and not Revenge of the Nerds.

[supple pollen]

I ended up getting a job in a particle physics lab probably partly due to Real Genius.

[inland jungle]

my educational and occupational trajectory was determined by playing the Three Mile Island reactor simulator on an Apple II in the early 80s

[unreal flax]

Speaking of Real Genius, I'm happy to say I was responsible for this: https://www.wired.com/2016/02/google-shot-laser-60-miles-just-send-copy-real-genius/

[inland jungle]

heh, nice

[distant raven]

https://tenor.com/view/fire-the-laser-austin-powers-laser-surprised-shocked-gif-17302164

[supple pollen]

I still have a fondness for playing with big industrial lasers

[latent jungle]

yes… but… what part does the pew pew pew?

(other than, all of it)

[supple pollen]

Uh, most of it, yeah

[latent jungle]

hehehehehehe, I mean, cool.

[latent jungle]

pew pew PEW PEW PEW PEW pew

[rain remnant]

https://tenor.com/view/shark-laser-beams-club-style-haul-style-haul-gif-gif-12797722

[woven bluff]

TEENSY 4.1 is in stock

[woven bluff]

eye protection is a must

we have ultrafast pulse lasers (pulse power ~0.2TW, continuous power 20W max) not as nearly as powerful as industrial cutting lasers, but is enough to tear a hole in air.

your retina will become sieves before you know it

[rain remnant]

crunch labs camp video had lazers on recent one

[elder peak]

I just remember the lab when I was in college that had a sign on the door that said "DO NOT LOOK INTO LASER WITH YOUR ONE REMAINING EYE"

[rigid kraken]

I remember a quote from my instructor when we were using the lasers: "Only look down the barrel of the laser if you want to know what it smells like when your brain catches fire".

he then proceeded to take a watermelon, put it in front of the laser and carve a hole through it within 2 seconds.

[supple sky]

I rarely come catch up on this part of the discord but this did not disappoint 😂

[jade wedge]

Lasers are terrifying because they can easily cause permanent damage that you can't notice. So you think you didn't get hurt, and do it again, and again, and again until suddenly your vision rapidly decays.
This is because your visual cortex is extremely good at lying to you to fill in gaps.

[dry pelican]

What type of laser? Looks like it might be HeNe

[supple pollen]

Yes, that's a Spectra Physics 127 HeNe. The ones above are argon ion lasers, the one below is helium-cadmium

[onyx kernel]

Hey. I need to read the charge level of a battery (which is connected to a charging circuit). However im using an esp32 so i cant just connect it parralell to the battery because the TP4056 has a 4.2V charging voltage.
I thought of using a voltage divider but wont that drain the battery?
I also thought of using a mosfet to turn on the voltage divider every now and then to check the voltage, but thats were it gets sketchy and overly complicated.
Is there a easyer option?

[cursive sentinel]

I wouldn't consider that overly complicated. And if you use 100k resistors the drain will be pretty minimal.

[onyx kernel]

but the voltage that the esp has to measure would be very low too wont it?

[cursive sentinel]

Depends on your resistor selection.

[supple pollen]

The resistors determine current, not voltage. The analog input on the ESP is high impedance, so a high resistance isn't going to affect the measurement very much (unless there's a lot of electrical noise to contend with)

Well, the ratio of resistors determines voltage, their total value determines current.

[cursive sentinel]

*Moderately high impedance. ADC inputs can be pretty dynamic depending on the sample/hold amplifier works.

[onyx kernel]

now it clicked. ok i got it 🙂

[inland jungle]

yeah, voltage dividers can play wonky going directly into MCU ADCs if the resistance is too high. You could maybe add a unity buffer op-amp between the divider and the MCU.

[cursive sentinel]

It all depends on the characteristics of the ADC in question and the resistor values.

[inland jungle]

yup

[cursive sentinel]

The same is true for opamps, though. Excessively large resistors tend to exacerbate DC offsets (Vio, Iio, and Ibias).

[onyx kernel]

i think ill just use this guy https://www.arduino.cc/reference/en/libraries/ads1110/

[inland jungle]

Still need a voltage divider, no?

single-ended input range is 2.048V

[supple pollen]

Seems that way, yes

[inland jungle]

could use a true 5V single ended ADC with I2C out and level-shift down to 3.3. Assuming you have a 5V supply

[supple pollen]

May not even need a level shifter for I2C

[inland jungle]

the datasheets I've looked at tend to specify 0.7Vcc min for logic high

unless the MCU can tolerate 5V on the pullup

[cursive sentinel]

That's what the I2C spec defines, but not all ICs actually adhere to those thresholds.

[inland jungle]

yeah, I was talking about the ADC IC datasheets

[supple pollen]

I've seen that for things like LEDs, but most modern ICs use CMOS levels, which specify 0.5Vcc for logic high

[inland jungle]

hmm, I seem to always see 0.7xVcc, though the more sophisticated ADCs allow separat analog and digital VCC

[supple pollen]

There is generally a deadband, so it's really 0.5Vcc + deadband, which may amount to 0.7Vcc, I dunno

[inland jungle]

yeah, maybe that's a conservative margin of error

[cursive sentinel]

Well, .5V as the typical threshold, but the VIH/VIL levels are closer to the supplies than TTL.

[inland jungle]

yeah, I'm not sure if it's convention, engineering margin, or what, but nearly all the datasheets I've looked at use 0.7Vcc as the min High threshold, and 0.3Vcc and the max Low threshold

so maybe 3.3 is close enough for a 5V ADC

[cursive sentinel]

It usually is... at room temperature, in a lab.

[rough wharf]

Howdy, I was looking for suggestions on a good motor controller/motor combination for a self rising desk (threaded rod approach for high torque)

[cursive sentinel]

I would recommend taking a look at LDO for the motors. For the controller, it's tough to say exactly.

[supple pollen]

If you're looking to do it on the cheap, look for surplus big (3 meter) satellite dish positioners. They have the motor and gearing put together in an assembly already.

[unique venture]

Hey guys, I am hoping to use ~100 Stemma QT cables , enough that buying official "stemma qt" cables is really inefficient

I found what looks to be the same thing - JST SH 1.0 Reversed cables (link below), but before I order a big batch, I was wondering if that sounds correct to anyone or if anyone thinks that "Stemma QT" might actually be something else

This is the item: https://www.aliexpress.us/item/3256801109294283.html?spm=a2g0o.cart.0.0.151938daUCUAiV&mp=1&gatewayAdapt=glo2usa

thanks for looking

[left grove]

there is a guide on the adafruit website about the specs of stemma and stemma qt and there they just tell you what the plugs are 😄

[broken zenith]

https://www.digikey.com/en/products/detail/vybronics-inc/VLV152564W/19203005 What's the best way to connect this to a PCB? Is it possible to pick and place?

It doesn't look like you can P&P it actually 😦 is there a compact, cheap LRA/ERM the you can do that with?

[flat roost]

Every time I've seen a LRA (either cynindrical or coin), it's been connected with tiny wires to a PCB.

[broken zenith]

Hmm, OK. I'll see if I can work with that.

[flat roost]

like this

https://www.precisionmicrodrives.com/ab-009

[flat roost]

I read through the DS, there is a note that it contains a strong magnet so be sure to place it on the PCB carefully to avoid interference

[broken zenith]

Ok, thanks for the info! My gears are turning...

[flat roost]

References I can think of in consumer electronics with LRAs are smartwatches, video game controllers, cell phones.

[unique venture]

Im looking at the guide and I can't tell what's what:
https://learn.adafruit.com/introducing-adafruit-stemma-qt/technical-specs

Is the Stemma cable the JST PH or the JST SH? both are described as being the cable, but as far as I know they're different?

ahhh i think I get it, Stemma cables are different

[left grove]

Stemma = big = JST PH

[unique venture]

there's Stemma (JST PH 2mm) and Stemma QT (JST SH 1mm)

[left grove]

Stemma QT = small = JST SH 1mm

[left grove]

exactly

[unique venture]

stemma QT is the newest / main one to standardize on right?

[left grove]

I think they might be electrically the same. Just different connectors 🤔

Looks to me like all their newer boards use Stemma QT

[rough wharf]

Honestly that's really clever haha

[cursive sentinel]

StemmaQT is compatible with Sparkfun's Qwiic spec.

Actually it goes one step beyond that and adds level translation.

[inland jungle]

it does?

[cursive sentinel]

Most StemmaQT boards have built-in 3.3V regulators and MOSFET-based level translators.

[tough matrix]

regulators yes, but level translators???
e. g. here is Adafruit Feather RP2040 - no level translators on SDA/SCL lines https://learn.adafruit.com/assets/116896

neither do any of qt py boards

[cursive sentinel]

On the sensor boards, not the controllers.

[supple pollen]

That makes sense, now that I think of it. Somebody has to decide the bus voltage, and the controller seems like the logical choice.

[distant raven]

Well, not all have level shifters. But in theory with I2C, if you’re pulled up to 3.3V on the sensor breakout, so long as the host isn’t pulled up to 5V, it should be fine.

I know most every SPI board Adafruit sells level shifts

[supple pollen]

Since SPI actively pulls high, that seems reasonable to me

[cursive sentinel]

It can get unreliable as the clock frequency goes up. And I2C has no error detection/correction mechanisms.

[distant raven]

I didn’t say it would work well, but it would mostly work lol

[cursive sentinel]

Yes, assuming those pins are used only for SPI.

[supple pollen]

That's an odd thing to say. If there's an SPI device driving them, they're going to get driven to that device's supply voltage.

[distant raven]

Especially given most i2c sensors work at 100-400kHz, it isn’t a huge concern

[distant raven]

You’re referring to hardware spi?

Software spi can be a bit more finicky

[supple pollen]

Hardware, software, it doesn't matter. It actively drives the signal both high and low, in contrast to I2C, which is passively pulled up to whatever voltage the pull-up resistors are connected to.

[distant raven]

Right, open drain vs push pull

[cursive sentinel]

I'm talking about situations where the pins may have alternate modes. SD cards come to mind. SPI is the 'fallback' mode on those. If you actually want to read or write significant amounts of data, you would use SDIO, which is a bidirectional bus.

[stuck sluice]

need panel mount connectors to transmit pwm data so i cut apart a usb cable and used the 4 ports in it to send pwm data. the issue is that the signal is getting attenuated, when checking on an oscilloscope the max voltage is below 500mV when it should be 3.3V. The signal exists but it's attenuated significantly. I have two channels with each of them having their common grounds and one of them is totally fine while the other is attenuated. Any ideas?

any ideas what could be causing this issue? I checked the resistance on the total wire and it's below 1ohm using a multimeter

[supple pollen]

I'm guessing a wiring error.

