help-with-hw-design

[flat vigil]

The broadcom chips on the pis at least

[distant raven]

Did those include hardware design examples?

[flat vigil]

No, I don't think so. I'd just use a compute module if you want to design something with it

[distant raven]

I’m seeing on broadcoms support page they have a $2500 minimum order

Under their “terms of sale”

From my understanding, Broadcom has some of the most restrictive terms of sales policies of any major ARM vendor

[flat vigil]

for ML stuff I'd look into the tpu boards from google

https://coral.ai/products/

[pale umbra]

You are probably aware that it used to be much more restrictive before cypress bought them?

[flat vigil]

didn't cypress only buy the wireless business?

[distant raven]

Yeah, they bought the wireless business

[pale umbra]

well I assume this is the broadcom wireless business

[distant raven]

Broadcom still owns their SoC business

[pale umbra]

otherwise skerr wouldn't be able to order at all from them sinfce not even adafruit could

broadcom is (or used to) be super restrictive about their chips and youd need a whole legal department to handle their demands

[distant raven]

This is from the Broadcom sales support website for their SoCs

[pale umbra]

and sales of over 100m$ a year

[distant raven]

https://docs.broadcom.com/doc/broadcom_terms_conditions

$2500 minimum line order

[pale umbra]

Im a bit confused if that is cypres or old broadcom website

since they also sell wireless stuff on it

[distant raven]

Copyright notice is through 2022

https://www.eetimes.com/broadcom-sells-wireless-iot-biz-this-time-to-synaptics/

They sold the IoT business

Not the wireless (cellular) business it appears

It also appears to be the rights to, not the business itself

[pale umbra]

digi-key have a few of their 5g boards out of stock at 200$ ish

but the sheets are signed by cypress but says broadcom everywhere so Im very confused

and digikey list the manufacturer as broadcom (cypress semiconductors)

[distant raven]

Purchasing the rights to use something means they have to maintain the Broadcom branding

Or likely means anyway

[pale umbra]

but you said they bought the iot part

a 5g chip isnt iot right? just stuff to make android/iphones ?

[distant raven]

They bought the rights to

Broadcom still owns the IP, they’re selling usage rights

[twin hedge]

I'll get the large GPT-2 model running on my Pi 4 when I get back home.

Here are the requirements listed by the Github repo.

So, what processor would be good for running GPT-2?

[distant raven]

Any ARM A72 more than likely

Cheapest one in stock with Digi-Key https://www.digikey.com/en/products/detail/LS1027AXN7PQA/12140524

This will be considerably harder to route for than the iMX8

[twin hedge]

Wouldn't a quad-core be better?

[distant raven]

sure, i'm just thinking ease of routing. Quad core typically come in larger packages which means more pins.

[twin hedge]

Well, if it's what you recommend, then I'll use it. Should I use DDR3L or DDR4 RAM?

[distant raven]

DDR4 will be faster and potentially cheaper given it's still fairly new

[twin hedge]

I'll be running a medium GPT-2 model (to start), which needs about 3GB of RAM. Could I use 4GB of RAM in the system? Also, what ICs should I use (like 32Gbit or 8Gbit)?

[distant raven]

I'd use two 16Gbit RAM modules

well, you might actually only be able to use one. i'd have to find the data sheet and verify

[worldly schooner]

Do you need to design a SBC for 4GB ram? 4GB is pretty standard for SBCs these days.

Try an off the shelf SBC and see where it falls short?

[distant raven]

yeah, technically if you can snag a Pi 4 4GB model, you should be fine.

[worldly schooner]

I was thinking a Radxa Pi Zero, for the “almost thumb-drive size” form factor haha

[distant raven]

in terms of the NXP chip, you can use up to two DDR4

there are two chip select lines for access. per the datasheet which is one of those locked behind account access 😛

[pale umbra]

sorry to bother you but this is very relevant to my interests. I though of trying to find things like that but I though that 99% of the value of chatgpt or the much better models available in the API version was the training but the software itself was very easy to make and the algorithms have been known since the beginning of AI in the 1950s ?

For the 1950s my source is my teacher back at the university 10 years ago who was a researcher in the field of AI who worked mostly on computer vision and decisions systems

[unreal flax]

I mean, sort of. The basic concept of neural nets has been around for a while, but large "deep learning" models are relatively recent. Kind of the difference between a horse-drawn carriage and a racecar. They both have wheels, but...

[supple pollen]

In the 1950s, they had "perceptrons" and the like, which aren't really comparable to convolutional neural nets. Then again, I was playing with what I called "synthetic synapses" in the 1980s, which was the same sort of notion as neural nets.

[faint narwhal]

I'm having trouble finding a source for getting a female version of plug:

I need to get these wires to attach to pins on this PSU:

Larger picture of PSU:

[supple pollen]

That's an ordinary IEC C13 plug, they're pretty common.

[faint narwhal]

AH That's exactly what I was looking for! Thank you!!

[supple pollen]

Ah, that's the C14 male receptacle for the C13 female plug

[faint narwhal]

ahh ok

[distant raven]

Data centers often use C13 to C14 cables and have power strips with C13 plugs instead of standard outlets

[faint narwhal]

ooh interesting

Any ideas on how I could safely connect to this from that female connector above?

(I'm repair a missing part on an arcade machine)
From pin 1 to pin 3 its ~15mm
and the size of the white piece is 6.5mm 19.5mm

[supple pollen]

That looks a little like a Molex KK connector with a couple of skipped pins.

[faint narwhal]

i think the molex female pin size is too big
these pins are ~1mm diameter

[supple pollen]

There are a variety of these, I'm thinking the 0.156" KK connectors (which are popular in arcade machines)

https://www.molex.com/pdm_docs/sd/009781061_sd.pdf

[twin hedge]

I've gotten the small GPT-2 model running on my Pi.

[pale umbra]

I have like 5 spares of those in my own from my dead PSUs over the years

[fervent lance]

Dear community,

I have some NeoPixel strips with 144 pixels which can get really hot if much brightness is required from them. They are still within their weatherproof sleeves.
Do you guys think that it's worth removing the sleeves for lengthening the lifetime of the LEDs?
I'm willing to use these as main lights in some rooms for as many years as I can. I don't really need the weatherproof feature, but I would like to keep it for safety reasons.
I was also thinking about opening up just the back part of the strip so the back of the PCB could get some airflow.

What is your opinion guys?
Can I get significant lifetime increase by removing/cutting the sleeves?

[dry jetty]

Can I source 3.3V at a few mA from an IO pin of Feather M4? What would be the pain limit? I probably need less than 5mA times 2 pins.

[supple pollen]

I'm not used to having heat issues with NeoPixels. When you say 144 pixels, do you mean 144 pixels per meter, which would be pretty high density and I could imagine would create some heat. I agree that heat could impact their lifetime, so it might be worth doing something about it. However, I'm not sure what impact it is: you do say "significant lifetime increase" and I'm not sure how much you'd gain. I'm wondering if aiming some airflow down the sleeves would work.

[supple pollen]

Yes, a few mA is doable. You want to stay below 8mA.

[worldly schooner]

I believe you’d have to basically disable the weatherproofing in order to get air inside?

Perhaps some heat sinking would work. Affixing the entire strip to something made of aluminum or copper would help wick heat away from the LEDs.

[supple pollen]

Since it's indoors, I was guessing the sheathing wasn't being used for waterproofing, just impact resistance.

[scarlet jay]

https://tenor.com/view/f-bi-raid-swat-gif-11500735

Hello guys!

I have finished a few (inspired but modified) patterns for one device, could you check out and review what is incorrect so that I can finish them fully? 😄

[rustic linden]

On the first page, you tied VDD and GND together

Same with the 4th

The +3V3 net in the first, and the +5V net in the fourth

[worldly schooner]

Not sure what sw3 is trying to accomplish there, but I suspect those connections aren’t right.

[twin hedge]

So I know what processor I need, the RAM ICs I need, and the requirements for gpt2-medium. What are my storage medium options though? Should I use eMMC?

[rustic linden]

Also you've wired the batteries wrong in page 1, unless you were going for dead short

[scarlet jay]

Thank you a lot, I'm on that one! 😄

[unreal flax]

A combination of eMMC and SD Card is pretty common for SBCs, yeah.

[scarlet jay]

something into this little beauty

[twin hedge]

A combination? I was going to just use eMMC. Also, should the SBC use a male USB-C connector? That would allow me to get 7.5W to use for the system.

[unreal flax]

You can use only eMMC if you want, but you should think about how you want to load things onto the board. An SD Card sometimes makes things easy because you can use a PC to write an image to the card, transfer data files, etc.

[twin hedge]

I'll do that then.

So will I use high-power device or USB-C power? Also, I was hoping to add a WiFi IC so it can connect to the internet. What IC can I use?

[distant raven]

You’d find it easier routing to use a module that has all its components and a U.FL or trace antenna.

Maybe a UBlox module

[distant raven]

Just want a sanity check. I'm looking to add this TVS diode to my FPGA feather because ESD protection is probably a good thing. If I am understanding the Datasheet correctly, I just send each USB trace the corresponding pad (basically passing through) and tie the third to ground?

https://assets.nexperia.com/documents/data-sheet/PESD5V0C2BDF.pdf

[worldly schooner]

Looks right to me.

[distant raven]

this thing is tiny

the whole thing the size of an 0606 capacitor

actually... smaller

0.225mm between pads

pads are roughly 0.25mm x 0.1mm

[supple pollen]

I have a similar one on a SIM card socket, but with more leads.

[distant raven]

it's so so small

like.. getting into 10005 sized components

soldering will be... a pain

maybe I should go for a bigger one lol

[supple pollen]

I used a TI TPD4E004DRYR if you're curious

[balmy tide]

@distant raven any particular reason you're making an FPGA board instead of using development board?

[distant raven]

https://www.crowdsupply.com/oak-dev-tech/icyblue-fpga-feather

it's a product i'm developing

[balmy tide]

Oh okay

No specific tasks

Have fun with it 🙂

[distant raven]

it's mostly for helping people learn in a familiar ecosystem 🙂

[balmy tide]

Ehh okay

[distant raven]

FPGA are not for everyone

but they are fun and can help demonstrate a lot of computer architecture and digital design concepts.

[balmy tide]

Not an elitist, but i kinda feel like FPGA shouldn't be treated similar to mcus

[distant raven]

they're not MCUs, they are tools for developing custom logic circuits. There isn't any particular reason why you shouldn't use an FPGA in a feather format

[balmy tide]

I know what they are

[distant raven]

I don't think anyone is out there treating them similar to MCUs

and the feather format doesn't specify what should or shouldn't use it.

[balmy tide]

I don't think they quite fit in the maker community

At least not yet

[distant raven]

why?

[balmy tide]

But idk maybe, it's starting to get integrated more

[balmy tide]

I don't see any use case for them other than educational purposes

Don't get me wrong educational purposes sounds like an amazing use case but doesn't strike me as useful as other things.

