help-with-hw-design

[proud summit]

I'm trying to connect a USB C to a USB 2.0 port of a microcontroller(STM32H7)I'm not sure what to do with the "CC'' and "VCON" Pin. Any help or guidance is appreciated 🙂
NOTE: This is the primary USB port to flash the microcontroller with code

[worldly schooner]

For usb2, tie the corresponding data pins together to your D+/D-. This gives your usb c port its reversibility.
Each CC line should be pulled to ground with a 5.1k resistor.

[proud summit]

Thanks, what about the TX,RX,CC and SBUS pins ? Can I leave them unconnected ?

[worldly schooner]

Yeah, you only need those for usb3.

And VBUS is your 5v source, so whatever you need for that coming in.

[proud summit]

so what does USB_ID do ? I've seen USB2.0 designs that use USB_ID

[worldly schooner]

Are you planning to use any usb host functionality?

[proud summit]

No, dont think so

Its to program the microcontroller

[worldly schooner]

If not, don’t worry about it. USB micro uses that for OTG, which you won’t be needing.

[proud summit]

In your knowledge is it possible to program STM32H7 with a usb 3.0 connection ? Till now all the designs that I've come across use MicroUSB even though H7 supports usb 3.0. Is there a reason for this ?

[worldly schooner]

Cost haha

Usb2.0is a much easier and cheaper interface to design for

I’ve never seen anyone try to program with usb 3, simply because there really isn’t a need for that speed

using 2 or 4 data lanes only serves to increase complexity without any much benefit

Though I personally also have very little direct experience with stm mcus, to be fair.

[proud summit]

This board that I'm designing is actually a Flight Controller. I'm planning on using it for HITL simulation as well.

[worldly schooner]

Not familiar with either of those applications, but just in case one of us is misunderstanding something, programming is the process of flashing code onto a chip, not sending commands in real time.

[proud summit]

usually in a type C design are the two different TX and RX pins connected to the same UART bus or different ones ?

[proud summit]

its the only port that I'm planning to have on the FC. It'll be used for both Flashing the code and also for the serial output. just like any arduino would

[worldly schooner]

Not familiar with uart over usb3, but the two lanes are typically connected to independent pairs?

I wouldn’t know the proper SBU configuration to make that work, though.

[unreal flax]

I don't think those are UART signals at all, but high-speed USB lanes.

[proud summit]

yes, they are USB lanes. I just wanted to know about the RX and TX pins in general.

If I'm going to use USB C might as well take full advantage of it.

[hushed blade]

Unless I'm misunderstanding something here, you're conflating things. UART has actually TX/RX pairs. USB doesn't, it has a differential pair. There's no dedicated transmit or receive lines at all.

[unreal flax]

The extra super-speed USB lines are additional TX+/TX- and RX+/RX- differential pairs on top of the basic D+/D-.

[proud summit]

i think I din't make myself clear. I'm trying to use this USB to program a H7 and also use it as a serial output for data. just like any arduino with USB. I initially only planned on using the USB 2.0 features by just connecting the differential pairs to the microcontroller. Now I'm trying to make use of all the features that USB C has to offer(assuming that H7 supports all the features). That's the reason for my question regarding UART. I'm aware that USB differential pairs and UART pairs are not the same.

[hushed blade]

👍 I came into this conversation late, seems I misunderstood what was going on

[inland jungle]

I need to switch 24V, 50mA into 3 different high side lines with a common ground using 3.3V MCU I/O. Any recommendations? Optocouplers? High side mosfets? looking for low $, low part count

[distant raven]

Is it likely that a comparator would be able to detect a very sharp voltage pulse if it was sufficiently high enough above the threshold?

The pulse would be more like a spike, very steep edge

[unreal flax]

The datasheet will usually specify that sort of thing, with like a minimum amount of time for the voltage to be above the threshold.

[distant raven]

One that I’m looking at says 16ns logic to data time

I’d have to measure the pulse width to see if it’s even that wide

I might just remove the input stage that converts rising/falling edges to pulses

[dry bough]

https://cdn-shop.adafruit.com/product-files/4960/4960_SK6812MINI-E_REV02_EN.pdf

hi! this image on the datasheet is incorrect for the sk6812mini-e

[distant raven]

Make sure you don’t compare WS2812 and SK6812 footprint

[dry bough]

seems like 欧思 made an oopsie

[dry bough]

does the ws2812 have a similar physical design?

[distant raven]

Sk6812 footprint rotate the ws2812 footprint by 90°

[dry bough]

yea but this isn't a ws2812, right? (im not too sure sorry)

[distant raven]

I believe clockwise. Very very similar but deceptive

[dry bough]

but its from the same datasheet

and the pinouts are clearly wrong

how can the chamfered lead by both gnd and dout

unless im missing something really basic

[distant raven]

Probably out the chamfer on the wrong pi

[dry bough]

what does that mean haha sorry

[distant raven]

Either way, general pinout is the same. Go by the notch on the package, not the actual footprint you solder to

Notch always points to ground for sk6812

[dry bough]

i understand that, but the datasheet is wrong then

shouldn't somebody change it?

@distant raven not changing it would only lead to more people being confused

[distant raven]

Adafruit doesn’t make the data sheet

[dry bough]

no i understand that

perhaps a notice or something

[distant raven]

And usually most people look for index pin marking on the physical package

[dry bough]

im just making suggestions

[dry bough]

the "most people" you refer to are people who know the basic design principles of the sk6812, what about people who don't know?

[distant raven]

I don’t necessarily think it’s that much of an issue,

[dry bough]

alright then, ig more people will just figure it out the hard way 😮‍💨

i mean idrc, i just don't wish other people to run into the same issue i did

and be confused asf

[distant raven]

I mean, it’s kind of clear from the data sheet that there is a notch on the package. Land pattern is kind of subjective as many people don’t follow the suggestions to a T

[dry bough]

thank you for the info skerr!

[distant raven]

@unreal flax you mentioned you worked on VLC systems before, yes?

More long range if I recall right

[unreal flax]

Yep, we did 100km links between high-altitude balloons.

[distant raven]

Okay, so just on a general get your thoughts.. right now I’m trying to make a Python test library to read a SPI ADC but I don’t think it’s fast enough (Python reading the adc) to extract the datastream I’m transmitting. I’m considering a comparator with a narrow propogation delay to just bypass using an ADC. Does this seem like a reasonable idea?

I know I’d ultimately be limited speed wise by how fast the comparator latches, and how quickly it sets data to the value of the latch. And the overall delay of the comparator.

But I’m not really concerned with super fast speeds

Just 10-100kbps

Maybe 1Mbps at the most

[unreal flax]

Yes, that sounds quite reasonable. Many optical receivers are basically an AC-coupled photodiode (to remove the background light) with an op-amp acting as a comparator to convert the signal into a binary stream.

You do need to design your bit encoding to avoid long strings of 0's or 1's, though.

[distant raven]

Right, I figured that would be the case.

I’m just a little lost of using an ADC to read the bitstream when the filtered Photodiode output is sufficiently square

[unreal flax]

Yep, almost all those ADC bits won't be useful, as you'll just end up thresholding the signal in software.

[distant raven]

If I simply pass it into a comparator I can more quickly read without needing any significant conversion

The comparator could compare smaller voltage levels and translate those into system level logic

Thanks for you help, I needed the sanity check.

[unreal flax]

No problem, any time!

As a first pass you ought to be able to more or less just feed the signal right into a UART pin, to get the free oversampling and bit synchronization.

[distant raven]

That’s a great idea

I’ll give that a shot. The transmitter side is easy, the receiver has been illusive for me

But I think I’ve got some good ideas now

And this will make it easier too when I move all this to an fpga

Basically just need to proof of concept this on a microcontroller first

[hollow steeple]

Hi so I tried to make a custom RP2040 board and it appears to power up ok, and it looks like it does try to talk to the flash chip, but despite grounding the flash select on power, it still never shows up as a drive when plugged into the computer. Not really sure what to investigate next.

[distant raven]

Do you have an oscilloscope?

Usually the crystal controls whether USB gets set up

[hollow steeple]

I do, this is the project that made me finally get one since I couldn't figure out what was wrong

[distant raven]

If you can probe the crystal to see if it’s generating a 12MHz clock

[hollow steeple]

I did worry about that since I deviated from what the HW doc suggested for the crystal, probed it and saw much wavy, didn't think to see if I could actually measure the clock, let me try that.

[distant raven]

Usually just probe the return capacitor on the XOUT trace

You should get like 12MHz +/- like 10-30ppm depending on the crystal accuracy

If you don’t get roughly 12MHz, you might check the connection/solder joint

[hollow steeple]

I'm showing 12 MHz

[distant raven]

Okay, that’s good

[inland jungle]

Here's an example schematic from a transmitter/receiver from Mikro that I cribbed for some tests:
https://download.mikroe.com/documents/add-on-boards/click/fiber-opt/fiber-opt-click-schematic-v201.pdf
receiver uses two op-amps in series, first one sets sensitivity and 2nd gets driven to the VCC or GND rail for a 'digital' output

might have to tweak the op-amp setup if it doesn't have enough bandwidth for your needs

[distant raven]

I’m using a two stage op amp from a design derived from https://OpenVLC.org not terribly far off from that

I’m going to spin a few designs to test and see which ends up being better

[inland jungle]

I couldn't make much out of the schematic on their github, do they really have a transmitter/receiver circuit somewhere in there?

[supple pollen]

By and large, it is not a good idea to use an op-amp as a comparator, and as the frequency goes up, it gets even worse.

[distant raven]

They provide Altium files

[supple pollen]

Oh good there's an online viewer for those of us who can't run Altium

[distant raven]

Basically the OpenVLC design converts the filtered photodiode digital looking signal into pulses that are read by an ADC

It’s an interesting design

[inland jungle]

What's a better high speed comparator then? A Schmitt trigger logic gate?

[supple pollen]

There are purpose-made high speed comparators (some CPUs have them built in as well). A Schmitt trigger gate is less versatile: it acts as a comparator with a single input with hysteresis and a fixed set of trigger points, while a dedicated comparator has two inputs like an op-amp so you can configure the threshold (and hysteresis, if desired).

[hollow steeple]

Would adding a little bodge to try to spy on old school USB data lines with a scope introduce too much noise?

[knotty tiger]

can you describe a bit more? and what speed USB?

[hollow steeple]

Full speed (RP2040). Thinking maybe being able to look at the data lines might provide a clue to why it won't show up as a storage device.

But the internet seems split on how strict you need to be with USB

[knotty tiger]

it might work, if you manage to keep the loading balanced (one probe on each of D+ and D-, assuming a 10MOhm "10x" probe, and scope set to differential?)

ideally, you'd use a specialized bus analyzer, but those are really expensive. turning on some kind of verbose logging on your host device might be easier

[hollow steeple]

Oh that's a great idea, try to see if it even shows up as any kind of device on the host.