[vast flume]

Have a bit of an obscure question. I'm interested in programming an stm32 board in rust. The problem however is that I am using a Segger J-Link programming connector. Does anyone know how to actually upload the compiled code using this connector? The only thing I could find required installing a new driver for the j-link that would cause it to no longer work with any application

[flat vigil]

what does it currently work with? usually I use jlink's gdb server

[vast flume]

This is the quickstart I had tried https://github.com/TeXitoi/blue-pill-quickstart/tree/master

[flat vigil]

openocd should be able to communicate with a jlink

[vast flume]

What is that?

[flat vigil]

openocd is listed in those instructions

[pale umbra]

what's up with function generators not longer being sold? only 6 choices on digikeys all obsolete. What's supposed to be the upgrade?

i mean stuff like max038c and xr2206

Im not sure if waveform generator or Direct Digital Synthesis are the same thing for these xr2206 kits

[distant raven]

Look up Arbitrary Waveform

There’s a ton of options

https://www.digikey.com/en/products/detail/SFG-205/7606530

[pale umbra]

thats 970$ thought and the xr2206 was like 2$

[distant raven]

Are you in Canada? It’s US$687

Anyway, Siglent has US$379 arbitrary waveform generators

[pale umbra]

yeah 970 CAD$ Just looking to make that xr2206 kit except I want a chip still being made and that kit is 6 US$ for everything including the chip. Also digital should be more fun. But it seems that functionnaility changed name or something

I feel like we are talking apples and oranges. Im not installing a radar or making my own radio I want just to undesratndw waveforms generators/PLLs and I saw a 6$ kit with xr2206 on aliexpress but the chip is outdated and hasnt been made since around 2009

[distant raven]

Probably new old stock

[ashen fox]

does Stemma call out a maximum I2C speed?

or specify one at all?

[latent jungle]

No (based on this page): https://learn.adafruit.com/introducing-adafruit-stemma-qt/technical-specs. But since I2C wasn't really ever meant to be run through a wire, 400kbit/s and faster is going to depend on wire length.

[ashen fox]

awesome. adding board def for stemma on thingy53 in zephyr and it came up

[distant raven]

Slightly more nuanced than wire length though 🙂

[ashen fox]

summing my pin capacitances rn

[latent jungle]

Yeah, but in a "plug and play" ecosystem, you really only have one thing you can change.

[distant raven]

You can change pull ups

With a little ingenuity

[dry pelican]

Usually running it at 400kHz max is fine

[distant raven]

Also adding a buffer can help a lot

[dry pelican]

3.4MHz is the fastest, but the signal probably gets so slewed that it's useless for stemma

And 100kHz is kinda slow

[latent jungle]

I'm not disagreeing. I'm just saying that for the vast majority of users it's going to be easiest to say: the speed limit is the length of the pre-made wires you're using with the pre-made modules to plug into your pre-made microcontroller board.

[distant raven]

There are buffers for i2c with stemma connectors so that falls into plug and play

Also, another nuanced point is going to be your choice of operating. Using CircuitPython/MicroPython will most assuredly limit you to 400kHz

Arduino/RTOS/Bare Metal will enable faster speeds.

But so few applications use more than 400kHz, and most even suggest using them at 100kHz

[dry pelican]

I would think that since I2C is usually implemented in hardware, you could go as fast as you want on CP or MP

As long as the underlying code supports it

[distant raven]

For CP/MP it isn’t the hardware that limits you so much as python running in a constrained environment

Plus as far as I’m aware, for circuitpython it only supports up to 400kHz anyway

[dry pelican]

Yeah I guess you won't get any gains from going faster than 400kHz. But you could probably run the clock faster if you wanted to modify the underlying CP code

[distant raven]

A faster CPU definitely will help as long as those CPUs are stable at those speeds.

I’m particularly interested in the ADS7138 because it supports turbo comparator mode and 3.4MHz i2c 😬

I have a breakout board made for it and a partial Arduino/Circuitpython driver

[ashen fox]

writing a driver for the ads131m08, it will theoretically run at 25MHz SPI

surely these dupont wires i have running from a dev board will support that right? hehe

[distant raven]

Probably okay given that SPI is push pull

[ashen fox]

ah i see you reacted to my board in the show and tell channel yep i like it! did you design yours

[distant raven]

I did 🙂

I design a lot of neat little boards, but not as cool as the things that @latent jungle makes

He makes some amazing boards like some Apple ][ accessories/replacement peripherals

I made this little feather wing for an FPGa feather I’m developing

[supple pollen]

You can make one out of a microcontroller these days, so the dedicated chips aren't popular any more.

[latent jungle]

That's cute. Is it doing traffic light logic?

[distant raven]

Yup!

[latent jungle]

oh. it's a video. lol

[distant raven]

https://github.com/skerr92/ice5lp4k_examples/tree/master/TrafficeWing

[cursive sentinel]

They're level shifters, not buffers. For I2C a buffer has to detect a rising edge and actively drive it high.

[latent jungle]

There are I2C repeaters that can be used as a buffer.

[cursive sentinel]

Yes, but that's still more complex than the level translators Adafruit uses on their sensor boards.

[distant raven]

https://www.digikey.com/en/products/detail/PI6ULS5V9617AUEX/6237571

The naming is weird

Level translator repeater lol

Also more interesting: i2c accelerator . https://www.digikey.com/en/products/detail/LTC4311CSC6%23TRMPBF/1814489

[cursive sentinel]

Yeah, that's a repeater, but Adafruit doesn't put those on every sensor board.

[cursive sentinel]

Yeah I have a few of those.

[distant raven]

https://www.adafruit.com/product/5159

Hotswap I2C buffer

😛

[cursive sentinel]

Yep, I have those too

[distant raven]

This is what I was referring to when saying i2c buffers

[latent jungle]

Oooo Hot Swap.

[dry pelican]

Those active terminators are also cool

[distant raven]

It’s more like a null modem. But it can be used to buffer signals and reduce bus capacitance so you can do longer runs

[cursive sentinel]

The PCA9615 is another interesting one. Differential I2C driver with hot swap support.

[distant raven]

And then you also have “redrivers”

https://www.digikey.com/en/products/detail/PCA9509GM,125/1756565

Which is the same thing as a repeater

But the name is more fun

[cursive sentinel]

You could also use these as I2C extenders...
https://www.analog.com/en/products/MAX96724.html

[distant raven]

But repeaters more often that I’ve seen just get used as buffers rather than voltage level translators

One startup I worked for, we were having a bus capacitance issue and we’re going to use that PCA9509 as a buffer

It went belly up when the grilling market collapsed last summer 😬

[cursive sentinel]

You missed a pun.

[distant raven]

Up in smoke? Lol

[cursive sentinel]

It got roasted when the grilling market collapsed

[distant raven]

Oh lol

That’s a good pun

[inland jungle]

sparkfun uses the PCA9615 to drive their RJ45 I2C bus

[cursive sentinel]

Yeah

I just set it up in my printer enclosure last night.

[distant raven]

I definitely enjoy the dozens of names for the same product.

[cursive sentinel]

Since I'm using Adafruit sensor boards, I'm running the bus at 5V.

[cursive sentinel]

Which?

[distant raven]

I2C level shifting repeaters

There’s like 6 different names for the same thing on Adafruit and it depends on the brand selling

Using the word dozens was an intentional hyperbole

[cursive sentinel]

Different chips though right?

[distant raven]

Different part numbers but they all say “level shifting repeaters” in the datasheet

[cursive sentinel]

Ah

[distant raven]

I even found one for HDMI

[cursive sentinel]

Yeah because HDMI was stupid enough to use I2C over the cable length with SDA and SCL in a single pair.

[distant raven]

Well, the name in the datasheet for that says “2 wire serial bus extender for hdmi”

PCA9507

18m of i2c over HDMI

So effectively 60ft if i2c distance

Fancy

I have no need for 60ft of i2c but neat that you can do that

[dry pelican]

Does anyone know an inexpensive synchronous buck-boost converter (like the TPS63xxx series) with a 22v or more maximum output voltage?

[distant raven]

this might work? https://www.digikey.com/en/products/detail/TPS55288RPMR/13212451

22V max output, 36V max input, nice beefy current output

16A 😬

I imagine that’s at a lower output voltage

[dry pelican]

That's something I'm thinking of using for another project

It's also kinda expensive

for what I'm aiming for

[distant raven]

You’d probably be hard pressed to find something cheaper

[dry pelican]

might be just 16A switches (which is still very beefy)

[distant raven]

Digi-Key narrowed down to like 8 results from like 1063

[dry pelican]

hmm. Weird that the type of IC I'm looking for is uncommon

I think the TPS63xxx is that cheap because there are a lot of low voltage applications for efficient buck boost converters

[distant raven]

https://www.digikey.com/en/products/detail/MPQ4210GU-AEC1-P/11570485

Higher max output

But super beefy for the price

100W

[dry pelican]

I've actually run into these chips before lol

Monolithic power seems to have some great stuff

[distant raven]

Yeah, I do appreciate their designs a lot

[dry pelican]

I also ran across the MPQ8875. That thing is a beast

[distant raven]

I was just about to share that one lol

[dry pelican]

6A advertised output (or switch current)

[distant raven]

$5.79 a piece

[dry pelican]

I2C control

30v output

[distant raven]

Mhmm

Very nice all around

[dry pelican]

just no current limit

and you can't use it without I2C

[woven bluff]

OK I burnt 5 BQ25606...I'm about to give up this project..

first two 450C 10% hot air solder, third 300C 33% hot air solder, forth and fifth 350C 10% hot air.

[dry pelican]

I've used 260C with low melt paste

And I haven't fully killed any chips that I know of

I think on like 20-30%

[distant raven]

Applying low temp paste around a part can make a world of difference when removing ICs

[dry pelican]

Huh. Good tip

[ashen fox]

i am a hotplate enjoyer now it's so much easier

[woven bluff]

we have hot plates in Chem lab, I'll try next week.

[scarlet jay]

how can make +-30V converted to +-3mA output 2 poinjts

[vast flume]

I'm confused, you're asking how to convert a voltage to a currnet value?

[unreal flax]

The stupid answer is a 10k resistor. 😅 But there's probably more to the story than that.

[crude ocean]

I was pondering a nice bowl of ohm-slaw

[ashen fox]

What's your preferred method of battery reverse polarity protection?