[distant raven]

treating FPGA like professional tools is kind of gate keeping people from learning them and using them. Yes, they have great educational purposes. They also have great utility in implementing logic that just isn't all that well handled on COTS microcontrollers. Plus RISC-V and other open source soft cores expand that even further.

We talk about there not being enough professionals knowing how to do ASIC design, these can be great tools in changing that.

[balmy tide]

Also, i do hope some usecases arise for them. And willing to participate if it's something i could

[distant raven]

well, the goal I have with IcyBlue is to expand what people can do and provide specific add on boards to make those use cases more obvious.

so things like a traffic light controller feather wing to help demonstrate sequential logic in a very visual way

[balmy tide]

Wanna continue at #fpga ?

[distant raven]

sure

[twin hedge]

I know what you mean. I wanted to run RISC-V Linux on an FPGA, and check out other interesting soft cores, but I kept getting told that FPGAs were too difficult for beginners, and eventually I scrapped it. I want to get back to it, and if someone knows a good FPGA board, I'd be happy to get it so I can mess with RISC-V.

[distant raven]

There’s lots of great boards out there. Orange Crab can run Linux I believe

https://1bitsquared.com/collections/fpga/products/orangecrab

[worldly schooner]

While it’s true that the prerequisites for FPGA development are very different from those for traditional coding, it really shouldn’t be that heavily gated as a technology. The more prominent reason for its reduced popularity among hobbyists, in my opinion, is the cost for performance, as most applications in the DIY realm have little use for the advantages FPGAs offer over the budget microcontrollers we have now.

The only DIY project that comes to mind that really takes advantages of the strengths of an FPGA is MiSTer FPGA retro emulation, and boy do I want to get my hands on that one day.

[twin hedge]

I remember this board. I wanted to make a board with similar parts, but like I said before, I getting told it would've been too difficult, and I threw the idea away.

[twin hedge]

If i remember correctly, the Analogue Pocket does this as well.

[worldly schooner]

It does, though I don’t personally consider that DIY. It’s an absolute beast at what it does, though.

[distant raven]

I generally think it comes down to how information around FPGA is shared. It’s rarely disseminated in ways that is easy to digest which tends to make it harder to work with.

And also the price tends to put it out of reach for a lot of people

Especially since the “ramp up” time to do meaningful things on an FPGA also makes that price feel even more expensive

[hexed mural]

FWIW I had a good experience with learning the basics on a Cypress PSoC5- it's an ARM code with a CPLD baked in, and you can pretty easily write Verilog for it. The IDE is Windows-only, though.

[worldly schooner]

I’d love to see competitive pricing in the FPGA market. Problem is, the one company that’s actively working in this direction doesn’t have an open-source tool chain behind it yet.

[distant raven]

Lattice?

I mean technically Lattice chips can be programmed with open source tools

[worldly schooner]

I was thinking Efinix actually. Haven’t heard from Lattice in a while…

[distant raven]

Lattice can be programmed with open source tools thanks to a dedicated community

I also think that because most of the FPGA brands being mostly focused on commercial products, little has been put into accessibility

[worldly schooner]

I hear Renesas has a super-low-cost FPGA in their pipeline too, but no idea when that’s coming to market.

[distant raven]

There was mention of some fairly decent sized LUT count fpga costing less than a dollar from Renesas

I shared it probably a year ago? Haven’t seen anything about it since

[worldly schooner]

Yeah those. Their page talks about their ForgeFPGAs, but there aren’t any actual chips yet AFAIK

[distant raven]

There are apparently engineering samples if you contact them

[worldly schooner]

Aren’t PSoC5 devices basically unobtanium right now? I backordered a PSoC 5LP last year, and they delayed my estimated ship date yet again.

[balmy tide]

Tbh cypress approach seem more appealing and useful to me

Tho never been around to doing something with it

[distant raven]

You can buy old Cisco modems and pull PSoC5 out of them lol

Saw a post about it this morning

[hexed mural]

Oh dang, maybe? I haven't looked in a while, booo. This was a few years ago when they had a $10 dev board that you just plug straight into your USB port. Came with debugger and all

This one: https://www.infineon.com/cms/en/product/evaluation-boards/cy8ckit-059/

But yeah, OOS. Dang

[worldly schooner]

Yeah that’s the one I ordered haha

Need it for a course I want to take

[hexed mural]

I have a few hanging around if you want one. They were all free from some trade show years ago

[worldly schooner]

Oh dang, that is super tempting.

[worldly schooner]

Even if you have excess, I couldn’t take that for free. If you’re seriously offering though, DM me. I’ll offer something worth its MSRP at least…

[balmy tide]

I thought about using one of these for 3d printer but i don't think it has enough io and idk what purpose FPGA would serve

Also unobtainable thing

[fervent lance]

Does anyone here have experience with this Recom TO-220-ish switch mode PSU?
Recom 78e5.0-1.0
https://eu.mouser.com/ProductDetail/RECOM-Power/R-78E50-10?qs=T0XSgvH75d4EsZCU9I5HVA%3D%3D&gclid=Cj0KCQiA3eGfBhCeARIsACpJNU9fzKmYwAyvA8WRkAuxN4cUz6PiLIDyDABJoJ-BsEWIMoAZ1YLJEhoaAth1EALw_wcB

Can they be mounted in parallel for double power?

[worldly schooner]

Which PSU?

[fervent lance]

Argh, I forgot to post the link.

[hexed mural]

Oh that's not a bad idea! Maybe driving the steppers super fast or something with the CPLD. Also you can run like half the I/Os at a different voltage from the main core, which is kinda nifty

[balmy tide]

Well you can generate steps really fast but you still need to process gcode as fast

Tbh doing that sounds miserable

[hexed mural]

True. It's an 80MHz ARM which isn't bad, but is a bit aged by today's standards I suppose

[supple pollen]

I've used a similar one.

[fervent lance]

How did you determine if it was suited?
I looked at this one's datasheet but found no mention.

[supple pollen]

The data sheet for the one I chose stated 5V and 1A, which was sufficient for my use case.

Here's the relevant data sheet Mouser provides: it's pretty clear, and in large text

If you're curious, you can see it on my board toward the left (black box between some of the ICs)

[fervent lance]

The question isn't how much current one can deliver, but whether I can put two in parallel for double output.
Am I missing something obvious to a real engineer?

[supple pollen]

Normally it is not recommended to parallel regulators. Often you can split up the load and use each regulator to power part of your circuit. However, if that's impractical, you would need to choose a regulator capable of higher current.

[fervent lance]

I was afraid that was the answer 🙂

[supple pollen]

Fortunately, 2A and 3A switching regulators in similar sizes are not hard to find.

[fervent lance]

Nice, I'll have a look around. I got these some years ago and hung on to them because they seemed practical, I just never stayed up to date on the selection.

[supple pollen]

They are handy, and 1 amp is plenty for the sorts of things I usually build

[fervent lance]

In this case, I'm supplying two LEDs with 375 mA and one Q5 flashlight module @ 5V, so 1A won't cut it 😦

[supple pollen]

I might consider a current regulator from the raw supply to power the LEDs instead of using a voltage regulator (and presumably a current limiting resister after it)

[fervent lance]

Good ideas, I'll keep them in mind 🙂

[fervent lance]

Is it possible to solder a female USB port (specifically https://www.adafruit.com/product/2225) to a XIAO rp2040? I've been searching all over the internet but haven't seen anything unfortunately

[distant raven]

Not without serious modification

[fervent lance]

What type of serious modification are we talkin about?

Nvm, found a way to avoid that issue lol

[tulip topaz]

Not advised to do so. These are 1% accurate on the outout, so good chance one device will outdrive the other and overload. Best practice to use a single regulator, properly sized.

[fervent lance]

Thanks for the elaboration, I found a small drop in DCDC good for 3A in a box in the office 😊

[scarlet jay]

https://tenor.com/view/hungarytravolta-confused-where-meme-gif-8774583

Hello!

So, I have analog swith.

How to put it together correctly?

[earnest folio]

Can you give a little more context?

Like, what kind of switch? Hall effect? Optical? What kind of output? Does it have a part number? What are you switching?

[scarlet jay]

analog and frequencies/electric outputs 🙂

[supple pollen]

This is the 74HC4052? It has two sets of 4 signal inputs, each with 1 signal output. There are also two logic inputs, these select which of the 4 signal inputs is routed to the signal output. There is also an enable input, if it's high, there is no connection between input and output, if it's low, they're connected as described.

[scarlet jay]

okay and the logic input is connected to my board if im thinking corectly? 🙂

[supple pollen]

Presumably the chip is on a board already. Normally the logic inputs are connected to I/O pins on a microcontroller to allow it to select which inputs to use.

[scarlet jay]

https://tenor.com/view/kizo-poland-polish-rapper-polska-gif-24409654

[scarlet jay]

okay but then do I need to have an option for every pin on the board or if I choose to have a single output for all the four of them, is that possible? 🙂

[distant raven]

while the GIFs can be funny, they can detract from people being able to help effectively so it's probably good to keep them to a minimum.

[scarlet jay]

sorry

[supple pollen]

I don't understand what you're asking. You don't have to use all four inputs if you don't want to.

If it's the board I think it is, you don't even need the multiplexer at all.

[scarlet jay]

SAMD51

I thought I can preprogram 4 signals and have 1 output at the end with logic selecting in between them.

[distant raven]

you could technically control the output in software

without needed a multiplexer

which is what I think Madbodger was trying to suggest with the capabilities of the SAMD51

[scarlet jay]

So I technically just need an output straight from the board and I change its frequency type inside the board?

[distant raven]

right, you would adjust your DAC output in software and just use one pin to output your analog signal

[scarlet jay]

oh, seriously and I can have multiple if I have them saved in memory?

[distant raven]

right

[scarlet jay]

like different types of frequency? 😄

[distant raven]

yup

[scarlet jay]

WOW!

Now that has changed some stuff massively.

[fervent lance]

Yes, they are 144 pixels per meter. Thank you for your answer!

[fervent lance]

Nice idea, I might give it a try.

[frozen sinew]

Hey. I'm trying to make my circuit low-power and I want to deactivate the USB to Serial IC when no USB is connected (only battery). Will this method work with the transistor on GND? Should it be on the VCC instead?

[balmy tide]

this could cause issues depending on circuit

[frozen sinew]

How so, if I might ask?

For this particular IC, I know that FireBeetle does the same for their ESP32-boards. https://images-na.ssl-images-amazon.com/images/I/A1VO7RyKHzL.pdf

[balmy tide]

you already know why because you have already made SER_GND node

you change USB and serial's ground potential, by minimal amount, I expect it to work %90 of the time.

[frozen sinew]

Yes, that makes sense. There is a diode drop as well.

