help-with-hw-design

[supple pollen]

A web search for "eagle jst-sh" brings up some possibilities

[limpid nest]

hmm I was including adafruit ha, classic misstep

Seems like something that should be corrected in the official adafruit.lbr, given how much they use the 4 pin SH connector

[supple pollen]

I use the AdaFruit, SparkFun, and MicroBuilder libraries when I can, but otherwise I roll my own (bodger.lbr) or start looking for other possibilities

[limpid nest]

I couldn't find microbuilder online except from weird sources

[supple pollen]

You could, of course, grab the Eagle files for a board with it and extract the part definition from there.

[limpid nest]

I tried that

Isn't working in Fusion

[supple pollen]

I haven't tried in Fusion, I'm stuck on old Eagle 7.7 since they stopped updating the licensed version.

[limpid nest]

Yeah

Fusion is....idk, fine? I don't like it but I also don't despise it

[supple pollen]

I guess you could grab an old Eagle version, use it to extract the library, then import that, but it's likely more trouble than it's worth.

[limpid nest]

Yeah

Easier to make my own

The really annoying thing about Fusion? Every. Single. Google. Search about it ends up with results for the 3D modeling side.

[supple pollen]

The first four hits I got were all for Eagle

[limpid nest]

If you search for Fusion specifically

You can add a bunch of search terms to get it to work but it's annoying

[supple pollen]

Ah, I (intentionally) didn't search for fusion.

[fervent lance]

my ex-uni pcb business use altium and you have to export it using a specific dialogue with 7 windows for them...

so basically 5k$ for one pcb before you even make one...

[cosmic plaza]

Hi folks, can I get some design review feedback on this:
https://oshwlab.com/logan.p.herrera/tuning-fork-audio-amplifier_copy

Next step is JLCPCB fab + assembly

one style question: is there a convention for signals that leave the sheet in a flat design like this?

[cosmic plaza]

Also I built and characterized a 4th order filter (linked is 8th) to validate the TI calculator. Turns out it really works! Though with more peaking than expected. I think it's to do with the extra DC blocking cap + filter input impedance

[heavy jasper]

I’ll try to take a look tomorrow evening!

[cosmic plaza]

Thanks @heavy jasper !

[pure gorge]

Im trying to use an MCP23016 I2C expander in a schematic, but I cant tell if I need to use the INT/interrupt pin. In the past I've had to solder I2C connections with an INT pin but is it safe to leave it floating?

[worldly schooner]

The INT pin is an output from the MCP230xx to your MCU. It should be fine to leave it floating.

[pure gorge]

Ok, thanks

Hopefully I dont fry anything and waste a board purchase, this is my first board

[distant raven]

do you think it's permissible to use a 100nF ceramic in place of a 10nF ceramic if it's just used between a reset line and gnd?

i know it'll affect timing of the reset line but i'm not sure if it's enough to make a huge difference where it would negatively affect operation

[heavy jasper]

If there’s a switch on that reset line and you don’t have some series current limiting resistor going into the switch, make sure the added capacitance isn’t more likely to weld your switch contacts over time.

(In general the series current-limiting resistor helps prevent that regardless of cap size)

[distant raven]

it has a 10k pull up if that helps clarify

[heavy jasper]

Not so much worried about the 10K pull-up, but say you have a huge cap charged to 3.3V, then you short that cap to ground via a reset switch. The cap will have a spike of current out into the switch that I’ve (rarely) seen fuse tiny switches before.

If you don’t have a manual reset switch (or anything else pulling the line low) and just have the 10K pull-up, then it seems pretty reasonable.

[distant raven]

It’s this board in question

Reset button on the line for manual reset

Reset traverses between two IC

[bright thistle]

Hello!

[distant raven]

Hi

[bright thistle]

I have a bunch of buttons individually connected to gpio, what alternatives would you suggest in order to reduce the number of wires going to the Pi (I2C, USB, SPI...)? :)

I want to try designing a suitable PCB, but don't really know which method I should use 😅

[limpid nest]

There are a bunch of SPI and I2C GPIO expanders. But you'll have to have the traces for the buttons somewhere

[bright thistle]

I think I could use the I/Os from my boards where the buttons are collected, then :)

[fervent lance]

I've used 3-to-8 decoders in small projects.

[alpine slate]

Hey guys! How "hard" is it to learn PCB design? I need a PCB with really only a place to solder a rpi PICO on and a bunch of LEDs connected to itl

[limpid nest]

It depends on your goal and needs. That doesn't sound terribly tough

Check out kiCad for a free software that's basically enterprise grade

[alpine slate]

Really just a bunch of LEDS that are placed in a specfic manner

[alpine slate]

thanks!

[limpid nest]

How are you sourcing and sinking the current?

[alpine slate]

ngl i have no idea how all this works im pretty new to this

[limpid nest]

Hmm OK, just thought I'd mention that it's not really usually feasible to power "a bunch" of LEDs from a microcontroller directly. A small number is possible, but also not recommended.

[alpine slate]

im talking about these small leds btw

[limpid nest]

Yeah, it's still not usually best done. There are ways to do it though, LED driver chips are not hard to find

[tough matrix]

typically, you can use 1 pin to source or sink power to 1, maybe 2 LEDs
if you need to power 10 LEDs, you do need an led driver, or maybe just a mosfet

[limpid nest]

Yeah LED driver is the "right" way to do it, but you can get away with a Mosfet without much trouble if you just need to blink N LEDs and have them be all on or off at the same time.

[modest grail]

i have a board i made a mistake on and im trying to rerout the trace. what do i use to seal it and bond it to the board?nail polish? kapton tape?

[distant raven]

You could. Make sure to trim the bad area

[modest grail]

I already cut the old traces and verified they wer cut with a multi

[distant raven]

Kapton tape should be sufficient then

[modest grail]

How about for a more perminate fix

[limpid nest]

How sturdy and/or long term are screw terminals?

[worldly schooner]

With ferrules? Very.

Industrial installations within a proper cabinet stay functional for decades if done right.

[limpid nest]

That's great

Thanks

[steel valve]

What solder paste should I use for hotplate soldering? I am soldering small components onto a 15x30mm pcb

[distant raven]

What kind of application is the board?

[steel valve]

The board intends to be a breakout for a sparkfun artemis bluetooth module

[distant raven]

Okay, you’re probably okay using low temp

[steel valve]

And has a charging circuit on it

(Runs at 3.3v)

[distant raven]

It will be like Bismuth, Tin, and Copper usually

[steel valve]

Where would I buy that paste at

[distant raven]

I’ve bought mine off Amazon

But Digi-Key or mouser would have it

[steel valve]

What's the name of the paste you use

[distant raven]

You can buy a syringe for like $8

I buy ChipQuik

[steel valve]

Do they have paste jar versions?

I am going to use a stencil

w/ a squeegee to put the paste on the board

[distant raven]

I forgot about inflation.

Yeah they sell jars

[steel valve]

https://www.amazon.com/Solder-Paste-Sn63-Pb37-clean/dp/B07BH5LP5G/ref=sr_1_6?crid=1XGJI6WJHMKEI&keywords=solder+paste&qid=1651287378&sprefix=solder+past%2Caps%2C158&sr=8-6

Is this it?

[distant raven]

Chip Quik NC191LT50 Smooth Flow Low Temp Solder Paste Sn42/Bi57.6/Ag0.4 T4 50g Jar https://www.amazon.com/dp/B08KWLLFT4/ref=cm_sw_r_cp_api_i_C2VZWYP18QBFQKA65HXJ

I recommend lead free if you’re not sending things to space

[steel valve]

What's using lead free solder do differently

[distant raven]

Well, it doesn’t have lead in it which is better in general

Low temp melts even lower than leaded solder

The only real reason to use leaded solder anymore is in space application

[steel valve]

Ok just making sure cause the surface finish of my PCB is HASL (With Lead) from JLCPCB

[distant raven]

Oh gotcha

[steel valve]

Since at that point there would already be lead involved

[distant raven]

Yeah, if there’s already lead it wouldn’t hurt to use lead

[steel valve]

Ok so https://www.amazon.com/Solder-Paste-Sn63-Pb37-clean/dp/B07BH5LP5G/ref=sr_1_6?crid=1XGJI6WJHMKEI&keywords=solder+paste&qid=1651287378&sprefix=solder+past%2Caps%2C158&sr=8-6 would work then

Cool

[distant raven]

Yeah

[steel valve]

What are other useful tools for hotplate soldering

Aside from having a squeegee

[distant raven]

Hot air gun to help with heating the top

[steel valve]

Nice I got one of those too, cool

Oh yeah how do I hold a stencil in place?

[distant raven]

Masking tape and usually spare PCBs

[alpine slate]

Okay I just checked and i need round 50-60 SMD Leds or maybe even more. Is this achievable? Each need to be accessed individually

[limpid nest]

Just FYi, that's a good amount of power

Off the top of my head, you could probably do it with the AW9523 chip. But actually getting those chips and then soldering them isn't easy

With multiple of that chip

Each one can do 16 LEDs

You can have up to 4 before you need a multiplexer

So 64 LEDs

[alpine slate]

okay maybe ill take my 2nd config then lol

[limpid nest]

You CAN get the chips

It's just not easy and the process of soldering them is harder

But it would be the most compact way to do it.

They are tiny tiny

That's my pinky finger

[alpine slate]

oh yeah thats tiny

[limpid nest]

They are lovely chips

Consider them

[alpine slate]

but i also wouldnt be able to power them by usb would i lol

[limpid nest]

You'd struggle to power 60 LEDs over USB except at very low brightness

Over computer USB that is

[alpine slate]

they wont be all on at the same time btw

[limpid nest]

Hm well that would change things

[alpine slate]

how many of them would you expect to work at the same time via computer usb?

[limpid nest]

computer USB is about 500 mA

which MCU do you want to use?