AFAICT host seems to have no idea a USB device was connected

[knotty tiger]

what's the device, exactly?

[hollow steeple]

Custom RP2040 board I tried to make. Powers up ok, appears to try to talk to the flash, crystal looks ok, but never shows up as a drive when connected to a computer.

[knotty tiger]

so not running the UF2 bootloader? is BOOTSEL set up correctly?

[hollow steeple]

AFAIK, should I post some schematics?

[knotty tiger]

yeah, probably would help

[hollow steeple]

I've tried with and without R2.

I've since removed R8

[knotty tiger]

you said the crystal oscillator starts up? i think the datasheet says that's initially off at reset. if so, maybe enough of the bootloader is running to start it up?

[hollow steeple]

Looking for a mention of the startup timing of it, but yeah I've measured it at 12MHz so that part seems like it's working.

Ah yeah it seems to say so in the full datasheet

[knotty tiger]

i don't know what the PLL lock detection parameters are like; maybe the clock isn't stable enough for it to lock, and it won't enable USB because of that? are either D+ or D- getting pulled up by the RP2040 during the reset sequence? (low-level full/low speed attach signaling, i think?)

[hollow steeple]

I haven't looked at D+/D- yet, hence the original question 🙂

I can try to measure the clock again to look for variance, the scope was always reading 12MHz but some of other numbers did vary slightly.

[distant raven]

RP2040 uses usb1.1 so you’re not really going to hurt it much sniffing the lines

[knotty tiger]

yeah if the host isn't seeing it at all, it probably won't hurt to see if it's getting as far as turning on a D+/D- pullup

[hollow steeple]

huh before I was seeing a solid 12, now it's bouncing between 11.99 and 12.01

[distant raven]

That’s within reason

Crystals have natural variance

[hollow steeple]

For measuring the data lines, would the expectation be they stay at the high voltage or am I looking for a blip?

[unreal flax]

Is that 1k resistor on the crystal in the reference design? That looks unusual to me at first glance.

[hollow steeple]

It is

The second consideration is the maximum ESR (equivalent series resistance) of the crystal. We’ve opted for a device with a maximum of 50Ω, as we’ve found that this, along with a 1kΩ series resistor (R5), is a good value to prevent the crystal being over-driven and being damaged. You may have to adjust these values if you choose to use a different crystal.

[distant raven]

Next thing would be checking the solder joints on the SPI Flash

[hollow steeple]

I've seen it try to message the chip, I think that part is ok

[distant raven]

I’ve found if SPI flash isn’t properly mounted it can cause USB issues

It’s weird all the little things I found that can cause it not to enumerate

[hollow steeple]

So plugged in and "idle", I should be seeing 5v across vbus, D+ and D-?

'cause I've got nothing on D+/D-

[knotty tiger]

one of D+ or D- should be pulled up to 3V3 when the bus is idle

[hollow steeple]

Seems like the problem is related to that then, 0v on both

[knotty tiger]

yeah, it might be time to break out a debugger, then

[distant raven]

Time to check to make sure the usb connector is properly seated too

Sometimes hard to tell

[hollow steeple]

I think I checked that originally (I set this project aside from a while when it wasn't working), but I will check again.. I've definitely fried a chip before because the connector was offset and I didn't realize.

[distant raven]

😅

[hollow steeple]

Just hobbyist things 😛

[distant raven]

I’ve fried a many chip

[hollow steeple]

In my defense, it was the first chip I'd managed to kill, I was pretty happy with that

[distant raven]

Don’t worry, it’ll climb with time 🙂

[hollow steeple]

Connector seems ok, pins have continuity with the resistors, no bridges

[hollow steeple]

Unfortunately I don't have one.

[distant raven]

Did you use NOR SPI flash?

[hollow steeple]

Yes, W25Q128JVSIM which should be just a larger version of the reference design used chip

[distant raven]

oh 128.. hmm I think the RP2040 can only support up to like 32MB SPI Flash

[hollow steeple]

128MBit, actually sorry it's the exact chip used in the reference design

The device chosen here is an W25Q128JVS device (U2 in the Figure 7), which is a 128Mbit chip (16MB).

(Also it says 16MB is the max it supports)

[distant raven]

yeah, 16MB

[knotty tiger]

hm, according to my reading of the boot rom source, if CSn is low, it shouldn't probe the flash before initializing USB

[distant raven]

for what it's worth, the Pico only has 2MB last I checked. because when I got samples for the RP2040 from RPi when it first came available, they gave me (and others) the same flash chip used on the Pico

which is the W25Q16JVUXIQW

tiny chip

[hollow steeple]

Yeah I wanted to be able to store a couple small sound files without a SD card on-board, so I maxed it out.

[distant raven]

Oh I see

hmm

[hollow steeple]

The adafruit and sparkfun ones have larger flashes onboard IIRC

[distant raven]

yeah, they put 8MB

there might be one with 16MB

but I think most use 8MB

anyway, only other issue I could see is that the RP2040 isn't making good electrical connection on all it's pins.

[hollow steeple]

D+/D- signal integrity probably would be an issue after the point I'm at, yeah? I realized after I wasn't very careful about following the HW design guide for that

If so that seems like, since nothing seemed to stand out on the schematic, that I'm probably down to yeah either reseating the RP2040 maybe, or trying to assemble a new one.

[distant raven]

yeah, essentially

[knotty tiger]

it would be good to know if someone with more experience can confirm whether the bootrom is supposed to probe the QSPI at all before initializing USB if CSn is held low. i think it's not, but i don't have one to poke at, and have just been reading the source code

[hollow steeple]

What difference would that make?

[distant raven]

so on boot, the bootrom does check SPI flash for data and/or a program to execute. if nothing is executable in flash, then CSn is pulled low.

unless the CSn is manually pulled low on powerup

[knotty tiger]

if you're not seeing USB being turned on, and you are seeing QSPI being probed, and if i'm reading the bootrom source correctly, it means QSPI_CSn is being read as high by the bootrom

[hollow steeple]

I'd have to recheck QSPI being probed. I know I've seen it, I don't recall if that was with or without the bootsel condition.

[knotty tiger]

though it should also give time out on probing the QSPI at some point and fall back to the UF2 bootloader, unless the flash contents somehow pass the CRC check

[distant raven]

but a fresh chip would always cause RPI-RP2 to show up

there shouldn't be anything on a fresh flash chip.

[hollow steeple]

Oh interesting, didn't realize I could ignore it altogether at this stage.

[distant raven]

on initial powerup for the first time, you will see RPI-RP2 without pressing boot

[knotty tiger]

so something is preventing USB from being initialized, then, and i haven't found the code that does that yet

[distant raven]

same thing if there isn't anything on flash if you used the flash nuke uf2

[hollow steeple]

Ok I did see that on a new official rpi pico, but wasn't sure

[distant raven]

bootrom checks crystal drive strength and speed to set USB PLL before initializing USB and USB MSC

[hollow steeple]

I guess I'll try to reseat and if I get nothing, assemble a new one.

I have one board I did just the battery bits to test those separate, I can finish building that one

[distant raven]

additionally, you could still technically get it to pull up without SPI flash too, you can program it to execute from RAM

[knotty tiger]

i guess if the USB PLL isn't locking for some reason…

[distant raven]

but yeah, reseat first. if you have a digital microscope, you can inspect the joints

[hollow steeple]

I keep eyeing them, but haven't gotten one yet

[distant raven]

yeah, probing the pad of the RP2040 footprint wouldn't be enough to verify continuity unfortunately.

it's why I wish they made it a TQFP instead of a QFN

[hollow steeple]

Yeah, that's been a challenge as I've gotten more ambitious with part selection. I went from THT to SOIC only initially.

[distant raven]

I probes right off the edge to spike bit of the receiver before the ADC, read it as a digital input and inverted it. Beautiful signal.

[hollow steeple]

Alright well I screwed up the reseat so I'm going to do a new board, thank you both for your help; I'll check back when this one works or fails.

[distant raven]

Comparator will clean it right up and help for further distance transmission. I’m glad to have had the realization and asked about it. So much gained from an itching question

[hollow steeple]

rofl it worked

Thanks again argonblue and skerr. Brand new build shows right up, guess I did something wrong with that first build, but it's my most complicated design yet so I just assumed I did something wrong design time.

[distant raven]

Great work!

[inland jungle]

So, what's your final circuit?

[distant raven]

photodiode -> 4.25MHz GPBW filter -> edge to pulse amplifier -> comparator -> digital input of host

[lone basalt]

i'm having some issues tracking down my drc error
https://cdn.discordapp.com/attachments/914235414041202758/1000506514827641002/unknown.png
https://cdn.discordapp.com/attachments/914235414041202758/1000506515125456906/unknown.png

[distant raven]

what happens if you click on the error?

some CAD programs will show you where it is if you click on the error

[lone basalt]

I tried, it does nothing

[distant raven]

hmm

what program are you using?

[lone basalt]

Easy EDA

[distant raven]

oh, i'm not super familiar with EasyEDA

[lone basalt]

I think it's only been happening after i put the gnd traces on the bottom lay

but without that the grid lines overlap

feel free to ping me on reply

[inland jungle]

are you using a dedicated comparator IC, or a regular op-amp as a comparator? When I looked at comparator pricing, they seemed pretty expensive

[hushed smelt]

oh i'm using easyeda right now actually

measuring trace thickness of the bluefruit sense and laughed out loud. lady ada did the trace width in pi. 3.14 😛

[distant raven]

Yes, for my purposes a general purpose one used for coax signals with a 40ns delay is more than fast enough

It cost like.. $2.30

[hushed smelt]

my easy eda defaults to 10 mil and i'm like why is it so big, how big is it on the sense. so apparently i can go as low as about 4 mil and be ok.

hmm what are you doing to get errors in easyeda? granted i just loaded mine up for the first time in 2 years today. i'm not getting any errors even with auto router.

i tried using Eagle. ok i gave it a fair shot, i wanted it to work, would have made my life easier to use the same program as adafruit but halfway through working on a board it crashed. all my work was lost. it doesn't even say i have a recently opened last file.

brand new install of Eagle and it crashed. i've had issues with easyeda sometimes but it's never lost all of my work.

[distant raven]

Huh, I’ve not had many issues with eagle crashing

[hushed smelt]

sigh, i thought const name= arg.pick etc.. was the error he was getting. lol

[distant raven]

But I use it on Mac which probably makes a difference

[hushed smelt]

yup it closed the workspace i was using and then the background program whatever that thing is too. straight to desktop, no errors, it was like i closed it myself.

hmm does ctrl+c close the program or something?

i was attempting to duplicate a footprint.

[distant raven]

Alt + f4 does lol

Alt C might close programs

[hushed smelt]

@lone basalt in easyeda if your rat lines aren't exactly in the center of a via or the via is off center somehow it'll throw a DRC error.

one single connection being a hair off is frustrating, been there.

it doesn't tell you which one is having an issue? :/

if you're doing manual routing avoid right angles if possible. i think lady ada did a video on that. you want electricity to flow like water through plumbing.