*with onboard charging, critically

[ashen fox]

Found an AD appnote, AN-171

https://www.analog.com/en/app-notes/an-171.html

[scarlet jay]

and also if you have bipolar dac with 3v3 input and at least 12v output (the higher the better)

[molten aspen]

Question regarding 74AHCT125: It's a 4 channel IC that can be used for logic level (namely, I'm using it to 3.3V to 5V logic level conversion) but the question remains; It has output enable pin for every channel which is enabling when pulled to the ground (active low).

Question: Do I need to pull it down with a resistor or is it enough if I just connect GND directly to those pins to enable the channels? AKA need for pulldown resistor?

[cursive sentinel]

You can ground them directly.

The one benefit of using pulldown resistors is that if you do decide you want to drive those pins high you can.

[molten aspen]

Okay, great! Thanks!

[molten aspen]

Quick question related to Schottky diodes; What is the difference between Schottky Barrier Rectifier and Schottky Barrier Detector and Switching diodes?`
https://www.onsemi.com/pdf/datasheet/mbr120vlsft1-d.pdf
https://www.onsemi.com/pdf/datasheet/bat54t1-d.pdf

[supple pollen]

I'm not sure myself, but generally I think of "rectifier" as referring to more power supply applications, and "detector" and "switching" as referring to signal handling.

[molten aspen]

Yeah, something like that I was thinking too, Onsemi's nomenclature is quite...interesting

[supple pollen]

This convention appears to go back to the vacuum tube days where a 5U4 was considered a "rectifier" and a 6AL5 is referred to as a "twin diode"

[molten aspen]

Okay

[supple pollen]

As for the confusing nomenclature, many modern firms are an agglomeration of various companies they've acquired over the years, with their own part numbering schemes, data sheet formats, and ways of wording things.

[molten aspen]

Regarding diodes for protection Pico's power pin (prevention of current flowing out from Pico back to the system through same pin when powered through USB), would something like this work? Albeit, it has way over the top amperage rating for the use-case; 1A whereas I need only maximum of 150mA at worst, but it has low Forward voltage of 0.32V at 0.1A.
https://www.tme.eu/Document/2989649f1dde670a1cc3c74637d70943/1n5819hw.pdf
Or have I understood the sheet wrong?

[supple pollen]

That seems like a suitable choice to me

[molten aspen]

But was my reasoning also correct? I'm trying to learn what specs to look for and if reasoning is correct :3

[supple pollen]

Yes, your reasoning is good. If you're optimizing for low voltage drop, you'll typically find parts with higher current ratings, which are loafing at your design current.

[molten aspen]

Ah, good to know. Thank you very much once again 🙂

Having community like this is nice way to learn things like these, even though you have Internet (and its abudance of information, some not so correct also) and books 🙂

[supple pollen]

It can be very helpful to talk through concepts to sort out your thinking

[molten aspen]

Exactly!

Also regarding diodes; When doing low voltage stuff, e.g. 3.3-5V stuff, wouldn't Schottky diode be the ideal over the normal diode, due the low voltage drop? (unless you really need to prevent the leakage of current). I'm thinking if I should replace the normal diode with Schottky diode in the open collector schematic, as in the pic (pic is from EEVBlog, not my picture)

MCU_Vcc is in this 3.3V.

I'm just thinking that normal diode might work but is there any reason why not to use Schottky in this use case if space is not a constrain?

[supple pollen]

Yes, the Schottky is preferable to minimize voltage drop. It is also possible to use a small MOSFET as an ideal diode to reduce the voltage drop even further.

[molten aspen]

Okay, good! Regarding these low voltage usecases (MCUs etc), are there other use cases than that very little current leak -requirement where normal diode should be used instead of schottky? If we are going with an absolute maximum of 1-2A of power consumption?

[supple pollen]

For most use cases, you can just use a Schottky everywhere: they're fast and low voltage drop. Sometimes you need to deal with details like leakage or junction capacitance, which can drive the choice of part in odd directions, and of course if you need to work with higher voltages, other diodes are better choices. But if you're just looking for a generic, use-everywhere part, a Schottky diode can fill that rôle well.

[dull perch]

I noticed a mistake in one of my featherwing designs that could be fixed by connecting EN to VBAT when powered on. I don't imagine this is a problem but am looking for confirmation. Any thoughts?

[supple pollen]

I see a lot of boards using a resistor to EN, which will pull it up to enable it by default, but have the option to pull it down without short-circuiting the power supply.

[molten aspen]

Okay, great, good to know 🙂 Cause that was the thing I was thinking, time to write that down for future reference! 😛 Thanks once again!

[vast flume]

Shottky diode

[distant raven]

Adafruit does this on all their feathers

[dull perch]

Thanks folks

[scarlet jay]

@worldly schooner @long wraith @supple pollen do you guys think you coukd help me find bipolar DAC chips/ boards? I Can do anything for you in return 😉

[worldly schooner]

I think for your application, it would be worth contacting an applications engineer from someone like Analog Devices.

[scarlet jay]

Tried anything, have Been through the whole linkedin 😄

[worldly schooner]

Please do not ping people without recent conversation.

[distant raven]

I'd suggest hiring an Electrical Engineer. Create a job posting somewhere.

or at the very least using the help forums that every manufacture provides.

[supple pollen]

This is after you said you didn't want an MDAC approach? One popular trick is to use a unipolar one and just shift the zero reference. So 0 to +5V becomes -2.5V to +2.5V

[scarlet jay]

@supple pollen okay, sounds realistically reasonable

can you send me an example? I have been brainwashed 🙂

[supple pollen]

That would depend on how it's deployed. In some case, a midpoint voltage is derived with a resistive divider and a unity gain buffer (to lower the impedance).

[inland jungle]

https://www.ti.com/lit/an/slaa869/slaa869.pdf?ts=1689475542380#:~:text=The unipolar to bipolar output,of the bipolar output voltage.

[dry pelican]

Those TI app notes are really useful

[ashen fox]

https://neatcircuits.com/doctor_pedal.htm this is also used in guitar pedals

[scarlet jay]

I'mThank you guys, Im going to check this thing out 😄

[molten aspen]

A connector for DS18B20, I'm debating whether to to go with RJ10 or 3.5mm jack

RJ10 is otherwise good but I need to solder an extra wire as it can't handle the default 23awg

[supple pollen]

That works fine for AC signals, but you need a different approach for DC signals.

[supple pollen]

I'm not sure what RJ10 is, possibly a 4P4C modular connector or something? There are some other possibilities like some of the cable mounted JST connectors as well if you want a latching version.

[woven bluff]

I finally found out why I burned so many chips...

the 2x2mm inductor was soldered 90degree off..

[inland jungle]

that'll do it

[woven bluff]

hot plate is a nice way to work single sided smd

but the magnet from stirrer tend to suck the magnetics off the board... that's how I found out.

[inland jungle]

I got one of those little smd hot plates that works pretty well for the sizes I'm working with

[woven bluff]

what is Voltage Dropout for a LDO?

[unreal flax]

That's how much larger than the output voltage the input needs to be. Like you can't usually give an LDO 3.301V and still get 3.300V out... gotta provide at least 3.4V or whatever for it to work properly.

[woven bluff]

thanks

[cursive sentinel]

What happens exactly when you run below the dropout voltage depends on the regulator. At best you will get a slightly lower voltage at the output; but more likely you will lose regulation. When this happens you will still get an output, but it will be load-dependent and likely have a fairly high resistance.

At worst, you'll end up with oscillation on your supply.

[latent jungle]

Yeah, a case where the phrase “it stops working” doesn’t mean there’s no output.

[cursive sentinel]

Undefined output is worse than no output.

[latent jungle]

Exactly

but it just now occurs to me that people usually equate “drop out” to “stop working.” Which is true, the regulator has stopped working. But in this case, it means it isn’t regulating as expected, not that it shut of the output.

[distant raven]

I like LDO that have minimum voltage for 3.3V being 3.3V 😛

Like the TLV75533 which has 238mV drop at 500mA

So I really only need 3.5V or so to get 3.3V when drawing the 500mA limit

It’s pretty close to 1:1 at 3.3V input for lower current loads

For what I’ve measured anyway

YMMV

[cursive sentinel]

The real question is what it does under changing loads.

Also 238mV is significant at 3.3V.

[tropic harbor]

Before I buy all the stuff and learn the hard way, can a neopixel jewl fit inside of a neopixel 12 ring?

[supple pollen]

Looks like yes https://twitter.com/w8lid/status/1140019618691657729

Mike Diehl (@w8lid)

@adafruit NeoPixel Jewel fits perfectly inside a NeoPixel ring 12. It’s almost as if they planned it that way. 😎

Twitter · Jun 15, 2019, 10:13 PM

[tropic harbor]

Thanks!

[distant raven]

It’s based on a 5V input so probably not nearly as significant. 😛

[supple pollen]

I'm used to more headroom, this one uses about 85V across a trio of triodes

It looks to me like this supply can deliver on the order of 350mA, which doesn't sound like much, but at 225V, that's over 75 watts.

Then again, that 85V of headroom is burning up 30 watts in the pass tubes themselves (along with 24W of heater power). The entire instrument consumes over a kilowatt, making an efficient room heater.

[cursive sentinel]

(30+24)/1000 = 5.4%

Not a very efficient room heater.

[woven bluff]

Carnot cycle heat pump (AC) has efficiency > 100% as room heater, plus it makes popsicle at the same time.

[supple pollen]

Eventually all, that kilowatt is dissipated as heat, so it's ultimately 100% efficient at converting electricity into heat, like a resistive heater, computer, or most other electric or electronic devices.

[woven bluff]

and our universe maxed out entropy and ended up heat death

[supple pollen]

I figure I can get a 1kW ceramic heater and get just heat out of it, or a Tektronix 555 and get not just the same kilowatt of heat, but also a high-end oscilloscope.

[woven bluff]

finally got BQ25606 working, this little chip can battery power your 5V rail and serve as transparent UPS.

I'm going to add this to most of my projects

[dry pelican]

How long does it usually take for Oshpark to make a board?

[cursive sentinel]

Probably several hours to a day per panel. There are a number of chemical processes that require time.

[dry pelican]

yeah I know, just wondering what the avergae turn time is

I found an EEVblog forum post for unofficial turn times

from 2013 lol

[inland jungle]

my last few orders have been ~8-10 days from orderto shipping

[supple pollen]

It also depends on the volume for that process. Different processes (4 layer or After Dark) may take different amounts of time to get enough orders to fill a panel.

[distant raven]

After dark has been surprisingly fast lately

~9 days has been my turn around lately

[woven bluff]

mine took a week, regular purple

[ashen fox]

The nPM1300 is a nice pmic as well. have a board I've designed w/ it

[last rapids]

hello! anyone have experience or know anyone who has experience with schematic/pcb design?