[alpine slate]

P2040

RP*

[limpid nest]

Are you using a particular devboard?

[alpine slate]

pi pico

[limpid nest]

ok, so that takes ~85-93mA, we'll go with 100 mA to be safe. So you have about 400 mA for LEDs and LED drivers

[alpine slate]

okay thank you!

[fallen pulsar]

has anyone tried to use crystal-less usb using an stm32 and stm32cube? I can't for the life of me figure out how configure it properly

im using a stm32L433rb

[upper moss]

Hi all! On an MCU board where there is enough room, what is the preferred STEMMA I2C connector? Regular? Or is everything moving to STEMMA QT? As I read the docs, it says that QT is being used to keep sensor boards small. But I also heard that everything is moving to QT....

[distant raven]

most of all the newer Adafruit boards are moving towards STEMMA QT

[upper moss]

thanks for some feedback! I was thinking of going that way but appreciate any input!

[limpid nest]

The biggest downside i see of stemma QT is you can't pump as much power down the lines.

But it's not meant to be a power transfer method

[elder peak]

Yah, often times you don't necessarily want to even use the same power rail either.

Like, either it's a sensor with low power requirements or it's something like my LED driver boards where you might be driving it with a completely separate power supply.

[supple pollen]

It might be easier to use a multiplexed chip (or charlieplexed for even more LEDs, but more confusing wiring)

[limpid nest]

Oh yeah I forgot to mention that my method gives you dimming, which is one of the big advantages alongside pcb size. There are other options

[pure gorge]

What should the L1 inductor be on the first page if im using a single cell lithium battery? https://www.ti.com/lit/ds/symlink/tps61322.pdf?HQS=dis-dk-null-digikeymode-dsf-pf-null-wwe&ts=1651376981641

[unreal flax]

Have a look at section: 9.2.1.2.3 Inductor Selection.

[pure gorge]

Oh, I didnt know there was a section about it. Thanks

[warped atlas]

VFD Question for y’all: I intend to make a project that uses 8 of these tubes

Each one has a filament, which must be driven with 2.2V AC (in the kilohertz, exact freq doesn’t really matter)

I intend to achieve this w/ a 555 + transformer

would I have to power EACH filament with its own mini transformer circuit, or could I just wire them all in parallel?

[inland jungle]

how much current do you need to drive them

[supple pollen]

Parallel should be feasible

[heavy jasper]

@cosmic plaza I'm not sure how to check stuff/no-stuff in EasyEDA, but e.g. it seems like you have R1 and R8 in parallel with both stuffed. The comparators are mostly about just having a variety of options by the looks of it, but I'm not sure that the COMPA one (that just compares INA directly) will be super useful for the phototransisor models, given that both the pullup voltage and the comparator reference voltage is 2.5V (so the phototransistor outputs will either be decidedly less than reference, or right around reference in a Voffset-sets-output kind of way).

It looks like you're using a micro-USB connector, I'll make my standard pitch for a type-C version instead (especially if you're already getting this board professionally assembled or are willing to do hot-air work, given your choice of BGA microphone)

For the type-C version, if LCSC sells the power-only type-C connector that could work and be easier to solder

22uF is also kind of a bunch of capacitance directly on a USB output - it's probably okay - especially if you're just plugging into a power brick instead of a laptop - but if you had a laptop that was particularly stringent about inrush / current-limiting specs (coughmacbookcough) then you might run into some problems.

[cosmic plaza]

Thanks @heavy jasper

all components are stuffed

the R1 and R8 combo is because the microphone output impedance is very nearly the required resistance for the first stage filter

so it gives some flexibility in prototyping microphone present (remove 270 ohm) vs signal generator present (remove 0 ohm)

however the phototransistors have other impedances

and the filter response is already looking sensitive to precise input impedance 🤔

[cosmic plaza]

what level of capacitance is obviously fine? or is it "none unless active inrush limiting is present"

[heavy jasper]

I’m actually not sure - but I knew of a case where 10uF of ceramic was too much.

[cosmic plaza]

In layout, 5V and 2.5V are distributed in a power layer (Inner 2)

all mixed up and interdigitated

with 1 mm tracks

above continuous GND plane

does anything about that layout look concerning?

[cosmic plaza]

actually I think I have some CAD goggles confusion

this PCB isn't that big

[pure gorge]

What should be used to give the backlight 19v on this display? Seems kinda high for an LED https://www.orientdisplay.com/wp-content/uploads/2021/10/AFK320480A0-3.5INTM-spec.pdf

[worldly schooner]

Generally, these type of display backlights are driven by current sources. You can reference https://learn.adafruit.com/assets/24414 from Adafruit's DPI TFT Kippah (2454) for what they use for such displays.

[slow plover]

I think the RP2040 is neat, and it's cheap enough for me as a hobbyist to consider designing boards that use it directly. As a non-EE who's largely self-taught, but able to understand most of the concepts involved re: high-frequency signals, am I likely to have much success trying to adapt the reference design for a specific application?

It seems to me that most of the difficulty is likely to be from RP to flash, USB, and crystal

from what I understand, it's mostly a matter of ensuring proper impedance/timing?

[worldly schooner]

If I'm being totally honest, the hardest part is probably not even impedance/timing, as the signals from the RP2040 are still considered low-speed and the impedance should be fairly tolerant thanks to its series resistors. Right now, the two hardest parts are going to be 1) selecting and sourcing materials, as the exact parts on the reference design might not be in stock at the time of this writing, and 2) debugging your board after it comes back from fab, because even the most experienced engineer will need a couple of bodge wires and a second spin to make sure everything is perfect.
That being said, if you've never designed a board before, it may be easier to start with using a surface-mount Pi Pico (or QtPy) in place of the RP2040, as all the necessary circuitry is already there for you. That will save you the headaches of dealing with the crystal and power supply circuitry, and save you a lot of time in terms of a complete design. From there, integrating an RP2040 is much simpler, and it narrows the areas of potential failure to the passives and components directly attached to the RP2040 chip.

[distant raven]

It’s actually pretty easy all things considered

Tons of designs that you can easily reference for your own board.

But the USB peripheral is probably the least spectacular part of the RP2040 other than the load of capacitors you need for all the inputs

USB PHY1.1 is pretty slow considering plenty of micros have USB PHY 2.0

[worldly schooner]

In its defense, having a lower-speed USB peripheral makes designing custom boards far easier, especially for new PCB designers?

[distant raven]

But the Pico is a much better deal considering at hobbyist quantities, you’ll pay more than $4-5 for the components

[worldly schooner]

Honestly, most of the cost on the Pico-equivalent circuit is probably the external flash. I think in hobbyist quantities, that comes out to almost half the cost of a Pi Pico...

[distant raven]

Unless you’re buying 500-3400 RP2040, you’re still going to pay $1-$1.50 for just the RP2040, $0.50 for a cheap crystal, probably $1 for the SPI flash, $1-2 for the capacitors and resistors, and the PCB itself will be at least $2

Not Including shipping

[slow plover]

I've done some PCB layout, but nothing that actually needs any attention to timing

[worldly schooner]

I would exclude the PCB cost since the discussion is for a specific application, so it should come out about equal.

[worldly schooner]

Don't think the RP2040 needs any particular attention there, either.

The USB peripheral is slow enough that a roughly matched paired trace is rarely cause for concern, especially on smaller boards.

[slow plover]

I'm thinking about basically laying out a "module" that I could plonk onto a board and just route pins for IO

[distant raven]

They recommend 27ohm resistors on the data lines to control drive stability

[slow plover]

for cases where the physical footprint of a pico or QTpy are inconvenient

[distant raven]

I’ve done a number of modules like that

[worldly schooner]

The only way I could see either of those boards get a significant downsizing would be reducing the pin spacing from a breadboard-standard 0.1" to something like 1.27mm...

Otherwise, there are a lot of existing designs that already basically do that.

[distant raven]

Yeah, the RP2040 stamp from Solder Party does that

[slow plover]

to clarify: what I mean by "module" is a pre-routed section of PCB layout that I could integrate onto a new PCB without starting from scratch

[distant raven]

Ah, that might be more difficult

I’m not 100% sure CAD programs support that

[slow plover]

I could always just make it into a component w/ the layout as a footprint

[distant raven]

Sure

[worldly schooner]

That sounds more of a CAD thing than it is a custom design. I'm wondering if you could just import the entire Pi Pico design to do exactly that...

[slow plover]

I have been working in DipTrace, don't think there's a file for it

will investigate

[grizzled wave]

I am assembling a Geiger Counter, it involves the following subsystems: DC Voltage Regulation Circuit (5V), Oscillator Circuit, High Voltage Regulation Circuit and Output Pulse Circuit.

I have completed the assembly of the first subsystem in sequence, checked for continuity and a voltage test for the 9V battery supply as well as a regulated test of the circuit.

I am currently "done" building the second and third subsystems, but my issue is with the High Voltage Unregulated Test, the circuit is not displaying the 470-850V reading that it should be, it instead is reading for 0.02V, I have no idea why. Bear with me, I am a beginner and I am not skilled in soldering by any means, so some tips on how to identify problems would be a great help if you are willing!

I've checked the joints for continuity, with some joints linked to capacitors and diodes + the farthest right pin of a power mosfet beeping, these I believe are problems that I had thought were a result of bad soldering, but I had redone the solder (to my knowledge) and beeps persist.

I will attach a picture of the PCB front and back, please let me know if more angles are needed! Thank you again for those able to help!

*Added a picture of the schematic

[sick geyser]

https://www.sparkfun.com/news/2561 Consider the USB traces on this beauty

[spice turtle]

First obvious thing is to make sure the 4025 (U1) is in the correct position and its not backwards. Theres a polarity pin on the IC itself and one on the board. They should match.

Also is that the full schematic? Because it doesnt seem like it gets power either. Check pins 14 and 7 for +5V.

Also how are you measuring the voltage? Your meter may not be able to read up that high without a voltage divider.