[distant raven]

Right angle trace is a myth

It’s more for 10GHz+ signals

It’s less mechanically sound though

IBM researchers debunked it 🙂

[hushed smelt]

yeah had something to do with trying to avoid an antenna situation and adding through hole via's to knock down reflectivity... something like that.

[distant raven]

For the average hobbyist, it won’t matter. There’s a lot that goes into what makes something more likely to be an antenna than you’ll need to worry about. Right angles don’t necessarily mean something will reach internal reflection thresholds to be an issue as there are rounded and triangular antenna shapes too.

Most of the issues you see with right angle traces is it gets peeled up easier during processes like applying solder mask

That and they don’t look as good

Lol

[hushed smelt]

i haven't given up on my mistake of buying an soic8 chip

[distant raven]

Soic8 is a pretty good footprint

Nice and easy to solder by hand 🙂

Easier than this anyway:

[hushed smelt]

that little uson8 chip needs to turn into an soic8 chip

[distant raven]

Oh hehe

[distant raven]

USON-8 are super fun

[distant raven]

Retrying my smöl part picture

[hushed smelt]

yup that's about right

little (*&%@#$%$%

[distant raven]

And I’m going to solder it lol

Once I order boards that is

[hushed smelt]

ha yeah ok gl with that

by hand or hot air?

hot air would probably send that thing flying

[distant raven]

Hot plate

[hushed smelt]

yeah i was thinking about it but i can't do that to my entire feather sense

and it's right next to the nrf52 chip, gotta pray no solder seeps under it.

what are you putting it on?

[distant raven]

In between C1 and C2

[hushed smelt]

oh my god

https://tenor.com/view/good-luck-gif-24090093

my problem doesn't seem so bad now, thanks

[distant raven]

I’ve got to make a variant of this that take up to 24V and breaks it down to 5V 5A that goes to the 3.3V regulator pictured and the JST connector pictured

[hushed smelt]

treehouse terminal?

sounds fun

what kind of module are you using i don't know that one with the pads in the center.

[distant raven]

ESP32-S2 Wrover

[hushed smelt]

oh and a tft. plant monitoring?

soil stuff?

[distant raven]

Grow terminal

[hushed smelt]

hydroponics?

never heard of a grow terminal

[distant raven]

Controls day/night cycles for indoor agricultural setups

Monitors the environment

Provides real time statistics

And will work with an app to see everything

[inland jungle]

it's weed, isn't it 😛

[hushed smelt]

sounds cool. great application for technology. work smarter not harder.

lol

i wasn't going to say it but why not

[distant raven]

As well as battery backup to provide a limited amount of grow time/monitoring in case if power outages

Sure, you probably could

[hushed smelt]

i've done hydroponic tomatoes and lettuce. lettuce does really well with hydroponics.

[distant raven]

I was thinking more local produce

[hushed smelt]

any of the leafy greens like spinach, kale, etc.. do really well.

[distant raven]

With the American south west turning into an even bigger water wasteland, localizing food production in an environmentally friendly way will be important

[inland jungle]

I'm doing some deep tech startup development, and about half of the specialized equipment we get is either used a lot in cannabis production, or brewing beer. We're always waiting for the feds to bust down our doors based on our equipment purchases 😛

[distant raven]

Hahaha

I’m not very familiar with deep tech

[hushed smelt]

true, i've seen the pictures of lake mead slowly turning into a ditch instead of a lake. humanity is growing so fast we're drinking rivers dry

[inland jungle]

eh, just means hardware development of technology that isn't currently on the market

[distant raven]

Ah

[inland jungle]

takes more time and money than usual startup timelines

[distant raven]

Yeah, I’m kind of in the weeds of development for my product line. It has to scale well and that’s hard to do with hardware

Especially for connected agricultural tech

[inland jungle]

yup, manufacturing scale up is tricky

[hushed smelt]

make everything modular.

easy to say, hard to do.

[inland jungle]

and modular only takes you so far, you can scale linearly with modularity, but it's hard to do orders of magnitude scale up...x10, x100, x1000

[distant raven]

Initial product with use ESP32-S2/S3 but I will need something far more capable for large scale warehouse based applications

[hushed smelt]

welp i'm gonna go eat some dinner and watch some dystopian apocalypse movies to prepare for the inevitable. hmm maybe interstellar, that dust bowl part was on point.

[distant raven]

Trying to avoid PLC geared applications because it tends to be so single ended

And also very monolithic

[inland jungle]

yeah, at the moment we don't have a dedicated embedded guy so all the R&D for sensors/comms/control falls on me

[distant raven]

Sounds like a good time

I was at a grill startup from May to June doing Embedded

[inland jungle]

it's fine for playing around, but when it comes time for an actually deployed system we need to get a bit more serious

[distant raven]

But investors went cold so series B was out of the question

So I got laid off along with the rest of the dev team

[inland jungle]

investors are taking a collective #@$@ over the recent market meltdown

they stopped talking to us about the same time

[distant raven]

Yeah, except Menlo Ventures apparently is trucking straight along

But no one will invest in hardware or physical product companies right now because of the fears over a recession

[inland jungle]

yup

[distant raven]

So bootstrap it is for my agrotech startup

[inland jungle]

they also want to keep money on hand incase their current investments need additional funds to keep going

[distant raven]

Yeah

I’m working on two startups at the same time, and going back full time to the job I left to work for the grill company I was at

Good times

One is a VLC comms startup and then Grew Labs, my personal agrotech startup

Both hardware/embedded

[inland jungle]

what stage is the VLC startup at?

[distant raven]

We’re pitching seed right now, trying to get some defense research contracts right now

Defense isn’t ideal but if you can get it, it usually pays well enough to get off the ground

[inland jungle]

yeah, we're almost entirely gov. funded, SBIRs and similar

[distant raven]

We started with a nasa funded SBIR

But didn’t get phase 2

We’re considering pitching for another SBIR

[low anchor]

Could I use a MOSFET to control 12v @ ~65ma using a 3.3v GPIO pin?

[unreal flax]

Yes, you could do that with a P-FET, I believe. You'd want one described as "logic-level" to have a low threshold gate voltage for 3.3V control.

[heavy jasper]

You'd either want an N-FET on the low side, or a P-FET + N-FET combo for a high-side switch

[unreal flax]

Thanks for the correction. There is really something about the transistor types that my brain absolutely refuses to cache...

[heavy jasper]

For the high-side FET, a P-FET is on with large enough negative gate-source voltage. Source is 12V, so gate needs to be pulled low to turn the P-FET on. You can do that with an MCU using an N-FET (which helpfully also inverts the signal so driving the MCU output high still turns the power on), then the resistor from 12V to the gate of the P-FET keeps the switch fully off when the MCU doesn't specifically turn it on.

The pull-down resistors on the inputs of the N-FETs are just to make sure the load is turned off when the MCU is in reset (and generally its outputs are in some high-impedance state).

For the low-side circuit that directly keeps the output off, for the high-side circuit it prevents the N-FET from pulling the P-FET gate low, keeping the output off.

[lone basalt]

do you think i can ignore it, it doesn't seem too significant or obvious after checking every thoughhole on the board

[proud summit]

I'm designing an FC based on the STM32H7 series. I'm trying to have 8 PWM outputs(4 Channels per timer). Does the below configuration look good ? Is there anything I should change ? Any information related to this would be helpful 🙂 I've attached the timers section from the datasheet as well.

[hushed smelt]

@lone basalt i wouldn't ignore a trace connection drc error. it's not a warning, it's an error, and a significant one, do not ignore it. ask for help in easyeda forum.

Just about finished with the uson adapter board.

the soic took up more room than i thought. had to make adjustments to allow a stackable header to be used on the feather, it takes up more space than just a header pin.

using 6 mil traces which is just enough to auto route, feather sense uses 3.14 mil so the extra width should do ok for the additional trace length.

@unique patio where there's a will there's a way. might be circling back to that PR after all.

[distant raven]

Did you verify the footprint of the uson8 chip?

Sometimes they have a thin narrow ground in the middle of the footprint

[hushed smelt]

i used the datasheet, short of unsoldering it i won't know.

hmm those pads are awfully close together, might reduce them a bit.

was planning on use regular flux, should i get smd paste instead?

i've never used the stuff.

[distant raven]

Flux, paste, heat

[hushed smelt]

footprint lines up, can import eagle files into easyeda now. whew that makes it much easier.

it's going to overhang the feather board a little 🤷 don't care as long as it works.

[distant raven]

Okay cool, looks good

[hushed smelt]

tried ordering with oshpark but pcb under minimum size. went with jlpcb. $23.74 for 5 boards.

probably take a month on the slow boat but w/e, i'm in no rush

[woven bluff]

is that a good place to put ground return via for SMPS ?

should I add more ground via near the input decouple cap (lower right)?

[hushed smelt]

bottom layer ground is green? you can via to anywhere you want on the gnd layer.

unless you're dealing with a component that requires a separate analog gnd or something

[woven bluff]

this is 2MHz

[hushed smelt]

above my head then

[woven bluff]

the layout is according to recommended by LT3467datasheet

[hushed smelt]

if your design has to work within a recommend layout then contact the company who made the layout?

[signal topaz]

are you using trough hole vias in pads?

[woven bluff]

yes, lower left pin has to be pulled high

is it bad to put via on pads ?

[signal topaz]

i mean from DFM (soldering) point of view using trough hole vias in pads is not good idea, unless those are filled and capped

[woven bluff]

I hand solder those

[signal topaz]

you probably will be fine then, but generally this is something to avoid

[woven bluff]

I mostly relay on surface tension of solder and my fingernail to keep component in place.

[signal topaz]

about the GND vias: personally I would just place vias for every pin, unless the vendor says to not do it. I do not believe in those single point connections for high frequency SMPS.

but you probably will be fine. Note that the other package for LT3467 has vias under the package in the recommended layout, probably as well due to thermal reasons as well.

[woven bluff]

I don't understand the "star ground" concept either.

Texas Instrument often tells you to use it

[signal topaz]

those techniques work good for low frequency, at high frequency it is impossible to steer return currents, especially on boards with thin dielectric

if datasheet says to do it, then do it. But very often it makes no sense to over engineer it. Also sometimes recommended layouts might be good for 2 board layout, but not optimised for multilayer boards

[woven bluff]

at least in SMPS I always use AD/LT/MI product and avoid TI. TI expect you to do error amplifier filter, slope compensation and feedback blanking etc by yourself.

[signal topaz]

why are you not using capacitor in the feedback loop?

[woven bluff]

I designed for 1A output but I only need 100mA. I don't think filter is required at this current. Feedback filter can impair regulation if not choose carefully.