[unique patio]

are you asking to pay someone, or are you asking if you can ask a q? If the latter, go ahead and pose the q

[last rapids]

either I think, (quick backstory; Im and artist working in video, tech, sculpture, ive been using a bunch of different ways to loop videos on different types of single board/micro computers) I recently found this "Turing smart screen” it’s a hardware monitor for pc but ive been using it to display videos on it with seamless loop I like using it but want to possibly learn to make it myself and customize it to my own needs, I have no expierence in designing PCBs so I wouldnt know where to start...

[woven bluff]

you would much better off with a tablet

[last rapids]

it sure would be a lot easier but that wouldn't work for the projects I'm working on

[woven bluff]

you can remove the frame, then you basically got a SBC with a screen

[unique patio]

are you saying you want to replace the circuitboard in the pictures above (as opposed to adding another board). ... hoo boy, Turing Smart Screen uses an E8051, derivative of an ancient microcontroller.

have you considered a Raspberry Pi with external screen (HDMI or LVDS)?

designing your own microcontroller board is a big deal, and I think would distract from your eventual goals.

there are many off-the-shelf solutions

[last rapids]

hahaha I thought that would be a big endevour! I am open to other solutions!, I was just thinking of using a pi zero with lcd gpio top to accomplish the same thing ( I think )

BUT please do let me know your solutions!

[unique patio]

are you just looking to play GIF's, or higher-bandwidth videos? We have a simple project: https://learn.adafruit.com/using-animated-gif-files-in-circuitpython
actually many: https://learn.adafruit.com/search?q=%2520animated%2520gif

https://www.adafruit.com/product/3932

https://www.adafruit.com/product/4499

we have our own products too, but many out of stock due to shortages which are now being resolved

[last rapids]

I think the gif would be too compressed an wouldnt give me the resolution needed, although i havent tested it too much and maybe it would work, im open to the idea of using animated image style file formats though

the videos I would be using would be anywhere between 30 seconds too an hour+ though

are there microcontrollers that could handle the bigger files?

also thank you for your help!

[unique patio]

Raspberry Pi would be fine. Now that they are finally becoming generally available again, worth considering. See https://rpilocator.com/ for who has it in stock. What country are you in?

[last rapids]

On the topic of using a pi for video looping, do you have any experience with hello_video?

[unique patio]

nope, sorry

[fervent lance]

Also consider hackerboards, their database is old, but it's usefull to have an idea

[proven lark]

I am looking at the documentation for the SN75176A chip (https://www.ti.com/lit/ds/symlink/sn75176a.pdf). I am trying to drive the device to output signal over A & B. I have the two signals sent out over D and DE appropriately, but I am a little confused about RE. From my understanding, it looks like I need to drive that line high to disable reading and enable writing?

Specifically looking at this section for my information.

[supple pollen]

The chip includes two units, one transmitter and one receiver. It only accepts one signal (D) to transmit, and DE enables the transmitter. /RE enables the receiver separately. Both the transmitter and receiver are connected to the same A and B pins. If I read it correctly, the transmitter ignores the /RE signal, and you can enable either or both the transmitter and receiver separately.

Presumably, you could connect the D and R pins together, and the DE and /RE pins together, and use them as a "direction" input.

[proven lark]

This is the signal I provide it (shown as the image below). Where yellow corresponds to D and the blue corresponds to DE.

I am expecting something similar to:

I am actually getting:

Given the state machine:

• D=1 & DE=1 -- This looks fine, clear high signal
• D=0 & DE=1 -- Extremely noisy
• D=X & DE=0 -- Stable, but does not return to zero

Which does not correspond to the documentation

[supple pollen]

It looks as if the signal driving your D input does not have sufficient logic swing.

The "return to zero" depends on how you have your signal terminated. Are you using a differential probe?

[proven lark]

Just a standard probe

I have ground and signal hooked up to the A & B lines

[supple pollen]

Oh, that's probably your problem. You don't want to ground out either of the signal lines.

[proven lark]

cool

[supple pollen]

You can hook one probe to A and one probe to B, and set the scope to "A - B" mode (in some scopes, this is done by inverting B and using "A + B").

[proven lark]

Interesting

And hook up ground to common ground?

[supple pollen]

Right.

It's a way to probe a differential signal (which is what this is) without a differential probe.

[proven lark]

Yea, trying to find out if my scope (DS-1250C) even offers that

I have never dealt with differential signals before, so I had no idea that they had a specialized probe for that.

[supple pollen]

Yes, press the "MATH" button, then choose "Ch1-Ch2"

[proven lark]

Interesting.

Yea, so it was working the entire time. I was just measuring it wrong. 😛

Also, was curious about reading since I will need to tackle that next. I need effectively read 3 states:

• +A & -B (A is high)
• -A & +B (B is high)
• A=0V & B=0V (both likes are 0 volts)
  But the diagram shows:

Looking at the diagram, there really appears to be two states I can determine from this, the two states I stated earlier when the diff pair is active. But I do not see a way to check that third state: when the lines are zero.

[supple pollen]

Right, the R output is undefined at those times.

[unreal flax]

Typically you'd have pullups/downs on the line to put it into a known state when nobody is actively driving it.

Or to rephrase, it's not a usual requirement to sense the third state, only to make it well-defined so that you can reliably sense the start bit when someone does start transmitting.

[supple pollen]

I think the termination resistors put it into that 0V state, which does make it tricky to detect from a 1-bit lead if it has transitioned to a "mark" or "space" state.

Presumably the protocol would allow for that, so it could pick up correctly on a start bit

[unreal flax]

This sort of configuration is sometimes used, for example:

[supple pollen]

This is the usual application

[hasty solar]

hey folks, i'm monitoring humidity in my greenhouse with the humidity meter in my SCD-41 CO2 sensor, but i'm realizing it's super inaccurate within the range i usually keep my greenhouse within (90-100% RH). are there any chips that read with high accuracy at very high RH values that aren't extremely expensive?

(by high accuracy I just mean ±2% or so)

[hasty solar]

seems like the AHT20 is a good bet? https://www.seeedstudio.com/Grove-AHT20-I2C-Industrial-grade-temperature-and-humidity-sensor-p-4497.html
plus it's an I2C sensor so i think i can just chain it onto the STEMMA QT breakout board my SCD-41 is on

SCD-41's humidity sensor is even worse than i thought :(

[hushed smelt]

How do you know you keep your greenhouse at 90-100% RH? Are you comparing the SCD-41 against a known highly accurate hygrometer?

[hasty solar]

no, it's not a matter of me knowing it's getting wrong readings, it's that i'm using it to drive a humidifier to keep it within that range, which is ineffective when it's not rated for/accurate within that range

the readings it gives at the listed range are anomalous (holding steady for hours at 96% and slowly dropping from very minor air leakage then suddenly dropping to 85% in minutes with no change in conditions, for instance), so i know it is indeed giving inaccurate readings outside of its rated range

[supple pollen]

Note, at that range of humidity, you have to deal with "condensing" vs "non-condensing", and of course, condensation.

[distant raven]

I like humidity sensors with built in heaters to evaporate any condensed humidity

[unique patio]

We have quite a few of these: https://www.adafruit.com/search?q=humidity A number of them say they go up to 100%. Check the product descriptions for range and accuracy. Some products are more accurate than others at high humidity. Also then check the data sheets. The SHTxx line are well-respected. Some are in fancy cases.

[distant raven]

Like the HDC1010 from Texas Instruments

[hasty solar]

definitely like ordering adafruit sensors when i can just because the chainable breakout boards save me a ton of hassle

[distant raven]

I have an Adafruit compatible board I’m working on for it

[hasty solar]

i ordered adafruit's AHT20 board for the time being, but definitely let me know when you have that board finalized! i'd be excited to make the upgrade

[dark grail]

thats using the openmax api, which is now deprecated

[gleaming crest]

Does anyone have a recommended PMIC? Looking for one on digikey that can charge an 18650 from USB 5v or 3v3 MCU power (one or the other, I don't need both) so I can power it when it's not plugged in

Edit: also ideally available in a form factor that I don't need a solder mask for (MSOP is ok but no QFN pls)

[supple pollen]

You can't charge a lithium cell from 3V3 without a boost converter

[gleaming crest]

That makes sense lol 3.7v not thinking rn. I am just looking for something that doesn't have some kind of issue that makes it terrible, like the time someone I know got a IMU that had a read-only sleep register that made it unusable

[gleaming crest]

Probably gonna go with BQ2409x

[onyx kernel]

how bad is cutting trough holes, i feel like there is a propper way of dooing it but i don't know how. This PCB should sit on another PCB and those two contacts will be soldered to the pcb below (just like with esp boards)

[hushed smelt]

The proper way would be castellations which put copper in the entire crescent shape. Without a proper castellation you'll just be left with the copper trace on the top pad and milled FR4 hole (not what you want).

Castellation costs a little bit more because there's more work to be done by the pcb manufacturer. Work with them to ensure your castellations will work how you want.

[rigid plume]

Just wondering if I hooked up my leds correctly

[unreal flax]

Yep, that should work, though the resistor value strikes me as on the low side, so you may end up drawing more current than you really want.

[molten aspen]

Being utterly utterly paranoid about MOSFETs (first time using them in my own project): Wouldn't this work basicly?

+3.3V comes from GPIO pin that is pull-down in order to keep it off during initialization. The two connectors resemble the load; fan in this case.

Have I wired it correclty and understood that the chip I'm planning, is suitable?
https://www.tme.eu/Document/35c9ffb7cd05261537b310e0613f3b53/FDC6401N.pdf
(Figure 2., with 3.3V the resistance drops to around 0.05Ohms, enabling the pin).

(FDC6401N is a dual N-channel MOSFET).

[latent jungle]

If your load is going to be inductive, like a fan motor, then you need a protection diode for the FET. I would also ground the Gate and Source of the unused FET.

[molten aspen]

The other FET is connected similarly as the first one. For the inductance, this is with PC fans that have very low inductance (I might be totally wrong, correct me if I'm wrong).

[supple pollen]

That FET has parasitic body diodes that will at least provide some protection from the spikes generated by inductive loads.

[molten aspen]

These are mainly 12v fans with average of 0.1A consumption

[latent jungle]

Low inductive is still inductive. I would still add a diode. Don’t rely on the body diode to protect against an inductive load.

especially on a FET that small.

[molten aspen]

Okay, great! Thanks for the feedback!

[molten aspen]

Little confirmation: We are now talking about flyback diode configuration; diode between positive and negative pins?