I was also thinking about powering the CH340C with the V_USB directly. Might not need a MOSFET in that case, and the IC will only be active when there is a USB connected

[scarlet jay]

Hey guys, I think I'm finished. I know there are some mistakes but yould you proofcheck everything to help me tell if everything is correct? 🙂

[distant raven]

Curious why you’re using a USB to serial device instead of using native USB? PA24/PA25 are the USB pins for the samd devices

[scarlet jay]

oh, didnt know, will fix, thank you 😄

[distant raven]

https://www.mouser.com/datasheet/2/268/60001507A-1130176.pdf

Be sure to check your pin usage with the data sheet

[scarlet jay]

highly thankful and will proceed but what do you think about everything else? is it correct by all the means? 🙂

[distant raven]

I’d just make sure you follow the data sheet

That is your source of truth for design

[scarlet jay]

and if I have that 3V3 already over there to give it power, wont that help? 🙂

[distant raven]

Sure, just make sure you’re following the data sheets for the parts you’re using and you’ll be fine

[sleek ice]

I need the smallest possible board to drive this display. https://www.aliexpress.us/item/2255800358721473.html?spm=a2g0o.order_list.order_list_main.5.667c1802fzmf5E&gatewayAdapt=glo2usa&_randl_shipto=US. I want to try not to use the included driver board to save on space.

[distant raven]

That’s probably the smallest driver unless you spin your own board

Which might be hard

[tulip topaz]

Without USB connected, the USB should be suspended and power draw will be about 100 uA. You could make 3.3V from 5V to power the chip from Vusb.
Another option, if your microcontroller can source 20mA at 3.3V, you can just use a GPIO output to drive VCC. You would have to sense Vusb to know when to turn it on.

[frozen sinew]

Didn't quite understand your first paragraph. The USB->Serial IC can handle the 5V (Vusb), so I don't think I need another regulator to 3V3.

Your second paragraph; that's a good idea. Unfortunately, I don't have a single GPIO to spare 😅

[balmy tide]

Your reset and boot circuit seems weird. Any reason for disabling button?

[scarlet jay]

not in particular but if I coded it incorrectly I always have a way out 🙂

Does"EP" on SAMD board mean the same an "EN" mod on ESP32? 🙂

[rustic linden]

It depends on the boards

[balmy tide]

You should try net tie for that, sometimes routing could make it not possible to cut the connection or make it hard.

[scarlet jay]

@rustic linden Thank you! 🙂

[rapid moon]

hello! I'm new with embedded systems and am trying to connect my board with platform.io but its not shwoing up on my devices.. is there anything i can do to fix it?

[limpid nest]

Do all button cells have high esr?

[stuck sluice]

hi guys i have a couple of the sht40 breakout boards. I want to use them on the same I2C bus but there's an issue of them having the same address. I know that there are no hardware pins on the sensor itself to change the address but is there a way I can do it through software?

[stoic birch]

Is there a chance my dupont connectors cant carry enough voltage and will in turn lead my rpi to have low voltage issues?

Also does anyone know if there is a quick connect for these types of buttons?

[supple pollen]

Basically.

[supple pollen]

I don't think they do, which leaves you with either an I2C multiplexer or multiple I2C buses.

[stuck sluice]

yeah

[limpid nest]

Thanks

[limpid nest]

If I'm looking to have 4 PWM channels, cap touch, deep sleep, in a low power package that preferably supports arduino (I can survive with c++) and programming over usb, where should I look?

I'm thinking use an independent pwm chip? So it can survive sleep? Or is that not an issue?

[distant raven]

Maybe a 32u4?

https://www.digikey.com/en/products/detail/ATMEGA32U4-MU/1914603

Plenty in stock at Digi-Key

[supple pollen]

I'm thinking TI's MSP430FR line, along with Energia, which is an Arduino port for it.

[limpid nest]

Thx

[limpid nest]

Taking a look now. What made you pick that one?

[distant raven]

The FRAM on the 430F is fantastic non-volatile memory

[limpid nest]

Ooh

[distant raven]

And great for low power

[limpid nest]

I see it has a ultra low power mode

An*

[distant raven]

Yeah, just a handful of uA

Possibly even lower than a uA

[limpid nest]

So I'm looking to do some n00ds work with 4 separate strands. If I want to dim and change dimming progressively of the 4, am I better off not sleeping when that's happening?

[supple pollen]

The aforementioned low power mode, the evaluation boards are inexpensive and come with a USB programmer you can re-use for your own chips by replacing the jumper blocks with jumper wires (just two signals plus power and ground is enough), they support capacitive touch, some of the Launchpad boards support power monitoring for low power analysis, I've used them before.

[supple pollen]

I suppose it depends on how fast you're fading and which sleep modes you're using. In general, anywhere you'd call delay or wait for a clock tick, sleeping is going to be more efficient.

[unreal flax]

Note that the power draw of the noods themselves will be way more than the MCU, so you won't save very much power by sleeping even if it's possible.

[limpid nest]

That was the algebra I was going to do next

[balmy tide]

Also generally while sleeping timers stop so pwm modules won't be available

[scarlet jay]

Hey guys, how do I add a bootloader/ISP on my board? 🙂
it a final board but I jut want to make sure that in case something will be up I can still change it 🙂

[supple pollen]

Basically, add a connector for the SWCLK and SWDIO signals, possibly along with with power, reset, ground, etc.

You'd presumably want to use the standard connector, and obtain the matching ISP programmer (SEGGER, Black Magic, JTAGulator, etc.)

[scarlet jay]

thank you.

[urban grotto]

basicly developing an nrf52840 dev board at this point, how vital is the 32khz crystal? If I plan on ommiting it and having it derive from the built-in 32Mhz crystal, do I need to like re-compile a bootloader?

[unique patio]

The nRF52 bootloader does not care about the 32 kHz crystal. Various programs do, like CircuitPython.

I determined that very briefly by seeing if there are any mentions of "xtal" or "crystal" in the board files.

[copper lintel]

Hey folks, does anyone know where I can find the Eagle library file for the KB2040? I found the board and schematic files, but I'm really after the library file so I can use the castellated KB2040 for my own PCB. I followed a tutorial to turn the source files into a .lbr file, but I dont think I did it right, or the video wasnt what I needed.

[rustic linden]

Oh that reminds me, does anyone know how to make castellated pads?

[distant raven]

Large via or put header holes on the end of the board

[supple pollen]

There should be a ULP that can export the parts in a schematic as a library

[silent plover]

Hello someone know how I can fix this problem? Without using straps (vias)

I try to connect what is in red together and what is in green together the green corresponds to an encoder
What is in red is the +5v

Here is the routing for the + 5v

And here is the routing for the encoder

[copper lintel]

I tried that, but to no avail. Not sure what I did wrong. If all else fails I think I can try to make a custom part, but I'm worried the pads wouldn't be in the right spot if I go that route. I'm pretty new to Eagle, so I'll tinker with it a bit more to see if I can get it working

[scarlet jay]

https://tenor.com/view/hmm-ok-okay-then-gif-24387952

Well hello, I've finished my board with a little quality on top. Now there is an issue, I will buy the debugger but somebody told me you have to have a microchip on board/bootloader. Can somebody explain to me in a short amount of time which of the solutions to choose? 🙂

[distant raven]

At the very least, you will need some kind of CMSIS-DAP tool to load your initial boot loader onto your board

From there you can program it over USB

[scarlet jay]

no problem, I have been through the whole research but everybody puts something on their board.

like a little chip that is also a debugger but I dont understand how.

[distant raven]

They usually load the chip with a CMSIS-DAP tool that does the same thing. It’s generally only on evaluation boards

It is rare to see that on production boards

[scarlet jay]

okay, so hypothetically if I were to build the device above, use an ATMEL ICE or JLINK BASE on the pinout, is everything safe and sound? 🙂

which is like the last "YES" that keeps me from yelling "Jeronemo" and moving into creating the board, wiring, soldering, programming, debugging and at last prototyping something amaing.

[supple pollen]

You still have a USB-serial chip you don't need, the LEDs should have current limiting resistors (I don't know what the resistor to ground is for, it seems superfluous). Note that you're using an RGB LED but you have all the dice hooked up to the same transistor, so it will just act as a white(ish) LED where you can change the brightness, not the color. VOUT is miswired. And the SAMD chips are hard to get due to the chip shortage, you might want to look at a SAME chip. Your EL_OUTPUT/tDCS circuit is miswired and won't work.

[scarlet jay]

thank you a lot. blessed be.

[worldly schooner]

Do you have two of the same schematic on a single board?

@scarlet jay are these two separate devices or two identical circuits on the same board?

If it’s two of the same device, you only need half that schematic and make multiple boards. If they’re a single device on one board, you can possibly use a single power source and microcontroller to control both halves?

[scarlet jay]

sorry, 2 versions.

[scarlet jay]

https://tenor.com/view/sweating-nervous-wreck-gif-24688521

one question is all it takes falling in love with mey...possibilities.

if I have VOLTAGE COMING FROM TOP at 3.3V and I customize the voltage output, will the current output be customized as well? 🙂

[supple pollen]

It looks like you have the voltage drop across a resistor (proportional to current), plus a base-emitter drop, in the feedback loop of an op-amp, seems like it would work within the voltage compliance. However, 3.3V can only push 1.6mA through 2kΩ, if those are silicon transistors, that drops a little, and the load impedance could limit it more.

[urban grotto]

Is there something like a simple "blink.uf2"? I want to test that my bootloader is even capible of installing a uf2, and am struggling hard

somthing that works on nrf52840, ideally without requiring an external 32Khz crystal

[urban lark]

Okay, I'm a muppet, I just got so excited that a solar panel could power the adafruit reverse tft feather (S2), after testing both battery and solar, then just solar, then removed both and as I stuck the battery back on I got the polarity wrong... Smelled that old electronics smell and it's no longer responsive. What component(s) have I blown most likely, or what's my best way forwards? Seems the neopixel now lights, but screen doesnt, maybe she lives in some respect, yay!

[supple pollen]

Interesting, the Neopixel is powered by the MPU on that board. I'd guess the likeliest parts to be damaged would be the ones connected to the LiPo cell: voltage regulator, charge controller, power steering MOSFET, and fuel gauge.

[urban lark]

I kind of dangerously assume the 3v regulator is okay, hence neopixel and mcu connected pin power

[urban lark]

is there a boost for the tft? like if its battery powered.
I'm not worried about lipo bit working, but its LED does briefly flash on when board is powered (like normal but maybe briefer?). The neopixel should change from white to red as code runs and display starts, but it just remains white.

[supple pollen]

The TFT is powered by the VSENSOR net, which in turn is provided by a second 3.3V linear regulator (not boost) that I admit I hadn't noticed initially. Both regulators are the same AP2112-3.3 type.

[urban lark]

is there a likely candidate fuse/diode I could have demolished, or is it likely to be more severe? I wonder what previous fixes have been required (my searching so far has been unproductive)

[supple pollen]

The ones I listed seem like the likely ones to me. The MOSFET acts as an ideal diode. To narrow down what's working and what isn't, I'd reach for a voltmeter and thermal camera.