You would also have to build everything up to R5 on the top part of the schematic in order to read something

[upper moss]

Be careful! Would be my first reaction. Although your multimeter (I assume you are using a multimeter) might show 0V, that doesn't mean it is actually 0V!

Fo example, say you set you multimeter to measure DC Voltage, and try to measure an AC Voltage, the reading can show something near 0V. And it will probably be correct regarding DC, but it doesn't alert you that there is 240V AC present. If you now assume there is no Voltage present, and continue your endeavour with your fingers, you are going to be in for a shock. Pun intended.

Make sure that you have selected the right unit and range you are measuring. Also, if the HV circuit is pulsed, you multimeter might not be fast enough to react to the frequency it is pulsing. That could also throw your readings off.

[supple pollen]

My first step would be to see if the oscillator circuit is oscillating. If it's in an audio range, you might be able to do so by just hooking the speaker to it. If not, maybe A/C couple its output to your meter with a capacitor and look for AC voltage.

[limpid nest]

Is there anything gauche about using a standoff for a wider thread (unthreaded, female/female) than the hole it is being used for was drilled for?

Hopefully that makes sense

[worldly schooner]

An unthreaded clearance hole shouldn't care about the exact thread specs, as long as the clearance hole still clears them?

[limpid nest]

That's what I thought

[worldly schooner]

The more important part is the flat surface on either side that physically holds the PCB in place AFAIK.

[limpid nest]

Yeah that's all up to snuff

[unique patio]

Some meters show "0L" for "overload", which is not that obvious. But the rest of the suggestions by others are very good.

[slow plover]

interesting! I've only seen it for "open lead" (i.e. open circuit)

[upper moss]

Here we go... OL meaning discussion! 😄

I vote for "0L" Zero inductance 😛

[still sphinx]

i want a switch on the edge of my board, something like this (doesn't have to be exactly like this), i didn't know how to explain it so i made a crude mock-up. I cant find anything for this on mauser or digikey, but i have seen stuff like this before, and i know for a fact that something like this exists

[supple pollen]

Slide switch?

Look at the Nidec CUS and CSS series, the C&K AYZ series, and maybe the E-Switch EG series.

[unreal flax]

That kind of mid-board edge mount is the tricky bit. I've seen that for push-button tactile switches, but not slide switches offhand.

[still sphinx]

I found a tht mount that has its leads on the side, I could use that, but it’s only rated for 5v and I’m running 9 thru it

[ember laurel]

@still sphinx https://lcsc.com/product-detail/Slide-Switches_XKB-Connectivity-SK-3392M-G1_C319017.html

looks like you could mount this one that way - and secure it on the sides

2.1mm tall though - i suppose you want 1.6mm?

[still sphinx]

Thanks! I did manage to find a 1.6mm one, but it’s smt so I’m going to use copper strips across the solder pads

[limpid nest]

Anyone know of LED driver chips that are just hardware? No programming?

[distant raven]

Yes, gimme a second I have a suggestion

How much current do you need?

[limpid nest]

hmmm

total?

Unsure

[distant raven]

Yeah

[limpid nest]

One sec

[distant raven]

This one does 350mA

[limpid nest]

hmmmm

[limpid nest]

I need 9 times 70 mA

minimum

[distant raven]

Okay

[limpid nest]

What should I search to find a software free driver?

[distant raven]

You could probably use two or three to drive sections of LEDs

[limpid nest]

Yeah

[distant raven]

These can also be PWM’d to dim the LEDs too

[limpid nest]

Nice

If I were involving PWM I'd use a different chip, but I'm looking for something to drive N LEDs from just a power source, in "parallel"

let me look at that one. Thanks

[distant raven]

I found this part with the Digi-Key mobile app. https://www.digikey.com/en/products/detail/ILD8150EXUMA1/12761788

[limpid nest]

Lol I tried to highlight the text I'm tired

[distant raven]

I think a lot of controlled LED applications use LEDs. Though you could just use a transistor and drive it in linear range to act as a switch

[limpid nest]

Oh I've got the switching all set.

It's possible to generate a signal that's like an oscillating PWM frequency so that the LEDs seem to "breathe"?

[distant raven]

What I mean is the transistor will act as your driver

[limpid nest]

Oh

[distant raven]

Yeah

[limpid nest]

I knew it was, but I can't remember ever learning how to do that

[distant raven]

Look up dimming LEDs on the Adafruit learn site

[limpid nest]

I know how to generate a PWM frequency, 555 timer

Let me check that out thanks

and thanks adafruit.

[distant raven]

Education +100

[limpid nest]

I feel like the pip boy mascot, but not nearly as violent.

Oh wait, wrong bethesda property?

So I searched "Dimming LEDs" and got a bunch of stuff but I'm not seeing something that perfectly fits

[limpid nest]

Oh, I meant with hardware only

[twilit mango]

You're looking for code to PWM up and down?

[limpid nest]

noo

I was unclear

[twilit mango]

Heh ok

[limpid nest]

I'm trying to see if I can make LEDs "breathe" with just hardware

[twilit mango]

Ahh ok.

I've only done it with code.

[limpid nest]

I feel like it really shouldn't be that hard, I just don't know

Yeah code is trivial (relatively)

But there's gotta be like an RC circuit thingy I can do

[twilit mango]

Probably, yeah

[distant raven]

https://www.electronicshub.org/pwm-led-dimmer-using-ne555/#Block_Diagram_of_LED_Dimmer_using_555

[limpid nest]

I figured the 555 would be involved, but I don't like linking the cathodes of LEDs

I also need something that will just run when powered and not require human setting of something like a pot. I am probably best off just using a cheap MCU and programming it

[distant raven]

That would be the easiest route

[limpid nest]

Yeah that requires you to set the PWM frequency with a pot right?

I'm looking for something that electrically does what a dwell mechanism does. Thanks for the driver above, I'll probably use it in the end and abandon the breathing mechanism

[distant raven]

You could have a push button to set a counter for your PWM frequency

[limpid nest]

It's gotta be completely human free unfortunately

So I have to get super old school or break down and program something

[distant raven]

Or just a timer that slowly increments a counter for setting your frequency

[limpid nest]

Yeah that would need programmable silicon though right? I'm seeing if I can avoid that

TBH programming something is the optimal solution, but I've gotten myself curious

[distant raven]

It would probably be pretty complicated to do it without a microcontroller

[limpid nest]

Yeah

To use a 555 for example you'd need something that varies a resistance with time, no?

[distant raven]

It’s doable, but probably more trouble than it might be worth outside of education

[limpid nest]

Yeah

[distant raven]

You could use a counter that you use a say, 4:2 decoder to trigger different paths for your 555 resistor

That you would use to control the timing

Then use another 555 in astable mode to clock the counter every 30 seconds or something

[limpid nest]

Hmmm

Interesting

Probably better for everyone just to grab a QT py tho

[distant raven]

Oh, easily

[limpid nest]

plus a QT py can change the effect

or can be changed

When you found those PMIC LED drivers, what did you use to make sure they would work without like a SPI or I2C signal?

[distant raven]

But if all you have is a bunch 555 , counters, and decoders, you could do a lot

[limpid nest]

I'm just using PWM dimming for now

as a parameter

[distant raven]

I’m not sure if there are any ones that support i2c or SPI

[limpid nest]

Any ones that are hardware only? That's what I was gambling on, but I'm still getting ones that are serial based

[distant raven]

I think most if not all are able to operate without a microcontroller

You just need a pwm source of you wish to dim them

[limpid nest]

Hm OK

thx

[frigid current]

Hey all, looking for some help with a PCB I'm working on. I got prototypes and they're not working as expected

I'm trying to make a USB-C switcher

Here's my circuit diagram

I'm using a rotary switch (center) to have one input USB-C receptacle switch between two output receptacles. I thought I could pass the CC pins through to allow for native USB power delivery negotiation. But when I soldered up the boards, that didn't work at all

Luckily I don't really need that feature, the only thing I want about USB-C for this project is the nice plug form-factor

So I soldered on some resisters to pull each receptacle's CC pins to ground:
(this tells the USB host to treat the device like USB-B, ignoring USB-C-style power delivery negotiation and other fancy stuff, and just connecting using normal VBUS, GND, D+, and D- pins)

The circuit digram should now be:

(I know ideally these should be 5.1kΩ resistors, and ideally there should be one for each CC1 pin and one for each CC2 pin, but I think using one resistor for both CC pins usually works fine, and I think the resistance is pretty flexible)

Now when I plug in my keyboard to the input, and two computers to the outputs (Host A and Host B), only one of them will work. And what cable I'm using starts being an issue: if I plug the keyboard directly into my MacBook, any of my 3 cables works, but if I'm going through the PCB, my computer seems to need a certain cable that came with my keyboard on its side of the system, and won't make the connection if I'm using either my fully-featured 10gbps USB-C cable from my hard drive, or my MacBook charging USB-C cable. Again any of these work when connecting directly.

Any idea what's going on here? I have a few ideas

• maybe the fact that I'm sharing ground between all three receptacles isn't good, and is messing up various things

• Before I grounded the CC pins, maybe the fact that B8 and A8 weren't connected was making PD negotiation over CC not work? Although I thought that would work fine...

[ember laurel]

I need TEA1716T/2,518

The world has gone crazy - lead time for this chip is 76 weeks right now

[supple pollen]

The Raspberry Pi people made that mistake, and it caused them a fair amount of trouble.

[supple pollen]

Lots of people are building power supplies, and increasing regulations are requiring PFC

[still sphinx]

https://www.ebay.com/itm/194288857705
I found this if you don’t mind paying inflated prices

[frigid current]

I wouldn't implement it this way for a final, but for this test I thought it would be enough. Maybe not. I'll set up a breadboard with properly grounded USB-C receptacles (https://www.adafruit.com/product/4090) and see if I still have a problem

Any idea about why the first setup didn't work though? I thought passing through CC would be fine

[ember laurel]

Says only 9x10 available

I need 525 units

[worldly schooner]

If this is a passthrough, you could instead skip the resistors and pass CC1/CC2 directly? The power negotiation is usually handled by devices, not cables, anyways.

Though you definitely should keep the two separate from each other, otherwise your negotiation circuits are going to have some funky readings.

[frigid current]

@worldly schooner that's what I tried first, and it didn't make any connection at all

[frigid current]

See this circuit diagram

I thought this would basically make the devices oblivious to the fact that there's anything in between them, but something about that didn't work.

Do you think common the common ground between the two hosts and the keyboard would mess things up?

[worldly schooner]

It's probably the crossing of CC1 and CC2? Try keeping those signals separate?

[frigid current]

Ooohhhh

Aw crap well my Rotary switch doesn't have enough poles to be able to do that 😦

Actually I don't think it's a problem to cross CC1 and CC2, see this article:

https://www.allaboutcircuits.com/technical-articles/introduction-to-usb-type-c-which-pins-power-delivery-data-transfer/

Connecting the USB Type-C cable creates a current path from 5-V supply to ground. Since there’s only one CC wire inside the USB Type-C cable, only one current path is formed.

So if there's only one wire, connected the two just makes sure that that wire is passed through no matter the plug orientation

(and admittedly plug orientation detection will be disrupted)

[elder peak]

tl;dr: Sourcing power factor IC's not everybody's cup of TEA today.

[ember laurel]

Slow clap

[limpid nest]

How weird is it for a chip to have a punch of external passives but the datasheet lacks any guidance for picking those passives?

[fervent lance]

I opened an electronics for the enclosure but the PCB had mounting one and one of them had metal around it

was it some sort of debug port/test point ?

[worldly schooner]

Mounting hole with metal?

That's pretty standard for any kind of metallic enclosure, as the plated hole can serve as a connection to earth?

Sometimes PCB designers just default to plated holes for reducing the likelihood of screws scraping fiberglass particulate from within a PCB hole.

@fervent lance Usually test points are fairly small circular pads, it's unlikely they would use a mounting hole for a test pad.

[still sphinx]

Oh boy

So like a reel?

[limpid nest]

EAGLE users. Is the pattern tool just awful in EAGLE too? I didn't use it a lot before I switched to Fusion. It's truly bad in Fusion.

[distant raven]

Yeah it’s not great but better than fusion in my experience

[limpid nest]

Oof

I legit don't think it's going to save me ANY time to pattern something that should absolutely be patterned, if the tool worked.

[supple pollen]

My approach is save the object, write code to generate the script or XML to generate the pattern with the objects, and re-import/run the modified file. I know that's not the approach for most people, but it puts things in the realm where I have absolute control, which is how I like it.

When I ultimately make the switch to KiCAD, I can just embed my ideas of reality directly in the code (and submit pull requests in the hope that someone else wants that behavior)

[limpid nest]

I think when I have a serious pattern to do, I'll try htat

[supple pollen]

I built a custom footprint the same way, and even numbered 3/4 of the pins correctly...

[pure gorge]

This is one of my first PCB designs so go easy on me, but any feedback or things I did wrong here? I know I probably messed something up on the CP2102N or ESP32, and I could probably squeeze the whole board into two layers if I try hard enough.

[unreal flax]

I have not done a thorough review, but at first glance I'm seeing a general lack of decoupling capacitors on the various power-rail pins on each chip. Also, the C1 and C2 caps on a couple of the serial lines are unexpected, so those should probably be double-checked.

It's not obvious to me that you're going to be able to flash the ESP32, since the various boot-mode pins don't seem to be hooked up in any controllable way.

[pure gorge]

Ok, i was following the ESP32 datasheet for the typical usage and probably got quite a few things wrong

This was originally supposed to be with the RP2040 but i switched to a faster processor with more SRAM

[unreal flax]

Espressif publishes reference schematics for a lot of their development boards, so those are a good source to copy.

[limpid nest]

Anyone have a good sense of how much heat a single neopixel puts out at full blast? The datasheet just basically says "it depends and it's important you get it right!"

[unreal flax]

The canonical current consumption is 60mA, so if you power it at 5V and all the energy went into heat, that would be 0.3W. Some of it will escape as light, but it's probably not more than 50% efficient, so I'd ballpark something like 0.2W of heat at full blast.

[limpid nest]

Hmmm ok

Thx

[inland jungle]

the voltage drop across a diode is typically 2-3V, so even if you power it with 5V, you'll be dissipating some of that across the current limiting resistor

[marsh nest]

OK, need a sanity check.

https://www.bourns.com/docs/Product-Datasheets/70adm.pdf
Looking through this connector's datasheet. And... There's nothing I see about what pins are actually connected?

so, the drawings they have are of a 4-contact connector, with 7 pins. 4 are aligned to the contacts, and the other 3 are on the other side and offset. Logic dictates that the four aligned pins are the ones that are connected, but. I've seen some illogically designed packages in the past.

[steel valve]

How do SMD RGB LED work exactly? Thinking of adding onto my project

Do I need to use resistors or caps on it?

What ends up being its power draw since it can be multiple colors?

[marsh nest]

Depends entirely on the LED.

You have ones that are common-cathode, in which you drive the anodes differently to mix your colors.
You have ones that are common-anode, in which you drive the cathodes separately.

For these, they're effectively three different LEDs - you'll need different resistors and each color will have a separate power draw.

There's also ones that have built-in controller chips, which can use a variety of communication protocols. Ones like those found in the Neopixels use a single wire for communication, but there's others out there too.

You can also get separate ICs that will control your LEDs for you over i2c, SPI, or one of a pile of other protocols. These may or may not require external resistors.

[steel valve]

How's the Neopixel one work?

Does it communicate via I2C or something

[marsh nest]

Nope, not i2c, it is it's own method of doing it

[steel valve]

So a custom library then

[marsh nest]

Thankfully, Neopixels are common enough that there's plenty of libraries out there for them already

[steel valve]

Does mouser carry single ones? I keep finding kits or strips of them

[marsh nest]

Mouser? No idea. Adafruit sells them, though, and you can get neopixel compatible individual things on places like Amazon too

[steel valve]

Are there all in one ic's that can charge a lipo, and also take the lipo and convert down to 3.3v?

[distant raven]

That seems to be the case. You could make two test designs and test continuity to verify for certain

[distant raven]

I think Texas Instruments makes something that does this but it’s not likely in stock

[marsh nest]

Some careful multimetering of the actual part could work too, before I go and order any PCBs.

[distant raven]

Probably, probing might be difficult if it’s a small part

[marsh nest]

It's a 4mm pitch connector, so plenty large enough. Annoying, possibly, but shouldn't be too bad

[distant raven]

Oh gotcha

[distant raven]

I have a curious moment. I see that they have these magnetic 3 pin connectors but they are right angle. I feel like it would be reasonable to be able to solder the pins SMD style to do a right angle connection. it wouldn't be a high torque set up, just a connector for an LED PCB onto a base to get power/gnd and data for neopixel designs. https://www.adafruit.com/product/5360

[supple pollen]

It seems to me that would work nicely

[pure gorge]

Does anyone know where i can find a footprint and symbol for this? https://www.adafruit.com/product/4932

[supple pollen]

You could probably extract them from an AdaFruit board that uses the connector

[distant raven]

I think solder party boards use that

[pure gorge]

Do you know of any adafruit boards that might use it?

[supple pollen]

Not offhand, alas

[pure gorge]

Ok, i found an alternative part that has a library

[fervent lance]

What is the part # for the the female battery connectors on the feather boards? I'm looking for that with wires attached or solder points. I want to connect them to a lipo battery recharge unit so I can easily plug in my lipo batteries and change them.

[marsh nest]

You're looking for something with wires that you're going to plug a lipo into?

It's a 2-pin JST 2.0mm pitch connector. I think they're JST-PH connectors but I'm not certain.

I do know that not all of the ones you can get match the battery colors with their wires, though

[steel valve]

Can I add in pads to a pcb for a circuit of parts that aren't on the pcb currently

Like an optional component area

Or would that make an open circuit

like I want to add an optional usb to ttl programmer circuit to my pcb

Since I have the space for it

[unreal flax]

Yes, that should generally be harmless.

[steel valve]

Ok cool

Cause I am unfortunately going to have to have components be placed on the bottom of this two layerp cb

Since this time I am going to go two layer rather than four layer

So since I am going to be using bottom space I am going to cram as much useful stuff down there as possible

Like possibly adding a battery monitor ic or something down there

Since those are generally OOS but might be back one day

At least in the form factor I would want it in

Most fuel gauges on mouser have circle pads on the bottom of the ic

And I have zero idea how this is supposed to be soldered

The only thing that fits the criteria is a bit too overkill and pricey

[steel valve]

Actually, what kind of components can be hand soldered?

(Of SMD stuff)

[distant raven]

Sure, BGA is just harder because you can’t see shorts under the package

But you can do it

[steel valve]

What about these type of smd components?

The only pads are near the edge

Which (should) mean it is simpler to solder cause it is easier to see bridges?

[distant raven]

Oh yeah, those are pretty easy

I do caution that anything less than 0.5mm pitch would be very unworkable without a stencil and very good solder paste application

[steel valve]

Oof that was just a placeholder image it does have a gnd pad in the middle oof

Ok well I guess my only option if I want to see how much is left in the lipo is to use that pricey one

[supple pollen]

You can also use "Coulomb counter" type fuel gauge chips, or monitor the cell voltage to get an approximate state of charge.

[distant raven]

Reading voltage isn’t a terrible way. I think a few Adafruit boards do that like the MagTag

[steel valve]

Couloumb counters on mouser seem to be even more expensive

At least in the form factor of choice

[distant raven]

Do a 100:1 voltage divider and tap the center

[steel valve]

100:1 voltage divider?