[glacial gale]

so i was looking at filesystems and I was thinking if maybe specific filesystem ops like read, write, open, close could be accelerated through hardware (e.g. an onboard FPGA next to the ssd)

then you would write minimal drivers for the FPGA, and mostly program the FPGA itself somehow take read, write requests into some optimised sequence of instructions to the SSD microcontroller

is this an idea or am I high

[unreal flax]

Mostly high, I think. 😉 The SSD exposes a particular set of allowed operations, so the OS driver for it is already trying to do the best job it can of translating filesystem commands to the underlying block device.

[glacial gale]

right, but the cpu is doing a lot of the work of the driver right? For something like ext4 or any b-tree+ fs, you gonna have to send a req to the ssd, get back a block, read that with the cpu, determine the next block, send a req, etc. But instead, you could just send something like:

struct Request {
  request_type: ["read" | "write"],
  offset: int
  bytes: int
}

and let the fpga figure out

[distant raven]

Fairly certain cacheless SSDs kind of do this already

[heavy jasper]

The FPGA then still needs to do those read/writes to the SSD, which is the big chunk of the time. If you're specifically trying to reduce CPU resources so you can spend them on other things, dedicated accelerators for filesystem handling start to make sense (especially for compression, encryption, RAID, zfs, etc). But for more standard tasks the FPGA would likely just be a latency addition

[distant raven]

Not sure cacheless is the right term here though..

[unreal flax]

Or maybe the way to think about it is that 8-core CPUs that are mostly idle is common, but FPGA coprocessor boards are not, heh heh.

[distant raven]

Are FPGA regularly used in SSD memory controllers?

[heavy jasper]

Maybe for R&D, but no - commercial SSD memory controllers these days are primarily ASICs. They're coming back for some networking stuff, but again that's a lot due to the standards being a bit early and not completely locked down; at scale as soon as folks know what they want they tend to want to ASIC-ify it as soon as they can because FPGAs are power hungry and expensive, and even the mind-boggling costs of doing ASIC development become quickly worth it

[distant raven]

That’s what I figured

We use FPGA a lot in the work I do

For networking that is

[heavy jasper]

see e.g. Nvidia smartNIC / bluefield, Intel/Google Mount Evans, etc

[distant raven]

I am also planning of using an FPGA for VLC transceiver. A small one though because it’s not terribly data intensive yet

[glacial gale]

hmm i suppose the cpu does run at higher freqs and can do multiple io reqs at once. ASICs wont be programmable but vould maybe be modular so you can plug a new ASIC driver in

[bright thistle]

uhmmm... guess I need to think this a bit more through... 😅

wonder if I can make it with just two layers, or if I should consider using more

[supple pollen]

Two layers is cheaper and faster to fabricate, and easier to modify if that becomes relevant. I'll generally put a fair amount of effort into routing to get to two layers. That one looks totally doable, you may need a via or two, but that's going to be a lot cheaper than going to 4 layers. I'd probably increase the trace width too, it looks like there is room in most places. You could make it more compact as well and save a little money if you felt like it. It looks like it might route easier if the Pi connector was facing the other way but I'm guessing that would put the rest of the board pointing the wrong way.

[bright thistle]

pointing the wrong way is relative 😄

it's (mostly) for the sake of doing it, and because I like to have bundled wires, instead of loose ends, and also because I really like modular stuff, I feel it to be more "failproof" and repairable.

would putting the Pi connector on the middle be a good idea as well?, or would you keep it on the side?

[supple pollen]

It might be easier to route if you put the LED and button connectors on one side and the OLED on the other. Note that you're allowed to cheat by not connecting to every single ground pin, you can skip a few if they're in the way of things you want to route.

[hushed smelt]

yeah use via's to go straight to the ground plane.

[bright thistle]

ah, I see!
thanks!
I tought it'd be better not to have floating ground, TIL. 🙂

[bright thistle]

that's the plan, bottom is ground plane, top is 3v3 plane, and signals

[supple pollen]

You don't want to completely float, you generally have to connect to at least one ground pin, and more is better.

[hushed smelt]

if they're all going to share a common ground that's fine but if they require separate grounds to separate modules that would be bad.

[bright thistle]

nah, only OLED here is powered, the other two are signals from the other board, which has its own USB power

[supple pollen]

I'm not sure what you mean by "require separate grounds" here, but there's only one ground net on a Pi.

[hushed smelt]

let's say your pi GND pin is your main ground pin, you can via that straight to the board so the entire board becomes an extension of the pi gnd system

you can then via straight to ground from the other plugs which will immediate tie into pi gnd... if that's acceptable for your project goal.

some modules might require separate analog ground, more common with audio modules.

if they combine with a digital gnd it might cause noise

[bright thistle]

if it gets too complicated, I may even make a micro board just for the OLED power and "F a duck!", if I may quote Jumpin Jack Flash XD

[hushed smelt]

i'm sure it'll be fine

trace width looks a little thin, always go for the thickest traces possible. only use thin traces if you absolutely need to for space.

[bright thistle]

methinks, but do i really need a 3v3 plane for only one connector? wouldn't it make routing overly complicated?

[bright thistle]

I'm going with default settings for now, will thicken them as next step 🙂

(edited) (edited) (edited) yeesh... today I can't type...

[supple pollen]

I wouldn't bother with a plane for that.

[hushed smelt]

bad practice to via directly under a pin too. increases risk of a mechanical problem combining a through hole with a via. put via directly beside pins and use a short trace to the pin on the layer you want.

possible yes, people do it, not recommended unless you're in a real pinch for space.

[supple pollen]

With through-hole it's not a problem, the pad is on both sides to start with, so just connect up whichever layer is most convenient.

You can think of through-hole pads as big vias.

[bright thistle]

yep, noticed most fabs use via-ed TH pads anyway

[supple pollen]

I'm not sure what you mean by "via-ed". Do you mean plated through?

[bright thistle]

yes, sorry.

term slipped away

[hushed smelt]

pretty sure most of those traces running through the double sided pin connector instead of around are going to fail a drc check

[bright thistle]

absolutely, that's why I'm keeping this board last XD

[supple pollen]

You can generally pass DRC with a single trace running between 0.1" pins with reasonable pad sizes and 10/10 spacing.

[bright thistle]

closing for now, gotta prep for work.

cursed this costume of exchanging labour for currency, so to exchange it for goods and services!

[supple pollen]

I still exchange labour for currency, but since I work from home, I no longer wear the costume 🙂

[bright thistle]

Lucky B 😂

Mine is a tad difficult to have it remotely done, unless you can pilot a bot located 30k ft from the ground in a tin can 😂

[hushed smelt]

dynobot strikes again

New emoji for oshpark and boards in discord

[elder peak]

I am mostly visualizing you showing up to work as Dr. Strange. Don't correct my mental picture plz.

[supple pollen]

Well, yeah. This is me.

[hushed smelt]

that's cool!

dr strangebodger 🙂

[woven bluff]

you could use a rotating LED bar

so the magic can be displayed without the rim

[limpid nest]

If a chip, a FET driver in this case, has a V_IH min of 2.4V, I can safely drive it with 3.3V right? I wasn't questioning that until I saw that all of the datasheet's assumptions about use case are with 5V logic

[woven bluff]

almost all modern IC with digital inputs accept 3-5V

except those 12V logic devices

[limpid nest]

hm

Also if I'm driving a P-Channel FET, and I want my default behavior of the FET to be non-conducting, then I need the gate to be pulled high in most configurations, right?

[supple pollen]

I did buy one of those LED hologram generators, but it uses an undocumented proprietary format for its images, and the software is DOS-only so I can't run it.

[supple pollen]

Yes, normally P-channel FETs are driven with a gate driver or level shifter, unless they're switching the same voltage as the driving logic.

[hollow steeple]

Is it possible a boost converter would act like it had a short when enabled if there was no output capacitors or load?

[woven bluff]

which one is hot, inductor or transistor ?

[hollow steeple]

It's an all-in-one, so I can't tell

[woven bluff]

inductor and switch together ? Then it should be a complete module, which means the module is broken.

[hollow steeple]

So even if there's nothing on the output at all, load capacitor or load, it shouldn't cause a problem?

[woven bluff]

a complete product should of course works with empty load

[hollow steeple]

Ok, I didn't know if the capacitors on the out were more about smoothing or critical to function, so I probably messed up assembly then since its so tiny. Thanks!

[spice turtle]

The caps generall smooth things out, all you would get is a square wave if they were functioning correctly. If its hot, its probably broken. They generally run pretty cool

[hollow steeple]

It was brand new so I'm going to assume it was an assembly issue. I've used the IC in a prior design with success so I don't think it was a design issue.

I accidentally dropped it so it's gone forever, saves me from trying to figure out if I killed it

Welp as with everything on this board so far.. start over and then it'll just work, thanks confirmations

[spice turtle]

Thats sounds most likely

[supple pollen]

With some modules, the output capacitors are required for proper functioning. Some of them are even particular about the ESR, which can be neither too much nor too little.

[golden spade]

Hi! is it possible to have a lower input than output on a buck converter? I want to use a buck converter to limit a voltage in an efficient way

[unreal flax]

No, a buck by definition reduces the voltage, so it can't have an output that is larger than its input.

[golden spade]

yes that is true, what I mean is would the buck cut off the output entirely or would it just output the input with a small voltage drop

for my application I would want it to output the input which would be lower than the voltage it was designed to output

however I am afraid it would just disable the output

[knotty tiger]

i think it depends on the design of the regulator. check the datasheet

[golden spade]

I have tried to find it but did not now what to look for

the buck in question is the TPS54560

[urban lark]

Hi, is it possible to do jtag if missing the Test Mode Select (MTMS) pin being broken out on QTPY ESP32-S2? I may well be wrong, just looking at schematic ( https://raw.githubusercontent.com/adafruit/Adafruit-QT-Py-ESP32-S2-PCB/main/Adafruit QT Py ESP32-S2 Rev C.sch ) via https://www.altium.com/viewer/

[woven bluff]

is the charging current limited to 100mA in Feather boards ?

[fervent lance]

@woven bluff Yes. Pretty sure there are SMT resistors that determine charge rate.

[woven bluff]

how do I change it to 1000mA ?

[fervent lance]

I don't know what you have, there.

[woven bluff]

M4 Feather express

[fervent lance]

https://learn.adafruit.com/assets/57242
from:
https://learn.adafruit.com/adafruit-feather-m4-express-atsamd51/downloads

[woven bluff]

great, a 0603..... this charger system is so out of date..

can I just short it ?

[fervent lance]

While it trends in the right direction (10k to 100 mA; 1k to 1000 mA) .. those are within a range wholly outside of a dead short.

You would read the datasheet for the charger chip and adhere to its advice, I would think.

https://www.digikey.ca/short/qdr7jjpn

and

https://ww1.microchip.com/downloads/en/DeviceDoc/MCP73831-Family-Data-Sheet-DS20001984H.pdf

[hushed smelt]

What’s wrong with 0603? I don’t have any charger issues with batteries on my feathers.