[urban lark]

no thermals here, but I can do some testing with multimeter

just curious if there was a "oh he's gone and done the classic, tell him to just replace diode 7 and the ap2112"

[supple pollen]

That would be cool, but I don't have any direct experience with that particular mistake with that particular board. I've cooked plenty of things, just haven't replicated your situation. Alas, I don't know which chips are susceptible, but a little probing should show if the MOSFET is passing power and the regulators are producing the 3.3V. The charge controller and fuel gauge are a little trickier, but if they're not getting hot, they may be okay (and the board would presumably work without them, just without their functionality)

[urban lark]

Thanks madbodger, will report the solution if I remember

[vast flume]

Hey guys, quick question

Making my first STM32 board. Do all the voltage input pins need to have their own decoupling caps or can they all just share one set?

Not too sure about the reccomendations for the proximity

[unreal flax]

The usual pattern is to give each pin a 100nF for local decoupling, but it's fine if they share a single larger 4.7uF bulk capacitor.

[vast flume]

Ah ok awesome

You guys are so helpful hahaha

[unreal flax]

As an aside, if you're doing a 2-layer design, you might want to put the ground pour on the back of the board where it'll be less interrupted by all of the other traces.

[vast flume]

I usually do a groundplane on both the front and back

Is that generally not a good idea?

[distant raven]

No, I’m two layer you probably should have it on both sides

It’s not as important on 4L+ because one internal layer will be ground

Sometimes with 2 layer design, you will need to have a top ground pour to help stitch together the ground if you have a lot of stray traces going through both layers

[vast flume]

I see, thx

Also with the BNO080, is there any reason why I wouldn't want to use the internal clock?

Also one last question, what state should this pin be in if I don't want to reset the chip and just want it to stay on?

[supple pollen]

"active low" means you would set it to a logic low level to reset it, so you'd set it to a logic high level to not reset it.

[vast flume]

OK

So if I just have it connected to 3v3 it'll be good?

[unreal flax]

Sometimes it's an advantage for a sensor to have its samples synced to some other reference clock, like if the processor really wanted exactly 100 samples per second with no drift. But if you don't care about that, the internal clock should be fine.

[vast flume]

Perfect thx

Yeah for my application I don't really care about how long is between each sample as long as I know what that time is lol

[scarlet jay]

might I get your advice on this little HW? I want to make it rather smaller and the PINOUTS especially into a smaller versions, how am I capable of achieving such masterpiece? 🙂

PLUS: Could oyu please check on my (hopefully final) board? 😄

[supple pollen]

This will short D- to ground. You don't want to do that.

[scarlet jay]

okay.

[urban lark]

So having probed it a bit, I think the diode is unhappy, but to be fair it could be anything (no obvious shorts and nothing hot when probed with my lip).
Curiously the white-neopixel is shown before the display is initialised in the factory sketch, but it never progresses, when on USB-C.
I then tried battery and it runs normally!
Adding usb-c is fine and it keeps running (started on battery).
Having just usb-c doesn't work (only white neopixel), and adding the battery while usb is connected causes no change, but resetting the board it boots normally. Battery guage shows 4.2v/98% which is about right or just unhelpful.

This suggests to me it's the battery fuel guage chip, or charging chip, or diode. The mosfets looked alright with multimeter.

[supple pollen]

At least the diode isn't too hard to change

[urban lark]

My thoughts exactly. But seeing the screen alive was a sigh of relief.

[unique patio]

Arduino can use the UF2 bootloader to load a program (in Arduino style, not as a UF2).

[supple pollen]

This P-channel MOSFET isn't going to work here.

[distant raven]

BJT would work better, yeah?

(Me who has no experience driving buzzers lol)

[supple pollen]

An N-channel MOSFET could serve too, it's mostly a matter of taste.

[distant raven]

The mosfet would probably be better power wise?

[supple pollen]

This series capacitor will keep the chip from getting any power: you want a parallel capacitor here.

Hooking a LiPo cell directly to a GPIO pin like this will damage the chip, as the cell voltage is higher than the supply voltage.

Shorting 3V3 to ground is going to cause major problems. Also, you don't want a capacitor here, it will block the current flow through the LEDs.

[supple pollen]

Yes, MOSFETs are generally easier to drive, and modern logic level ones can offer lower voltage drop than BJTs.

I have no idea what is happening here. The op-amp output goes to an X which I think means "no connection" but touching a wire marked "induction" which doesn't seem to go anywhere. Then there's the pair of inductors that seem to have a combination designation and part number like "L410µH", the input signal seems to go to both the positive input (via R43) and the negative input (via R37). And the power supply pins aren't connected? I'm not quite sure what this is intended to do, but I doubt it's going to have the desired functionality.

[worldly schooner]

@scarlet jay I suspect that capacitor belongs between 3.3v and GND. Not 3.3v and LED.

[scarlet jay]

@worldly schooner Thanks G!

@supple pollen Thanks, m8 once again as I've made this a little bit of a board review.

https://tenor.com/view/oh-really-hmph-surprised-gif-18286648

[scarlet jay]

Based on the schematic I upgraded the file (I just want to say that this sole project will take another 2 months to deliver so plenty of time to get it together.

https://tenor.com/view/your-welcome-gif-23991412

[urban lark]

@supple pollen so I wonder what causes this then... The right side... It's a working unbroken one, but at some point last night it freaked out, then left side shows after reset. Seems to show no i2c device so fuel guage freaked? (The NO one is the broken one) -- Turned out to be recreatable by pulling battery out.
Oh and the broken one, so far only identified the orange charge LED is gone, and main shottky is iffy, but think we're still able to charge right now. More magic smoke testing the charge ic, but still working afterwards (replacement on order + fuel guage).
Turns out the fuel guage on broken one responds if battery powered, otherwise not there and misbehaving. Ordered from

[scarlet jay]

Sup Gs!

Welcome to the wonderful world of P.A.T.R.I.K!

I think I finished my file but I might be wrang!

https://tenor.com/view/saloon-enter-crowboys-gif-24849059

[distant raven]

D- here should be GND.

you're shorting EN and D0_BTN to ground with that bottom trace, you don't get any button input with this.

as mentioned before, this will block any DC voltage. tie one lead of the capacitor to VDD, tie the other to ground.

i'm confused why you have two connectors with a battery lead.

[hushed smelt]

Think I'm going to have to give up on trying to tie the 3W amp's I2C GND to the 20W (12V) amp's I2C GND. Too much noise and the ground potentials are too different. Only found 1 place online that has a 20W I2S amplifier and scooped it up this morning. Adafruit doesn't have a way to power more than a 3W speaker from I2S.

The 20W analog amplifier works amazing by itself with a line level input. From the 3W I2S module... tons of noise. 😦

[silver marsh]

Are you feeding the out of the 3w amp into the 20w amp?

[hushed smelt]

maybe 👉 😅 👈

The 3W I2S amp doesn't have an AGND so I can't tie them together. 😦

Tying the 20W GND to the feather produces wicked hum.

[silver marsh]

The output of the 3w amp is inherently noisy and not ground-referenced. Both of the 3w amp’s output pins are floating. Neither can be tied to ground without damaging the amp.

Should be using an I2S DAC instead.

[hushed smelt]

Yeah that was kind of my thought. There's noise either coming from the feather or the 3W I2S module... the GND is very very poor.

I have it soldered the best I can, it's just the GND for the module is really only designed for a 3W max.

[silver marsh]

The 3w amp is designed to only drive a speaker.

[hushed smelt]

yes yes i know i'm trying to cheat 😦

Can you explain more about the I2S DAC and how I might go about that instead? Is there a different board I should be using?

I tried the ESP32-S3 first but that doesn't have any MP3 capability which I really prefer.

[silver marsh]

The only possible way to use it in the way described is to get an old speaker output transformer and wire it backwards. Would still be pretty noisy.

[hushed smelt]

ok, I need the 20W of power for the 2x 10W speakers. How would you do it?

I have almost every kind of Adafruit board at my disposal.

Prefer using the feather form factor as it's going to be used with an airlift featherwing and LORA featherwing.

ohhh it's because i'm using something that's already being amplified where as a regular DAC isn't?

i tried lowering the gain on the I2S with a resistor but of course that made little to no difference.

[silver marsh]

I have some of those old transformers, so I’d start with those. Some pro audio DI’s use them, so that could be a source.

Also https://www.adafruit.com/product/1381

[hushed smelt]

I completely missed that one and didn't know it existed. Thank you.

I kept searching for I2S and amplifier.

I legit need that as a featherwing. I might design an adapter.

[silver marsh]

Don’t know if there’s a CircuitPython driver though.

[hushed smelt]

ack

Seems there is whew. https://docs.circuitpython.org/projects/vs1053/en/latest/api.html

[silver marsh]

That’s a pretty old one that looks like it has limitations. Worth a shot though.

Then there’s this:

[hushed smelt]

That one is discontinued though 😦

[silver marsh]

Another option would to build one from scratch using an available DAC like https://www.digikey.com/en/products/detail/cirrus-logic-inc/WM8524CGEDT-R/5036730

[scarlet jay]

okay and do you think you could possibly check "EL_OUTPUT"? 😄

also if you want to check some of my latest news about the project, check out https://twitter.com/SlachtaPatrik

Im going to give away a few

[vast flume]

How should I know if I need a switching power supply on my board?

It runs at a 3.3v logic level but would a switching power supply increase the battery life if I'm just using a 1s lipo since it can drain the battery lower?

[scarlet jay]

yes.

[vast flume]

Awesome that was what I suspected but just wanted to make sure

And then on the other side of the spectrum, how do I decide if I need to use a switching power supply for stepping down voltage? If I'm powering a 3.3v controller off of a 4s for example

[scarlet jay]

since i just researched that but I would make it a little different, why dont you put its connectors into microchip and control it so that when unnecessary, it will run on a lower pace?

like D+ and D-

put it straight from battery to controller and make an extra layer of code (if im correct) and form a standby mode

[unreal flax]

A typical switching supply might be 80% efficient, but a linear supply would discard all the excess voltage from a 4S pack and be only about 20% efficient, so a buck would be a good choice there in general.

[vast flume]

Ok perfect thanks

[woven bluff]

what is the connector used in stemma qt?

S4B-PH-SM4-TB?

I don't recall stemma qt being this huge

2mm pitch?

[worldly schooner]

That's the JST PH connector, typically used for battery and Stemma connections. Stemma QT uses JST SH, and the connector you're looking for would be SM04B-SRSS-TB(LF)(SN)
https://www.digikey.com/en/products/detail/jst-sales-america-inc/SM04B-SRSS-TB-LF-SN/926710

[woven bluff]

What is Stemma

[worldly schooner]

Alternatively, there are compatible clones like https://www.adafruit.com/product/4208 which aren't JST originals, but are slightly cheaper.

[worldly schooner]

STEMMA is the predecessor to Stemma QT. https://learn.adafruit.com/introducing-adafruit-stemma-qt

[woven bluff]

ok

thanks

[copper lintel]

Sorry if this is a dumb question, but on the QTpy and Xiao boards, can I connect a 3.7v LiPoly battery to the 5v in pin to power the board? I dont want to supply power from the 3.3v pin, as I have a device that needs exactly 3.3v coming from that pin, and connecting the battery there would screw with it.

[supple pollen]

Let's see the QTpy has an AP2112K-3.3 regulator, which has a dropout voltage of 1-1.3V. I would have expected a lower dropout voltage so it would work with a lithium cell, but that doesn't appear to be the case.