[supple pollen]

That's one configuration, and one I'm fond of, as it keeps the circulating currents local. However, it's not the only possibility.

[molten aspen]

Well, that is the easiest one cause I have just a nice deadspace near the positive and negative lead in the PCB!

Thanks once again! 🙂

[urban lark]

I'm trying to use the Enable pin, on an esp32 whcih is pulled by a 10k up to 3.3v on the board, as a reset pin for the esp32spi library. Would the internal pull-downs on the raspberry pi pico be enough to pull that to ground near 0v, or would I likely end up in mid-voltage territory as the internals pull ups / downs are probably 10k?

[latent jungle]

Why would you need a pull-down if the enable already has a pull-up?

[unreal flax]

Typically internal pulls are weaker, like 50-100k.

[urban lark]

weaker? 10k < 50-100k?

[latent jungle]

Also, the internal pull-up/down are much weaker than 10k:

[supple pollen]

Some of the internal pulls aren't a strict ohmic resistance, but a current source/sink (such as 75µA).

[unreal flax]

An open circuit is infinite resistance, which is the weakest pull, heh heh. Smaller resistance = stronger.

[latent jungle]

weaker/strong = how much current it allows. big resistor is smaller current, so it is weaker.

[urban lark]

Oh got you, backwards learning like electron flow compared to voltage direction.
yeah more aware of the current limits, but this is a minor blip to reset an external esp32, should be ok 🤞 .

[supple pollen]

It's true that if you have a pull-up and a pull-down, you'll tend to get an intermediate voltage, however.

[latent jungle]

No, nothing like that.

[urban lark]

its funny in my head the strength would relate to watt rating

[latent jungle]

but, regardless, if the enable pin already has a pull-up, why does this matter? You should be using the IO pin as an output, which means the pull up/down resistors aren't enabled.

[latent jungle]

Okay? So bigger resistor means less wattage, so weaker.

[urban lark]

two disconnected boards, the esp32 has a pull up to allow the reset button (a pull direct to ground) to work.
I'm using a separate pico to the same point as the switch connects, labelled Enable on the pinouts, so have to contend with that 10k pullup to 3v3

[supple pollen]

Meh, not really. I have a drawer full of 1Ω, quarter watt resistors. And another drawer full of 150kΩ, three watt resistors.

[urban lark]

I did think that, you can go big physically and dissapate more wattage, but the ohms rarely relate to the watt rating

[latent jungle]

I was speaking in terms of increasing the resistance which reduces the current from a fixed voltage source. So, less power burned.

Not "in general"

[supple pollen]

Ah, as in less wattage dissipated/wasted. That I'll agree with.

[latent jungle]

Is the separate pico connecting to that button's point to use the signal as an input or to drive it as an output?

[urban lark]

drive, and not directly to the button, but to the Enable pin on the sparkfun esp32 Thing

[latent jungle]

but it's at the same node as the button? so if you drive that pin high while the button is pressed, you get a short? (I think that's what you're trying to solve)

[urban lark]

looks like it makes most sense to go with some transistor/driver to pull it hard to ground, no biggie but wondered what the normal go-to was and wanted to avoid additional circuitry.
No ignore the button, it will never be pressed. It just illustrates the circuit on the esp32 board for the Enable pin.
Related to esp32spi specifically it wants a reset pin so some inverting output on the rp2040 side makes sense too rather than modifying the library.

[tribal knot]

Hi, not sure if this is the right group for this question. What's the cheapest way to control the speed of a cheap 220v AC fan with an esp32 board? I'm hoping there's a speed controller (or similar) that I can buy off the shelf.

[late lily]

Most of them are not possible to speed control without a variable frequency drive, ie, a glorified AC-DC-AC converter that can adjust its frequency from mains frequency

They run synced to the mains frequency

So altering the voltage or using a triac to alter the duty cycle tends to just reduce the motors torque and cause it to stall out if it cant maintain speed rather than vary the speed

[supple pollen]

Unless it's a universal motor, tapped motor (like multi speed fans sometimes are), or perhaps some other lashup (unfortunately, none of those are likely)

[last rapids]

im fine with using buster, i just want its perfect looping capability for h.264 files

do you have experience with it by anychance??

[dark grail]

ive not looked into perfect looping

vlc can do fully hw accelerated decoding with the new api's

[last rapids]

hmmmm

[last rapids]

also @dark grail you mean that hello_video is part of OMX?

[dark grail]

yeah

https://github.com/raspberrypi/userland/blob/master/host_applications/linux/apps/hello_pi/hello_video/video.c

yep, omx appears several times in that source

and its using the omx stack for decode and render

[pale umbra]

is it true that modern smd leds with very low amp uses (like 500 microamps) have very little voltage drop ? Like someone was asking in a stream if they could use a led to check if tx/rx was transmitting/receiving in uart I told them they couldnt because leds have like a 2v voltage drop which will cause uart to fails and there isnt much point since they wont see the led flashing 800000 times per second. Id like to try this out but I dont know of a device that can hardware clamp 0 to 800000 down to 0 60 ?

[brisk shell]

does anyone know of a good cheap 6-7 inch display? doesn't really matter for the resolution, i just need the display pretty much

i've looked everywhere but i can't really find one for under 40 dollars

[pale umbra]

color/mono ? not oled I hope ?

[brisk shell]

not oled

mono will work

[pale umbra]

also what do you mean by good ? as in hand-holding to easily use it? reliable (not really possible under 40$) ? works with specific tools like arduino ide ? arer you going to use an MCU with it ?

[brisk shell]

i'll be wall mounting it in a case, using it with a raspberry pi

thanks for helping by the way!

[pale umbra]

Waveshare has several if you dont require very precise touch, color LCD which are usually much cheaper than OLED and available in many sizes like https://www.waveshare.com/7inch-HDMI-LCD-C.htm. They hav a couple of models including one with case. This one is 44$

but they arent as easy to use as raspberry pi or adafruit stuff but easier than aliexpress stuff

[brisk shell]

oh thanks!

[pale umbra]

there is a grid at the end, there might be cheaper ones depending on the amount of direct interface you dont need/hand-holding you dont need. The grid has like 15 models so you can find exactly what you need and you can check the wiki too for the code/how to use them to see what you are comfortable doing vs the price

ie: just to clarify there is also a dpi touch-less 30$ one but I dont know your exact needs. HDMI wont use the pins on the pi. but DPI will

[dark grail]

DSI doesnt use any gpio pins on the rpi

[pale umbra]

what pins does this go over then?

looks like it attach to the standard pins header of the pi to me

[dark grail]

thats likely DPI

which uses a majority of the GPIO pins

[pale umbra]

oh I see I made a typo

[dark grail]

DPI and DSI are very different interfaces

[proven lark]

I am using the SN75176A chip (https://www.ti.com/lit/ds/symlink/sn75176a.pdf) to read ARINC data. The chip provides only one pin for reading incoming data; which is fine for reading RS485 data since that protocol does not use return to zero as far as I know, but I need to detect when both lines are being held low for ARINC. So I was considering using two diodes on A and B to hook up to a gpio pin to detect if one or the other differential line is being held high to detect the invert of the zero state.

The chip I am using technically uses abs(A-B) < 0.2V at the "zero" state. Which is technically not something that the basic diode idea does. But I do not know if that is necessary for a protocol like this.

I also do not know if there is another chip out there that will assist in performing nearly the same features as the one I have, but with the ability to read the three states that I need (+V, -V, & Zero).

[supple pollen]

Does "held low" mean -V or zero?

[cursive sentinel]

You could use a couple of comparators for that.

[proven lark]

Sorry, both lines are held zero (not low). There are three representable states for ARINC (A - B): +V, Zero, -V. While no data is being transmitted, the line has a zero differential across A and B such that abs(A-B) < 0.2V. Ideally A ≈ B ≈ 0V in that case. When a bit needs to be shifted out, either A is +V while B is -V for a 1. And opposite for a 0, where A is -V and B is +V. Between every bit, both lines must return to "zero".

[supple pollen]

The diodes (you'll also need a pull-down resistor) might work then. There are a few possibilities. Note that if the signals are required to be floating/isolated (it won't be with the diodes), it becomes a little trickier, but there are ways.

[proven lark]

Yea, I am guessing if the signals are floating, then that is where I bring in comparators as Herr Brain has suggested.

[cursive sentinel]

What is ARINC by the way?

[proven lark]

https://en.wikipedia.org/wiki/ARINC_429

[supple pollen]

Basically, there's a termination resistor that will pull them close to one another, but perhaps not at a specific voltage. With the diodes and a pulldown, it would quickly discharge remaining (positive) voltage (via the diodes).

However, I wouldn't apply a diode hack to anything mounted in an aircraft!

[proven lark]

No a real aircraft

Do not worry

What would a better method other than a diode hack be?

I only mentioned diodes because my limited EE brain just defaults to what I know.

[cursive sentinel]

Getting an actual ARINC429 transceiver.

[proven lark]

Which unfortunately costs $29K

I am not joking by the way

I am just looking for something that is beyond a diode level hack to get this working.

But not something that will make me cry when I go to bed at night

[supple pollen]

I'm thinking of something like a floating zero crossing detector, basically a bridge rectifier driving an optoisolator.

[cursive sentinel]

Yeah I'd need to look into it more. The RS-485 approach might work. If not a couple of comparators would likely be the best approach.

[proven lark]

My current approach looks something like this:

[supple pollen]

You probably want to connect RE# to something on the receive side

And DE as well!

[proven lark]

To ground?

[supple pollen]

You'll want to enable receive and disable transmit (at the receive end), so ground both of them (via a resistor if you like, like you did on the transmit end for some reason for RE#)

[proven lark]

So if I just have U1 hooked up (the transmitter), I get a good wave form that I can see on the oscilloscope. Adding a resistor (R2) brings down the voltage and makes ground float a bit higher. And Adding in U2 (the receiver side) with R1 makes the zero float a significant ammount.

[supple pollen]

I suppose the "hi-Z" mode has some offset (like TTL gates, perhaps)

[cursive sentinel]

With differential signals it usually does in order to prevent oscillation.

[proven lark]

This is to say that I am reading this with the math mode on the O-Scope. Where A-B is enabled to view the signal.

[supple pollen]

The bridge based difference detector would work anyway. The diodes may or may not, depending on how high it floats and the threshold of your GPIO.

[proven lark]

And to the best of my knowledge, I will leave #RE low to enable R to recieve. Unless I am mistaken. So should both of those be hooked up to ground on the RX side?