[distant raven]

Yeah, Adafruit does something like that on the MagTag for reading the battery voltage

[steel valve]

How accurate is doing a voltage divider to read it?

[supple pollen]

Most boards use a 2:1 voltage divider which brings the maximum 4.2V or so cell voltage into a safe range and is easily done with a pair of resistors of the same value.

[steel valve]

What I use is a ldo to get the thing to 3.3v

[distant raven]

Oh sorry, it’s 1Mohm/1Mohm

[steel valve]

unless you mean 2:1 voltage divider to read battery left

[supple pollen]

Yes, the supply voltage is 3.3V, so the 2:1 divider makes the maximum cell voltage (which is more than 3.3V) safe to read with an analog input.

[distant raven]

Though they tap VBAT rather than the regulated supply

https://cdn-learn.adafruit.com/assets/assets/000/096/946/original/adafruit_products_MagTag_sch.png?1605026160

[steel valve]

So pretty much I have it like this:

USB -> USB to TTL -> MCU
USB -> MCP Charger -> LiPo
Lipo -> 3.3V LDO -> MCU
Lipo -> Voltage Divider -> MCU Analog Input

[distant raven]

Schematic for reference

Yeah, basically

Dang that link doesn’t seem to render for me

[steel valve]

epic embed fail

[distant raven]

Yeah haha

https://learn.adafruit.com/assets/96946

[steel valve]

What are other useful things to put on a pcb

Aside from a Charger, LDO, USB To TTL, Voltage Reader, and a button that puts the mcu to sleep and disables everything else

[supple pollen]

LED for testing?

[steel valve]

Oh yeah forgot to mention yeah that is already there

When the button is held for x amount of seconds, the MCU changes the color of a RGB LED and does stuff accordingly

Like for example holding it for one second turns it on/off (sleep), holding it for three seconds makes it go into pairing mode, and this all makes the led change color or flash

I just have not decided what rgb led to use

I would like to use one that doesn't require external resistors and the work is done via the mcu using a library, such as a Neopixel, but I can't seem to find anything that does that on mouser that is sold in singles

Since I don't need 10 I need like five

[supple pollen]

Resistors aren't hard. If you're making a PCB, you can even use a resistor array to reduce part count.

[steel valve]

Yeah I was thinking of using a resistor array

Since I do have some sections that have multiple resistors

Like for example the USB C requires two 5.1K resistors for pulldown on CC1 and CC2

Also, why does a USB C require that value specifically?

Tryna see if I can use a resistor that I use in other circuits for the USB C

Since it would save a bit to just use a resistor I am buying a lot of

[supple pollen]

That value was presumably arrived at by a process of dodging values that would be near what circuitry would draw, as well as values already used by various manufacturers to signal an assortment of semi de facto practices for pre-existing charging capabilities, as well as not drawing too much current.

[distant raven]

USB C spec says 5.1K is a standard value that will negotiate the max current of 3A, pull ups of 10k and one or two other values negotiate lower current on the up stream side (source).

https://www.st.com/resource/en/technical_article/dm00496853-overview-of-usb-type-c-and-power-delivery-technologies-stmicroelectronics.pdf

ST covers it nicely

[inland jungle]

wouldn't you get a messed up reading unless the input impedance for the measurement was >>1MOhm?

[distant raven]

Look at the schematic

The idea is the voltage divider is center tapped because it evenly splits the voltage, and the 1Mohm resistor from VBAT to tap is a very small current so it doesn’t damage the MCU pin

[inland jungle]

that's not the issue. If the input impedance of the MCU pin is <1MOhm, then it's essentially acting in parallel to the bottom resistor of the divider

and changing the effective voltage division

[distant raven]

It works for the MagTag

Besides impedance and resistance are different even if they have the same units

They imply different things as you consider different variables like frequency of the voltage, and whatever the imaginary component of the associated CL components

[inland jungle]

sure, and typically analog inputs are measured by charging up sampling capacitors through the input impedance

[distant raven]

Which is likely why Adafruit includes a 0.1uF capacitor as it changes the impedance of the battery tap

[inland jungle]

yeah, that might solve the issue if the sampling caps are much smaller

[supple pollen]

Correct. However the input impedance is indeed very high, so it doesn't affect the measurement much.

[steel valve]

What copper pour clearance is recommended for two layer 0.8mm thick boards that do RF stuff

[supple pollen]

For radiating components (like chip antennæ), the data sheets will have guidelines.

[limpid nest]

For EAGLE/Fusion, does anyone have experience doing a "pad in pad" footprint? e.g. a footprint that both a thru hole and an SMD pad? I'm trying to recreate a QT-PY footprint that I had some issues using

[unreal flax]

Normally I'd do that as an SMD pad plus a via.

[limpid nest]

Via in the library editor?

That's a good idea

Thanks, I'll try that

[still sphinx]

I used this https://www.instructables.com/PCB-Coils-in-KiCad/ python code to make pcb coils in kicad, and i added the output to the pcb file, but nothing happened.

[unreal flax]

Oh, good point, I'm not sure if the library editor has a via tool. The alternative would just be to create two pads and tie them both to the same schematic pin in the device connection dialog.

[limpid nest]

Yeah that's what I was thinking

thanks

Hmm, I'm here:

I'm just editing a footprint I had issues with to see if I can fix the issues that way. This is what shows up when I use New in the package manager. Should I just name it like QT_PY_MOD?

Should a PWM input pin have a pullup if it's also an EN pin that enables the chip when it's high?

[unreal flax]

That sounds unusual. Wouldn't the chip be disabled whenever the PWM signal goes low?

[limpid nest]

Yeah it's confusing

well

It's an LED driver

So maybe that's normal

[unreal flax]

Oh, I see. So the PWM is basically like flicking the light switch on and off really fast, basically turning the chip on and off. Up to you whether you want it to default to be on or off.

[limpid nest]

Hm ok that's what I was thinking

Thanks

[steel valve]

The chip antenna is outside of the pcb, which is why I ask

[supple pollen]

So you're asking about clearance from the externally mounted chip antenna to the PCB, or traces/components on the PCB, or what?

[steel valve]

Ok so pretty much, I have a breakout PCB for a RF Module, the Antenna & Copper Keepout Zone is hanging off of the breakout

And am trying to decide copper clearance for the copper pour on both sides of the breakout pcb

[steel valve]

How many volts does a digital pin output?

Is it the same voltage that the mcu is powered with?

[distant raven]

Usually

[limpid nest]

Ooh careful there

Just that

[steel valve]

Because I believe the mcu I am using bucks.

[limpid nest]

Most (adafruit) MCU boards take 5V, but the actual MCU uses a regulated 3.3. Easy to get mixed up

[steel valve]

The NRF51822 by adafruit runs from 1.8V to 3.6V

And I am supplying 3.3V

So I assume it is being regulated down to 1.8V by the module?

Pretty much I have a button on my PCB that tells the NRF51822 module to go to sleep, and set all EN pins of components on the PCB to off. I want to signify this by having a RGB LED flash or change color depending on what happens

[distant raven]

Usually pins output the voltage level of the main supply. The board input voltage is usually the same as USB, it’s then regulated down. Some pins are 5V tolerant for inputs on but don’t generate 5V signals themselves

nRF51822 should have 3.3V logic digital IO.

Or whatever the supply voltage is in that 1.8-3.6V range

[steel valve]

Ok so then I'll just assume it's 3.3V since that is what I am providing it

What is forward current on a led exactly, is it it's power draw?

[limpid nest]

Kind of

[steel valve]

Or how exactly do I know how much power a led will use

[distant raven]

Individual LEDs draw anywhere from 10-40mA with 20mA being standard for a fully driven single color

[limpid nest]

This isn't a perfect formula, but you can calculate the Fwd current with I_Fwd = (VSupply - VLED)/R_Series

It's the current thru the LED

[steel valve]

Yeah pretty much I want a LED whether it be RGB or a normal LED in my pcb, but I am tight on power

[limpid nest]

You don't have to run an LED at its max to get "good enough" brightness in most cases

[steel valve]

I am also trying to make the component placement not ridiculous

[limpid nest]

Heh, that can be troublesome

[steel valve]

So I keep swapping between using a RGB LED and Normal LED

[limpid nest]

normal LED is easier for sure, you can control one with just a transistor and a pin

[steel valve]

If I could use a Smart RGB LED like a WS2812, I would go with that but mouser does not seem to carry RGB LEDs with Drivers in them in singles

Aside from one

And it is unclear if it only runs on 5V voltage or not

https://www.mouser.com/datasheet/2/244/SMD-LX0707RGB-TR-1524526.pdf

[steel valve]

I'm confused at what that means

Is it saying that the max supply voltage is 5V?

[limpid nest]

It says typical is

I wouldn't stray far from it? Since it doesn't give a min or max

[steel valve]

Yeah, that's why I am going with a normal RGB LED or just a single LED

Since I can communicate everything with both

Aside from not having a low power indicator

[limpid nest]

Yeah it's easier to just get a FET that can be switched by 3.3V or lower

[steel valve]

But that would probably not be used since power constraints

[limpid nest]

To switch the LED

The FET itself takes almost no power

[steel valve]

No no no like the low power indicator

[limpid nest]

ahhhh

[steel valve]

Pretty much:

It is red when the battery is low on power
It is green when awake
It flashes blue when pairing

[limpid nest]

I can see why you wanted an RGB LED

[steel valve]

I'm a little tight on space on the top layer of the pcb, which is where the rgb led would go

And I use 0603 components

[limpid nest]

hm

well RGBs can be small, 3.5mm square

[steel valve]

Which is why having an all in one RGB LED would be nice

Since I don't need the three resistors

[limpid nest]

I think adafruit sells a 3.5mmx3.5mm neopixel

[steel valve]

I could however use a resistor array

Which would save space and be cheap

But from my understanding they aren't the easiest to solder

[distant raven]

Easy with a stencil and solder paste

Or easier

Neopixel are a great option for RGB

5050 and 3535 sizes are easy to work with

Usually have max current draw around 50mA, though typically draw about 1/3 of that for a single neopixel

[steel valve]

I'd be running them really low then

Cause the lipo I am going to use is 110mA

[distant raven]

Yeah, plus you can drive lower intensity on each channel and get decent colors

[steel valve]

I just wish they could be bought individually instead of having to get a 10 pack

[distant raven]

So instead of making it red (255,0,0) you can do (122,0,0) and get a lower intensity

Never hurts to have extra. I’ve personally bought two reels of 1500 each lol

[steel valve]

what circuit do neopixels use?

Do they require external caps or anything

[distant raven]

You can put a 0.1uF ceramic capacitor between VDD and GND

Large chains of neopixel typically like a resistor on the data line

[steel valve]

Yeah I only intend to use a single one

At least per pcb

And I intend to make two pcb

[distant raven]

Yeah, no passives really needed then

[steel valve]

I am making it so my PCB never fully turns off like most fancy ones do.

I can disable the charger, usb to ttl programmer, multiplexer, and ldo (which I wouldn't do cause that would break stuff)

From my understanding, LDO are not energy efficient and require a lot of energy

How do I figure out how much the LDO is drawing?

Or for designs that go to sleep mode rather than turning off, would it be better to use a buck.

[unreal flax]

An LDO discards all the extra voltage between the input and output. So if you have a 7V input and a 5V output, it will be 5/7 = 71% efficient.

[steel valve]

Actually, what does EN do with a LDO

Or I mean

What is a practical purpose of it

And then in that case, the efficiency of the LDO is 80% min since a lipo is going to it

And the efficiency increases as the lipo loses power

[unreal flax]

You can use it as an on-off switch for (sections of) your circuit, for example. Disabling the LDO will also disable anything that it normally powers.

[steel valve]

So in the case that the LDO powers everything it should always be on

[unreal flax]

Yes, it's typical to just tie EN to the input voltage so the LDO comes on automatically.

[steel valve]

When a MCU goes into sleep mode, do sensors connected to the mcu draw power still

[unreal flax]

Yes, generally the sensor will not know that the MCU is asleep. Sometimes sensors have their own sleep modes which can be activated by a command.

[steel valve]

Yeah, the sensors are potentiometer-like (variable resistors)

So they'd just draw power too

[unreal flax]

If they're like pots, then they would have the resistor current flowing through them all the time, independent of whether the MCU is reading the result or not, yeah.

[steel valve]

Luckily the majority of them do not go through the mcu normally

10/15 go through the mux

Which can be disabled

[unreal flax]

I don't think that would matter either. As long as the sensors have power, they'll draw current, even if the mux isn't reading them.

[steel valve]

The stuff going through the mux require a voltage divider that is at the output of the mux to be readable

Would they still draw power while being half of the voltage divider circuit

[unreal flax]

Oh, good point, I had forgotten you put the divider on the output side. In that case, they would indeed not draw current when the mux is de-selected.

[steel valve]

Ok cool thought so

There are seven voltage divider sensors that I assume would always be drawing power then

What would be the best option for cutting them off?

And does a neopixel draw power when the led is turned off?

[unreal flax]

One option for the voltage dividers would be a FET on either the high or the low side, so you'd switch them on right before taking readings, then disable them again.

The Neopixel would draw some power, since it needs to be monitoring the data pin and echoing signals for the other LEDs in the string, but I'm not sure what its quiescent current draw looks like.

A quick search online suggests about 1mA minimum current draw, though it may vary with the specific type of pixel.

[steel valve]

Yeah that is what I found too

I might have a simpler way to do this

Pretty much on my PCB there are a crap ton of through holes on the edge

And they have unevenly distributed voltage holes

So I have on the top layer of the PCB a long strip that goes around the pcb in a loop-de-loop

And I could put the mosfet there

That would completely disable every external sensor

Yeah since my first pcb failed cause I put the wrong type of switch on it and the usb c hole was a bit too small so this time I am adding in everything I want, and making it two layer

The top layer solely controls power excluding the rf module

It contains the charger, ldo, led, button
The bottom has everything else, mux, programmer + breakout pins for an external programmer, etc

[steel valve]

I have two options, have a trace go straight to its destination on the top layer, but a chunk of gnd pour is gone, or make a trace go underneath the pcb and come back up with a via to get to its destination but there is significantly more ground pour.

What is the better option?

I assume it is the via?

[steel valve]

Also, what is more important, a signal trace being shorter or having more ground plane (in a compact two layer pcb)

[supple pollen]

As in most engineering decisions, it depends. If the trace is carrying a high speed signal, it may make sense to optimize it. If the ground pour being lost is under a bunch of traces carrying high speed signals, it may be worth avoiding that.

[fiery parcel]

Hey all, I'm using a PN532 receiver for one of my projects but I want my antenna to be in a specific shape. What are the chances I would be able to design my own with PCB design?

[worldly schooner]

https://www.electronicdesign.com/technologies/passives/article/21769333/electronic-design-welcome-to-antennas-101
tl;dr not impossible, but also not ideal. You could certainly make a simple one if you have enough technical understanding of basic RF concepts, but it's not a trivial task and the returns for it are limited. Unless you're really particular for ideal aesthetic over ideal function, I can't say I would ever think it's worthwhile, but it would make for an interesting experiment.

[fiery parcel]

Yeah. Basically, I have this toy that has a hexagonal base that has the main board, speaker, battery and NFC reader

The antennas I’ve been using previously are too big, or too small. I’ve been wondering as well if I can detach the antenna from the adafruit NFC breakout. That would work too.

[worldly schooner]

https://www.campusidnews.com/understanding-rfid-part-4-the-black-art-of-rfid-antennas/#:~:text=This is because the wavelength,wave is approximately 22 meters.

[fiery parcel]

Thanks for sharing!

[steel valve]

The NRF51822 uses a 16MHZ ARM Cortex processor so I am prioritizing ground since I can't via stitch the ground in this case due to the limited area to do so

[ember laurel]

does anyone have any good tips for an LDO that handles 40V input, 9V output, and a very low load?

(and also doesn't require a ton of current on the adjust pin)

I was using the SY6345, but it has disappeared from the face of the earth

I really only need to pull some 3mA

LM317 doesn't handle regulation when loaded with less than 10mA current draw.

at 10mA (through lower adjust resistors) it starts pulling quite a bit of current, due to the large drop I have.

I'd prefer to avoid a buck for this

[pure gorge]

Is there anything wrong with this schematic? I'm just making sure before I get it manufactured

I also don't know if the resistor values are right for the notes, I used this website to calculate them roughly: https://ohmslawcalculator.com/555-astable-calculator

[ember laurel]

1. Your Electrolytic capacitors seem to be reversed?

10nF seems like an MLCC.

But the 10uF - is that also MLCC, and not-polarized?

[pure gorge]

Theyre not electrolytic, theyre ceramic. I just used a polarized library since I couldnt find it unpolarized.

[ember laurel]

Ok. What is this thing supposed to do?

basic synthesizer? 🙂

[pure gorge]

Its a stylophone business card

[steel valve]

In PCB Design, how safe is it to keep a crap ton of digital signals close to each other

That traffic jam of digital signals (ignore the 90 degree turns on the bottom fixing that) pretty much can only be 0.127mm apart from each other

Otherwise this isn't gonna route

[limpid nest]

Low speed it's fine

I do it literally all the time

[steel valve]

When is something considered high speed

[limpid nest]

Hmmm

I've heard GHz quoted?

But I think it's a band of speeds

where it begins that is

[steel valve]

Yeah the NRF51822 runs on a 16MHZ processor so it's relatively low speed

[limpid nest]

I'm usually doing a low speed SPI or I2C bus

[steel valve]

But anyways yeah on the bottom layer is the signal layer, and it's just really funky looking

I'm using like everything the nrf51822 has to offer aside from a few pins and those just are not accessible in a two layer board

[edgy jungle]

So I have a Pi Pico being powered via ~5V to VSYS, however, only ~1.2V comes out of VBUS. Any ideas why this is?

[supple pollen]

There's a diode to isolate Vbus from Vsys

[edgy jungle]

So that is what's causing this voltage drop? Is there anyways to remove it?

Does it cause issues if I short VBUS to VSYS?

[limpid nest]

Had a wacky idea. Mounting a QT PY in the middle of a board but on pins so that the USB port is accessible. This is so that the USB port isn't too close to an enclosure wall to be used.

Wacky or reasonable?

[distant raven]

Seems reasonable

Or.. I think the qt py usb data pins are exposed on the bottom

[limpid nest]

let me see

[distant raven]

So you could probably use pogo pins to break usb to discrete headers

[supple pollen]

It could cause voltage to backfeed into the USB port if it's plugged into USB while powered via VSYS

[limpid nest]

I'm looking at an ESP 32-S2 bottom atm and don't see that

[distant raven]

Ah it might be just the m0 one

[steel valve]

Does I2C work on any digital pin?

I am adding an optional I2C pad since the routing of my pcb is complete

So I am using the last bit of spare space to do something useful

And the NRF51822 doesn't have any specifications on requiring a specific pin to do so, so I assume it is programmable

[distant raven]

Yeah, you set those in the firmware

[steel valve]

Why does a decoupling cap go to the analog power supply (AVDD)

[limpid nest]

That doesn't seem to be a decoupling cap

[steel valve]

What's it for then

(Datasheet BTW)
https://cdn-shop.adafruit.com/product-files/4076/4076_specification.pdf

Is it a filter?

[limpid nest]

Oh I see now

I was wrong, I think it is

[steel valve]

Why does the analog vcc need its own cap?

[limpid nest]

I would think to smooth any input voltage ripple

[steel valve]

Ok so it's kinda like a filter then

[limpid nest]

llike a band pass filter? ish?

[steel valve]

kinda

I think it is so it can filter out RF SIgnals

Since this is a RF module

And AVDD is very close in proximity to the antenna

But anyways, now I have everything done aside from the voltage going to the components

I have a LiPo battery that has to go to two things, a capacitor from the charging IC, and a voltage divider circuit so the voltage can be read

The LiPo + has to go through the resistor of the voltage divider to get to the capacitor

Would issues occur from this

[distant raven]

Looks like AVDD is for generating an analog reference voltage

[steel valve]

But that's what AREF does

[distant raven]

I don’t see an aref on the board

Let me look at the data sheet real quick

[steel valve]

Yeah this is the sheet

But there are multiple AREF

https://infocenter.nordicsemi.com/pdf/nRF51822_PS_v3.1.pdf

And this is the NRF51822 datasheet since the adafruit sheet sucks

[distant raven]

Nordic uses a 1nF capacitor on AVDD

[steel valve]

So I assume that means the module board itself has no cap and wants an external one?

[distant raven]

My guess is Adafruit probably expected more noise on AVDD or didn’t include one on the module.

[supple pollen]

It shouldn't go "through" (series) at all, the voltage divider is normally in parallel.

[steel valve]

This is my only ref image of the internals

[steel valve]

wdym

[supple pollen]

There should not be a series resistor to the LiPo

[steel valve]

The arrow is the direction the lipo is entering

The three diode line is the voltage divider for reading the voltage of the lipo

And the two to the right are the caps that go from the charging ic to the ldo

[supple pollen]

That picture is singularly uninformative, and I don't see any diodes at all. Do you have a schematic?

[supple pollen]

Like I said, there is no series resistor, the connection goes from the LiPo to the regulator.

Oh, do you mean the capacitor across half the voltage divider?

[steel valve]

yeah

[supple pollen]

That just forms a combination voltage divider (to sample the voltage) and an RC filter (to smooth out variations in the voltage). It should be totally fine.

[steel valve]

The LiPo connects like this

It goes through the bottom pad of the first resistor of the voltage divider in order to connect to the capacitor of the charger

[supple pollen]

The charging current, however, does not go through the resistor, it just goes through the pad.

[steel valve]

Ok cool

Then another routing question is, can the capacitor for the charging circuit connect to pin 1 of the ldo, that is connected to its respective cap

[supple pollen]

Yes, those capacitors are effectively connected together anyway. Often the capacitors will be physically close to the various units, so they can give the best effect.

[steel valve]

Ok so it doesn't matter which order it is connected (doesn't have to be linear)

[supple pollen]

Generally not. Since you have RF signals nearby, things can't be too far, but it doesn't need to be linear.

[steel valve]

Yeah everything is as close as it can be while being easy to assemble

[supple pollen]

That's a good thing to keep in mind!

[steel valve]

Now I just need to ground stitch, and then add in a mosfet that the mcu controls for the sensors

Since ten of the variable resistors go through a multiplexer, which can be disabled when in sleep

But five of them do not go through the multiplexer

And they draw ~1.3 milliamps

Which would drain the battery in only 100 hours or so assuming it was at max charge

[supple pollen]

So you'll use the MOSFET to switch the power to the variable resistors?

[steel valve]

Yeah

When the MOSFET is enabled, it allows the voltage to travel to a long trace where the variable resistors are

[untold moat]

@supple pollen You have used the Sparkfun Artemis Module in the past from what I read, what issues occurred with it that made you not like it?

[supple pollen]

The board package was incompatible with the board. I tried installing a dozen or so different previous versions, but none of them would talk to the board.

[untold moat]

wdym by board package being incompatible with the board

[supple pollen]

I mean I was unable to load software onto to board to run (even the basic blink sketch)

[untold moat]

Ah

Is it because of the Sparkfun Bootloader?

[supple pollen]

Presumably the bootloader changed over time, and there's no support for the early version (which I have)

[untold moat]

Why not ignore the bootloader and program via SWD

(or jtag)

Is the bootloader a requirement?

[supple pollen]

I'd have to find a way to hook up SWD (no connector is populated), find something to produce the signals, find the right software, etc. There's a limit to the amount of trouble I'll go to when I have plenty of other projects to do, and the board doesn't bring any particular functionality I need. I'm spending more effort on trying to get my gimbal driver board working, as it has abilities I can use.

[untold moat]

Isn't D21 and D20 SWD Pins?

But yeah good point it is annoying to have the external software

[supple pollen]

A bootloader makes the board so much easier to use (SparkFun promised "blinking in 5 minutes": after several hours with no success, I threw in the towel)

[untold moat]

So it is just because yours is an original version

[supple pollen]

Presumably.

[untold moat]

By the way how did you get it on your board?

Like soldering it

Isn't it annoying to place it down

[supple pollen]

I didn't, I just bought the SparkFun Artemis Nano

[untold moat]

Ah

I have been trying to find people actually succeeding in putting it onto a board since there are a bunch of pads and if one gets messed up it could cause issues

The only thing I have seen is someone placing solder only on the pads they intend to use

Which would solve the problem

[supple pollen]

I suspect you'd need good soldermask, stencils, and hotplate/reflow to reliably solder the little carrier board

I had planned to use the entire Nano as a daughterboard (the same way I use Teensy, Trinket, and ItsyBitsy boards)

[untold moat]

I intend to use the module itself since it seems to be the best RF module available on paper

[supple pollen]

So the Artemis joins the nRF chips in the "parts I no longer care to use" category

[untold moat]

What do you use instead for RF?

[supple pollen]

I've gone with the TI and Cypress chips

[untold moat]

I keep hovering between modules

Because Apollo3 and nRF are annoying

[supple pollen]

Heh, agreed

[untold moat]

Segger is annoying to use, and nRF modules are constantly OOS

(Also not as power efficient as others)

Apollo3 modules are typically hard to put onto a pcb

Have another crap sdk

But are a good bang for your buck considering how powerful and low power they are

^ Also are in stock consistentlyt

[supple pollen]

The TI CC2650 is pretty power efficient. The RF430 chips are ridiculously power efficient, but they're NFC, not RF

[untold moat]

Any sort of Arduino support for the TI CC2650?

[supple pollen]

Hmm, I actually don't remember. I used the "Energia" port of Arduino to the TI MSP430 chips, which was pretty painless.

[untold moat]

Has an all-in-one module been made for that chip

[supple pollen]

Yeah, they call it "SensorTag"

The last ones I used were the CC3100 modules, which are less power efficient, but fairly easy to integrate with a separate CPU

[untold moat]

From what I am seeing, it is not in stock at the moment, oof

parts shortage moment

[supple pollen]

The CC3100 has a goofy footprint

[untold moat]

Wow that really is goofy

[supple pollen]

Worked, though

[untold moat]

I am not experienced with pad only pcbs such as the artemis-- should I apply solder paste to only the pads I intend to use, or should I put it on everything

[supple pollen]

I'm not sure what the best practice is there. I'd kind of lean toward putting paste on all the pads (for better mechanical stability and levelling) but I'm no expert on such things

[untold moat]

Yeah, I have not been able to find anything on google for these tricky types of components

The closest I have been finding is bga component soldering

Since that (should) be similar in how it works

Pretty much anything with those solder balls on the bottom

The only other idea I have to make it simple to place would be is to make an aligner of some sorts

Like a 3D printed jig that matches the outline precisely

Then there would be a very low chance of failure

[supple pollen]

A bunch of guidance here (specifically for QFN and SON packages, but referenced for other modules): https://www.ti.com/lit/an/slua271b/slua271b.pdf?ts=1652254845469

They get very detailed in the recommendations for some modules, one specifies "pads 56-63, 79% solder coverage, pad 55, 77.5% solder coverage". Wow.

That's for a 0.125mm thick stencil, specifically

Stuff like this actually happens, happily

[untold moat]

Ok so it is more forgiving than I thought

I was thinking if it is slightly out of line the whole thing would fail

[limpid nest]

Also a good representation of me putting my socks on

[supple pollen]

It's cool to see it happening: https://www.youtube.com/watch?v=_5lksMvmqQc

[untold moat]

When is an external analog to digital converter a better option than using an analog multiplexer

[unreal flax]

One reason would be that you need higher resolution than the internal ADC can provide. Or if you need to simultaneously sample multiple channels instead of doing them sequentially. Or if you need to sample a differential signal.

[untold moat]

At what speed is that beneficial, I intend to use a 96MHZ MCU in my project.

[unreal flax]

Speed is kind of orthogonal to those considerations.

You might be speed-limited by the internal ADC, or by the communications bus (SPI, usually) to get the readings from the external one.

[untold moat]

I intend on reading from multiple analog sensors, but unfortunately, the mcu does not have enough analog channels

Which is why I would need to use either an External ADC or Multiplexer to do so

How much faster would reading channels all at once be compared to reading them sequentially

[unreal flax]

It wouldn't really be any faster, since you'd still have to read out all of the data the same way. It's just that some applications care about exactly when the reading is taken, instead of having data that's misaligned in time.

[untold moat]

Ok, so a multiplexer would be better for my usecase then.

[unreal flax]

Yeah, if the performance of the internal ADC is otherwise fine for your needs, a multiplexer would do the job.

[untold moat]

Yes, the internal adc is 14 bit

Which is more than enough

[steel valve]

Speaking of which

What happens when a potentiometer goes through another half of a voltage divider

Essentially, I want to put both my half voltage-dividers and full voltage-dividers in the same multiplexer

At the output of the mutiplexer is one resistor

Which does the voltage dividing for the half voltage divider sensors

[unreal flax]

That would cause a nonlinear distortion in the potentiometer mapping, since you essentially have a three-legged divider between the two halves of the pot and the extra resistor. It's something that you could potentially correct for in software, I think, but it's not something that would be typically done.

[steel valve]

So, bad idea basically

[unreal flax]

Easier to avoid if you have the option, yeah. It would probably be slightly simpler if you left one side of the pot disconnected and just used it as a rheostat instead.

[steel valve]

Would it still function the same

[unreal flax]

No, it would be different because of the extra resistor, just easier to correct for in your code.

[steel valve]

No as in using the potentiometer as a rheostat (two pin)

And then connecting it to the mux

Or did you mean that

[unreal flax]

Yes, that's what I meant. The readings would be different than a normal potentiometer, just less complicated than the three-sided divider scenario.

[steel valve]

Ah

I am trying to make cutting off the sensors in the pcb simpler than doing a loop-de-loop

And then having a mosfet

[unreal flax]

The rheostat mode would do that, at least, since no current would flow through the pot when the mux had it de-selected.

[steel valve]

In order to interface this with a microcontroller, supposedly I need to use a pull-up resistor on STAT (so that I know it is charging, when it is done, bla bla)

https://cdn.sparkfun.com/assets/learn_tutorials/6/9/5/MCP738312.pdf

How do I figure out what value?

[unreal flax]

Are you using it with a LED or just as a GPIO input?

[steel valve]

A GPIO input

I use an RGB LED so I am using it as GPIO

[unreal flax]

The easiest would just be to use the GPIO's own internal pullup setting, then. But if you want an external pullup, 10k is typical, or higher if you're especially power-sensitive.

[steel valve]

Do most gpio have internal pullups?

[unreal flax]

Most, yeah. It's not guaranteed, but it's a good bet if you're using a mainstream MCU.

[steel valve]

If a mcu has an internal pullup

Is there any benefit to having an external one

Just power efficiency?

[unreal flax]

The internal pullup is generally too weak for things like I2C buses, so you'd want an external one in that case, for example.

[steel valve]

Do I need to use 3.3V from an external USB to TTL if my project is always running

[supple pollen]

Generally not even that, the internal pull-ups source about 75µA for many common chips

[supple pollen]

I'm not sure what you're asking. The old TTL logic family runs on 5V.

[steel valve]

Most modern external USB to TTL have an external 3.3V/5V Jumper

Do I need to use them if my pcb is already powered

[hidden night]

I think that’s for setting the logic level voltage so you’d probably want to match it up with whatever your board uses. That’s how it works on the one I picked up

[steel valve]

Well the thing has a jumper to set the logic level voltage

Which wouldn't that mean if my board is already running at 3.3V and is powered I wouldn't need the USB to TTL 3.3V external jumper

[limpid nest]

re: thermals. My inclination is the 4 X (Smaller thermal size) is better than 2 X (Larger thermal size) if the width of the smaller thermals is such that 4 x (One Width) is greater than 2 x (One Width) for the larger size. Does that make sense and is that true?

[steel valve]

If a MCP73831 has no LiPo battery connected and is being powered, what happens?

Does it just make a 4.2V output anyway?

Also, what happens if a LiPo battery is connected to a MCP73831 and the MCP73831 is being powered, and the LiPo is connected to the rest of the pcb?

[supple pollen]

Some of those adapters have two jumpers, one for logic level and one for supply voltage. Since your board is already powered, you don't need the supply voltage, so you'd just set the logic level correctly.

[supple pollen]

The ones I've seen supply some voltage, then see the voltage go up, so they shut off thinking the (nonexistent) cell is fully charged, but then the voltage goes down, so they switch the voltage back on, and just oscillate like that.

[steel valve]

So would I need to modify anything to make it so that the PCB can be charged while running

The PCB never turns off, it just goes to sleep

Pretty much.

USB C (5V) -> MCP73831 (4.2V) -> LDO <- LiPo (4.2V-3.2V)

Would it be a better idea to put in a switch instead

So that when USB C is applied, it uses the 5V of that to power the rest of the pcb

Rather than the LiPo

So it isn't being charged while being used

[supple pollen]

I really don't know how to keep a cell charged while it's powering something. I might use steering logic to power from USB when it's available and charge the cell, and run from the cell when USB voltage isn't available.

[steel valve]

What does this symbol mean

(D4 and D3)

Are they schottky diodes?

This is apparently a voltage select circuit

[inland jungle]

yeah, schottky diodes

used because they have a lower voltage drop typically than normal diodes

[steel valve]

How does this circuit work exactly

[inland jungle]

VIN gets powered by V_USB or V_BATT, but you can't backfeed the voltage from V_USB to V_BATT or vice versa...keeps your battery from frying your USB port, for example

[steel valve]

Wouldn't that make both USB and LiPo able to power the circuit simultaneously

I assume a mosfet is the better option then if I only want one powering everything at a time

[distant raven]

It is

Schottky diodes are great and all, they collapse and recover very quickly which is great for ESD protection on usb lines, but it’s suboptimal for voltage selection. A p-chan mosfet is much better

Adafruit does this on all their current feathers

And more importantly what we’re seeking to block is sending current the wrong way on the USB line since when USB is connected we are very likely charging a LiPo, the mosfet acts like a diode in those cases. It’s just better overall

[steel valve]

What's the schottky diode for in the adafruit circuit

Since it has a mosfet, wouldn't that make the schottky useless

[distant raven]

Still ESD protection

Plus current protection as well. Limits the overall draw of the feather

[steel valve]

Ok so its best to add both in a pcb like that

[distant raven]

Yes

[steel valve]

USB C PD can request a specific voltage & amperage, from my understanding this is configured by the CC1 and CC2 pins

[distant raven]

Yes

[steel valve]

How do I figure out what resistor to use?

[distant raven]

Typically you put 5.1K resistors to ground

This comes from the usb c spec

[steel valve]

Is USB C minimally 5V? Or can usb pd set it to be lower

[supple pollen]

Nope, the Schottky protects from Vbat going to Vbus, the MOSFET provides a low-loss connection when Vbus is unpowered.

[steel valve]

So it's preventing reverse voltage

Why is a resistor used on EN?

Does it not work with V going there normally

[limpid nest]

What's the pt of the 100k to ground?

[steel valve]

Yeah that too

Also, for a p-channel mosfet and shottky diode like the ones they use, what parameters do I need to use to get one that would work for its purpose?

[supple pollen]

That pulls the MOSFET gate down when there's nothing connected to Vbus.

[limpid nest]

Ahhh

that makes sense

[supple pollen]

That sets a default of "enabled", but lets an external signal override it if someone wants to.

[steel valve]

Oh I always want it running

So I would ignore the resistor

[supple pollen]

Yeah, you can just tie the enable pin high.

[steel valve]

And R12 is always required

[supple pollen]

Yes, otherwise the MOSFET will just have its gate float randomly when nothing is connected to Vbus

[limpid nest]

yes

[steel valve]

Also why is it 100K?

[limpid nest]

to limit current

note that there's a +5V or whatever it is tied directly to it

[supple pollen]

It's high enough to not waste too much power when Vbus is powered, and low enough to keep random noise from affecting the gate voltage.

[steel valve]

Ok that makes sense

Then, do the mosfet and shottky have specific parameter requirements?

[supple pollen]

Generally, yes.

[steel valve]

On mouser there's a crap ton of parameters that can be selected for them

[supple pollen]

The easy way out is to look up the parameters of the ones in the AdaFruit design and look for similar units.

That is assuming that your use case is roughly equivalent.

[steel valve]

What exactly do the parameters mean

And yeah both are wireless boards

That are bluetooth low energy

[supple pollen]

There's a decent writeup of how to interpret those specifications here: https://www.embeddedrelated.com/showarticle/809.php

For your use case, you can safely ignore some of them like minimum operating temperature and Gate Charge. The "Vgs th - Gate-Source Threshold Voltage" is problematic: it tells the gate voltage at which the transistor just begins to turn on. For switching purposes, this is not a particularly useful figure, you need the gate voltage at which the transistor is solidly on. But in general, lower threshold transistors will tend to be solidly conducting at lower gate voltages, so it's one way to sort through stuff before going into datasheets and looking at the conductance graphs.

[limpid nest]

Digikey often offers the "just turns on" and "fully turned on" voltages

[limpid nest]

Having a weird Fusion/EAGLE issue. I mirrored a part on my PCB but on my 3D PCB, all the silkscreen still appears on the non-mirrored side of the board.

Anyone have guidance?

Nevermind, it just decided to sort itself out

Weird!

[steel valve]

what about schottky diodes how do i pick the right values for those

[steel valve]

What trace width should I use for USB C VBUS?

[limpid nest]

how much current are you expecting?

[steel valve]

a very low amount of current

[supple pollen]

You'll need a reverse voltage of at least 5V and enough current to power your circuitry

[steel valve]

I am a bit confused on how USB C negotiations work, does setting it for 5V 3A for example make it so the USB knows that it can provide up to that amount?

[unreal flax]

You may need to clarify the question a bit. What is "it" in this context, and how is "it" setting things?

[distant raven]

USB PD is pretty complex and usually requires some knowledge of what kind of upstream USB PD supplies your connecting too. Usually because the max negotiable current is determined by the upstream CC pull ups, downstream pull downs for usb c cables

I shared a ST doc the other day that explains it pretty well

https://www.st.com/resource/en/technical_article/dm00496853-overview-of-usb-type-c-and-power-delivery-technologies-stmicroelectronics.pdf

Plus some software controls too

[steel valve]

How does routing capacitors work exactly, is it linear, where the trace has to go through the capacitor before reaching its destination, or does the trace just have to cross the capacitor when it is near its destination

[limpid nest]

how do you mean, through?

And it depend on what the cap is for

[steel valve]

Through the pad of the decoupling capacitor*

[limpid nest]

It has to touch the pad

[steel valve]

Does it have to touch the pad before it gets to the pin

Or I mean

Directly before the pin

[limpid nest]

The trace?

[steel valve]

Yeah

[limpid nest]

Yeah it has to touch the pad, or else it won't be connected

[steel valve]

no no no

ima think of how to explain