[copper lintel]

Thanks!

[worldly schooner]

The QTPY does have VBAT, which allows you to still use the USB port with a battery soldered. If you use the 5V pin, the 5V can flow into a LiPo and charge it in a dangerously uncontrolled fashion.

Oh, exactly 3.3V would be tough for the LDO.

[copper lintel]

Doesnt need to be exact, but I probably shouldn't hook up the part Im using directly to the battery, as the datasheet says no more than 3.7v

[supple pollen]

It might be okay then if it's content with 2.6V or so.

[glacial gale]

what ways of monitoring and reacting to user input is there other than interrupts and polling? Are there ways to do so very cheaply/efficiently?

[woven bluff]

what's the pull-up resistor value for sdcard?

[supple pollen]

Interrupts and polling are the cheapest and most efficient options. The other approaches tend to involve shift registers, latches, and dedicated input chips.

[supple pollen]

I suspect it depends somewhat on the supply voltage you're using, but I doubt it's critical. Maybe 10kΩ would work?

[distant ivy]

Hello everyone, I am trying to use microsd cards with RP2040s however I am experiencing problems, and I have no clue why. I have tried three different readers 3 different and sd cards (64gb sandisk, 16gb sandisk, 32gb intenso) 2 different pis and with circuitpython and cpp. I can't get any sort of communication going with any of the sd cards, the arduino SD.begin just returns false and the circuitpython code (where I recomplied cpy with the debug flag in the sdcardio module) returns this: ```cmd 0 [00] arg= 0 [00000000] len=0 data
cmd 0 [00] arg= 0 [00000000] len=0 data
cmd 0 [00] arg= 0 [00000000] len=0 data
cmd 0 [00] arg= 0 [00000000] len=0 data
cmd 0 [00] arg= 0 [00000000] len=0 data
Traceback (most recent call last):
File "main.py", line 8, in <module>
OSError: no SD card

[unreal flax]

How carefully have you checked the hardware connections? This smells like a wiring issue of some sort, either physical (loose connection) or logical (good connection to the wrong pins).

[distant ivy]

I switched around the readers a few times (one dedicated and two on tft screens) with different pin layouts

I also smell wiring (metaphorically)

But like

ahhhh

[low anchor]

What kind of connector could I use to connect two PCBs that are parallel with each other, edge-to-edge?

[rustic linden]

do you mean inline? because parallel is one above the other, which in i would suggest your standard pin headers

[low anchor]

Oh yeah, inline. Like on the same plane

Although I suppose I could use right-angle headers & a right-angle socket

[rustic linden]

I'm personally a fan of card edge connectors, but other than that I'm not sure

[rustic linden]

from experience they're not usually the same height?

[low anchor]

Ah hmm I'll look and see what I can find

Issue I'm having is it looks like most card edge connectors are for one card to plug into a board like a PCIe connector

The other thing headers would be good for is if I end up needing to put a cable between the boards

[rustic linden]

that's totally fair!

[supple pollen]

That's what I did. There are board-edge-mount solutions, but they tend to be a little more finicky.

[dry pelican]

Is it possible to solder one of these with hot air? I'm using USBC in my next project (which funnily is a hotplate for reflow soldering)

the stakes might be an issue

or maybe not

but the plastic melting could be an issue

[distant raven]

Yup, it is

I use these all the time

[dry pelican]

just fill the holes with paste?

[distant raven]

Yup

Or leave the aperture open for a stencil

[dry pelican]

and how do you not damage the plastic?

[distant raven]

Ladyada talked about this this last desk of Ladyada

It won’t damage the plastic because it’s rated for like 400°C

[dry pelican]

I've had issues with qwiic connectors that I soldered with hot air

they didn't make contact to the cable

[distant raven]

Your hot air should not exceed like 300°C when soldering

[dry pelican]

I suspect it's because the plastic melted a bit

the contact resistance is a few kohm

[distant raven]

Probably soldering too hot with your hot air

[dry pelican]

I soldered it at 260C

[distant raven]

These things can be soldered with nearly pure tin paste which melts close to 300C

It should talk like, maybe a minute for the solder to melt

[dry pelican]

maybe I blasted the connector with too much hot air

[distant raven]

Same connector

I used hot air on this board

[dry pelican]

Why does the board have no soldermask?

Or so it seems

Wait maybe it’s OSHpark’s after hours transparent solder mask

[distant raven]

Yeah, after dark

[dry pelican]

What would be the best way to use a PCB trace as a temperature sensor? I’m making an inexpensive hotplate that is powered by USBC PD. I think I would just use a voltage divider (with the trace as the bottom resistor) and a high gain amplifier that feeds into the ADC of the RP2040 I’m using. Maybe I should use an op amp set up as a differential amplifier and us a Wheatstone bridge.

The heating element will be a low resistance trace at the top of the heating board, and the temperature sensing trace would be a higher resistance trace on the bottom

[woven bluff]

did you assemble this board yourself?

[supple pollen]

While copper does have a temperature coefficient, it makes a pretty poor temperature sensor. It's a lot easier to use a diode, thermistor, thermocouple, or even a dedicated temperature sensor.

However, you might be able to make the heating trace do double duty: as it will have a largish current flowing through it, it would have a commensurate voltage drop: a much easier signal to measure, so if you measure the voltage and current, you could calculate the resistance, and thereby the temperature.

[distant raven]

Yes

[woven bluff]

with reflow?

[distant raven]

Yeah

[woven bluff]

I don't have reflow... and always end up meting the plastic parts with hot air

[woven bluff]

I use THT version of this, the leads are shorter than the thickness of PCB, so the back is still flat

[crimson ocean]

hi im new here where would be the best place to ask for help regarding custom made pcb problems 🙂

[rustic linden]

here!

[crimson ocean]

looking at this it looks like to my limited knowledge that the digital pins have pull up resistors if the vcc is lets say 6-12v would that not damage the esp32 ? (esp32 max voltage 3.6v)

[dry pelican]

What is VCC? 12v?

[crimson ocean]

It would be between 6 and 12v depending on the battery

[dry pelican]

I was planning to integrate a current sensor anyways, so that might work. I found the INA237 IC which has a 16 bit resolution, so it should be able to do the job. The only problem with the method is that it would only measure the temperature when the hotplate is heating

I was also thinking of using an INA219 with a wheatstone bridge that has the temperature sensing trace on the bottom of the PCB as the unknown resistor. Thie INA219 would just read back the voltage difference, and the resistance could be calculated

[dry pelican]

I don't think you would need the pullup resistors on the ESP's side. The ESP should be able to pull those outputs low. You would actually want pull down resistors

[crimson ocean]

Thank u just so I learn something the 12v vcc pull up resistor would damage the esp32 right

[dry pelican]

most likely

[crimson ocean]

Thanks thought so

[dry pelican]

I'm honestly not 100% sure (because of protection diodes) but I wouldn't risk it

[crimson ocean]

On my current pcb I had pull up resistors linked to 3.3v but I was getting brown outs on the esp32

[dry pelican]

that's weird

brownouts would be because of esp32's power supply

[crimson ocean]

Yer I tried adding caps did nothing when I removed the resistors it worked

[dry pelican]

what value were the resistors

[crimson ocean]

10k

[dry pelican]

connected to esp output pins?

[crimson ocean]

Only brown outs on my l9110s I have 4 on the board

[dry pelican]

oh

hmmm

[crimson ocean]

I don't run them all at the same time one at a time to test

[dry pelican]

that's a weird issue

if removing pullups solved it then I guess that works

[crimson ocean]

Yer lol been trying to work it out for 2 days

[dry pelican]

Adafruit's example doesn't use pullups

maybe it's just how the inputs work

[crimson ocean]

Can't understand why 3 l9110s work but this one brown out the board

[dry pelican]

the chip is not very well documented

pic of board?

might be decoupling

make sure decoupling caps are close to chips

each one should have 1 or 2 caps

or more

RP2040 for example likes to have 11 caps for good stability

[crimson ocean]

Yer I think that's the problem lol as it goes vcc the l9110s then caps

And the other 3 go vcc caps driver

[dry pelican]

you may also want some chunky electrolytic decoupling caps

[crimson ocean]

yer i was thinking of that but i was trying to work out the problem 1st

[dry pelican]

because the motor drivers draw a lot of current and parasitic inductances can cause voltage to drop when there is a rapid current change

decoupling caps provide a nearby current/voltage source

[crimson ocean]

was trying to keep the board as small as possable 🙂

[dry pelican]

also you might want thicker traces going to power signals

like VCC to drivers and motor output traces should be 1mm thick

at least 0.5mm

or just have larger value caps

[crimson ocean]

what i dont understand is why it browns out when the driver is not even on surely it would not draw that much current to brown out the board

[dry pelican]

like 10uF as well as 0.1uF

[crimson ocean]

yer i have both

2 for each motor driver

i think maybe i doint have 0.uf and i just used 2 10uf

[dry pelican]

those look like 0603 caps

[crimson ocean]

let me check 🙂

[dry pelican]

a 10uF one would maybe only be able to take 6.3v before blowing up

that's usually their rating iirc

[crimson ocean]

they are

[dry pelican]

that could cause brownouts

[crimson ocean]

https://www.lcsc.com/product-detail/Multilayer-Ceramic-Capacitors-MLCC-SMD-SMT_Samsung-Electro-Mechanics-CL10A106MA8NRNC_C96446.html

[dry pelican]

if your caps have low rating

no they're 25v

that should be fine

[crimson ocean]

what i dont understand is why top right and bottle right and bottle left all work fine

but top left does not

[dry pelican]

just make traces thicker

[crimson ocean]

and the only thing that fixed it was removing the resistors 🙂

[dry pelican]

that's weird

[crimson ocean]

i will in the next version 🙂

[dry pelican]

not all traces, just higher current carrying traces

[crimson ocean]

the bottle left even gets its vcc by going past top left

[dry pelican]

and add ground plane

[crimson ocean]

yer this was jsut a quick prototype 🙂

as i was not 100% sure this would work

[dry pelican]

the pcb does look pretty nice

[dry pelican]

but it definitely needs thicker traces for power and a ground plane

[crimson ocean]

yer defo

i know the traces are a mess lol my new version is way better but im trying to get the top left motor driver to work 1st befor sending off the new version

[dry pelican]

I guess try some bodge wires and bodge caps

[crimson ocean]

i tryed holding them down but im guessing i need to solder them

[frank onyx]

so im trying to control a small DC motor (one of those plastic gearbox DC motors) or two with a PWM signal from an ne555+potentiometer. I have the NE555 circuit working and I'm able to generate the PWM signal. I'm trying to choose a suitable MOSFET but im not sure if there needs to be special considerations made since its going to be PWM controlled. I'm not really sure what im looking for in the datasheet. does this seem like a reasonable choice? https://www.lcsc.com/product-detail/MOSFETs_Alpha-Omega-Semicon-AO4485_C51499.html
Further, I'd like to have direction control as well, and I'm planning on having a switch that swaps the output of the mosfet and +V.
This is the approach that I've taken with previous attempts using motor driver ICs, but they seem to work great at low speeds but die when switching directions at high speeds- which I wasn't expecting since they have inbuilt flybacks, as motor driver ICs. But these were only rated for ~2A or so, which is why im thinking to move to general MOSFETs and add an external flyback because they have much higher current capabilities for the same or cheaper price.

this is the circuit for the MOSFET. the PWM signal being generated by NE555 isntead of the Arduino (used arduino for convenience)

and this is the circuit for the current iteration of the board, which is killing motor ICs at higher speed direction switches

[dry pelican]

the killing ics problem can maybe be solved by adding a capacitor across the motor and adding some flyback diodes to V+ and GND on both motor terminals

Also. Having a mechanical switch seems like a problem

You can use the motor driver to switch directions

And it should be able to handle it better

A single mosfet cannot make the motor switch direction. For that you need an h bridge

Another thing that could kill the driver ics is a voltage spike on the input caused by parasitic inductances

To limit that, you need more decoupling

[dry pelican]

@frank onyx

[woven bluff]

any comment of this SMPS would be helpful before I send it to fab

[supple pollen]

That feedback path is a little meandering, could pick up noise

Also, you probably don't want silkscreen under the thermal pad of Q1

[dry pelican]

That fly back diode is connected wrong. Should be connected reverse biased across motor.