[cursive sentinel]

Or in the case of some AC coupled interfaces the inputs are weakly pulled to opposite rails and the line (AC coupled part that goes through the cable) is double terminated and biased to mid-supply on one end.

[proven lark]

Do I need to run the signal though some sort of an amplifier after the SN75176AP? Or should the chip as is be fine?

[cursive sentinel]

The RX on a differential receiver is already an amplifier.

[proven lark]

Sorry, for the TX end.

[cursive sentinel]

So no not really unless you need to do some significant common mode rejection or level translation.

I would have to look at both specs in more detail, but I doubt it.

[proven lark]

From what I have seen, the output voltage was around I think 4V across A & B. And I need 10V.

[cursive sentinel]

Ok yeah in that case a diffamp might be the way to go.

[proven lark]

I am trying to decipher this table shown on the top of page 5 to understand which voltage diff output I should be expecting to understand if this (1) is normal to what I see on the scope given my resister pair and (2) what I should look for for the diffamp.

I see V_{OC}, |V_{OD1}|, and |V_{OD1}| , but not really sure which I even look at for this diff V.

[supple pollen]

It's a 5V chip, it's not going to give you 10V. Adding amplifiers would mean having to implement all the right driving characteristics (the chip's job). In your case, I'd select a different chip that supports the voltage swing you need (and you might be able to find one that generates the "not driven" signal you're looking for, as you'll need to replace the receive chip as well)

[supple pollen]

Your pair of 75Ω resistors gives a load of 37.5Ω, which is lower than the lowest one in the data sheet, so behavior is not guaranteed. You could be overloading the transmitter.

[woven bluff]

can't you use logic analyzer to read ARINC?

[cursive sentinel]

Sure, but it would still need to be level shifted.

[vast flume]

Would there be any major disadvantages to using a linear regulator for stepping down 16v to 5v instead of using a switching regulator? I know that they're less efficient, but if I'm just trying to power an stm32 off of a 4s lipo is it a big deal? Thinking of using this one https://www.digikey.com/en/products/detail/stmicroelectronics/LDL1117S50R/7102072

[cursive sentinel]

Efficiency is the main disadvantage. For your use case it's probably fine.

[vast flume]

Perfect thanks! I've been trying to use a switching regulator but its caused so much trouble lmao

[dry pelican]

If your current draw keeps the regulator's dissipated power under its maximum dissipated power, it's fine

[latent jungle]

If the STM32 is the only thing you are powering, it will be fine. But it'll be dropping 11 volts. Assume maximum power it can dissipate (without serious heat sinking on the PCB) is 1 watt. That means you're looking at about 90 to 100 milliamps maximum current for whatever else you plan to power with it.

[vast flume]

Yeah STM32 and then some other stuff with similarly low current draws

[latent jungle]

on the other hand, also consider, if your circuit averages 10 milliamps, that means the linear regulator is dissipating 110 milliwatts or wasting 22 mAh of the battery's capacity (at 5 volts)

[woven bluff]

LDO in battery-powered application...think carefully

newer MCU even has build-in SMPS

[proven lark]

Yea, looks like this chip, the HI-8585, might be a a better match for my needs. Unfortunately, this thing is out of stock at every site I have checked thus far.

[spice zenith]

How likely am I to get "noise" if I leave one of the ADC inputs floating on a trinket M0 and then read the ADC? Could that produce a low grade source of entropy if I looked at the LS bit?
The problem I'm trying to solve is to seed a 32 bit PRNG with some entropy every time the sketch starts running. Even 16 bits would be enough just to get different values from one run to the next.

[proven lark]

What are these transistors? The symbols for the base on those transistors looks like a schottky diode?

Never mind, found it: https://en.wikipedia.org/wiki/Schottky_transistor

[opal rune]

It's not as reliable in my experience but good enough if you want something somewhat unpredictable. It tends to float within a narrow range of values, so you're likely to seed it with the same value at times.

[supple pollen]

I do this all the time, it's probably random enough for my purposes. You could add a thermistor, RTC, or count time until the first input as well.

[spice zenith]

@opal rune @supple pollen Sounds good. Most likely what I'll do is read it at (e.g.) 1 ms intervals, and then take the ls bit. Then take pairs of bit values and use Von Neumann debiasing to get the actual random bits I need. https://mathoverflow.net/questions/152107/proof-of-von-neumanns-debiasing-algorithm.
Best case it'll take 64 ms to generate 32 random bits, average time will be ~128 ms, worst case it could never terminate. I'll deal with that if it ever happens. 🙂

[woven bluff]

any tips for rolling your own battery-powered CC/CV bench DC supply?

I find DAC injection to SMPS chip FB node somewhat unstable

[supple pollen]

You're thinking computer control?

[dry pelican]

There are ICs from Monolithic Power and Texas Instruments with I2C control

Pretty powerful ones too

But a method like DAC injection is more versatile

Maybe use a mosfet or BJT in its linear region as the bottom feedback resistor

Would that even work?

[woven bluff]

digital control by MCU

[woven bluff]

burned a DAC (and the MCU with it), a SEPIC regulator, and a current sensor trying to do that..

[woven bluff]

IMU is great for that

[dry pelican]

Nah reverse bias a diode and use the junction noise to generate random numbers

[cursive sentinel]

The trick is not making it too temperature dependent.

[inland jungle]

if I try the ADC LSB trick, the Von Neumann algo will hang unless I add a wire (or touch it) to inject a little bit of pickup noise.

QTPY rp2040

[supple pollen]

Yeah, a little antenna will get you some signal to work with.

[inland jungle]

Lampert circuit is a nice little RNG https://eprint.iacr.org/2016/884.pdf

[inland jungle]

let's say I had a stream of random 1's and 0's coming out of a circuit, how would I chop that up and turn it into a UART byte stream?

other than using an MCU 🙂

[latent jungle]

Look at the packet/protcol for UART and build a circuit with discrete logic to frame it correctly. Or use a very cheap microcontroller that has that hardware built-in already.

[inland jungle]

well, yeah, just wondering if anyone has a ref to that sort of circuit, I googled a bit and came up empty

I would have thought that there might be a standalone IC to do it, but can't seem to find that either

essentially a serial bit stream to UART packager

[latent jungle]

If you found one still being made, it is going to cost significantly more than an ATTiny.

[inland jungle]

yeah, lots of solutions generally boil down to that 😛

[latent jungle]

Actually. I can't think of why such a chip would exist. Whatever is generating the data would need to do it in word-length chunks. So that circuit would know when to generate the start/stop bits, so the logic to add those bits wouldn't take much more.

oh, duh, something like the 8250 UART chip or 16550 with built-in buffer. They used to be used in PCs 30 years ago.

https://www.ti.com/interface/uarts/overview.html

[inland jungle]

AFAICT, that does parallel -> serial UART, so it could work if I added a serial->parallel shift register to my bit stream

an ATTiny with built-in UART seems like the way to go

[supple pollen]

That's probably the easiest way in the modern era. While you can get separate shift register and UART chips, they'd need some support circuitry. With modern MCUs being "cheap as chips", it's not unreasonable to use one as a dedicated UARTish thing.

[spice zenith]

Well worth knowing. And easily solved, I'll just hang a few inches of wire on the pin, that should do the trick. 🙂
@dry pelican Will a reverse biased diode work when I only have a 5v power supply: will that be enough to get noise from the diode? I have a memory that you need to get it somewhere near its breakdown voltage,

[supple pollen]

Most diodes are damaged if you exceed their reverse voltage, so you want an avalanche/zener diode. There are low voltage ones available. Another option is a forward-biased LED. Some posts in an audiophile forum describe testing low voltage zeners vs LEDs for noise, and shows some IR LEDs (IR204) driven at low current (1mA) produce about 3.7µV of noise. Red/orange/yellow/green were low noise, but blue ones (EL204UBD) produced about 4.6µV at 1mA. Low voltage zeners (Temic BZX55/C2V7 2.7V) produced about 1.1µV at 1mA, higher voltage zeners produced more, but wouldn't work from a 5V supply. Oddly, the post didn't specify the bandwidth, which is odd since they were concerned about audible noise.

[dry pelican]

I've also seen circuits that use a reverse-biased BJT to generate noise like this: https://synthnerd.wordpress.com/2020/03/09/synth-diy-a-white-noise-generator/

Or https://www.analog.com/en/analog-dialogue/raqs/raq-issue-154.html

[woven bluff]

this is when I really want a auto router

[inland jungle]

of course, there is also this product: https://www.adafruit.com/product/4351
which can generate RNG over I2C (along with a bunch of other crypto stuff)

[supple pollen]

It's awkward when one end of a pair of connectors has the connections parallel, but the other end has them crossed over, so you can't make it easier by just turning a connector around!

[elder peak]

Yeah, it makes a person flip out.

[scarlet jay]

I need somebody who Can take a healthcare hardware made out of 3 boards and put it into 1 😉

[dry pelican]

What are you trying to do?

[distant raven]

It’s been relayed before that making a medical device is not appropriate for this server. You have also been advised that you need to hire a professional because of the liability related to medical devices.

[scarlet jay]

I get it

[rigid plume]

Hi folks I was just wondering what your opinion was on this ESP32-S3 based dev board I'm making, is it acceptable to put the DI GPIOin of an RGB LED (like neopixel) to the header?

[distant raven]

seems a bit different but I don't see a reason why it wouldn't work.

[rigid plume]

Sure just thought there could be an issue with external connections etc

I've tried it here with my prototype, and when you connect it to GND it stays on as white, and to turn it off you need to shortcircuit it (it has internal ic)

[fervent lance]

Fun fact: many hospitals and the kind, often try to hire random technicians to repair machines with insane trade policies to try to cut down on cost

[latent jungle]

if the pin is floating while powered, it is probably picking up noise that the shift register interprets as a command

[distant raven]

there's a notable difference in laws for repair vs developing untested devices.

[dull perch]

In one of my current projects, I've got a Huzzah 32 controlling a 3D printed car chassis (one 6 V TT DC motor and one MG90s servo) via BT. To get 5 V, I'm using the TPS61023 and I'm driving the DC motor with an L298D. Programmed in Arduino. Seems to work fine until a point where moving the servo crashes the Huzzah 32. I've got a switch on the EN pin and cycling that does not reset the Huzzah; instead I have to remove the battery. I suspect there comes a point when there is too much current drawn from the TPS61023, but I'd like to confirm, since the servo does not appear to be stalling and it can happen whether or not the DC motor is operating. Any thoughts on debugging?

[dull perch]

Hmm, I might have seen the problem: my servo limits are set a bit wide, and the arm hits part of the car chassis, causing the stall.