Ohhh it’s 100ma, didn’t know that. I thought it was taking forever to recharge my triple 18650 pack. It took like 3 days. No wonder.

Is it a bad idea to put a bigger resistor on the feather charging circuit? Otherwise I would like at least like 500ma.

[distant raven]

You should make sure the charge rate is less than the battery capacity

So you shouldn’t charge a 500mAh battery with a 500mA+ charge current.

You can, and no one will stop you but it’s not advised

[crude ocean]

if I recall rule of thumb was half mAh for charge current ("0.5 c")

not sure how a triple pack affects that math

[fervent lance]

It's a hobby market. The vendor is in no position to be selling fast chargers with non-permanent mated cells to be charged. Lowest common denominator (100 mA is not the lowest, but low enough to be relatively trouble-free).

[fervent lance]

The way I look at it, if I can't put my finger right on the text that says I can do a thing (with tech like batteries that arrive in specially marked packaging) .. then I cannot do that thing.

It's best to have a safety plan that's well-thought out and critiqued by someone with no skin in the game.

[distant raven]

I think Adafruit covers it in one of their learn guides, you might be right. Sounds about right.

[fervent lance]

https://learn.adafruit.com/adafruit-microlipo-and-minilipo-battery-chargers/charge-rate

[hushed smelt]

6600/100 * 125% = 82.5 hours to recharge. Wasn’t exaggerating when I said it took 3 days to recharge.at 110ma. I can put it on a power boost and charge it in 8 hours no problem. I had no idea the charge current on feathers was that low. That means the 0603 is indeed a 10k resistor on the nrf52840 sense as well. This makes a lot of sense. Thanks.

My only concern modifying that resistor is if it will affect the rest of the system somehow. I’d like to hope not but I’m not great at board design. I’m not like you, if something doesn’t say it won’t do something explicitly, I’ll give it a shot lol. I’m more hands on and trial by error. If it dies, it dies.

[supple pollen]

I have learned a whole lot about how things work and how things break by trying stuff. That knowledge has served me well, and is totally worth the money I spent on stuff I ended up frying in the process.

[distant raven]

Changing it out for a smaller value resistor should be pretty easy with hot air and some kapton tape

[inland jungle]

what about piggybacking another resistor on top?

[distant raven]

Also a great option.

[woven bluff]

I think you are really going cheap on the charger chip

at least go for software programmable chip

like arduino MKR

even for hobbyist level, rarely someone would have 0603 resistors around

[hushed smelt]

i just got 200 0603's in 1k and 2k for 5 bucks on digikey. i will now "have some just sitting around" 😛

this is good stuff, i'm learning a lot about hw design. please keep posting up your board designs. i love looking at board designs.

all the stuff skerr is doing is fascinating, loving it .

[supple pollen]

I realized I could make the traces neater by using the ability to reassign the pins on this HT16K33 chip.

But then I realized I could make it even more compact by hiding the driver chip between the rows of pins on the through-hole display

[hushed smelt]

i understand the first one, very smart. the 2nd image is going to take a while to digest. 😛

ah piggybacking like the chip isn't even there. that's cool.

minimal disruption for the main traces going through

maximum board space savings 🤩 👍

[supple pollen]

It's for some Back to the Future style time circuits. I realized a single HT16K33 chip could run 3 alphanumeric characters, along with ten 7-segment displays, a few LEDs for AM/PM, colons and the like, and a 12-key keypad.

[hushed smelt]

https://tenor.com/view/whoa-glasses-off-amazed-wow-omg-gif-16482157

[supple pollen]

A huge improvement over my original MAX6954 based design, with its 40-pin DIP package

[hushed smelt]

ah multiplexer for seven segments

where's the chip on the new design on the back of the pcb?

is that your own custom uno board? that's cool.

i recognize the chip qwick synringe. i have the same style for hand soldering flux. i love using it.

dynobot need to go to H***. this isn't ok.

[supple pollen]

I had tried using a pneumatic solder paste dispenser, but it was problematic and I eventually got an iExtruder.

[hushed smelt]

never heard of that, looking up

[supple pollen]

https://www.i-extruder.com/en/

[hushed smelt]

an electronic dispenser?

didn't know those existed, neat. i only use flux once in a while. if i was making custom boards in quantity i might want to get one. great to know.

[supple pollen]

Yeah, it's basically a stepper motor pushing on a plunger. In quantity, I'd probably use a stencil, but for just putting little dots of paste wherever I want, it worked well, once I learned how to get solder paste with the right particle size (smaller particles flow better) and thin it with flux, both to flow better and avoid using too much.

[hushed smelt]

ahh you could use that with a 3d printer or cnc to automate panelized pcbs.

if you really wanted to scale up production

[supple pollen]

Yeah, that would be cool

[hushed smelt]

provides a way for hobbyists to scale to a small startup company if desired. a nice start to automation without needing a multi-bazillion dollar pick and place.

bookmarked for my ehh maybe someday folder.

though, i imagine just going with a pcb fabricator would make more sense except for assembly. some assembly costs are over the moon.

[distant raven]

Voltera V-One does this

[low anchor]

If I put an LED and a relay in series, how would I calculate the resistor required for the LED? I assume just compute the resistance of the relay and subtract that from the resistance calculated for the resistor if it's just the LED? Also, I'm not too familiar with voltage drop; would this mean that if the voltage drop of the LED is 2.2v, and my power supply is 12v, that only 9.8v would be applied across the relay coil?

[unreal flax]

It would be a little unusual to do this in general versus running them in parallel, since both the relay coil and the LED operate based on current, so it's not necessarily guaranteed that the same amount of current is appropriate to activate them both. For voltage drop across the coil, you'd also need to subtract off the voltage drop in the extra series resistor.

[spice turtle]

This is unusual as if you want to know when a relay is on, you would have the LED in parallel with the coil. Relays take a fair amount of current, much more than an LED. Your math is correct though. Voltage drops are subtracted from your main supply. For more info: Loop up Kirchoffs voltage law.

[supple pollen]

You may also have to allow for the inductive spike created by the relay coil when the relay is switched off: you would want to protect the LED from that spike.

[bright thistle]

So, capacitor?

[low anchor]

Hmm. My idea was to know when current was flowing through the relay, so that if it broke or was disconnected the light would turn off

[tacit panther]

I guess if this relay broke, it probably wouldn't be the coil side that would have the broken issue. It would more likely be the mechanics going bad on the contact side.

Putting an LED in parallel with the coil would at least let you know if power is applied to the coil or not, since power would also be going to this LED.

[spice turtle]

Having serviced relay output card's that have been in service for 10+ years, it won't be the coil failing, but rather the contacts themselves. They mostly fail open, rather than closed. I don't think you have to worry too much about this tbh

[inland jungle]

if you're trying to indicate whether the relay is open or closed, you would put the LED somewhere on the contact side of the circuit, not the power side

[spice turtle]

That would break isolation then, and they would need another contact. Ie if they were controlling 120 VAC, they would now have to power the led from the same source or use another (bigger) relay

[inland jungle]

depends on what they're switching, but yeah if it's 120V better to use a neon indicator light rated for 120V

[bright thistle]

bleah, cannot figure out a decent way to design the PCB 😐

[bright thistle]

is using freerouting considered cheating?

[supple pollen]

More like a diode

[supple pollen]

I have no idea, I don't know what "freerouting" is. It looks like you're pretty close.

[hushed smelt]

auto router isn't cheating. if it works, it works. doesn't matter how you get from A to B as long as you get there. 😉

@supple pollen some programs have a feature that will automatically connect all the traces for you. i can only speak for easyeda's auto router, it works very well but the way it goes about it isn't the most optimized way and certainly not the most aesthetically pleasing. you can sometimes tell from looking at a board if traces are connected manually or with an auto router.

[supple pollen]

Oh, I know what an autorouter is, I just hadn't heard of "freerouting"

[bright thistle]

An autorouter :)

https://freerouting.org/

Works with kicad

[bright thistle]

Believe it or not, it was my second guess, but then I thought, maybe a diode is more useful for reverse currents, not spikes... 🤣

[supple pollen]

The voltage generated by the collapsing field is generally of the other polarity, which allows the diode to steer it away. You could also design a snubber network with a resistor and capacitor, but generally a diode is easier.

[bright thistle]

ok, the way this autorouter works is... interesting.

it thankfully offers you the capability of interactively drag the routes it creates, but for some reason, if you click on it after dragging, it defaults to "get as tight as possible"

how I fixed the routing

how it reverts if I dare touching the routes again

dare I say this could be an acceptable tentative board?

only two signals crossing the main connector, and I bet it could be improved if I use different pins

[spice turtle]

Oh yes but as long as those arent high speed signals, Id say its OK

[bright thistle]

Nah, nothing high-speed (maybe just the display?)

And I fixed those two lines crossing the main conn

(fyi this is the display)
https://www.audiophonics.fr/en/screens-vu-meters/white-spi-iic-256x64-oled-312-screen-p-14604.html

[tacit panther]

I'm looking for a little insight on designing a board with a few traces that can handle up to 12A worth of current. I've been using this Saturn PCB Design calculator to mess around with trace weight and width options, and I'm getting about 306mils width can handle 12A with 2oz copper and a temp rise of 20C from ambient. I can get thinner traces if I'm willing to accept a higher temp rise.

I guess my question is, what is an acceptable temp rise and what other things might I have to keep in mind to handle this current requirement?

Should I look into the components on the board and see what temp they are rated for and keep the rise under these values?

[distant raven]

One trick that is often used to increase current carrying ability is leaving traces unmasked and applying solder to the traces

you use 40-70mil traces and then put a bead of solder along tthe length of the trace

you can do larger traces of course like 100mil

[tacit panther]

Oh that's brilliant. So I could cut some cost down by keeping to a smaller oz size for the copper and then hit all the traces with solder. Would you happen to have any idea how much thicker the solder would make the trace? I have access to a wave soldering machine, where I could run the board and hit everything exposed.

[spice turtle]

Based on commercial designs Ive taken apart (like power supplies); not much. Maybe 1mm at most but thats being generous. On top of that, Ive also seen copper wire added or soldered to the unmasked trace.

[tranquil narwhal]

Im designing a straing guage amplifier and was wondering if i could get some feedback, the top pots are for offset and then the other op amps are output filtering. Would you suggest things like a dedicated 3.3V regulator or anything else to improve signal stability

[bright thistle]

fsck. one pin housing wire snapped at the base of the crimp. 😐

and of course I don't have spares 😐

[distant raven]

Oof

[bright thistle]

welp, out to get a crimp kit.

[bright thistle]

{puts ash on his head} I admit hand crimping using a Channellock has not been a smart idea...

[spice turtle]

I would just make sure you have quite a few bypass caps around the main power line feeding your amps. I wouldnt think you need a dedicated 3.3V regulator, but there are some accurate ones out there. I think the spec you want is load regulation and its usually in %. Be aware that if you dont meet the min current demands of the regulator, it may operate out of spec as well.

[proud summit]

I'm working with a buck converter that can step down a maximum of 36v to an adjustable voltage between 3.3-10v. I need help with the voltage divider circuit. I need help finding the R2 value. any help is be appreciated 🙂

I've attached a screen shot of the "Setting the Output Voltage " and a part of "Electrical Characteristics" portion of the data sheet

Link to the full data sheet - https://www.mouser.in/datasheet/2/256/MAX20404_MAX20406-2887587.pdf

am I doing something wrong here ?

NOTE: I'm trying to get an output voltage of 10v

[supple pollen]

The feedback regulation voltage is 0.9V? That math looks wrong, 10 divided by 0.9 is 11.11, and 11.11 - 1 is 10.11. It looks like you're using 100kΩ for R2 which seems reasonable. Note that if you do the math in kilohms, you also get the result in kilohms.

[bright thistle]

I think it's 0.8 as per DS, but the vertical dash of the arrow makes it looks like a 9

[supple pollen]

That would make sense then.

[proud summit]

sorry for the blurry image. feedback voltage is 0.8 as stated by the data sheet.

[proud summit]

so does it mean I'd have to use a 1.15 MOhms resistor and 100k in series ?

[distant raven]

Looks like it

Though you could use a smaller R1

It says less than or equal to 100k

So realistically you could use a 1K there and use a smaller r2

Probably a 12K resistor for R2

Or whatever common value is close to 11k5

[proud summit]

are there a upsides to using one value over the other and what difference does it make ?

[distant raven]

Easier to find values

Usually gets you closer to common resistor values

You’re essentially changing the feedback bias to the internal looping. A stronger driver probably will change the operating frequency. Might help improve efficiency but the data sheet doesn’t go into it in depth

[proud summit]

actually this is for a flight controller with sensors including accel/gyro/mag/gps so a circuit with lower noise would be better. considering this what values would you recommend ?
@distant raven

i know these values don't directly relate to noise but the frequenzy of the switching regulator does

[distant raven]

I personally would go for lower values since they’ll lead to a strong more consistent drive, and the values would be easier to source. You probably would get better efficiency with higher value resistors though

Smaller resistors would probably improve drive strength but reduce overall efficiency since you introduce more resistance

Resistance is going to increase potential heat loss so it’s a consideration

You’re going to filter and smooth the input to your components anyway, so it mostly comes down to do you want more efficiency or stronger drive

[proud summit]

are there any down sides in using lower values while operating in high frequencies ?
the regulator has a maximum switching freqenzy of 3.0Mhz. I plan running it at the highest frequency possible as it has lower noise. correct me if I'm wrong

also what exactly do you mean by "efficiency" in this context ?

[distant raven]

Power loss due to switching

So higher frequencies have lower efficiency because more energy is lost

I might be backwards, hold on.

Nope I was on it

Anyway, smaller resistors on feedback divider will lose more to heat

So if you’re battery operated, that might not be ideal

[proud summit]

Yes, I think it might be better to go with a higher value as this is primarily to power the companion computer(Nvidia Jetson Xavier NX) which consumes 20W and
The entire system is powered from a LiPo battery.

[distant raven]

Also considering that the table of recommended values is based on a 100k R1, probably better to not have to recalculate all those values

[proud summit]

thanks and btw other than the one from the data sheet would you like to suggest any filtering circuit specific to my case ?

[distant raven]

I would imagine that the recommended power circuit for you individual components should be sufficient

Some devices have filtering based on LDO and DCDC supplies if there is a need to address such variance

[supple pollen]

It does mention a capacitor, so I'd probably put a small (100nF or so) capacitor from the FB pin to ground. There's a resistor calculator that will find standard values nearest a ratio here you can use to choose values (for instance a ratio of 0.115 yields 15kΩ and 130kΩ): https://nightshade.net/design-tools/resistor-calculator/

[distant raven]

Oh this is valuable. I need to save this link

[limpid nest]

so what are the major benefits of machined female pin headers?

[inland jungle]

compared to what?

[limpid nest]

Compared to "normal" headers that are half the price

[inland jungle]

much lower profile when you're plugging in machined male pin headers compared to the male/female square headers. I use machine pin combos for adding the qtpy boards to my carriers

[elder peak]

Well, I'm really glad we were able to pin down that detail.

[proud summit]

I'm working with a buck converter that steps down a maximum of 36v to 10v. The input capacitor section of the data sheet is a bit confusing. It suggests a 4.7u(ceramic) and an electrolytic capacitor with High ESR and also mentions that a low ESR ceramic capacitor should be used. Also the board is a Pi/Jetson HAT and the components are placed on the bottom layer(back side) as the top is reserved for connectors. So getting an electrolytic capacitor to fit would be a very tight call. Any help or information regarding this subject is appreciated 🙂

I've attached a screen shot of the "Input Capacitor" section, typical application from the data sheet and a circuit from an Evaluation kit that uses the same IC.

Link to the full Data Sheet and Evaluation Kit:
DataSheet - https://www.mouser.in/datasheet/2/256/MAX20404_MAX20406-2887587.pdf
Eval Kit - https://datasheets.maximintegrated.com/en/ds/MAX20406EVKIT.pdf

[signal topaz]

There are some ways to avoid electrolytic capacitor. This type generally is most cost and area efficient. You can look into the tantalum capacitors (but get familiar with the disadvantages and risks), or even some big ceramic capacitors. In case of big ceramic bulk capacitors, sometimes it is good idea to add discrete resistor in series, to increase the ESR for damping purposes.

[proud summit]

are there any down sides to using a tantalum capacitor

?

[signal topaz]

oh yeah, the biggest one is that their fail mechanism is very likely to cause a short-circuit 😄 So it is not so common in high-reliability systems. You need to take care of voltage derating when selecting one

[distant raven]

Okay, so the trick is you have to run it at higher frequencies to use ceramic capacitors

400kHz mode would require electrolytic caps

But 2MHz/3MHz mode could use ceramics

It’s all about the frequency

A combination of the capacitance and the capacitors material determines response, though mostly the capacitance

[proud summit]

what's the exact reason for this ?

[distant raven]

Ceramics can handle higher frequencies

Ah dang, TI link has a censored word lol

[distant raven]

I can sent you the TI link via private message if interested

[proud summit]

sure

[distant raven]

Wait, forgot about bit ly

https://bit.ly/3oEkUXk

There ya go

It’s a TI note about designing high frequency buck converters

Anyway, big takeaway is that smaller footprints enable smaller total area which means less issues using higher frequencies

Also more space for more cool things

And then of course dealing with the frequency response range of ceramic, vs tantalum, vs aluminum electrolytic.. the characteristic frequency decoupling of electrolytics and and other polarized capacitors is relatively low, usually far less than 1MHz.. so not ideal for high speed buck/boost converters

[proud summit]

how exactly do the frequency settings work ? The data sheet mentions an external clock source. Are they just PWM signals or something like an oscillator ? The Data sheet mentions about pulling the sync pin HIGH to force it into FPWM mode. How do you ensure that the frequency stays constant under different load conditions ?

[distant raven]

Look at Table 1: Recommended Components

They specify the components for specific ranges. So the LC circuit helps set the timing because the inductor and capacitor oscillate at a specific frequency

Since you’re likely only using one of these, I’d tie SYNC to ground

Sync is used usually for timing with other on board DCDC converters

It will generate its own PWM signal under loads when sync is pulled to ground as well as it mentions

External PWM signals would come from another converter or PWM source

Probably not necessary in your case

[proud summit]

Thanks for the help@distant raven

[marble locust]

hey all,
New to hardware design, any advise on what software to use for PCB design?
I am running linux btw 😸 :arch:

[heavy jasper]

I think either Kicad or one of the web-based ones is then the recommended tool

[bright thistle]

kicad is the way to go

[hushed smelt]

i'm actually a fan of EasyEDA online. works great but there's a learning curve like with any design software.

[hushed smelt]

my pcb's arrived today. 🙂

i need to get a macro lens for my phone :/

[supple pollen]

I ended up just attaching a camera to a microscope

[hushed smelt]

hmm i do have one of those cheapy usb cameras around here somewhere but was tired of fighting with it in software.

that's a nice microscope 👀

neat setup

[supple pollen]

It's a pretty cheap student microscope, but sufficient for my basic needs. Sadly, the expensive bit was the relay lense and adaptors for the microscope barrel and camera.

[hushed smelt]

and i think it's in like 480p scaled up to pretend to be 720p. cheap thing not even worth using.

[supple pollen]

Camera is an ancient one that takes full sized CF cards, but it's light enough to not strain the adapter, and has decent resolution. Older cameras like that just don't hold their value.

[hushed smelt]

i don't have the steady hands for a microscope setup, respect to anyone who can work at that microscopic level.

i still have an iphone 5 because i refuse to give up my 3.5mm jack 😛

[supple pollen]

I'll see that and raise you a Blackberry Key2 because I refuse to give up my physical keyboard.

[woven bluff]

why ESP32-S3 Feather so stingy, tiny flash and psram

[distant raven]

Not sure, appears there is only one variant of the mini module

[woven bluff]

WROOM doesn't fit ?

[distant raven]

I don’t think it does

[bright thistle]

Mwahahahah

[distant raven]

🤓

[eternal fox]

Random question i know... Anyone with a good idea of how i could make basically a "12 track disc player" but for SD-cards? I have a pi400 board that id like to use for many different things. Right now i can choose one thing and ive had the idea of just spamming out mini sd-cards with random OS's (Android TV, weare-os, retropi etc), and instead of doing dualboot it would be cool to have a row of SD-cards i could push down on to select that sd-card and have the pi400 boot from it at next boot.

Anyone with experience doing soemthing like this? Could there be a better way of doing it?
Can you boot from USB-drives maybe? Theres a cool Cyberpunk datashard idea in there somewhere...

[hushed smelt]

it's not a datashard unless it goes in your head though right? 😁

[supple pollen]

I think there's a "chip select" line used to talk to SD cards, so switching between them would involve asserting the proper chip select line. Some SD card sockets have built-in switches so you might be able to co-opt those to detect the card pushed on, or even route the chip select signal directly.

[hushed smelt]

some boards have multiple SPI bus capability but it's usually limited to 1 or 2, don't know about the pi400. switching CS seems like an easier solution and would scale better.

also could just get a bunch of sd card readers and put them on a USB hub that has on/off switch capability per port.

but datashards are typically loaded individually if you're going for cosplay idea.

running a wire up the neck with a ribbon cable under prosthetic latex and some prop magic could make a functional shard port on your scalp.

[bright thistle]

Wouldn't SmartMedia be more fit for the purpose, being flexible?🤣

(speaking of flash media, Sony still makes MemoryStick, along with the Select version, with a physical switch for choosing which memory bank to use! 😂)

[bright thistle]

Much better

[bright thistle]

Prototype assembled, I'll try to make it go live tomorrow, next stop, the screen (en route) and the casing (namely, a 1U rack box)

and since the final housings for the tags will be a bunch of plexy slabs, I was thinking of putting some thin felt lining into the slots of the frames holders, so not to scratch them

[hushed smelt]

i see why you wanted to make them yourself. makes sense.

[bright thistle]

them what? the tag holders?

[hushed smelt]

oh nvm those are just plugs, thought you needed them for a cable conversion.

[bright thistle]

ah, those! yes, at the end it was a cable conversion, from Molex SL to single header housing (not offered as a prebuilt, shame)

but, once I manage to make another PCB, I'll go with SL-SL cable once more

[hushed smelt]

if you're making your own pcb why not just find a compatible plug for the other side?

pin headers are probably dirt cheap in comparison but unless you're making them in quantity i don't see the sense in all the extra work.

though you did a great job on the crimps. i can't do it that nice. 😛

[bright thistle]

because "the other side" for now it's just the Pi header, once I manage to design that board, I'll make it SL again, but I'm in no rush, it's just for keep it tidy and modular 🙂

(btw, love that connector, it has the same factor as a multi-pin crimp housing, plus the polarization :3 )

[bright thistle]

believe it or not, it's my first time using that crimp gun (only used an RJ one), and it came with a nice explanation booklet

and I've still managed to make some duds XD

[hushed smelt]

well i meant i don't have one so i'd have to mash it with wire crimpers. 😛 looks awesome for a first try. maybe i'll have to add that to my ever growing list of do wants.

[bright thistle]

this was my previous crimper (no, really) XD

[hushed smelt]

yeah i can see why you'd want the real deal then lol. there's no way you're getting that done with vice locks.

at least i have a wire stripper that kind of has a little crimper built in but it doesn't work on small stuff.

you might as well have tried using a hammer 😛

[bright thistle]

this one, I suppose?

had to strip with electrician scissors, the ones with the notvh near the fulcrum

the one I got is an amazon special, I think the actual cost is some €25, with the remainder split between the two boxes of connecotrs

https://www.amazon.co.uk/gp/product/B07QV4GKF8?psc=1

[hushed smelt]

yeah i have something like that but designed for automotive, so much bigger crimpers.

it strips small wires fine but won't crimp those little things you're working with.

anyway, glad you finally got those things made like you wanted, you've been talking about it for a while. 🙂

[eternal fox]

Somehow im not confident enough in my own abilities to start playing around with SMBD's (Surface Mount Brain Device) 😛

[limpid nest]

Are there DC/DC converters that just don't require external caps? I'm not seeing mention of them on this datasheet.

And when is it NOT appropriate to connect VIN(-) to VOUT(-) to GND?

[hushed blade]

I would be surprised at one not requiring. AFAIK it's pretty fundamental to have them, and putting a big enough one inside the chip is pretty difficult/space prohibitive. There are cases where you want to isolate sections of your power, and I could see having a split ground, but that's beyond my personal uses. 99% of the time in hobby stuff you likely just connection all the -s to gnd. Do you have a link to the data sheet? I'm curious about the cap thing.

[bright thistle]

uhmmm... guess I have to clean away the flux from the sensors, they're shorting or having parassitic currents

is windex ok for the purpose?

[distant raven]

I would use as high of concentration IPA as you can get

[limpid nest]

It's a module

[distant raven]

Isopropyl alcohol, not Indian Pale Ale 🙂

Lol

[bright thistle]

suspected much, albeit I like fruit lambic more :3

[hushed blade]

Oh! If it's a module there might be one already in there. Hard to say. I'd scour the documentation make absolutely sure they don't mention one offhand somewhere, be prepared to possibly add one if your power seems flaky

[distant raven]

I’m not a beer drinker so IPA is always isopropyl alcohol 🙂

[limpid nest]

They mention an input one but not output, so I'm tossing a 1 uF on the output

[distant raven]

0.1-1uF won’t really hurt you, just sprinkle a few on for good measure lol

[hushed blade]

Too much capacitance causes "problems" that aren't problems for most circuits. It's pretty safe throwing on unneeded ones unless your circuit is particularly sensitive or you're trying to cut costs

[distant raven]

That’s why lower 0.1-1uF is usually better than throwing in day, 4.7uF or more. Parallel capacitance adds up fast

Obviously too much capacitance on output or input to a circuit can cause issues on power up. Taking too long can cause weird transients when it hits the target voltage trigging brownout detection or even safe mode on some microcontrollers/mpu

[hushed blade]

Yeah, way too much will cause problems, but throwing in a reasonably sized one where it's not needed likely isn't going to be noticed in your average circuit

It'll take microseconds longer to start up 😉

[distant raven]

One a light controller I’m working on, I’m one prototype I have a 1000uF capacitor on 5V and it causes a 1S blip/reboot on the 3.3V DCDC regulator lol

The 1000uF is for power buffering long strips of LEDs but when no strip is connected to pull the power, fun things happen haha…

[hushed blade]

Yeah, that's a monster capacitor. I've only seen that capacity in capacitors at least the size of my finger

[bright thistle]

anyhow, time to stash the board another month, I guess

I know its code works, so I won't need it for some time.
meanwhile, I can dedicate some for the main code and the other board

it did work on x86, let's see if it does on Pi as well

[distant raven]

I reduced it to 470uF in a SMD electrolytic for the final design. I figured I probably don’t need 1000uF if I do the design right

[bright thistle]

I've discovered the joy of using global labels for the schematic.
I'm never going back.

[bright thistle]

aaaand done.

just missing the silkscreens

[bright thistle]

no way I'm silkscreening the whole Pi connector!

[supple pollen]

Why not? A short program will do it for you.

[bright thistle]

no phisical space on the board XD

and it's more a joke than something useful

[bright thistle]

grrr, cannot find a part number for idc 2x8 2.54mm header

found
710-61201621621

[bright thistle]

Me: lemme just put M2 mounting holes, along with a pad so to use the chassis as ground

Also me: drilling and tapping looks like a PITA, I'll just use adhesive standoffs

[bright thistle]

uh? are JST XH connectors not made anymore? mouser doesn't even assign an internal part number

[distant raven]

that is weird. the search for other common JST *H connectors doesn't really yield the right results either.

[bright thistle]

I see digikey has it in heavy backorder (talking october)

bleah, BOM polluted 😛

[distant raven]

you might look on LCSC to see if they have them. they sometimes do

I ordered like 1000 stemma qt connectors from there when it was impossible to get them from anywhere else lol

[bright thistle]

oh, but I do have them (albeit generic), it was just for prepping the document

[distant raven]

ah

[bright thistle]

since I'm eventually opening up the project, code, hardware et al, I might as well make the BOM so to help anybody willing to replicate it 🙂

[distant raven]

sure, makes sense

[limpid nest]

Is a linear regulator an OK power source for an adafruit board? Something like a 7805?

[unreal flax]

Yes, though often Adafruit boards will already include a linear regulator on them.

[limpid nest]

I'm regulating down to 5V from higher

IIRC they include a 5-3.3V regulator, no?

or is there an X-5V regulator too?

[woven bluff]

anyone know what's the model of speaker on MagTag ?

I cannot even find the model on ECAD files

also It doesn't look like a speaker to me. I think it might be a piezo buzzer

[bright thistle]

yep, it's a buzzer

[woven bluff]

what kind ?

[unreal flax]

Yep, generally. The input voltage range might extend a little higher than 5V, depending on the board, but it wouldn't usually accept like 12V or something if that's what you're starting with.

[woven bluff]

it says nothing

[woven bluff]

still nothing

[limpid nest]

Yeah I need the pre-regulator I think. Thanks

[bright thistle]

says it's a 7.5mm one, so I guess one of those black button-like piezo

[woven bluff]

SMD piezo are very thin, It might be a magnetic buzzer

[bright thistle]

probably this one

[woven bluff]

it is

polarized, must be magnetic

what happened to RPi ? zero stock everywhere

[bright thistle]

currently made of unobtanium

[woven bluff]

they just stopped making them ?

[bright thistle]

nah, smart sand crysis still going on, albeit in a lesser form

this is what the schems say about it

datasheet
https://product.tdk.com/system/files/dam/doc/product/sw_piezo/sw_piezo/piezo-buzzer/catalog/piezoelectronic_buzzer_ps_en.pdf

sorry @woven bluff, but unless an ada designer for the board chirps in, that's all you get

[woven bluff]

that's not possible, PS series are all through hole mount

[bright thistle]

¯\_(ツ)_/¯

https://www.adafruit.com/contact_us

Choose "technical support"

[outer adder]

A beginner here.I wanted to ask how can I chain multiple sensors in the same bus,without using the same breadboard.I have seen some examples with stemmaqt,but I am using a normal mcp9808 with a breakout board from adafruit

[unreal flax]

For I2C sensors like that, you'd just use wires to connect SDA, SCL, power, and ground together along the chain. That's all the Stemma connectors are doing.

[bright thistle]

welp, full pinouts documented as well

[outer adder]

thank you!

[outer adder]

ok,thanks

[bright thistle]

Whoops, sorry, was talking about the work I was doing yesterday 😅

[outer adder]

aha,no problem,thought it was for me...

[ember laurel]

What went wrong here?

[unreal flax]

That looks like an AI-generated picture of a circuit board...

[distant raven]

Lol

Going to guess something got too hot

[heavy jasper]

Clear potted board that someone realized they had to do a rework under?

[limpid nest]

Digital panini press over here!

[low anchor]

Maybe this is a dumb question, but how do I determine how much current a 7805 regulator will draw from VIN? Maybe I could just compute the number of watts delivered (5W at 5V since it peaks at 1A) and divide that by 12V for the input, but that doesn't make sense to me because it's not 100% efficient

ahhh wait a linear regulator just burns off the extra voltage

[distant raven]

Yeah

[knotty tiger]

current in is approximately equal to current out for a linear regulator. it will probably sink some small amount of the input current to ground just to run the regulator itself ("quiescent current" might give you a ballpark figure on that)

[hushed blade]

Isn't that one of the things where there's usually a graph at the end of the data sheet?

[distant raven]

Sometimes they have them, yeah

[ember laurel]

The image is a Dall-E 2 generated alternate version of a board Im working on 🙂

[unreal flax]

That looks like an AI-generated picture of a circuit board...
Yay, 10 points for me! 😁

[ember laurel]

Yes - you just won a brand new Porsche 911 Turbo S. Where should I send the keys?

Should I send the keys to Keyes?

[bright thistle]

OLED module just arrived!

Documentation? Never heard of it.
Be thankful we tell you the controller.

gotta love how it's sold as I2C compatible, but actually supports only 3/4pin SPI, 8080 and 8060

and the datasheet is missing the last section!

[eternal fox]

Thast why i always buy stuff from Adafruit or a local supplier of their stuff, never any problems with documentation or help.

I know the struggle all too well 😄

[bright thistle]

scratch that, I confused the global index with the list of tables and figures 😛

[eternal fox]

"5 pack of super duper mega oled screen xyz i2c controller board brain iic audio" etc 😛

[bright thistle]

I mean, it's the module audiophonics uses for its DIY network streamer, and it uses a Pi as well, so at least I know it's compatible

[eternal fox]

Dont forget the tiiiiny hand clicking on a non-touch compatible "0.9 module, making it look huge 😛

[bright thistle]

nice, controller is up to 480x128, but this is 256x64

standard pin header... I'm refusing to use loose wires on this project, I'll use an IDC flat instead once I'm done with it :3

let's test it out... thankfully I've documented all the pin assignment already 🙂

[bright thistle]

it lives! but for some reason it whistles when showing pics

could it be it doesn't like gettin much data when using 3.3v? what if I switched to 5v?

The whistle

[bright thistle]

ok, it might be the buck converter

[bright thistle]

I just realized I might as well do a HAT, as there's only passive connectors on that board 😅

at least I'll get rid of the ribbon going to the Pi

[pure gorge]

Does anyone know what the dimensions are on the FPC connector for the iPad 1st generation LCD? (Pitch, number of pins, thickness, etc.)

I'm looking for a compatible ribbon cable

[eternal fox]

Strange question maybe but... Does anyone have a gerber of a USB 2.0 or prefferably 3.0 hub?
All i really need is a hub with 3-4 ports, USB-C female in and USB-A females out.
Its almost impossible to find such a hub without it being crazy thick, a fully fledged dock or in a really strange shape.

Any guides for how to make one is also, of course, very much appreciated!

[pure gorge]

Actually, nevermind. I might use the iPad mini LCD instead

[hushed blade]

There are USB Hubs in IC. They have example schematics in their data sheet.

Not as easy as a gerber file, but is a good start.

[limpid nest]

Pin assignments between a thru hole and an SMD 74AHCT125D are likely to be the same right?

[unreal flax]

Why guess? Look at the datasheet and be sure.

[limpid nest]

the datasheet only has surface mount

currently looking for another version

found one

[bright thistle]

Is this good for removing flux?

http://www.crcind.com/crc/CRCproductdetail.csp?division=industries&product=FG CONTACT CLEANER&ilang=en&plang=en

[low anchor]

https://www.digikey.com/en/products/detail/rohm-semiconductor/RE1C002UNTCL/5042785 would this transistor be able to switch 12v DC @ 70mA with 3.3v (under 40mA)? I think so based on all the specs, but the Vgs(max) concerns me - from my understanding, that's the maximum voltage difference between the gate (3.3v or 0v in this case) and the source (+12v). But that doesn't make sense to me given that it's rated for 20v source to drain. What am I misunderstanding?

[inland jungle]

your source should be near 0V, your drain is 12V

[limpid nest]

which I've always disliked since it feels like the terminology is backwards but I have gotten used to it

[hushed blade]

The terminology has been confused for as long as electricity has been a known thing: https://www.douglaskrantz.com/ElecElectricalFlow.html#:~:text=Electrical engineers say that%2C in,back into the positive terminal. (first google result I found that talked about it, not necessarily claiming every single claim therein is correct)

[slate ether]

This is High temperature LED circuit that works fine. I also have to input the same 10K PTC to another controller. But, due to the high reading at controller input, it reads inaccurate temperature because of the LED circuit throwing off the resistance being input to the controller. The thermistor used was KTY81 110. I did verify that a thermistor by itself (without an LED circuit) returns a reasonably accurate temperature on the controller.

I intend to use the single thermistor to trigger both the high temp cutoff on the controller and also the LED circuit, How can I optimize the circuit to produce the correct resistance value at the appropriate temperature for the controller?

The controller is off a shelf motor controller board that takes thermistor input and cutoff on high temperature as set by the user although there is no indication for that. To have the LED indication I came up with this circuit which will indicate High temperature. #help-with-hw-design

[eternal fox]

Sorry for the noob question: What do you mean with IC? All im capable of thinking of is Integrated Circuit 😛

[hushed blade]

Yup, that.

[heavy jasper]

If you just need the bare board in small quantities, you could try buying an off-the-shelf hub and pulling just the board out? If you’re in it for the experience of building the hub then yes, there are a variety of hub chips available that you can build a design around.

[bright thistle]

You know how some storage hats use an usb bridge for connecting to the pi?

Is there any regulation in place on where to place them and so on? If I use it only for power, is it still a Hat?

Also, is the EEPROM mandatory?

[limpid nest]

Are you selling the hat?

[bright thistle]

No, but I'm making it public design

Also, I don't think there's anything to configure

Five buttons, five LEDs, an SPI-4 screen and a LED Shim

All off the shelf and plugged in the "hat"

Ok, turns out I still have to provide EEPROM, even if I don't use it to configure anything

[bright thistle]

Uhmmm... Apparently I'll have to rethink a couple of details, power-wise

[distant raven]

Which controller?

[cunning wasp]

This is probably a stupid question but iI'm designing a mechanical keyboard with kailh choc v2's. They say that they support cherry low profile keycaps. I think that cherry keycaps use 19.05 spacing but I am unsure if it is just that they can work with it or thats what the spacing is.

[bright thistle]

?

[ember laurel]

I have a question about LEDs and storage temperature. How affected is the LED lifetime by a storage temperature outside of its absmax storage temp rating?

in my case, the LED may see occasional temperatures of 135C.

but when turned on, will only have ambient temp of 25C-ish

kind of specific question, I know, but I couldn't find that much info about this with my google-fu

[bright thistle]

my guess is, it'd at the very least damage the casing, then the driver (if any)

[ember laurel]

no driver on this part

We just did a test with our motors with built-in LEDs - running them through 300 sterilization cycles

That is, 2 bar pressure, 135C steam treatment for 15 minutes.

That's a pretty harsh environment.

after every cycle testing the motor and LED operation

motors were running fine after this all, but the LEDs started failing after 100+ cycles

(first one at 100, some at 200 cycles...)

I think we need to look at the LED supplier :/

[bright thistle]

probably shorted because of humidity leakage in the lens

[ember laurel]

yeah that's my thought as well - the lens material fusing not being correct

[bright thistle]

maybe encase it somehow, so to have a small insulating air gap?

[ember laurel]

some pics of the LED and how it is fitted in the motor

[bright thistle]

ok, so no chance to apply a gasket+window

[ember laurel]

no, I think we need to do an in-depth analysis of the LED under microscope

and look at the LED supplier

We may need to pony up a bit more for better LEDs

https://www.schott.com/en-tw/products/solidur-p1000293

something like these

the one we're using should be 135C rated

[bright thistle]

maybe ask the manufacturer directly if they have something

mil or med grade

[ember laurel]

but also - noone ever will actually sterilize these motors.

perhaps once per year max

They just wipe them down.

Only handpieces are sterilized.

(even though I think it is mandatory to sterilize motor after each patient...)

[bright thistle]

I mean, if the LED is easily replaceable, you could just build some check, and if a fault is detected, swap it

[ember laurel]

It is not so easily replaceable

requires soldering :/

[bright thistle]

so Schott it is

[ember laurel]

yeah Schott looks like they have proper ones - but will cost a bit more

and we'll need a special design anyway

[bright thistle]

if you need TO-LED, they also make it connectorized

same with Mini-LED3.0, size down to 2mm

oh, cool, if I want to make the hat, I have to shuffle pins around already, as I'm using GPIO6/16 😐

or is it only valid for legacy Pi1A/B?

[bright thistle]

ok, I understand j3 is the jumper for write protect (either via solder bridge or pins).
but what about PP3 and PP4?

same thing, different smell?

ok, they're test points, got it

[unreal flax]

I'd guess "PP" stands for "pogo pin", so they can easily flip the write-protect for automated flashing in a test fixture or something.

[bright thistle]

probably, and I'm going mad already at trying to place and route this XD

[bright thistle]

let's see if I'm able to make it a complaint HAT

[bright thistle]

question: if I have a display accepting up to 5V for power, will its SPI go to 5V logic as well, or will it stay at 3.3V?

[inland jungle]

if you're supplying it with 5V, the SPI will likely be 5V, if you're supplying 3.3, then the SPI should be 3.3

does anyone know of a simple I2C tachometer, frequency counter, can the seesaw do this?

[supple pollen]

Is there any reason you're using metal cased LEDs? Ordinary epoxy cased LEDs are cheap and very weather resistant.

[ember laurel]

They're running pretty hot - and need the thermal connection through the metal casinh

1W for that little led

(25000 lux)

[bright thistle]

Ok, so time to insert a LDO, yoink 5V from the Pi, and yeet 3v3 in the display

[inland jungle]

You can use level shifters

[bright thistle]

nah, it can work at 3v3, but I'd rather step it down from the 5v rail than try to power it from the 3v3 (already in use by eeprom and rtc)

at least I should be able to provide more current

please don't hurt me XD

[limpid nest]

Yoink and yeet are my favorite technical terms

[bright thistle]

wow, apparently not only the autorouter did everything right (and without vias!), I also managed to make a contigous ground pour on both sides as well

[eternal fox]

Sorry for the late reply
Ive already done the first version, however the board i bought uses a really thin PCB and ive already gotten some broken connector issues with it.

The base idea here is to build my own keyboard, i was thinking of using an existing design and apply it to the board i want to have made and use some internal wiring to hook up the keyboard to that hub (effectively removing any external cabled between keyboard and hub).

Problem is i dont even know what USB really needs on a component level (thats one of the things i wanted to learn bu studying gerbers).

Alternatively have a thin USB-hub gerber and just replace the cable>usb a/c male connector (to the upstream host) with a soldered on usb c Female connector.

🙂

[charred vigil]

I want to try to add a development board to my pcb but I dont want to add the height of header pins, or have them stick out the other side of the board. Is it bad practice to try to surface mount the board if it doesnt have castellated holes?

Like I was thinking of flowing solder into each of the holes onto a rectangular pad until they stick

[distant raven]

You could probably get away with it

Just don’t put a through hole in the pad

[supple pollen]

There are also shorter headers available

[inland jungle]

I've been pretty happy with machine pins and sockets

[bright thistle]

yay, finally found some tentative specs for the OLED I'm using (because the seller couldn't be arsed to state the manufacturer...)

A Vishay equivalent will draw 27mA@12V with 50% usage.

54mA@12V at 100%

about 197mA@3.3V

I'm using a Pi3B+, which should provide some 500mA on that rail, so I guess it's the panel's regulator, but I guess it wouldn't hurt to power it off the 5V with an LDO

[bright thistle]

Finally!

[modest grail]

this is an adaptor for my psp to upgrade it to usb c. its meant to be a 2 piece design and soldered/ glued together using the holes with no traces as solder pints to secure itdo yall think it will work or will it be too thin and likely to break?

[honest badger]

i have a really cheap soldering iron that came with a 1 amp fuse but the first time I plugged it in the fuse blew