[dry pelican]

Does this look good (especially on the current measuring ICs and USB C PD side (especially the INA219 used as a differential amplifier))?

I still need to add some things like decoupling and outputs

[dry pelican]

Why is the VOUT trace so thin if all the components leading up to it are high power? There are some other traces that I would make thicker as well.

[woven bluff]

0.5mm, it only has to carry 20mA max

[woven bluff]

yeah, but the decoupling cap is more important

[woven bluff]

there is no silk under Q1 pads, that's net name

[balmy tide]

Can't you decouple to the c1's ground?

[scarlet jay]

Hey guys, I need your help: I need this but other way around 😄

2 outputs 1 input

[worldly schooner]

What do the outputs do in relation to the input?

There isn’t such a thing as a single-input op-amp, as far as I know.

[scarlet jay]

thank you. Well...there goes a current from 1 board to this amplifier. This amplifier splits the current into 12 linear voltage regulators which all end in the same pin. This pin is the input of another amplifier which has 2 outputs. these two outputs then turn into 1 output which goes to deliver current.

@worldly schooner

[worldly schooner]

Um, what exactly are you trying to do, and why do you have all this splitting and merging?

[scarlet jay]

complicated but I am trying to make a very precise current output. Meaning it will be 0.0001mA precise in the end.

[worldly schooner]

Tenth of a microamp? Do you have a circuit reference?

[supple pollen]

What range do you need? If you wanted 0-1mA, that would be like 14 bit resolution.

Also, what is the load impedance? How much voltage compliance do you need?

[dry pelican]

https://hackaday.io/project/159693-precision-current-generator

Here's a simple precision current generator

with the 10th of a microamp requirement, you're probably delving into high precision expensive components territory

you would need a super precise shunt resistor

and probably temperature compensation with the transistor that controls the current

and a super precise op amp

and that type of precision is rarely ever needed

[scarlet jay]

@dry pelican @supple pollen@worldly schooner I need this basically.

[dry pelican]

You could just use a bunch of these circuits

with higher shunt/sense resistor (replace the 150 ohm one) of course

and then they're all controlled by the same voltage source (resistor divider/potentiometer)

[scarlet jay]

@dry pelican Thank you a lot! 🙂

[dry pelican]

Make sure to be safe with the tDCS

make sure the resistor is not too little that it passes too much current

although skin resistance is pretty high

[scarlet jay]

thank you a lot! 🙂

[woven bluff]

can you connect a MCU to cc pins of usb-C to get whatever voltage you want?

by emulating PD controller

[supple pollen]

I've only done a little reading about it, and my takeaway is that the protocol is complex and subtle, and analog in nature, so you may need both a DAC and ADC and possibly voltage-to-current converters to implement the PD protocol.

[rustic linden]

at that point wouldn't a dedicated PD chip be cheaper? or are they prohibitively expensive

[worldly schooner]

I would imagine that is an option, though an argument could be made for digipots as well?

[supple pollen]

I can understand the desire to save a chip, and it's something I've looked at in multiple projects. I've seen projects implementing a switching power supply with a few spare pins and some clever code instead of a dedicated switching regulator. I had one project where the notion was to implement an ohmmeter without custom analog circuitry. I ended up taking a creative approach instead, which garnered me a patent, but it did the job and didn't need any additional circuitry. So it's worth considering for PD, but I don't know enough about it to know if it's practical. Just running the CC signals to a couple of GPIO pins isn't going to cut it, and the chips I'm familiar with that have both ADC and DAC capability onboard don't have them on the same pins.

[woven bluff]

PD chips are all BGA or FN

analog in nature? to this date? whoever designed it must have brain cancer

[rustic linden]

that's not a nice thing to say

[woven bluff]

it's not, the same goes whoever named USB3.x protocols

[worldly schooner]

Oh, wait, no, PD isn’t something where you can select within a continuous voltage range anyways. You would only be able to select between several discrete voltages, and even then only if the supply supports those outputs.

[woven bluff]

I'm aware of that

I do have multi-voltage PD upstream

I just need it to cough up 20V

[worldly schooner]

Something like https://www.adafruit.com/product/5501 perhaps?

[woven bluff]

humm, Chinese chip

I don't trust closed source stuff at embedded level

the chip is small enough to hide in the connector

[worldly schooner]

The point is it does exist as a chip. If you want to find your own and don’t mind a slightly more expensive one, TI makes plenty such chips IIRC

[supple pollen]

I'm not certain, but it kinda looks like it to me

[worldly schooner]

Unless you want to go the diy route and use digipots as CC resistors you can freely configure on the fly. You’ll have to figure out your own startup sequence if you do go that route.

[woven bluff]

I just don't have the resource or skill to solder FN chips

I have a feeling they deliberately made PD complicated to sell permits for making PD chips

[supple pollen]

If the question begins with "Why did they...", the answer is probably "money".

[woven bluff]

I have seen the growth of opensource hardware in recent years, but it's still to slow.

[worldly schooner]

Modern fabs save a lot of money by not making QFPs. QFN and BGA are much easier for more modern processes.

[knotty tiger]

there’s an analog power negotiation on USB C CC pins, consisting of pullup and pulldown resistors. there’s also a digital protocol (USB PD) that happens on the CC pins

[worldly schooner]

Most QFN soldering isn’t really that bad though. Just extend the pads enough to have space for your iron, and use a paste syringe instead of solid solder.

[supple pollen]

You can always have someone like ProtoAdvantage assemble a custom breakout board for you.

[worldly schooner]

BGAs are best serviced with a hot air station at minimum.

[woven bluff]

paste is the problem, when heated, solder paste surface tension changes drastically

and the paste would spread everywhere before melting

[knotty tiger]

the analog negotiation doesn’t let you ask for more than 5V, nor more than the max allowed current for the device speed, unless the host advertises “Type C current” with its pullup resistor

[worldly schooner]

They generally flow on hot copper much more than they do soldermask. I don’t think I’ve ever had an issue when heating one pad at a time.

[woven bluff]

It's disastrous trying to find out which pins shorted underneath the FN chip

[worldly schooner]

It does take practice.

[woven bluff]

I think I have no choice but to include FN chips in my designs

[worldly schooner]

Hand soldered boards aren’t really the target consumer for most IC companies, unfortunately

[woven bluff]

they have never been

[limpid nest]

Tips for sending this signal several feet to an MCU, conditioning it?

[supple pollen]

Those can produce some high voltage signals, so conditioning might involve a current limiter (like 100kΩ series resistor) and voltage clipper (possibly a zener diode or two).

[limpid nest]

Ok there's a seller that sells them with that hardware. What about distance? Thx

[supple pollen]

I suppose distance depends on signal to noise ratio, desired performance, type of wire, interference environment, etc.

[worldly schooner]

Depending on your goals, it may be worth considering an mcu to digitize the signals before transmitting?

[limpid nest]

Yeah, I'm trying to get, "Is this thing vibrating above a background level, and how much more, roughly?"

[supple pollen]

You could have an envelope detector local to the sensor (it could be passive and not require a power supply) and then just send the DC(ish) level back as another option.

It's also easier for a processor to monitor a slowly changing signal like that in a polling loop than try to follow a waveform directly.

[limpid nest]

Why would the above mentioned conditioning board have a plated hole?

[supple pollen]

Most board houses offer plated through holes these days

It's not really a feature worth mentioning (it's like putting "can use an email program" on a resume), but many of the product descriptions on sketchy websites throw in random stuff like that.

[limpid nest]

Here's what I'm referring to. Also why is this active?

[dry pelican]

You might be able to convert the signal to a differential pair and amplify it at the other end. That removes all common mode noise. But the least noisy and lossless way is just to digitize it with an ADC or smth

[limpid nest]

My concern is sending the data over a wire

[distant raven]

@silk lark what thickness boards did you use for your PCB coin cell battery holder? I wanted to try it out on a little Bluetooth board

[silk lark]

0.6mm

[distant raven]

Sweet! Thanks 🙂

Is there a tab length that’s ideal for it?

[silk lark]

that also works for the microusb socket

[silk lark]

at least 1.5cm, shorter tends to break

[distant raven]

Makes sense, you did kind of a bone shape for yours? Or was it a straight piece?

[silk lark]

I did 1.2mm and it's barely long enough

of course it depends on many things, not all pcbs are the same

[distant raven]

True, I might make a test board to try out a few different styles

[silk lark]

I also left the contacts on top of it, so you can still solder a battery holder and use it normally

in hindsight, I should have rotated the battery holder footprint 90° so that the contacts are on normal part of the pcb

[distant raven]

Yeah, just in case the tabs break

[silk lark]

note that it also matters on which side of the board you put the traces

they may break when the pcb is bent if they are on the outside of the bend

@distant raven ^^

[distant raven]

Oh that’s good to know, thanks

[limpid nest]

Could the hole be for mounting?

[supple pollen]

I don't know, it appears to just have a resistor and (zener?) diode.

[cinder spruce]

is this circuit sane?

[rustic linden]

I don't immediately see anything wrong with it, but that doesn't mean that someone else won't see something i missed

[cinder spruce]

thanks. figured i'd get an opinion before i did a test

[spice turtle]

I have doubts that triac will actually turn off on the 2nd half of the wave due to the rectifier (it would be 1.4V above 0) . Id honestly just use a MOSFET with the isolator after the rectified DC.