[spice zenith]

@dry pelican @supple pollen and others. Yay for random noise! Just reading the ADC once a millisecond, and then taking the LS bit is enough. I got the Trinket M0 in the mail today, and with nothing attached to Pin 1 , I'm getting a useful set of bits. 4 runs at it produced the following:

Bitcount: 32
Iterations: 52
Bits: 68cbb5b8

Bitcount: 32
Iterations: 55
Bits: bea778eb

Bitcount: 32
Iterations: 65
Bits: f7d5f577

Bitcount: 32
Iterations: 59
Bits: fb82f24c

To explain: Bitcount is the actual number of bits extracted from the Von Neumann debiaser, Iterations is the number of bit pairs used to get those bits, and Bits is the hex value produced.
The trick is that I've set it to cap at 1000 iterations, so in the event it keeps on getting pairs of identical bits, it'll eventually fail out with less than 32 bits extracted. And I don't care in the least, that'll happen so infrequently it just won't be an issue.

[scarlet jay]

i need to move 0-30V to -30-0V on a scale

how to?

[unreal flax]

Possibly an op-amp in an inverting configuration, but I don't know of +/-30V chips off the top of my head. It might be easier to see if you can alter the signal earlier in its path, like inverting it when it's lower voltage.

[scarlet jay]

its a DAC output, it is already amplified

I need it to be inverted fully into -

or I can output 0-60V and turn them lower to -30+30 V, would that be possible or would dual power supply approach win?

So, to put it easy, lowering 60V to +-30V

Or convert 1 power supply to -30-0V and the other to 0-30V

[spice turtle]

Vote is for Op-amp as well
https://www.eeeguide.com/summing-scaling-and-averaging-amplifier/

https://redox-college.s3.ap-south-1.amazonaws.com/kmc/2020/Apr/08/xmSI5vY3bL46jo5bLAhF.pdf

[scarlet jay]

oh okay, thank you - that sounds realistically buildable 😄

[dry pelican]

That can be done with a switched capacitor converter IC like the 7660 or a dual rail supply converter (TI makes some)

Another solution might be h-bridge

Motor drivers are often available that can handle those voltages

Just supply it with your op amp supply that goes to 30v

[broken zenith]

If someone could help me review a PCB in #327254708534116353, I'd be really grateful 🙂 It's not too complicated, should take about 15 minutes.

I can also drop it in this channel as a PDF, but I find voice chat a little easier

Here's the schematic. I know that J3 is kind of funky, please ignore that 😉

[scarlet jay]

say the function

[broken zenith]

It's an RP2040-based game controller

[latent jungle]

@broken zenith fwiw, I have never found the resistor necessary on the crystal. It always seems to prevent the clock from running (for me, with multiple crystals.) But, I always put the pads and load it with a 0 ohm resistor.

And you’ll probably regret not having the RUN / RESET signal connected to a button at some point.

Especially if you plan to program via the USB bootloader

[vast flume]

Hey guys quick question. I am creating a circuit with a LSM9DS1 imu, but I cannot for the life of me figure out what the DEN_A/G pin is for

The datasheet says its the data enable pin but does that mean that I need to pull it high? I cant find any references to it in the datasheet lol

[dry pelican]

I guess. A 0 ohm jumper or solder bridge could be used for testing

[latent jungle]

searching for LSM9DS1 DEN_A, I found this thread: https://community.st.com/t5/mems-sensors/what-does-the-den-a-g-pin-on-the-lsm9ds1-actually-do/m-p/429602

so, you’re not alone in wondering. (and I’m not sure the response from the ST person answers the question)

[vast flume]

ST back at it again

Their response was terrible lol it makes zero sense

I shouldn't need a PHD to decypher the use of a pin 🤣

[latent jungle]

(yeah, sorry. I posted the link and then read it. I love that the next post is someone pointing out those registers don’t seem to exist.)

[vast flume]

🤣

I have no idea what to do 🤣. Should I just leave those pins unconnected?

[latent jungle]

at a minimum, break it out to a test pad so you can bodge it. Or, you can do like @dry pelican suggested. Or, you can combine that with a solder jumper (after the resistor). That way you have a pad and can pull it high or low.

In a case like this, I would do a pull-up to VDD with a ground pad nearby if I need to pull it down instead.

(or make it easier to leave it floating)

[vast flume]

Ah yeah that makes sense

Thx

[latent jungle]

Or, look at other boards to see what they did 😉

[vast flume]

Ended up looking pretty cool

[distant raven]

Controlling the crystal drive strength tends to be pretty important. I’ve never had issues as long as I kept ESR + series resistor at ~1050ohm

[latent jungle]

Yes, I agree. But something about the RP2040s amplifier/oscillator works different from other microcontrollers.

I’ve used 4 different crystals with the RP2040 and they all failed to oscillate with a resistor in series.

[distant raven]

What was your total ESR?

[latent jungle]

I’d have to go back and look. They all had different ESRs.

[distant raven]

I use a crystal with 18pF of load + 100ohm ESR

Combined with a 953ohm resistor in series, it works very reliably

[latent jungle]

It’s why I say to keep the pads for a resistor. There’s some combination that needs it. I just haven’t found it yet.

[distant raven]

Would you like the p# for the crystal I use?

https://www.digikey.com/en/products/detail/ECS-120-18-33-AGN-TR/9648931

[latent jungle]

The first one I used was the pn from the pico. It wouldn’t oscillate until I shorted the R.

[worthy knot]

I'm looking for a differential pressure sensor like the Sensirion sensor (SDP 810) but with micropython support. I want to measure the difference between inner und outer pressure. Or is it better to take two BME280 and use code for the difference?

[distant raven]

It’s an I2C device so writing a driver for it would be fairly straightforward

I’m not sure you’d get a cheaper solution using two bmp280

I’d just get the sensor and write a driver for it

[supple pollen]

You don't even have to write the driver from scratch, you can use a driver from some other language as a template, since all I2C drivers do is essentially send I2C commands.

[distant raven]

Yeah, if you wanted a quick and dirty solution that would work. But it doesn’t hurt to write a driver and put it on GitHub for other people as well 🙂

Lots of great options really

[wicked root]

Any tips for DIY non-conductive, cost-effective conformal coatings?

Context for my question is that I'm thinking to get some more of these https://www.adafruit.com/product/4701 for my makerspace to provide manuals and documentation for the equipment available (nearly full machine shop, woodworking shop, and electronics bench - it's awesome and terrifying to have un-restricted access to all of it). So, they need to have at least a minor amount of resilience to metal and organic particles. The Stemma/JST ports and pins, I'm less worried about and can likely print some TPU plugs for. The SMD electronics, I want to make sure are protected from shorts.

[left grove]

I know that sounds kinda stupid, but if you want to conformal coat something, I would just get conformal coating. On Amazon I can get a 200ml spray bottle for 10€. Though I don't really like spraying such stuff, you can spray it in a container and then apply it with a brush or something.
Personally I use mg chemicals acrylic conformal coating which comes in a little bottle and is a bit more expensive . (I got it because it has a bit higher temperature tolerance compared to the 10€/200ml one.) MG chemicals silicone conformal coating is a bit better from what I've heard.
Definitely apply it and let it dry in a very well ventilated place.

I've seen people use nail polish but I doubt that's cheaper

[wicked root]

Thanks!

[supple pollen]

Yes, but my point was that using an existing driver as a starting point makes that task easier.

[distant raven]

I wasn’t disagreeing that it would be easier 🙂

I absolutely agree

I really need to work on not using "but" so often. it can make a sentence read wrong

[low anchor]

https://www.adafruit.com/product/1766 would this be sensitive enough to detect a dryers ordinary vibration

[hushed smelt]

I would say yes but it's a switch not a sensor. Also might be better to get a temperature sensor and slap that on the side as dryers can get pretty hot.

Plus with a properly leveled dryer it shouldn't vibrate that much. Like the goal is to get it not to vibrate and walk around.

[unique patio]

could also look at the data from an accelerometer. You'd want to do some experiments to see what the data looks like with dryer on and off, and see whether there are characteristics when on that are easy to spot.Maybe attach the accelerometer as a flap on the side so it would swing around more than if it were firmly attached.

[supple pollen]

I suspect that vibration sensor might work. Or something like a pendulum (like a tilt switch in a pinball machine) or a force sensor with a weight. An accelerometer seems like a lot of complexity, but would likely work.

[faint rain]

I've wondered about just using a microphone. My dryer at least makes noticeable noise

[opal rune]

If it's extra shaky, I'd probably take advantage of a breadboard's flimsy connections, hook up an analog pin, and sample it.

[lime herald]

I’ve used “MG Chemicals 422B Silicone Modified Conformal Coating” on a couple of projects and it’s worked well. I was mostly going for moisture resistance. I got it through Amazon but it doesn’t seem to be currently available there.

I have a similar project for our hackerspace, though we’re mostly providing a control interface to Home Assistant through the tags. Same ones as you’re using. Happy to share it once it’s working and up on GitHub. Still designing the case for it.

[latent jungle]

I've never used conformal coating, so this might be a dumb question. But how do you apply to a board? Do you pour it on? Or dip the board in it?

[lime herald]

That’s a great question! I used a brush to apply a light layer of it. The bottle of it I bought included a tiny brush 🙂

Beware, you’d want to avoid getting it in the openings that some chips have (like air sensors or I2S microphones or similar devices).

[slow plover]

There are different types also

I use a spray-on type at work

[latent jungle]

Ahhhh okay. Now I remember seeing MG bottles with a brush in them. So that makes sense. I guess that's more difficult to get an even coat but easier to apply especially if you have stuff that shouldn't be covered.

[lime herald]

For what I use it for I'm not too worried about an even coat but I'm also just trying to improve the durability of some sensors around home and the makerspace I'm a member of, I'm not using it on professional products.

[latent jungle]

Makes sense. It's probably better than just covering the board with hot glue!

[lime herald]

I've been tempted to do that too 😉

[worldly schooner]

With sprays, definitely mask off areas you don't want covered with some masking or painter's tape. With a brush, you generally worry less as you have a lot more control over where it's applied.

The brushed conformal coat from MG is viscous enough to apply fairly precisely to specific target areas, but fluid enough to even itself out as it cures to leave a smooth, glossy finish.

[elder peak]

I guess we've got some real non-conformists here.

[inland jungle]

if I want to do CAN over Cat6, am I better off matching the cable impedance with 100 Ohm termination, or sticking with the CAN standard 120?

[latent jungle]

CAN or CAN-FD?

[inland jungle]

preferably CAN-FD

[latent jungle]

It seems like tolerances would cover the difference. But I’d pay close attention to the edge rate differences of Ethernet and CAN-FD in that case,

[dark grail]

i think 10mbit ethernet uses 20mhz as the edge rate (plain old manchester)
i think 100mbit is the same, 200mhz edge rate

gigabit is then just goes into RF territory, QAM on all 4 pairs, with rx and tx on the same pair at the same time

[latent jungle]

Well, edge rates are defined in terms of rise time, not millihertz (or megahertz)

[dark grail]

ah, i was thinking edges/sec, not risetime

[latent jungle]

Yeah, that’s fundamental frequency/effective data rate. Which is almost always orders of magnitude slower than the edge speed. which, in a cabled system, is (almost always) band-limited by the cables.

ie, a cat6 cable probably has plenty of bandwidth for a CAN-FD signal’s data rate. But, I have no idea what a typical CAN-FD rise time is… (although, I suspect it is commiserate with its data rate, but still worth checking before assuming it will work.)

[hushed smelt]

working on a birthday project box for my sister, open the box it plays a sound. found this that seems like it'll work as a mp3 player instead of a short sound clip

https://www.amazon.com/dp/B092WTSLKK

plan is to hack it apart, slim it down, and put it in the lid behind some backing material

while taking it apart i found this

an antenna that is not advertised in their product description

hooked up to the top left third pin of this chip.

any ideas if this is benign or if i should cut that trace just to be safe?

maybe they were experimenting with trying to get it working with wifi or BLE. I haven't even plugged it in yet. Requires plugging into a PC, downloading music to the onboard SDcard. Pretty sure cutting the trace won't affect the USB or SD card capability.

Just a really weird thing to pop open this device and find an antenna in there plain as day.

[unreal flax]

This appears to be the chip manufacturer from the logo: https://www.zh-jieli.com

[hushed smelt]

seems they specialize in bluetooth chips, but there's no advertising of bluetooth capability with that product

I've seen that logo before in something else but I can't remember what.

[hushed smelt]

Now I remember, it was the same chip I used in my bluetooth sleeping mask on a DW-CT14+ bluetooth classic amplifier board.

Was about 4 years ago https://forums.adafruit.com/viewtopic.php?t=148565

I'm positive there was some kind of CVE vulnerability or notice not to use them. It's why I stopped using them, can't find it now.

[dry palm]

Is it normal for all pins on a pi pico to be partially shorted to ground? Not enough for a continuity test on a multimeter to beep, but enough to register something happening

[distant raven]

There’s generally a diode

You can measure the diode path to ground

[dark grail]

chips tends to have a diode from all inputs to the vcc rail
if the chip isnt powered, then the voltage from the "ohm meter" will be trying to power the chip thru that diode, and it will let some current flow, which the "ohm meter" then mis-interprets as a resistance

[distant raven]

I general test connection on solder joints for the rp2040 by seeing if I can measure the diode between the pin and ground

[dry palm]

Thanks!

[slow plover]

are the load cells that Adafruit sells able to measure in both directions?

[supple pollen]

The load cells themselves can, as they're a balanced bridge configuration, but I don't know which amplifiers do

[vital elm]

tl;dr: If I power Adafruit's BNO055 VIN with 3.3V, will the I2C pins output 3.3V or 5V?

From Adafruit's resources on the board (https://learn.adafruit.com/adafruit-bno055-absolute-orientation-sensor/pinouts):

• The VIN of the board is 3.3-5.0V
• The I2C pins "can be used with 3V or 5V logic" - So It's 5V-safe and will read 3V fine.

From the chip's datasheet (https://cdn-shop.adafruit.com/datasheets/BST_BNO055_DS000_12.pdf), the chip itself cannot handle more than 3.6V. So Adafruit's board must regulate the voltage, which is expected.

It specifies that it will accept 3V or 5V, but I want to connect the board to a Teensy 4.1 whose digital pins are not 5V-safe, so I want to be confident that Adafruit's BNO055 will not output 5V on its I2C line if I power it with 3.3V. Is anyone able to point me to some sort of documentation that clarifies this?

[unreal flax]

The I2C pins will typically be pulled up to the power supply of the board. If you don't provide it with any 5V, they can't produce that voltage on their own.

[vital elm]

If that guarantee is true of the BNO055 that's comforting. I'm being cautious because I ran into a case recently where a board only took a 3.3V input, but outputted 5V. So I wanted to make sure this board didn't also do that.

The other board was a 3.3v/5v level shifter, but its only input power was 3.3v, and it outputted 5v power and 5v digital logic. It was specifically designed for that but I didn't realize when I used it that it actually increased the voltage. I had assumed I'd have to supply the 5v

[unreal flax]

That's pretty atypical, yeah. It would need to have incorporated a boost regulator or charge pump to do that.

[vital elm]

Thanks 🙂 I was pretty shocked when it powered my 5V device that I hadn't turned on yet. But it made sense after looking into it. It was the Rev Shifter, which was specifically designed to connect a 3.3v system (The Rev Expansion/Control Hub) to a 5v quadrature encoder.

[latent jungle]

Thanks 🙂 I was pretty shocked when it powered my 5V device that I hadn't turned on yet
Just because a "5V" device powered on, it does not mean the board was outputting 5 volts ...

[rigid plume]

Have I done this correct?

switching circuit from vbus (5v usb), and battery (3.7v lipo), diode to stop reverse current back to usb, trying to get 5v to be priority

[regal lodge]

is a 0.16mm trace width gonna be an issue? :3

(for data)

[latent jungle]

Seems like it works. FWIW, Falstad circuit: https://tinyurl.com/29fj24ru

[vital elm]

I'm aware. I measured it with a multimeter to see what it was, and it was 5V. Then I looked up documentation, which confirmed that it was intended to be powered by 3v3 only.

[molten aspen]

Simple but confirmatory question:
https://www.tme.eu/Document/35c9ffb7cd05261537b310e0613f3b53/FDC6401N.pdf
Is the Power dissipation wattage the wattage that the chip can handle or is it the classical: Maximum Drain-Source Voltage * Drain Current? In this case, as the Drain-Source Voltage is 20V and maximum drain current is 3.0A in continuous, 12A in pulsed...

Or is the power dissipation as the name implies: The amount it is able to dissipate heat? All I'm using it for now, is to have a software controlled ON-OFF switch.

[inland jungle]

You're not dropping 20V when you're conducting 3A

[molten aspen]

.... Good point.

[inland jungle]

There should be an rdson value

[molten aspen]

RDSON 70mOhm @ VGs = 4.5V
RDSON 95mOhm @ VGS = 2.5V

Oooooooh

[inland jungle]

What's your gate voltage?

[molten aspen]

It's not 20V, it bit less, thanks to that resistance of RDSON that is dependant on the gate voltage that I apply, right?

[molten aspen]

3.3V.

So I'd say that the RDSON is then somewhere 80mOhm, since it's between those two values.

[inland jungle]

Yeah, when it's on you're dissipating I *I *rdson

In watts

[molten aspen]

For watts, am I then using the voltage of the gate or voltage of the drain/source?

[molten aspen]

@inland jungle Regarding your phrase You're not dropping 20V when you're conducting 3A, is this line then more correct or am I missing the mark nice and wide? 😄

[inland jungle]

The voltage across drain source will be I*rdson when you are conducting, so really small

That 20V is what it holds off when not conducting

[molten aspen]

Ooooh, right, now I'm getting you!

So, in the sense of my application, that mosfet is still valid option for the usecase where I would like to have a switch controlled by GPIO that turns it either on or off, no PWM whatsoever? The current draw is maximum of 0.5A (drain current is 3A in the schematics).

(and yes, there are diodes, resistors and pull-ups done to the circuitry to handle the protection and state during boots)

[inland jungle]

Should be fine. If you're switching it fast with Pwm there are other things to worry about

For a slow switch it should work fine

[molten aspen]

Yep, it's pretty much very slow Every 2 weeks.

But great, now I understand what's the meaning of the RDSON. Thank you @inland jungle!

(I was bit worried cause I already got the PCB's from JLCPCB and spotted a magical 0.96W in the label of the MOSFET bag, immediately freaking out did I choose a totally wrong MOSFET for the task).

[inland jungle]

Well if you're running 3A with a low ish gate voltage, that might be an issue

[molten aspen]

Not going to be the case, the circuit is limitting the current to 0.8A

[inland jungle]

Yeah, so that's around 20mW

For.. 5

[molten aspen]

And less for 3.3V which I'm using so

And that is pretty much very bad scenario 😛 Having multiple fans in single output, while the system has 4 individual ones so

It's not a scenario that will occur that often.

[distant raven]

Sort of a one time advertisement:

If you’re US based and need help with prototyping a project, I can help. I’ve helped many small makers with design and small batch assembly for small projects. I should note that I do not develop devices for medical, high voltage, or defense. I have a web page to get started: https://www.oakdev.tech/prototyping.html

[tame kestrel]

Would it be possible to estimate how many of NeoKey 1x4 (https://learn.adafruit.com/neokey-1x4-qt-i2c) could be connected to a single microcontroller at a time.
I know the I2C limit is 16, but I am more concerned about the NeoPixels and their power consumption.

I may be wrong, but I image it would not be possible to connect 64 (16 x 4 per unit) NeoPixels on full brightness on such thin cable.

[distant raven]

It’s not recommended to drive more than 10-15 by the 5V pin of your microcontroller. Each color channel can pul up to 20mA each for 5050 sized LEDs, closer to 12-15mA for smaller varieties. So you can anticipate pulling 45-60mA per. Computers generally deliver 500-900mA on a USB port depending on if it’s usb 2.0 or 3.0. IT’s recommended to have a separate supply of you want to power any more than 10-15

[tame kestrel]

Thank you so much. That is probably why Adafruit's MacroPad has 12 NeoKey slots.

[woven bluff]

just use a load switch, and a electrolytic tank behind it.

[worldly schooner]

Most keyboards, especially the larger ones, do not run these LEDs at max brightness. Capping out at 10-20% can still offer a nice glow without overloading your circuitry.

[dark grail]

  Configuration Descriptor:
    MaxPower              260mA

the config descriptor includes a max-power rating

[Mon Aug  7 21:22:50 2023] usb 1-11.1.2: rejected 1 configuration due to insufficient available bus power
[Mon Aug  7 21:22:50 2023] usb 1-11.1.2: no configuration chosen from 1 choice

and if i plug a usb stick into a keybord, linux can reject it due to lack of power

i can imagine how you might have 2 nearly identical config descriptors, differing only in max-power

and the keyboard could disable all its RGB bling when the host selects the low-power config

but, ive never even seen a device with 2 configs in the wild