Oh, also R7 needs to be pulled HIGH, not low. So Pin 4 would go to ground, and then pull pin 5 (the collector) high with the resistor. Ive found you need a very large resistor (100k) in order for it to work

[cinder spruce]

i was going to use a power mosfet, but i killed them somehow 🙂 they conduct forever now. i have a triac and some scrs. but yes, they seem more difficult to turn off

[cinder spruce]

i'm curious about r7 though. is the impedance too high at the mcu input therefore this is backwards?

at this point i'm just trying to get this dc motor spinning again with what i have on hand

it's proving to be challenging for my electronics skill level

[spice turtle]

You have to have a certain amount of current flowing through the collector in order for the transistor to turn ON basically. I personally experimented with it, because at low values, I got nothing. I kept turning it up, and got closer to an actual pulse. It would look like a pulse, but not really. It has to do with the way a photo transistor works basically. The transistor is acting as an amplifier of sorts and it doesnt need much current

NOTE!!!!!!!!!!!!!! PINS ARE BACKWARDS. The model is correct, the schematic is not.

[cinder spruce]

cool. i can easily scope that part to make sure it's a good pulse

[spice turtle]

thats what I did

[cinder spruce]

but i'll go with pulling hi

[spice turtle]

this is what I mean by backwards

I never changed the schematic lol.

if you look up the datasheet, its correct.

Pin 3 is the collector 🙂

[cinder spruce]

cool

are there any simple ways to get that triac to turn off?

[spice turtle]

have it cross 0V. Someone else might have to pick that up as Im not 100% sure

[cinder spruce]

one part i don't seem to understand is how the gate wouldn't see both half cycles crossing zero

[spice turtle]

thats what the ZCD is for, you have to send it a signal every crossing

[cinder spruce]

i get that, so i fan fire it at whatever angle

[spice turtle]

well, what you would do is this:

I actually have a flow chart lol

So in this case, I am using a Timer to time when the gate is OFF. When the gate is output of my micro is ON, the gate is OFF. When the micro's output is OFF, the Triac is ON.. Its reverse logic.

So basically, you sense when ZCD happens, Start a timer, turn OFF the output (triac is conducting) and then when the timer expires, you turn the output ON to shut the triac OFF.

Make sure to keep the timers overflow value at less than 8ms (which is a half wave)

[cinder spruce]

the part i'm having trouble with now is understanding "I have doubts that triac will actually turn off on the 2nd half of the wave due to the rectifier (it would be 1.4V above 0)". is this the forward voltage of one of those diodes? and if that's the case, wouldn't we still cross zero at some point?

[spice turtle]

Oh. Yes. Its the forward voltage of two of the diodes

Thats why I said, Im not 100% sure, because, the AC wave does cross 0, but youve also got two diodes in series with it

BTW, big tip, when you test this out, use a transformer on the Input side, like a 5VAC or 12VAC transformer. That way you can poke around with a oscope

[cinder spruce]

found out about scope poking. was connected via a GFI which was nice

could have saved the scope

i have a large welding transformer. lol it's probably like 26v

[spice turtle]

that might be OK but I try an seek out something smaller

and that doesnt have a ton of current

[cinder spruce]

makes sense

[cinder spruce]

@spice turtle so based on the way the opto works, this is more appropriate(regarding the pullup)?

[supple pollen]

Basically two diodes, but that assertion is not entirely correct. The triac doesn't turn off due to voltage going through zero, it turns off due to current going through zero.

[cinder spruce]

gotcha. so in this case, would the triac turn off?

[supple pollen]

I think it would, but I'm not sure: it's not the diodes, it's the phase shift due to the inductive load.

[dry pelican]

the FBR kind of acts as a flyback though...

well flyback diode

not flyback transformer

I think it would work

but maybe fire up a simulator and see if there is a current zero-crossing with a large inductive load and an FBR

[cinder spruce]

this is with just a resistor

[supple pollen]

Yeah, totally fine with a nonreactive load. I know some phase control dimmers have issues with reactive loads, but that's partly because their phase timing comes from the line. If you're just using it for on/off control, it'll be fine. If you're trying to do phase control, you may have to adjust the zero crossing somewhat to align with the phase shifted load. Or perhaps not. I'm not sure.

[cinder spruce]

it's not near zero for very long either way though

[supple pollen]

Doesn't have to be.

[cinder spruce]

how should i define the inductor?

[supple pollen]

I don't know anything about your motor. Maybe try 1H to start with?

[cinder spruce]

it's a 120v 5000rpm ~300w brushed motor

the current shape gets a bit more pointy

[supple pollen]

Ah, that's a much bigger motor than I was envisioning. Hmm, 100mH gives an inductive reactance of about 38Ω at 60Hz. Add in a series resistance of 10Ω, should give about the right current.

[cinder spruce]

yeah. i don't know enough about inductance to simulate this apparently

get wacky current results

[cinder spruce]

last one

i'll have to figure out isolation so i can scope this

if i unearth my scope for things like this, will the electron gods come for me?

i gotta just find a transformer i guess

[dry pelican]

Looks like there is a zero-crossing

At least according to this sim

[cinder spruce]

nice

i guess i'll just give it a shot. thanks for the help everyone

[outer cliff]

What does this mean? It's from ATM90E26 application note. Is pin 1 Vdd and pin 28 GND?

[unreal flax]

Those looks like jumpers of some sort, which would let you select whether each pin is high or low.

[outer cliff]

Thanks, I did not think about jumpers

[spice turtle]

Yes 🙂

[woven bluff]

is there any compact option to get air plasma?

ZVS is too large, one big inductor...(two, if the transformer has no center tap)

flayback convertor without feedback? not sure what timing chip to use.

[dry pelican]

555 driven flyback circuits are often used by HV hobbyists.

[supple pollen]

Yeah, look at the circuitry in arc lighters

[woven bluff]

555 is older than me

[dry pelican]

Yes. You're basically building an arc lighter

[dry pelican]

Lol

[woven bluff]

I'd probably use a PIC timer

555 cannot control frequency and duty independently, and require a lot of passives

then there is the issue of gate driving...

[vast flume]

How ambitious would it be to design a pcb for driving a brushless motor at > 100 amps?

[supple pollen]

The advantages of the 555 are its robustness (it's not easily fried by noise on the power supply line or feedback into its output pin), its voltage capability (if you run it on 12V, you'll get a 12V signal on its output), and its current capability (it can source or sink 200mA).

In other words, you can use a 555 as a self-oscillating gate driver. Adding a transistor booster makes it an even better gate driver. There are, of course, gate driver chips available too.

My approach is, as usual, unusual

[supple pollen]

Seems ambitious to me, but I don't have a lot of experience with high current engineering (I'm more of a high voltage engineer)

[dry pelican]

The 555 is super robust, doesn't require programming like a pic (I've also found that microcontrollers ca produce inconsistent pulses). It also can do 300mA output iirc (so it's better for gate driving than a 20mA pic). Also, 555 can control frequency and duty independently. But you are right in saying it does need a lot of passives (and also some calculations to find values). A microcontroller is definitely more flexible. Your frequency will probably range from ~20kHz to ~200kHz. You just have to see what frequency works best. If you make one with feedback, you can use a royer inverter or one of those cheap single bjt arc lighter circuits.

[supple pollen]

A Royer inverter doesn't really incorporate feedback (that's what's on the CCFL driver in my pic), it's more of a resonant push-pull driver. I've added external feedback to mine to regulate the rectified DC voltage.

[dry pelican]

Maybe ambitious. At that current, you're probably using TO-249 fets that are just soldered to wires instead of mounted on a pcb. All the low power stuff would be on a pcb.

That's not a definitive answer. Just extrapolated from tesla coil circuits I've seen

[supple pollen]

I was thinking TO-3 transistors or monsters like an ET227, but yeah, many times in high-current circuits the transistors are chassis mounted and connected with bus bars and the like.

[dry pelican]

Some screenshots of the PD powered hotplate with PCB trace temperature sensing. I had an INA219 working as a differential amplifier to amplify changes in resistance of a sense trace at the bottom of a hotplate PCB, but I might cut that out since I can pulse the current through the main trace and measure the current with the main current amplifier suring the cooldown cycle. Also I can dynamically adjust the PD supplied voltage to be lower so the plate doesn't heat up much when I'm measuring its resistance during the cooldown portion of the curve. I can probably cut out some more parts (such as some decoupling caps) to save on cost, complexity, and assembly time. Are there any other routing things (specifically RP2040 power traces that may affect stability and digital signals crossing other signals (if that will interfere with anything)) or things I can cut out without sacrificing too much performance? I'll probably make the RP2040 decoupling caps 0603 or 0402 since they will be much less bulky and allow for easier routing. I have yet to do more routing, add the hotplate contacts and expansion header (to break out SPI, I2C, UART, some ADCs, and some GPIOs), add a ground plane, and finish it up.

Yes this is a repost, but I thought I should put it in the right channel

Basically the question is: how many decoupling caps can I remove from the rp2040 circuit?

[vast flume]

Can somebody help me figure out the values that I need for the parts for this voltage regulator? https://www.ti.com/lit/ds/symlink/tps5405.pdf?HQS=dis-dk-null-digikeymode-dsf-pf-null-wwe&ts=1677979851110&ref_url=https%253A%252F%252Fwww.ti.com%252Fgeneral%252Fdocs%252Fsuppproductinfo.tsp%253FdistId%253D10%2526gotoUrl%253Dhttps%253A%252F%252Fwww.ti.com%252Flit%252Fgpn%252Ftps5405

I guess the first question is how do I know what switching frequency I want?

[supple pollen]

That's an older chip, there may be easier to use options available. But as the data sheet explains, it's a tradeoff: a higher frequency allows smaller inductors and capacitors, while a lower frequency allows higher efficiency. There is also an effect on inductor ripple current, which depends on both the frequency and input voltage.

[vast flume]

Which chip would you recommend? I am stepping down a 4s lipo (~16v to 5v)

[supple pollen]

Here's one possible choice https://cdn-shop.adafruit.com/datasheets/MP2307_r1.9.pdf

Granted, it's an older chip too, but it's pretty easy to use.

The LM2576 is another fairly straightforward choice

[vast flume]

Thanks

So for the MP2307, I just use the component values listed in the datasheet?

[supple pollen]

That's the easiest way forward, and what I'd do.

I mentioned it because it's the one in this handy product https://www.adafruit.com/product/1385

[vast flume]

Awesome thx

[supple pollen]

Note that example is for 3.3V output, you'd want to scale the feedback divider to get 5V.

The data sheet gives recommended resistance values for several common voltages including 5V: R1=44.2kΩ, R2 stays 10kΩ

[vast flume]

Ok

[dry pelican]

I’m using TI’s “simple switcher” Lmr51420 buck in my hotplate project

[supple pollen]

That's a good choice too

[balmy tide]

Let's do some math, well not really do ourselves make computer do it.
https://www.4pcb.com/trace-width-calculator.html

[worldly schooner]

4 oz/ft^2? Don’t most proto orders default to 0.5, up to a max of 2?

[balmy tide]

I wanted to give it at least some fighting chance with extra beef

[worldly schooner]

It’s not a fighting chance if it can’t be ordered haha

[balmy tide]

Thats a casual 10cm by 8 cm rectangle (almost a square) for 2oz

[worldly schooner]

There are also alternative methods people use for high current applications, such as solder pour and external cooling.

100A I’d say is pushing those limits too, but on a short track using an aluminum PCB, it’s not impossible.

[balmy tide]

Sure let's make it 10 by 6

[worldly schooner]

The more difficult part is sourcing a connector rated for that current output…

[balmy tide]

Good point actually

[vast flume]

How are the commercial 100A esc's made then?

https://www.amazon.com/Waterproof-Brushless-Electronic-Controller-Connector/dp/B0BNVG64DT/ref=sr_1_5?crid=2NURSHBOFP1KD&keywords=100+amp+esc&qid=1678291688&sprefix=100amp+es%2Caps%2C265&sr=8-5

Like this one

[balmy tide]

At that current it's probably just a hole on pcb and ring terminal bolted to it

[vast flume]

Will I be fine with the inductor they choose in the datasheet?

[worldly schooner]

Basically more current needs more copper. That guy probably uses 8AWG wire and an aluminum PCB…

They’ll still run pretty hot, but they use enough metal to not melt itself.

[balmy tide]

For bldc you might be able to assume it'll be on only %66 of time

[vast flume]

Ah I see

[vast flume]

Thats true since only 2 phases are activated at once

[balmy tide]

It's very much doable.

Personally I wouldn't trust my skills (without supervision) with anything that could burn down my house

[vast flume]

Would there be any issues with using this inductor in my buck converter circuit?https://www.digikey.com/en/products/detail/abracon-llc/ASPI-0630LR-100M-T15/3059602

[worldly schooner]

Depends on the requirements of your buck converter circuit?

[dry pelican]

The other issue is driving the fets. Since you need 3 n channel half bridges, you need bootstrapping to drive the high side fets. And then there’s also the control. At 100A, something is going to go boom if you accidentally short one of the bridges or kill a mosfet.

But it is definitely doable. EVs need kA ESCs

There is a simple FOC project to put FOC control onto microcontrollers to make escs, so that may help

[cinder spruce]

how does triac gate resistor selection work? you'll see low values like 200R, so the actual gate voltage must be lower even though it's getting mains, otherwise the current would be high

apparently has to do with gate trigger current

"The value of the series gate resistor is a balancing act between limiting the peak current through the
opto, and allowing enough gate current to turn on the triac.
From the Fairchild Application Note AN-3003:
The max surge current rating of the optoisolator, ITSM, is 1 A for the MOC series of optos.
The peak voltage for a 120 VAC line is 120 x 1.414 = 170 V, so R = 170 V / 1 A = 170 ohms
minimum.
Round up to 180 ohms for a standard value.
The balance comes in here for the gate current and the line voltage to drive it. The minimum
voltage needed to turn on the triac is determined by adding up the gate current through the resistor
IGT, the triac gate voltage VGT, and the opto on-state output voltage VTM.
R x IGT + VGT + VTM = 180 ohms x 50 mA + 1.3 V + 3 V = 13 V. "

looks like i have an answer at least to implementation

so it seems with the moc3021m i have, the max surge current for the opto is 1a as well, and the VGT is 1.3v as well for the triac. 180r seems appropriate. how do i know whether a 1/4w resistor will survive this? after some more reading, i see the RMS power is low because the gate isn't drawing current for long

[supple pollen]

That one is made the same way as the "9999mAh" 18650 cells: by lying about the specifications.

[hexed mural]

"How many amps do you need? 1000? How fortunate, this tiny device happens to support exactly that much!"

[old dust]

I am having a problem where a pin on my rp2040 chip is internally shorted to ground does anyone know what might have caused this? I removed it from the pcb and checked the pcb that trace is not shorted to anything. but the pin its self is shorted on the raw chip.
My thinking is maybe the pin was momentary shorted to 5v or 3.3v or gnd and it fried the pin?
I dont know how a pin acts when its fried tho

[distant raven]

Did you probe the pin on the chip off the board?

[old dust]

yes I did I removed the chip from the board and thats how I found out that its internally shorted

[distant raven]

Okay, is the board one you made?

[old dust]

yes but I think I have found this on another design. I just am about to do more testing. I also had the theory that it could be code. Maybe trying to make a pin be both an input and an out put at the same time? the board has always run circuitpython. I am happy to fry more chips to try and recreate the problem if you want me to do more testing

[distant raven]

What is the purpose of your board? It’s possible there isn’t enough protection for over voltage/drawing too much current

[old dust]

they are keyboards. So the pins are doing matrix scanning using adafruit keypad lib (not saying its that coe that did it just saying thats the kinda code running on it)

[distant raven]

Are you tying your rows high? And if so what voltage?

I’d guess that the pin got more than 3.6V

[old dust]

umm what day to you mean? yeah the rows are pulled high then the columns are set as inputs. does that answer your question? and real quick let me see what the voltage is on a row should be around 3.3v

[distant raven]

👍🏻

There is also a possibility you got a bad chip. Happens every once in a while

[old dust]

sorry I had that backwards the column is pulled high at 3.27v and the row is an input. If it helps at all the pin that has gone bad is a column pin

[twin hedge]

Is there a way I can get a RK3588 chip?

[cinder spruce]

are there resistors in case the input was ever an output unintentionally?

i guess if you're pulling it high, there would be

[old dust]

I am just pulling high in code eg pin.value=True so not resistors on the pcb

[cinder spruce]

so if the pin was an output, it would short?

[supple pollen]

A search for "新一代旗舰应用处理器 RK3588" yielded the possibly useful page https://www.rock-chips.com/a/cn/news/rockchip/2014/0416/480.html

Oh, they offer English pages as well https://www.rock-chips.com/a/en/index.html

[old dust]

Umm the pin is an output pin and when a key is pressed it shorts to another pin that's set as an input. Does that make sense?

[cinder spruce]

well, hopefully it doesn't actually "short"

but i suppose i do

[dry pelican]

A good protection measure for next time might be to add some resistors to each output

Just in case an input pin becomes a low output pin

And then the high output shorts too ground and too much current goes through the one pin, probably killing both boils

Gpios

Not boils

[dry pelican]

Do you have a schematic? And which pin is shorted?

[old dust]

Yeah one min I'll pull it up

[broken zenith]

Generally speaking, is more capacitors for power filtering/decoupling on a microcontroller (RP2040) better? I'd imagine there's a line somewhere, but I want to play it safe.

[old dust]

its gpio 18 thats shorted to gnd

[dry pelican]

Higher resolution?

[dry pelican]

Probably. RP2040 seems to want 11, but that seems like a lot especially when it’s near more caps for voltage regs and stuff. I second this. What is the minimum number of caps for RP2040 when regulation and things are near?

[old dust]

sorry hope this is better

[cinder spruce]

looks like you'd have some issues if the pins were ever output

and a key was pressed

output low connected to VCC would not be good

well it could go either way if both are outputs

two output pins (a col and a row) could source/sink without being current limited if the pin states happened to be that way

[old dust]

Do you think it could kill a pin?

[worldly schooner]

Yes, absolutely.

[old dust]

Do you think thats what this is? (A pin being internally grounded) that's a dead pin?

[worldly schooner]

Overcurrent conditions to a GPIO is the most common reason for bad GPIO

A diode in itself is not a reliable resistive load. Depending on the I-V curve of the diode, you could be sourcing or sinking way too much current.

[old dust]

So how would I replacate the problem to test it? Just short 2 columns together and see if one dies?

[cinder spruce]

row output low, col output high if you wanted to try to kill another pin(s)

but, just add resistors instead 🙂

[old dust]

a resistor to every output or input?

[cinder spruce]

every could be output

[old dust]

true

[cinder spruce]

so all gpio

[old dust]

lol >.< I have seen that done in older keyboard designs that dont use diodes. I wonder why I am seeing a problem on this one in particular

[worldly schooner]

I don’t think that’s necessary. As long as every possible current path has a resistor, you don’t need one on every pin.

[old dust]

what value like a 2k?

[cinder spruce]

right

[worldly schooner]

2k sounds fine

3.3v minus diode drop puts you somewhere in the range of a 2.3-2.8v drop? A 2k should limit current to a little over 1mA.

[old dust]

another question is why do you think it happens on this gpio in particular? the other keys on the keyboard did not have a problem. longer traces so they get a little bit of voltage drop?

after the diode I am measuring 3.26v so maybe I should go a bit larger?

[cinder spruce]

if we're operating under the assumption that two pins were output and loved each other too much, it was simply that. but you don't want software conditions to be able to hurt the hardware

[dry pelican]

It happened again?

The simplest hardware issue would be some sort of solder bridge

But that doesn't seem entirely right

[old dust]

no it was not hand soldered it was pcba very low% in the prototype batch maybe 3% have the problem with this pin dying

[dry pelican]

Picture of PCB? Front and back?

[vast flume]

Sorry I was gone for so long after asking my question haha. But pretty much, I am using the MP2307 voltage regulator and stepping down ~16v to 5v, and I was wondering how I should select my inductor https://www.monolithicpower.com/en/documentview/productdocument/index/version/2/document_type/Datasheet/lang/en/sku/MP2307/document_id/503

This is an inductor that I found but I'm not sure how to tell if its a good choice

The datasheet just said to use a 10uH 4A inductor

[spice turtle]

Basically anything that fits. 10uH, 4A or greater, and then get the lowest DCR you can afford

in the package you want

[vast flume]

Ok perfect

[spice turtle]

Thats how I go about it. First is value, then amperage, DCR and package. Usually they run ~ $1 each

for the power stuff

[vast flume]

Yeah

In that case, does my schematic look good?

[spice turtle]

The one you chose up above is fine.

[vast flume]

ok

[spice turtle]

Why are C11 and C12 in series? Did you mean to put them in parallel?

Same with C15 and C16

[vast flume]

I'm not sure

One sec

This is the typical application circuit in the ds

Does that mean in parallel?

[spice turtle]

Yeap, they short handed it to show 2X but they mean to be in parallel

[vast flume]

Ah ok perfect

Thanks

[spice turtle]

so, 44uF total for Cout and Cin is 20uF in total.

you parallel them for low ESR

[vast flume]

Does this look better?

[spice turtle]

yes

[vast flume]

Awesome

Thanks

[spice turtle]

R6 doesnt have a value btw

Also what software are you using?

[vast flume]

Oops you're right

[vast flume]

Kicad

[spice turtle]

Draw out a reference (or a labeled line) past C16. Theres a lot going on there and you cant make out if its VOUT (I mean you can, but its easier)

[vast flume]

Ah yeah you're right

[spice turtle]

I forget what they are called though. In eagle they are XREF

[vast flume]

Like that?

[spice turtle]

Yea thats better

[vast flume]

Ok cool

[spice turtle]

Also, you dont have to call them out as ceramics. Its understood unless they are polarized

BUT

when it comes to making a BOM, you'll have that little bit too

[vast flume]

Yeah that was what I figured

I've messed up in the past with that

[spice turtle]

I still havent figured out a way to designate special caps and resistors yet

I just put a note on the schematic OR fill out a cart on digikey with a note

So even though your schematic is readable, and I understand why you did it the way you did, and this may be a personal preference, but I like setting my SMPS converters up so that my FB is shown like this: