What is the purpose of a 0ohm resistor?
#help-with-hw-design
light ermine
supple pollen
0Ω resistors as filters? That seems odd. Small inductors (especially the lossy "ferrite bead" style) would make sense, but I normally only use 0Ω resistors as option selectors or jumpers.
light ermine
TIL 🙂
supple pollen
I'll often see low-value (22-33Ω) series resistors to reduce reflections and the like, and in some designs they turn out not to be needed and are replaced with cheaper 0Ω resistors. However, I don't know if that's what happened here.
distant raven
I'm pretty sure that's what it is
silent niche
Hey all! I'm new to the server, but I have a PCB design that I'd love to have someone look over. It's for an LED art project - I have a previous design and schematic that worked great, but the new manufacturer has redesigned slightly and I'm no good with reading schematics (software guy). It's an ESP32 PCB with APA102 LEDs. Happy to pay for the time 🙂
silent niche
Hey all! I'm new to the server, but I have a PCB design that I'd love to have so...
I've had custom boards done made the past that wouldn't reliably flash (missing EN pin resistor, I think?) so this time I'm playing it safe and looking for a second opinion
distant raven
Post a high res of the schematic and board file and most of us would be happy to look
silent niche
Here's the design I got back today! I don't see anything obviously wrong with it, but I don't know what I'm looking for. It's supposed to be USB-C (says Micro, but I suspect the pins are interchangeable). We haven't used the Audio interface from the previous iteration, so it's more of a nice-to-have / future-proofing thing. It's going in a cool hexagonal LED lamp
distant raven
It seems okay, but I need to take a better look on my computer
heavy jasper
Re: series resistors, I think here they are putting them in as an option. Depending on your exact routing, you may need to tweak the value, or may not need them at all, but these data lines are fast enough that it'd be bad to not have the option to tweak the series termination, at least for an early rev of board. If you release a board with 0 ohms, do the design validation probing, and find that you're not seeing any issues, then you can make the decision about whether or not to remove them.
tough matrix
@silent niche welcome! I think you followed my link to this discord (I post on FB under my real name, Alexander K).
I tried looking at the schematics but got complaints about missing font and the wires are shown as yellow on white, which makes it almsot impossible to see 😦
silent niche
Thanks, yes I did! I don’t know if I can do anything about the font or the wire lines. I thought it was a strange choice. Maybe I can get the gerber files when the layout’s done
tough matrix
what EDA software do you use? doesn't look like Eagle - was it KiCad? Altium?
silent niche
I don't know, I'm afraid - I didn't produce this schematic. A supplier in China did, through our manufacturer
heavy jasper
Unless they have a very specialized USB-C header, the micro-B and C will not be directly interchangeable.
You need the pull-down resistors on the CC1/CC2 lines to properly indicate to a type-C host that you are a power-sinking device.
(This would unfortunately work with an A-to-C cable, but then not with a C-to-C cable with a compliant host)
cerulean crest
Or if using some active USB-C cables which shouldnt be an issue much but comes up every now and again
Sorry, hold on
correction, the resistors just wont work at all, I dont know what I said it like that
Just something to be aware of
heavy jasper
The resistors alone should be enough for a compliant host; the cable has its own active ID chip for identifying that they can carry more current.
cerulean crest
Im referring to something separate, reading the problem I think this is a unnecessary side note I added that I probably shouldnt have thrown in the ring
heavy jasper
Ah, OK.
(It would not surprise me at all if there were weird corner-case gremlins in the various options of negotiation for power delivery/active cables and the like)
And it would also not surprise me at all if people are not good at following the spec in a way that causes even more gremlins than the spec itself allows
cerulean crest
Yes, there are and is exactly what I was talking about 😜
heavy jasper
(see Benson's cable reviews early in the type-C days)
cerulean crest
Some cables will only work with PD digital interface if using a compliant charging brick
But if you get the shotty brick too it works just fine
So 👌
heavy jasper
I think e.g. the Nintendo Switch chose a somewhat obscure PD profile
so that not many chargers work with it except their first-party one
Except for some very high-end chargers that support fully variable PD profiles (instead of the cheap controllers that have a fixed number)
pearl tapir
Looking at the schematic, I didn't think his circuit was actually using a USB-C, so no programming resistors were needed.
I guessed his problem with flashing the board would be in the transistor circuit but I'm not familiar enough with the details to have a valid opinion.
heavy jasper
Only reason I mention type-C is that Elliott seemed to be under the impression it was.
cerulean crest
I think e.g. the Nintendo Switch chose a somewhat obscure PD profile
Well that doesn't surprise me, for all the work every laptop vendor has put in to have cryptographic signatures in barrel jack PSUs.... lmao. Now Ill stop interrupting sorry
heavy jasper
The transistor circuit is definitely.... a tad strange, but also without fonts rendering or more context, it's hard to tell.
pearl tapir
Only reason I mention type-C is that Elliott seemed to be under the impression i...
I immediately looked to see if it was USB-C as forgetting the resistors would lead to sketchy operation. The original circuit might be but this one doesn't not seem so.
If it was me, I'd put a scope on those transistors and see if the outputs had full swing and rise/fall times.
heavy jasper
So it does look like this comes from an espressif reference design: https://dl.espressif.com/dl/schematics/ESP32-Core-Board-V2_sch.pdf
But it definitely triggers my "please just use real logic gates for this" senses.
Some more info here: https://github.com/espressif/esptool/wiki/ESP32-Boot-Mode-Selection
cerulean crest
But it definitely triggers my "please just use real logic gates for this" senses...
Its pretty reliable, every single ESP32 board with USB programming uses that if it doesnt have a boot button
heavy jasper
There does seem to be one note from Espressif on this: "(Some third party ESP32 development boards use an automatic reset circuit for EN & GPIO pins, but don't add a capacitor on the EN pin. This results in unreliable automatic reset, especially on Windows. Adding a 1uF (or higher) value capacitor between EN pin and GND may make automatic reset more reliable..)"
And seems like no such cap here, so that's at least worth adding.
(at least pads for it)
cerulean crest
If you dont have that cap, its probably not going to work, some silicon revs are better than others but the cap needs to be there
pearl tapir
Pretty easy to test.
heavy jasper
It also looks like the audio amp is missing its own supply bypass cap.
(As best as I can puzzle out by only looking at the manufacturing information in the properties of the components)
pearl tapir
I didn't look at the data sheet and I agree that a bypass would be a good idea but if the output is differential as a lot of voltage boosting audio amps are, that tends to null out power noise.
heavy jasper
I don't see how it's differential, it's just a single output signal. Datasheet recommends a 100nF cap, which probably means "you'd possibly be fine without it, but may as well put it on the early rev and see if you need it later, you can always de-stuff"
I also can't tell where the 3.3V is actually coming from on this board.
I thought it was that big regulator, but the parameters seem to suggest it's a 5V regulator? Those may just be wrong
Yeah, the parameters I think are just wrong, when I click on the input of it, it directs to the +5V net
pearl tapir
Agreed. More bypass capacitors that you don't need and don't populate is far better than tacking them on later.
heavy jasper
Ah, I think this is an Orcad library organization thing.
The "Part Number" is the overall series one (which they made the 5V one probably because they added that one first) but then "Value" is HT7533, which is the 3.3V version
That's also a lot of LEDs (49; at a quick search they seem to be 50mA max per) to be running off just a micro-B, especially if the intention is to connect this to a computer to control. But that depends on the usage profile and how bright the LEDs are planned to be run.
It’ll also mean a bunch of cap on that 5V rail (which I see on page 2, though I don’t know what values/voltages they are) which can be a bit risky for working with a proper USB controller on the other end.
(e.g. you plug it in, the inrush current trips the over current limit, and the port turns off)
silent niche
This is a ton of info, thank you all so much! I’m working my way through it
We found that the EN capacitor was required for one PCB we did but not for the original version of this PCB. I don’t know why, it may be the specific ESP32 module that was used. But better to include it. I remember reading about it on that link that was posted
silent niche
That's also a lot of LEDs (49; at a quick search they seem to be 50mA max per) t...
There’s supposed to only be 37, thanks for spotting that. Yes, it’d be a bad look if this were all white, but I limit the total current in FastLED to something reasonable, and it’s worked great in the past
The enclosure uses a USB-C jack, with a USB-C wall wart. The previous iteration had a Micro USB jack. I would’ve guessed it’d just default to 5V and be happy, but thought it’s important enough to be sure of
heavy jasper
So, I believe a well-behaving USB-C host actually provides no power on VBUS at all, until it sees proper termination of the CC1/CC2 (and potentially after that a PD negotiation partner).
silent niche
So if I’m reading this right, as long as I only use D+/D- I’m basically at USB 2.0/1.1, using a USB-C connector
silent niche
That works for me. I really just want the C form-factor, of course
heavy jasper
You do need to wire both D+/D- pairs
silent niche
However, to connect a USB 2.0/1.1 device to a USB-C host, use of Rd[54] on the CC pins is required, as the source (host) will not supply VBUS until a connection is detected through the CC pins.
heavy jasper
Sounds right. Not sure where you found that specific quote; but I seem to remember looking this up in the type-C spec in the past
silent niche
Wikipedia
heavy jasper
I mean technically I think to connect any speed device you need the pull-down resistors
It's not like a USB3.0 device magically works without them
but I'm nit-picking there
heavy jasper
sweet beautiful rendered text
To confirm because of the block of text at the bottom: You are nanode LLC?
(you may want to also censor the address if you care)
silent niche
No no, they designed it for us a couple years ago
silent niche
Yup!
silent niche
No sweat. Good due diligence
heavy jasper
What type of machine do you use for development with this, and do you happen to have a macbook available to try?
Apple tends to be very scrupulous in their design of USB ports
such that you can't get away with as much inrush/overcurrent tricks with them as you can with most PCs
silent niche
I’m on a mac, never had any trouble in the past though
silent niche
You have to install a SILabs driver for it
heavy jasper
For reference I've heard of even just a 10uF ceramic directly on VBUS being enough to trip it out sometimes.
silent niche
For my other projects I use the M5 Atom Lite which doesn’t require a driver
heavy jasper
But they may have moved to the middle a bit
silent niche
I’ve definitely tripped the power and disabled the port before, as well as other dumb things with the USB ports
So I know what you’re talking about. But once the drivers are set up there’s no problem
heavy jasper
OK. Is the intent that this is reprogrammable? Or is the CP2102 just for initial programming to get a wireless-capable image on?
silent niche
That’s the hope, yeah
My software can flash OTA, but it’s reassuring to have the USB connection just in case
heavy jasper
(I ask just because if you don't need it to program after first run, could also just chuck some testpoints down on JTAG, flash it using a pogo pin manufacturing fixture, then call it a day, enabling you to save the chip)
silent niche
They’re actually having the chip suppliers program them for us, which is going to be great
heavy jasper
Ah yes, that is the other option
silent niche
I think the number of people who’ll be reprogramming it themselves will be the vast minority. But important to keep it hackable
heavy jasper
Yeah if that's an intentional feature of the product, well worth the cost.
silent niche
I’ll try re-exporting the PDF as an image tomorrow, that might help with legibility
So far this is very encouraging, thank you for all your help. I’ve got some good notes, main one being the EN pin capacitor I’d had in mind, as well as supply bypass for the audio. I haven’t used the audio yet, that’s more there as a nice-to-have. Very hard to make it look right. The USB-C, I’ll confirm with them. Sounds like it needs pull-down resistors on CC1/CC@
sweet rampart
Is there a basic featherwing part for Fusion 360 I can use? I downloaded the adafruit library but nothing turns up searching for "feather"
Something like a connector with a PCB footprint that matches the feather positioning.
sweet rampart
I'm just getting started with Fusion 360 pcb design. Is there a way to get the center of a PTH that's part of a 1x16 header? It'll show me the center of the header's position, and I can do math to figure out where a given PTH is, but I'd like to just select that PTH. (Is this a good place to ask questions like this?)
carmine scarab
Hey @sweet rampart - There appears to be some of that information on the "Introducing Adafruit Feather" learn guide
https://learn.adafruit.com/adafruit-feather/overview
There's a "Feather specification" section, which has measurements and other guidelines. Also, in the "Community Feathers and Wings" section, there's a link to the GitHub repositories for Adafruit Feather and Wings PCB designs (https://github.com/adafruit?utf8=✓&q=Feather PCB&type=&language=) in EagleCAD format. If Fusion 360 can import those (I believe I've seen that it can), you can use as those jumping-off points.
last bison
I'm trying to connect the backlight of a LCD display.
I got 4 pins to the LCD for the backlight, pin 2-5.
pin |
2 | led anode
3 | led anode
4 | led cathode
5 | led cathode
I guess something like this
12V -> pin 2
12V -> pin 3
GND -> R1 -> pin 4
GND -> R2 -> pin 5
How do I calculate the values for R1and R2?
round stump
@last bison Do you have a link to the part?
distant raven
If you had the possibility for 8 different i2c address, how might you set up your circuit that a person could select the address easily? I'm thinking DIP switches, but maybe that is too much freedom and I should just select one, maybe two (cut jumper)
elder peak
I would add the two footprints and default to the 0x10 address
supple pollen
I didn't know Eagle added an after dark view, that's cute
elder peak
They had to, otherwise it was a black mark on their record
distant raven
Actually you just set solder mask color to hex 000000
Alpha value to 66
Then set the substrate color to black
round stump
Why does everyone else's pcbs look so much fancier than mine
round stump
lol am I just a bad perfectionist?
heavy jasper
If this isn't 4 layers, you might specifically try to run the blue tracks further down
distant raven
might be the colors
It might also be that I use ground pours
This is probably as good as this gets for this design without ripping everything up and rotating the main IC 90 degrees
tough matrix
indeed, they seem too close
supple pollen
If you're going to really be a perfectionist, you could rotate it 45 degrees
distant raven
I dislike 45° rotations now for whatever reason
When I made my first Nano board, it was 45° offset like they originally were
But then I tried it on other boards and it just didn’t stick
supple pollen
Fair enough: art is a reflection of the artist
distant raven
Indeed 🙂
distant raven
darn savages developing ULP scripts for Eagle on Windows only. how could they 🙂
elder peak
If you move the QFP and R2, R3, and C2 over a bit, you could make it look like a Cheshire cat grin.
supple pollen
I didn't think ULP would be system specific, how disappointing
unreal flax
I didn't think so either. I haven't ever had trouble running them on Linux.
tough matrix
New bunch of boards..
distant raven
I didn't think ULP would be system specific, how disappointing
File path parts of the script
rain remnant
nice logic boards way to teach logic chips
supple pollen
It seems like modifying those bits of the script wouldn't be too difficult.
tough matrix
nice logic boards way to teach logic chips
yes, that's the plan. I am teaching some math circles and wanted to give kids something tangible 🙂
I'll do a full write-up later
silk lark
there was a board called maple
the blue pill killed it
one of the first stm32 boards
tired bobcat
im designing my first pcb for use with an ESP32, I was wondering if it's ok to use the entire bottom plane as ground and top plane as 3.3v? or is that bad for some reason?
ember laurel
you specifically need a 2 layer board?
pearl tapir
@tired bobcat Are you planning on 4 layers? If so, I'd recommend putting power and ground on inner layers close together as they will act like a very low impedance albeit small capacitor.
If 2 layers, pour power and ground anywhere you can is fine.
tired bobcat
@pearl tapir im doing a portable game console and it has alot of buttons
so theres a ton of gnd and 3v3 connections
its only 2 layers
can i just flood the top side with GND
and bottom layer with 3.3?
tawdry yoke
isn't that a gameboy sp pcb? I'm planning on making one too but out of a pi4, I have no idea on how to even start with pcb design
tawdry yoke
thought I recognised it
tired bobcat
how are u gonna power the pi4 lol
tawdry yoke
with a nintendo switch battery
tired bobcat
the pi zero only gets like 2 hours battery life out of 1000mah
id imagine a pi4 ud get like 30 mins lol
tawdry yoke
Hmm true, what would you suggest? I have the zero but it's a bit laggy I'm in the process of gathering parts for he SPW project
tired bobcat
i have 3 of the spw
and while they are pretty cool to show off
its kinda useless with how short the battery life is
tawdry yoke
yeah. I read up on something like that.
tired bobcat
also the boot times are terrible
im working on a esp32 version of that
im planning on only playing NES and SMS on it so it's good enough
but atleast it'll instant boot and save
and i dont have to worry about shutting it down
tawdry yoke
Esp32? Is that different than the pi zero? I don't know much of pcb and circuitry but I'm trying to learn it, my plan was to make a retro pi portable to play atari 2600 up to the ps1 but now I don't think that possible
tired bobcat
supple pollen
Yes, they're fairly different. The Pi zero has a powerful CPU, but runs a time sharing operating system, so needs to take time to boot and shut down, offers multi-user capability, compilers, graphics, and demanding computation, but isn't as good at real time things.
tired bobcat
im basicly making this but in the GBA SP form factor
supple pollen
The ESP32 is a microcontroller, so boots fast, doesn't need to do anything special before removing power, and just runs one or two simple tasks quickly.
silk lark
while the microcontroller itself boots fast, the emulator may still need some time to initialize and load the data
distant raven
Ugh DRC/ERC missed like three things on this SapWing
Nothing jumper wire couldn’t fix but still ughhhh
distant raven
also found a via that I must have accidentally moved and made a connection where one shouldn't exist lol
heavy jasper
What do you mean "DRC/ERC missed?" Was it a true miss, or is it just not fully-featured enough (at least without other scripting) to catch the errors? A check I've come to use very commonly is single-pin net check. I'm not sure if EAGLE has something like that built in; if not then there's almost certainly a script for it.
Very useful for discovering typos on net names leading to missed connectivity.
(since generally, if you're naming a net, you're running it more than one place, except for e.g. testpoints defined on the PCB only)
distant raven
Mostly that it didn’t see that a few connections were not complete
Grounds connecting to a ground plane but not to the other ground connections
A 3.3V cutoff
heavy jasper
Isn't the Eagle DRC a bit weird in that it never reports unrouted connections?
Or is that a very old thing?
(I remember always having to run rats and see if there were any unrouted nets reported, but it's also been ~8 years since I used EAGLE seriously)
distant raven
I use autoroute to check for missed nets
I guess I didn’t do that one last time before ordering
Oh well, ya live and ya learn
Still pretty happy how well the routing came out overall
This is really tight clearance for OshPark
But continuity is good on this traces
I think I did 4mil
tough matrix
@heavy jasper eagle DRC does report airwires, unless you specifically instructed it not to and then forgot about it...
@distant raven 4 mil traces... wow.
No, I understand that it is far from being record thin, but still, I never went below 8 yet.
and I finally understood that I really need to get myself a proper paste dispenser, the syringe is annoying and slow (for me), and stencils are somewhat expensive.
Looking at https://www.tindie.com/products/danm/dm-solder-paste-and-adhesive-dispenser/
supple pollen
I went with the iExtruder, but that one might work better
tough matrix
quick and very basic question on logic ICs
I am reading a datasheet on a CMS logic gate (MC74VHC1G32) and it has 2 electrical characteristics:
- output diode current
- output sink current
What is the difference - and which of them really tells me what is the maximal current that can be sourced from the output pin when it is high?
distant raven
@tough matrix did you ever get a chance to try those adjustable load switch?
I found one on my desk and remembered that you ordered some 😂
heavy jasper
"output diode current" note is rated for vout < gnd, and is the abs-max current that the low-side output body diode can sink if the output is externaly driven below GND
"source/sink current" is I believe the abs-max number that you're looking for, but even then on the maximum ratings are not the most useful (that's just what doesn't damage the part)
What you really want to look at are the high-level and low-level output voltages
for different output currents
pearl tapir
@tough matrix You may have copied those wrong but neither one is intended to tell what current can be sourced from the output. That's actually a function of what output voltage is acceptable. If you are not interfacing with a logic input, you may not care about noise margin.
At a Vcc of 4.5V, it looks like 8mA to me but I may not be looking at the same data sheet as you.
heavy jasper
Yeah, I would normally want to see curves here
but I guess not, they just spec in terms of "8mA max for a given droop that still meets ViH requirements of similar-series chips for 4.5V VCC specifically"
pearl tapir
I like to see the internal circuit.
At Vcc = 4.5V, 8mA output, high temp it can drop output voltage to 3.66V on the Mot part.
tough matrix
Thanks!
last bison
I trying to create a feather wing for a 480x480 screen. I was looking at RA8875 but my LCD have 18 bit RGB and the RA8875 only have 16 bit RGB to the screen. Should I switch to RA8876? I liked that the RA8875 has built in SDRAM. My board have some space constraints and I might not be able to fit a external SDRAM module for the RA8876. What is my alternatives? Any other chip with 18bit RGB+spi and built in SDRAM?
tough matrix
8mA is a little too low for me - I want to runs a couple of LEDs from it - but hopefully it will work
@distant raven I did test it - and it works great.
I just need to find a project where this switch could be used...
distant raven
Very nice, thanks! Let me know if you need anything else. Willing to send samples of any current and future boards (just to make sure inventory doesn’t go stale)
And that’s an open offer for anyone 🙂
distant raven
I'm going to be completely revamping my CP Sapling boards to get this to be even smaller
going 2 sided, but carefully two sided.
And I also need to finish my SAM E51 board at some point too. Polish the Icy Tree M0 + FPGA board too
lots to do this weekend
silk lark
I'm not sure how good this idea is, but it kind of happened last night. A super-flat version of PewPew, with bi-color LEDs and touch pads.
still need to finish routing the leds
tropic crane
Hello Chat,I come to you for the advice.
Does anyone know any Chinese assembly service?Im looking for some through hole assembly witch will not break my wallet.I order my boards from JLC but they only offer SMD assembly. I'm designing HDMI splitter/adapter board for Odroid and stuck using HDMI male connector like this( i coudnt find any SMD style).I have to jam PCB between the pin rows.
So preferably I'm looking for the service which will produce PCB,source the parts and assemble them.
distant raven
Through hole assembly is usually pricier
Might check Makerfab (not sure if I’ve got the name right)
tough matrix
you can also check pcbway, seeed studio fusion, and elecrow
distant raven
And of course, Sparkfun’s new service
tropic crane
Thank you ! I will check them out.
sweet rampart
Is this a good place for help with Fusion 360 electronics? I'm just learning how to use it (haven't really used Eagle before, either), and I'd like to copy the board outline from an Adafruit Eagle PCB to my PCB. Is there an easy way to do that?
unreal flax
Fairly easy, yeah. Fusion 360 should be able to open Eagle board files, and you can just copy/paste from the dimension layer to get the board outline.
sweet rampart
Fairly easy, yeah. Fusion 360 should be able to open Eagle board files, and you ...
I can't figure out how to select the outline
If I click on the edge of it, the property inspector talks about a line
unreal flax
That's correct. The board outline is just a set of lines drawn on the dimension layer.
That lets you create arbitrary shapes if you want, too.
sweet rampart
Putting aside the fact that it's virtually impossble to select all the lines and not something else, once I do that, how do I copy it? Command-C just boops
Hmm contextual-click->Copy copied something, but when I go to paste it into my design, it was a couple of the lines, and it doesn't preserve the position of the lines in the original (it lets me place them whereever)
unreal flax
Easiest method here would be to turn off all the other layers and just drag-select the whole board outlines. Then you want the "copy group" command.
sweet rampart
If I turn off all layers except 20 Dimension, I see what appears to be the board outline (with an extra line), but I can't drag-select it; it just boops at me
Says "Group is Empty!"
unreal flax
The drag-select needs to include the line endpoints, not just a section in the middle... could that be it?
sweet rampart
unreal flax
(I am assuming some equivalence between how Fusion 360 works and Eagle works, but I've only used the latter.)
sweet rampart
(I am assuming some equivalence between how Fusion 360 works and Eagle works, bu...
Fair enough
unreal flax
Sorry, at this point I'm not sure what to suggest.
sweet rampart
ok
unreal flax
Possibly you can do a multiple selection in the item filter?
sweet rampart
Possible, but I don't know how. Like I said, I'm new, and it's exceptionally difficult to google for answers.
distant raven
ahhhh it's done.
only thing i'm concerned about is potential conflicts with the SPI lines. (I have two separate CS pins)
the SPI flash + SPI programming for the FPGA
silk lark
those are really thin traces
distant raven
6mil
so not terribly
well, depending on the net class
power traces for other than 3.3V (power plane) and GND (plane) are larger
tough matrix
@distant raven the connector on the left is the microUSB?
the trace connecting to Vbus seems to go outside the pad, coming dangerously close tot he next pad...
tough matrix
ok. But why not have it go straight form the pad?
distant raven
can you circle what you're talking about?
tough matrix
distant raven
oh yeah, i just fixed that
😛
i dislike when it shifts like that when drawing traces
found another too
tough matrix
yep! nice to catch it before sending to fabrication
and you have areally tiny NeoPixel there... is it 2mm?
distant raven
so it's just a standard RGB LED
and then a 5050 neopixel
lol
or it might be a 3535
3535
tough matrix
oh... the RGB one is not a neopixel, but a dumb one?
distant raven
yeah
there's connections for it on the FPGA that are specified for it so i figured it won't hurt to use it
the LED is tiny too
nova pollen
@distant raven love the rendering, what CAD software are you using?
distant raven
Eagle
restive vale
Hello, total noob here. It's scary to ask because very likely I'm all wrong but anyway, I'm trying to learn about pcb design and I am trying to make a breakout board for an stm32 microcontroller. I am starting with the electric schematic using KiCad and I have a couple of doubts. So far I have this:
and one of my questions is if the part of my power filtering looks reasonable according to the datasheet of the microcontroller:
and my second question is about the reset switch I put there with the label NRST on my diagram, does it look correct?
unreal flax
I can't comment on the analog-supply inductor, but the caps look good.
The reset switch I think is backwards, though. The NRST is active low, so you want a pullup to 3.3V and the switch connected to ground, so it'll pull the pin low when pressed.
restive vale
The reset switch I think is backwards, though. The NRST is active low, so you wa...
Excellent! thank you very much EdKeyes
heavy jasper
The filtering there somewhat depends on what the actual inductor is
(e.g. is it an inductor, or a ferrite bead, and if inductor, what type)
In short, for most people, the juice isn't worth the squeeze on that filtering. Doubly true if your upstream supply for this is an LDO.
ST also generally recommends a 100nF cap from NRST to ground, and I've generally found that a good thing to add.
restive vale
@heavy jasper Great! thanks, I will update my schematic
heavy jasper
(I've had some bad experiences with ESD coupling into the board and causing spurious resets that were nicely solved by that cap)
For now, you might just make that L1 a 0-ohm resistor
And then if you want more filtering later, you can tune it with chip-size inductors.
restive vale
@heavy jasper Perfect 👍
heavy jasper
Any other chunks of schematic you want to check, feel free to keep posting. A bunch of folks here have done STM32 designs before as well.
restive vale
oh that's great, I will surely come back with another question tomorrow. What a nice group, I didn't expect such quick and good replies
thank you guys
heavy jasper
Thanks! We try to be helpful folk.
silk lark
the filtering caps are tricky, because the schematic doesn't show their physical locations, which is in this case important
distant raven
Just watched an insightful video about separating analog and digital ground planes
Turns out, it’s not necessary in a majority of cases
round stump
Does #help-with-hw-design #repairing-a-pcb-you-got-via-a-kickstarter-where-the-founder-grabbed-the-cash-and-ran? 😄
I backed this project - https://www.kickstarter.com/projects/lumin8/lumin8-l-4-waterproof-app-controlled-led-for-photo-and-video for an LED light. It arrived and I got to use it once or twice before it stopped working.
Since the founder hasn't replied to anything since October and every comment is yelling for refunds, I decided to investigate and it turns out the whole product is one giant PCB with a battery in a tube.
My first hunch is that the charger stopped working. The lights still turn on but the power indicator does a blinking sequence which I assume is low power and then it shuts off. I've shared some photos below, any thoughts on what I might look at/try?
Left side top
Left Side Bottom
Middle top
right side top (The black bit is the power switch)
signal topaz
Left Side Bottom
what is that black stuff on "Left Side Bottom" photo, on upper pin of 3-pin, cable connector? light reflection?
distant raven
at first glance it's like, "wow cool"
but the more I look at it, the worse it gets.
supple pollen
Wow that's a lot of power supplies. The brains of it looks like an ordinary ESP WiFi module.
distant raven
Yeah
round stump
but the more I look at it, the worse it gets.
lol does this feel like a cheap circuit for a price tag of like $120?
:rip: my wallet
distant raven
Perhaps. I’d have to have my hands on it to know
restive vale
Hello folks. I've updated my electric schematic according to what you recommended last night, swapped the weird inductor for a 0 ohm resistor in the filter part and fixed (hopefully right?) the active low reset. I also added an 8MHz crystal according to what I saw on a video and left the boot pin pointing to ground
I also added these gpio headers but not sure if these are the correct ones used in breakout boards:
unreal flax
The push-button is problematic, unfortunately. For one you have a capacitor blocking the current path to ground. And a pulldown resistor won't be strong enough to change the 3.3V direct power-rail connection.
Instead you want to have a pull-up to 3.3V, and connect the push-button to ground. The capacitor should be between the reset line and ground, so it acts as a filter similar to the other decoupling caps you have.
The crystal looks okay, though you may need to double-check the capacitor values versus the specific crystal part you're using. ST actually has a pretty good application note about picking a crystal and calculating those values, if you haven't seen it: AN2867.
restive vale
The crystal looks okay, though you may need to double-check the capacitor values...
great, will double-check this
restive vale
Instead you want to have a pull-up to 3.3V, and connect the push-button to groun...
like this?
unreal flax
Almost. The push-button should have a direct connection to ground, and the capacitor should be hanging on the reset line.
restive vale
ohhh I saw a component in the kicad library that is like a push-open switch
I will try that
unreal flax
You don't have to switch parts. Just connect the left side of the switch to ground instead of to the cap.
That way it'll make the reset line go low when the button is pushed.
restive vale
unreal flax
Aaaalmost. One side of the capacitor should be on the reset line, and the other side should be on ground, instead of interrupting the current flow to the reset pin.
(Remember, capacitors are open circuits at DC: they don't let current flow through them, so they will block a signal in the configuration you have.)
restive vale
unreal flax
There you go! 👍
restive vale
jesus christ, I'm sorry for all the annoyance 😅 I will study all this more thoroughly
unreal flax
This is a good circuit to understand and internalize, so if you're not sure why it's this way, please ask about it.
restive vale
Wow, thank you so much
distant raven
@unreal flax there are instances where a capacitor between ground and the button terminal is necessary for graceful resets. It’s rare but there are a few MCUs that need it.
unreal flax
@distant raven Not sure what you mean... I think that's what we ended up with above. Or do you mean the earlier AC-coupling series capacitor arrangement when you say "between ground"?
distant raven
no
on the Nano, they have the capacitor between the terminating side of the button and ground.
or the old nano schematics did
they don't in the new ones
i think it's an old practice that probably isn't used anymore
unreal flax
I must not quite be woken up yet, as I don't understand what you mean. The terminating side of the button should also be connected to ground, so it sounds like you're describing a capacitor between ground and ground.
distant raven
let me find an example
unreal flax
Or maybe in combination will a pulldown resistor on that side, to control the fall time of the signal?
unreal flax
No worries. Coffee time for both of us... ☕
ivory jasper
I've got an OR gate that selects between PC USB power and external power that prefers external power. It works great except for one problem: After connecting then disconnecting external power the PC USB +5V flows backwards through the gate. This would not normally be a big deal except I need to detect when external power is connected/disconnected at the microcontroller and it basically always thinks external power is connected after it gets plugged then unplugged. This is the schematic... Any ideas how to fix it?
distant raven
probably add a schottky diode
supple pollen
I'm not entirely clear why you have three transistors there, you could use two diode-connected MOSFETs. You can sample the external power on the upstream side of the MOSFET. I'm not sure what you mean by "flows backwards through the gate", as the gates are isolated.
ivory jasper
@supple pollen If I disconnect external power I get continuity from the PC USB VBUS line at the barrel jack (and can detect it at the pin I've setup for detecting external power)
Also, with just two diode-connected MOSFETs what determines which power supply gets priority? I chose this design based on the fact that it will always use external power if present
heavy jasper
It seems like Q2/Q3 forms a latch?
e.g. once external 5V is applied, it turns on Q3, which makes Q2 fully on (not just a diode) so now power can also flow back through Q2, keeping Q3 on even when the original source is removed
distant raven
CP Sapling revision b
4 new pins, small QFN for SAMD21
all on 2 layers
it took some magic and finagling
also, I just spotted a horrible right angle
also found another mistake
much better
bottom side
ivory jasper
@heavy jasper Yeah that's how it seems to work
I'm thinking a diode in the following position will fix it but I'm not sure. Nor am I positive it won't have a significant impact on forward voltage.
distant raven
also, LiPo Powerpack for the Qt Py. smaller footprint 🙂
heavy jasper
Depending on your budget, would you consider a packaged solution (e.g. TPS21123A)
ivory jasper
Your search, "TPS21123A" did not come up with any results (Google) which is actually impressive!
heavy jasper
Probably yes, somewhere
Now that you have a sense of that being a product that exists you can probably do some more digging
(I just picked a reasonable-looking one that was near top of search from a vendor I knew carried them)
ivory jasper
I'd rather avoid a non-basic part for this if I can though
I know there's commercial products using the schema I posted because that's where I got it from 😄
heavy jasper
On the other hand, odds of needing to rev to fix glitches tend to go down with designed-for-purpose parts.
(to present the other side of that story)
I don't think a diode where you're showing it would help your core problem (as soon as external power is applied, Q2/Q3 latch).
Because that latching can happen with just the body diode voltage through Q1
fervent lance
is this 4 layer setting right?
distant raven
That’s one way, yeah
ember laurel
I've got an LED that's rated for 3.2V Vf. How bad is it to run this one on 3.3V, but PWM'ed to max 97% ?
(I'm PWM'ing this LED no matter what, but I'd like to avoid a current limiting resistor if possible)
I guess diode's don't really like being hit with more than their rated voltage, regardless of how short the over voltage period is
silk lark
I think all they care about is the average energy you pump through them, as long as the voltage isn't something ridiculous
as long as they don't hit the thermal runaway, when they get so hot that their forward voltage drops, it should be fine
ember laurel
I guess running a PWM of 3000V at 0.1% duty cycle wouldn't be great
light ermine
Why do you want to avoid the resistor?
ember laurel
I've got an existing design, and products in field, that turn out to run the LED a bit too hot, I think
(I got new specs for the LED after assembling the units)
I can push an OTA firmware update to limit the PWM though
vivid roost
The spec is probably a "typical" forward voltage drop but it wouldn't be surprising if some were only 3 vfd. It is advisable to always include some current limiting in an LED circuit to protect both the LED and the pin driving it. Since rework isn't possible in your case lowering the PWM duty cycle is your next best bet.
ember laurel
I have a regulator for this LED alone
and drive it through some mosfets
some mosfets, because it is a special led - two leads, but two colors
change color by switching polarity
vivid roost
the V forward drop is an inherent property of the LED, regardless of supply voltage
vivid roost
you may get away with it in your case, just saying 🙂
ember laurel
(especially when pushing it hard with current....)
Yeah I'm just wondering in general, if hitting it with more current than it is rated for, with lower duty cycle, is even remotely considered reasonable practice.
vivid roost
best practice is to include a resistor. at minimum place pads for one and "select at test" a value to limit current when full on to 10 mA or so
even if only 33 ohms or so
vivid roost
most modern LEDs glow nicely at 10 mA, can you link yours, 312 mA seems an odd maximum current spec
ember laurel
I'm currently looking at adding a resistor for the next design, it will be small enough to also allow me to put a current sense opamp on it
it's sort of like this one:
https://datasheet.lcsc.com/szlcsc/1810261621_Everlight-Elec-ELSWD-J11C1-OLPGC-C5000_C282219.pdf
not exactly - but almost
I don't have the exact datasheet for the LED in question, unfortunately
vivid roost
OK I thought it was an indicator, not illuminator type. No wonder it runs a bit hot.
how are you PWMing it?
oh, mosfets
unreal flax
I'd tend to recommend a current-limiting resistor myself. An ideal LED with a forward voltage of 3.2V will suck down an infinite amount of current if you give it 3.21V. The only thing you're relying on is the small amount of V-I curve and whatever thermal effects change the forward voltage.
LEDs are fundamentally current-driven devices rather than voltage-driven.
ember laurel
either that or a constant current drive I guess
unreal flax
Yep, that's a great option is it's available.
ember laurel
I think I'll put a resistor on the next board revision
and branch for it in the firmware
(same firmware will run on new revision board)
pearl tapir
@ember laurel I've shipped millions of LED circuits. The LEDs lifetime is primarily limited by die heating, so the PWM frequency needs to be high enough to average the heat, at least 1 kHz, imho. Some LEDs have suffered from current density where the immediate part of the die below the wire bond was worn out. That's not common and only at the upper edge.
Then make sure the Vf is going to be withing reason. Typical 3.3V could easily be run at, say, 3.8V with the right PWM based on your data sheet. But it's risky as you have no monitoring of the solder point temperature. You can estimate the die temperature in a test circuit by measuring the solder point temperature and calculating using thermal resistances. Be sure to include the maximum ambient temperature.
But best practice is to have some current limiting. But it also can be limited by the driver circuit. A feed back loop around the mosfet can work very well to form a constant current source.
ember laurel
I'm running this at 30kHz
pearl tapir
That's fine. But as an FYI, I try to run everything below 9kHz as that is the upper limit for FCC testing requirements. You don't escape FCC requirements as I suspect that you have a clock somewhere that exceeds 9kHz but it's still not a bad idea to keep the unintentional radiators at a low frequency.
ember laurel
I need the 30kHz for my motor driver anyway
(LED is built into the motor)
motor driver runs FOC at 30kHz
but that's a good tip about 9kHz, thanks.
I'll need EMI testing for this thing according to IEC-60601 no matter what though
distant raven
ugh, I realized i bought an 0201 inductor rather than a 0402 lol
good thing the boards they are going on are bust and need to be reordered lol
solemn raptor
this resistor to ground causes some current to leak from the vbus right?
but i'm guessing the DMG341 gate needs to to see gnd?
distant raven
Resistor prevents the short to ground while pulling down the transistor when vusb is not present
The transistor in this case is used as a body diode to prevent flow back to the battery directly from usb
It’s really interesting to read about LOL
solemn raptor
thanks! i assumed when the vbus was powered vbat would be turned off?
but i guess if vbus is powered some current will go to ground so you don't get the maximum usb current but how much less...or am i wrong?
thanks Skerr
distant raven
It’s kind of that way too
It does turn off the transistor when vusb is on
Drives gate to 0 when vusb is off. But definitely the body diode is a big factor in this case too
heavy jasper
And a bit of further clarification: This could be done with just a pair of diodes, but there's so little headroom on VBAT that the diode drop could have a significant impact.
So here, when VBUS is 0V, the FET is fully turned on (it starts with just the body diode, but then Vgs == Vbus - (Vbat-Vbody_diode) == 0V-(3.7V - ~0.7V) == ~3V, which is enough to fully saturate the FET, turning it into just a small-value resistor instead of a diode.
But when VBUS is applied, Vgs == Vbus-Vbat == 5V-3.7V==1.3V, which does not turn on the FET, so it remains just a diode, which is reverse-biased and doesn't conduct backwards to overcharge the battery.
distant raven
Yeah, that’s the nuance I couldn’t explain clearly on the fly
heavy jasper
Where here, "s" is the source, which is the "IN" net of U2, and "g' is the gate of the FET which here is just VBUS.
(and for complete clarity and to finish my train of thought, P-FETs turn on with sufficiently negative Vgs, or (Vgate-Vsource)
A side note: this is one of the many reasons why it's pretty much always a good idea to draw body diodes on MOSFET symbols, at least for discrete parts.
vivid roost
That's beautiful, good luck with it! (How many amps will it draw 😉 )
inland quest
750mA
vivid roost
i remember 1970's 4K memory boards that drew almost an amp by themselves. It really looks nice.
vivid roost
WATT indeed 🙂
inland quest
Heh
supple pollen
The 4k memory boards on my first computer had four 7805 regulators apiece, I think they drew more than an amp (that 16k of add-on memory draws some serious current)
fervent lance
The ADXL375 vs a KX134, for a rocket, I wanna know which is better in terms of
a) Libraries (Arduino)
b) Communication
c) Basic wiring ease
I know that adafruit has a wonderful board made for ADXL
twilit mango
I don't know who else works on rockets, but the one person I do know does isn't around at the moment. Which is disappointing, because I'm certain they'd have a good answer. I don't know the second chip, I only know the first, and since Adafruit makes a board for it, the library support should be solid, and wiring it up quite simple. That doesn't fully answer your question though.
supple pollen
No, that's SRAM. That's one of the reasons it eats so much power.
twilit mango
Maybe this is a good place for this question. Do you think this PCB could be cut to make it shorter? https://www.adafruit.com/product/3776 There's a NeoPixel side-light rigid PCB bar too, which seems obvious that it can be shortened, but the DotStar one less so.
pearl tapir
@twilit mango The text for the connection names could be screened on the backside and the wires and pads moved all the way against the LEDs more or less. Does that help?
twilit mango
I'm not sure. Basically it won't fit where I am considering putting it, and I wanted to know if it could be cut. I don't intend to do anything crazy, simply slice the PCB if possible. (Not even sure I want to do this project, but figured I would ask at least.)
heavy jasper
It seems like yes, as long as you're quite careful. I would probably desolder one of the LEDs, then cut through the middle of where the LED was, then clean up.
And then probe everything to the aluminum to make sure that you didn't introduce a short.
twilit mango
That makes sense. I hadn't considered removing an LED, but that and the rest makes a lot of sense.
Thanks, folks!
carmine scarab
No, that's SRAM. That's one of the reasons it eats so much power.
Sounds like it should be PRAM - Power-hungry RAM!
I'm here all week folks, try the fish.
Seriously though, it's a marvel how much we can do today with so little (relative) power
supple pollen
SRAM is still used these days, and it's still bulky and power hungry, but considerably less so than it used to be.
supple pollen
The computer pictured above was originally just the front board, maxed out with all six 6810 128-byte SRAM chips. That's the first computer I learned to program. Page 0 was data, page 1 was stack, and page 2 was for my program. With only 768B to play with, I didn't have the option of using an assembler, so I'd write my programs in assembler, then hand-assemble them into machine code, translate that into Motorola S-records, then key them into the computer (the Mikbug ROM lets me talk to it via asynchronous serial).
It used 1/4" stereo jacks for serial connectors, I've built an adapter from that, via an FTDI cable, to USB, so I can use a ridiculously more powerful computer as a serial terminal to talk to it.
fervent lance
does adafruit make any other high g boards
unreal flax
Which ones do you already know about? There's this one, for instance: https://www.adafruit.com/product/4627
unreal flax
That one I linked may suit, then, as it's 100g/200g/400g selectable.
fervent lance
Can I get the eagle library of the H3LIS331
fervent lance
Or does this library for the sensor look right
tough matrix
snapeda has this part: https://www.snapeda.com/parts/H3LIS331DL/STMicroelectronics/view-part/?ref=search&t=H3LIS331
fervent lance
alight I'll doenload it from there
but is it right?
like the pad placement and all that
tough matrix
seems to be right; in general, snapeda is usually reliable.
But I'd verify that each pad is tied to correct signal by double-checking with datasheet
heavy jasper
C2 doesn’t look right
(You want it next to/parallel to C26, right now C2 has GND on both sides)
heavy jasper
It eliminates ground bounce.
ember laurel
or it is being used as an inefficient antenna, to pick up alien communications and put it to ground
heavy jasper
(But in all seriousness: the original ST reference design is not exactly setting the reader up for success either)
fervent lance
does it look right now?
distant raven
(But in all seriousness: the original ST reference design is not exactly setting...
Yeah, I definitely agree there. The thing that sort of makes me scratch my head is that most devices that do i2c have pull ups on the application diagram and this does not.
They would have done better separately doing SPI and I2C to make it more clear
distant raven
I didn’t see them on the schematic so I wasn’t sure if you added them or not 🙂
fervent lance
but does everything else look right?
distant raven
Seems to be
fervent lance
thanks
restive vale
Hello folks, I believe I found an error in my electric diagram for stm32 and wanted to confirm
I found in an application notes doc from ST some of the advises @heavy jasper accurately gave to me! And also I think I spotted an error I made
this is related to the NRST switch
according to that doc I believe I should remove a connection to 3.3V in my diagram
like so
what do you think?
heavy jasper
This is roughly where you ended up. I think you have an extra 10K ohm pull-up to 3.3V, which isn't needed, but also isn't going to massively cause problems.
in fact as I scrolled back down you pasted it.
restive vale
exactly!
heavy jasper
It would be a problem if you had a very strong pull-up there (if it was strong enough that the internal low-side switch couldn't cause the pin to go low)
I have run designs both with that resistor and without, and both have worked. If you want to save some space and BOM, feel free to remove it.
distant raven
prototypes ordered 🙂
earnest plover
What's the model of the tactile switch on the QtPy and Feathers?
The board design shows a KMR2 footprint, but the KMR2 switch has an oval-ish acutator, while the QtPy one is round
(yes, this might sound like a foolish question, but it's bugging me I can't seem to find the switch anywhere)
distant raven
it's likely the same footprint
earnest plover
yeah, what i figured, but none of the parts catalogs are suggesting footprint-similar alternates
keen elbow
Hi all 🙂
I want a PCB where i can Plugin Ethernet and Power (24v - 48v) to USB OTG to charge a Android Tablet and also provide Network
Usb could 4 or 8 Pins
I would pay for a custom PCB Design. I want to order it at jlcPCB. Does someone have the time to make this for money?
Best,
Daniel
heavy jasper
Hey, I’m not sure if this is an acceptable place to post consulting job requests ( <@&327289013561982976> to confirm). You might have better luck on the Adafruit Jobs Board.
fast tundra
Yeah, please post to the Adafruit jobs board: https://jobs.adafruit.com/
proper wasp
@keen elbow ^^^
distant raven
CAN Tree SAM E51 board
2 layers of goodness
Single side assembly too
Might add a SPI Flash footprint on the bottom
distant raven
probably a power LED too
distant raven
Almost forgot a reset button
i'm not 100% sure the headers are wide enough to fit on a breadboard lol
better check
distant raven
spacing
I wanted the battery connector and power inputs + the other stemma connector on that side
changed it a little
carmine scarab
CAN Tree SAM E51 board
Not sure how, but somehow, for a split second, I thought I'd read something like "TRANS AM" or some word play/pun on that. I think the yellow board threw off my brain!
distant raven
Lol
vapid grove
Also you can add a soda can drawing near the can bus port
I hope you understand why)))
tough matrix
Id much prefer cancan girls 🙂
ember laurel
how much current would you normally expect to draw from an MCU when charging a mosfet gate?
15mA sounds reasonable?
ember laurel
1mA it is
brisk spire
Anyone have a KiCad guide or template for having mounts for the casetlated and through hole pins on the Pi Pico or similar? The Pi docs and all my Googz seem to be fruitless for some reason (I got tha dumb likely)
Chapter 3 in this shows what Im looking for but alas- cant find their file in GIT https://datasheets.raspberrypi.org/rp2040/hardware_design_with_rp2040.pdf
or anywhere...
Dis
brisk spire
Im still hunting but no luck yet. Will be interested to see how you make out with your board
rain remnant
are you planning your own RP2040 board skerr?
quasi gust
I'd love to see a Pi CM4 and RP2040 on a single board with the USB connected
distant raven
That would be interesting
quasi gust
Yeah, you'd get all the general compute power of the Pi with the PIO of the RP2040 etc. You could tie the reset etc pins to the Pi GPIO so everything is controllable and programmable from the Pi
distant raven
It’d be like having a PRU on a RPi
quasi gust
Exactly! Combined with the awesome community of the Raspberry Pi. Honestly I'm hoping to see the foundation create a board like that in the future
distant raven
I’ll take up the task, mostly because it’s highly relevant to some research I’m doing
heavy jasper
@brisk spire Found it: https://datasheets.raspberrypi.org/pico/pico_datasheet.pdf
There's specifically a separate datasheet for the pico module, which includes on page 9 the SMT footprint.
brisk spire
Yep I had seen that and the other diagram but seems odd to have to trace it out manually when they are trying to get people to use them as modules. Thought it would be in a repo somewhere 🙂 Im not good in KiCad yet but I guess I will have to learn and just make it from scratch
heavy jasper
I suspect that if they’re not doing it themselves, someone will soon have it added to the main-line KiCAD libraries. That someone could be you if it’s not already done! But worth checking the pull requests in progress.
brisk spire
Good call on the PRs!
heavy jasper
Was there one already going?
unreal flax
I’m planning on knocking out my RP2040 board tomorrow
Is there some place where you can buy RP2040 chips? With a quick search I've only seen boards for sale.
heavy jasper
I think chips are coming sometime Q2
ember laurel
Could I ask for some opinion on this circuit, to drive a back-to-back LED?
it drives LED's that are connected like this
only one LED at a time will be active, but they will be PWMed.
so when driving the upper LED, UL1a and UL2b are on.
Only UL1a is PWMed, UL2b stays open.
it is a pretty powerful LED, drawing about 312mA
I'd perhaps like to charge these mosfet gates a bit quicker than the 10k pullup allows for
and also, I'm wondering if I shouldn't put an inline gate resistor there as well, to slow down the shutdown
LED1_H will be pulled low to turn off the upper mosfet
since it is a bit of current here, I don't want to turn on/off faster than I need to, to avoid unnecessary EM issues - the LED is attached to a 1.5m long cable
Qg for this mosfet is 7nC, but only requires 1nC to reach the threshold voltage
@heavy jasper I know you made a point of perhaps switching the upper mosfets to pmos
heavy jasper
or alternatively, a separate 5V buffer chip off that rail
(where you don't necessarily need a gate driver here, just some logic buffer with 5V power)
ember laurel
I'll still want the upper mosfets off upon powerup though - before the MCU has booted
hmmm yeah I guess
what's a good rule of thumb for the charge time vs frequency here?
I'm running this at 1kHz PWM
trying to keep the logic driving pin around 1mA seems reasonable to me
it will have to sink 0.5mA for the upper - current from the pullup there
heavy jasper
Sims are your friend here, and keep in mind that any shoot-through will likely be much worse from an EMI perspective (and a "survivability of your board" perspective).
ember laurel
yeah - I've been simulating this in LTspice
I just want to keep it as slow as I can, really
just wondering about a reasonable practice when it comes to timing vs frequency
the pullups will ensure 0.5mA charge current for the upper mosfets
with 1nC required to reach Vt for the mosfet, that gives 2us charge time
so, 250kHz
pulling it down to ground will be a ton quicker though, only limited by impedance and resistance of the MCU's current sink
(25 ohm-ish I guess)
My thought is to add a 2.2k gate resistor on the mosfets
gives me this in sim
without the gate resistor shutdown gets pretty darn fast, but sim doesn't look too bad anyway
heavy jasper
A note: You can also do FFTs from ltspice
so you can look at e.g. the frequency content of the output voltage/current running out to the LEDs
heavy jasper
That non-monotonic crunch near the beginning is a tad worrying
heavy jasper
(either your sim is going non-physical there, or there's some weirdness in the circuit that's probably worth investigating)
heavy jasper
You can also sweep the value of gate resistor by parameterizing it
and then see plots across different values.
(but this is getting a bit into the weeds)
ember laurel
🙂
heavy jasper
But it can be a helpful way of doing some initial tuning
ember laurel
@fervent lance your Vf is 2V
so it really depends how you are driving it. If you drive it with 3.3V, you'll definitely need a current limiting resistor
ember laurel
how much current do you need through it
30mA?
then you'd need a 43.3 ohm resistor, to reach the max sustained current - 30mA
heavy jasper
30mA on that LED is going to be blinding
fervent lance
15mA?
ember laurel
I was kind of expecting a PWM on that...
datasheet says 35mcd for 20mA
but yeah, if you don't do any PWM, I'd probably tune that resistor value by hand
to find an intensity you're comfortable with
I don't know the purpose of the LED right now, is it an indicator light, or are you lighting up a football stadium?
fervent lance
an indicator led
fervent lance
ember laurel
That does the trick yeah.
how do you drive it though?
with a transistor, or mosfet?
@fervent lance
fervent lance
my microcontroller, the mk20
ember laurel
how much current can it source on the GPIOs?
looks like 25mA
25mA source or sink
with 200R, you should thus be able to drive it straight from the MCU
200R will only give 6.5mA
fervent lance
distant raven
Is there some place where you can buy RP2040 chips? With a quick search I've onl...
Not yet. It’s basically a “design it before you can build it” scenario
ember laurel
What’s special about the 2040?
crude ocean
It is four bucks and has cute coprocessor units that Shenzhen I/O or TIS-100 may have already prepared you to use.
brisk spire
Was there one already going?
Nope notta one 😦
ember laurel
2MB flash is nice, but I think not so much ram
distant raven
The RP Sapling will use LDO power but I’m thinking i'm going to use a boost/buck or maybe just a buck on a larger module that I’ll mount the compute module
fervent lance
2MB flash is nice, but I think not so much ram
with flash expandable upto 16MB
distant raven
progress
distant raven
plenty of space. Near R28
lol
what clearances are you using?
(5mil) 0.127mm?
or (6mil) 0.15mm?
fervent lance
that's tiight
ember laurel
That’s enough. Wireless.
fervent lance
even more progress
what's it for?
distant raven
RP2040
lots and lots of decoupling caps
doing my first design with USB-C and RP2040
kind of exciting 🙂
fervent lance
cool
unreal flax
2MB of flash
Actually the RP2040 has no flash on-chip itself, just a small boot ROM. The Pico board has 2MB SPI flash on it.
fervent lance
I stand corrected
ember laurel
what clearances are you using?
0.09mm
I try to keep my tracks to .3mm for most things though
distant raven
dang 3mil spacing is pretty tight
distant raven
i organize text at the end 🙂
ember laurel
yeah
distant raven
I think I'm done
distant raven
which one?
ember laurel
middle pin
distant raven
it's for bridging another pin on USB-C
ember laurel
I just know that this would trigger a design rule violation with me
ember laurel
it just looks half-tented
yeah that's better
via to track clearance is usually lots higher than track clearance
distant raven
some end up being half tented because of clearance
ember laurel
do you have any solder mask between those USB-C pins?
distant raven
it's the sparkfun
ember laurel
not sure what your fab can do - but those pins are usually mighty close to eachother
if they have capability for mask between those pins, go for it
especially the GND/VCC pins on the sides are easy to bridge
distant raven
it's OSHPark
ember laurel
I've ended up usually changing the footprint a bit because of that
(a bit less than datasheet)
you could also add some stitching vias in the areas left of the MCU
to get a bit of ground fill there
ember laurel
ah ok
distant raven
Dang I forgot Stemma connector
The one thing I’m anticipating is it taking a few months for the Raspberry Pi Foundation to start selling bare ICs
supple pollen
@ember laurel I like that you moved the LED and tucked the components behind it: LEDs can be more visible if they're near the edge and stuff is covering a board.
ember laurel
the LED has a light guide over it as well, extending outside of the enclosure
works pretty nicely that way
supple pollen
Wait, what? How did you get the board fabbed and assembled in 3 hours?
ember laurel
3d pcb printer
you just press a button, and it prints silicon and all
that's actually the old revision.... squeezing even more stuff in :/
supple pollen
Silly me, I was gonna buy a Wazer but now I want that instead.
tough matrix
3d pcb printer
you almost got me 🙂
I was about to ask, where does one buy it?
ember laurel
it's a big LED driver
all it does is to light up that LED
unfortunately you cannot buy it yet, still have a bit to go before market
I think sometime in summer it should be ready
funny enough, people seem to think 3d printers can print anything
my kids were asking if I could print them a new ipad
tough matrix
I was asking about buying a 3d PCB printer.. that I'd love to buy 🙂
But I can well imagine that some years in the future, there will be desktop 3d silicon printers for making custom chips ...
ember laurel
and they'll be as reliable as inkjets
supple pollen
Inkjets don't work for me because I don't do color printing often. However a PCB+silicon printer is something I'd use pretty frequently, if it were obtainable.
ember laurel
totally with you there 🙂
I've made the decision to abstain from inkjets
only having black and white laser
if you need color prints, order them online
supple pollen
As have I. I have a dye sublimation printer that prints from ribbons. It's ready to print whenever I like, even if I haven't printed something for a couple of months. Yes, the supplies are a little expensive, but with my usage pattern, it's much cheaper than an inkjet is.
During the recent unpleasantness, I've been using it to print up custom postcards. The prints are durable enough to mail and don't get ruined if they get wet (unlike inkjet prints).
ember laurel
aha that is nice
that 15 minute wait for a print when you haven't printed anything for 2 weeks is pretty darn frustrating with an inkjet
I had that once when I really, really needed to print some documents just before a travel.
being late to the airport...
distant raven
I would love one of these with how many PCB I make: https://www.voltera.io/store/v-one
ember laurel
what does that thing do?
distant raven
It’s price point is steep, but it would be super convenient for 2 layer designs
tough matrix
pcb mill?
distant raven
Cures the resin, then you can use it to apply paste as small as 0402
And then also cook the boards
ember laurel
I bet the capabilities aren't stellar though...
distant raven
6mil/6mil I believe
tough matrix
most likely, it is just for prototyping, not for mass production... ut still very nice
How expensive is it?
unreal flax
@distant raven Does your RP board have any flash on it? I don't immediately see a footprint for a chip that screams "memory".
distant raven
It’s on the bottom
unreal flax
Gotcha. 👍
distant raven
Hey @heavy jasper question about USB
Someone mentioned on Twitter that I should put 5.1K pull-ups resistors on the data lines
But the raspberry pi rp2040 data sheet only suggests 27.4ohm series resistor for usb data lines
Which is the “more right” way to approach it?
Design guide from Raspberry Pi: https://datasheets.raspberrypi.org/rp2040/hardware_design_with_rp2040.pdf
heavy jasper
Were they asking for 5.1K or 1.5K?
I think 1.5K was the traditional resistor for D+ to indicate full-speed, but also almost all PHYs have that built in now.
And in fact, from that datasheet, specifically: "The USB_DP and USB_DM pins on RP2040 do not require
any additional pull-ups or pull-downs (required to indicate speed, FS or LS, or whether it is a Host or Device)"
Which to me says that any pull-up/pull-down that are needed are in fact built into the silicon.
5.1K is the standard resistor for the CC1/CC2 lines of USB-C, which is a separate thing (used to indicate that this is a power-sinking device taking standard 5V,3A)
supple pollen
The ways USB chargers signal various things adds another dimension, but that probably doesn't apply here.
distant raven
5.1K is the standard resistor for the CC1/CC2 lines of USB-C, which is a separat...
Does it default to 500mA if you leave them off?
Nvm
ember laurel
this is a pretty good reference for usbc CC setups
@distant raven
are you supplying or pulling current in this case?
ember laurel
how much current do you need?
ember laurel
normally what you would do in the UFP is to add the 5.1k pulldown, then measure the CC voltages with ADC
distant raven
That chart says 56K for 500/900mA max
ember laurel
500mA should always be possible though, regardless
see this pdf as well
in your case you can happily use 5.1k pulldowns and just draw the current you need
distant raven
Makes sense
ember laurel
3.6 in that pdf, shows which measured voltages would correspond to which current capability
if you need to power an induction stove or something from your board
distant raven
Lol hopefully mine doesn’t become an induction stove
digital moat
going out on a limb here, but would any of you kind souls happen to know what sort of connector this is? 5mm x 2mm 10 position. believe it is a programming header for an stm32f103
ember laurel
@distant raven secondary, unintended use
tough matrix
I got this request on instructables:
Sorry for disturb, thanks for sharing those amazing projects on Instructable.
PCBWay is a large circuit manufacturer. We could provide free PCB for your further forthcoming projects. Would you mind posting an introduction about PCBWay and some detailed product reviews in the Instructable blog?
This is the first time I am approached with a request like that, so I am a little unsure how to proceed. I would be OK posting any kind of reviews (except that so far I haven't used PCBWay, but I am willing to try), as long as I am free to write in my review what I really feel. Or is there an implicit expectation that I should only write positive reviews? (Clearly, they do not say so explicitly - that would violate all kinds of ethical norms, and if they did, I would toss in trash right away - but do they really expect that?)
vivid roost
First, congratulations on the Hackaday article, it popped up on my news feed this morning and I said to myself "That looks familiar..." 🙂
Take a look at 1:30 on this vid from Ralph Bacon, I think it's the kind of thing they expect in return. https://www.youtube.com/watch?v=aMS4XwEr8s0
If their service pleases you, no issue. If it does not you are no worse off than now.
fervent lance
@tough matrix I would see about developing a telephone relationship with them, and even then proceed with great caution.
Or just treat them as any other vendor, place an order and see if you like what you get.
If you find yourself in the position of wanting to praise what they do, and on top of that, they want to pay you to want to praise them, that's probably okay. ;)
fervent lance
yeah just make sure you're not restrained by anything regarding like what you can and cant say
but like ive bought from pcbway and its pretty chill
tough matrix
thanks everyone
I've written to them, let's see what happens
brisk spire
I've been working with PCBWay for some years now and can vouch- they are great. Good product at a great price. All sponsorships start with them asking for as much as they can get- just lay out your own rules and all will be great.
You get good boards and/or $$$ and they get targeted advertising while actually supporting makers. Good deal
distant raven
revamped thanks to everyone's insight (both here and on twitter)
distant raven
unreal pollen
I have 0 experience with PCB design but I just wanted to play with the Feather ProtoBoard and make a custom one. The problem I have with the proto board is that the individual holes makes things more flexible, but I'd rather have some lines of connected holes. I think what I need to do is add some sort of component to the schematic so that I can add connections to the holes that I want connected, but I have no idea where those components come from. I can see that it looks like it's coming from some adafruit library but the basic library that is available to download doesn't seem to have the right stuff. Anyone have any advice for how I might go about doing this or if I'm missing something? I can't imagine I need a custom component just to trace some holes together vs making them all individual, right?
distant raven
are you talking about the protowing?
unreal pollen
distant raven
ah okay, do you have in mind what you want to make?
unreal pollen
yeah basically this but picture the individual holes in the middle as individual lines of holes, not individual holes. So basically the holes in the middle would be 14 vertical rows of 5 connected holes.
Then you could do the same thing with all the holes on the left, connecting them in the opposite direction though, and having 5 rows of 4 holes
slightly more breadboard esque if you will
distant raven
well, it would in that case be as simple as dropping 10x 14pin header pins onto a schematic, and wiring those together. then on the board side of the CAD drawing you'd just make the actual trace connection.
unreal pollen
Ok, so I was assuming if I wanted say 4 of the holes connected, I needed like a 4 pin header added to the schematic
But it sounds like that is not correct
Or rather you're saying it doesn't have to work that way
distant raven
you'd use a similar process for any number of connected headers
unreal pollen
Are there libraries for n pin headers?
distant raven
usually sparkfun or adafruit have libraries for Eagle. KiCAD has pretty extensive built in libraries
unreal pollen
I already pay for Fusion so was using Eagle, is KiCAD easier or better?
distant raven
though, it isn't entirely difficult to make simple through hole custom part libraries in eagle
unzip that to the "libraries" folder for eagle
unreal pollen
Thanks 👍
distant raven
I used that tutorial to learn how to make custom parts and libraries for eagle as well
unreal pollen
So making custom stuff is pretty normal? Seems odd.
I would think anyone making components would want to have up to date libraries so you can place their stuff in your design
Or is it more that there are lots of custom ways to mount the same part?
distant raven
umm, it's more that not all parts you want have part libraries available
especially newer parts
I had to make a custom footprint for the RP2040
but it was easy thanks to the datasheet
unreal pollen
I would think Nordic might have a good library?
distant raven
they might, a lot of prebuilt libraries are found on either SnapEDA or UltraLibrarian
it's always good to check the footprint based on the data sheet though.
you'd be surprised that a lot of people make their own parts simply because it's the best way to know the part is properly translated from the data sheet to your board.
distant raven
it's what we're here for 🙂
now if you ever run into issues designing your board, there are many others with great knowledge to help out 🙂
they pop in a different times throughout the day, so there's usually at least someone to help you even if you have to wait an hour or two
distant raven
goal for this weekend is the RealTime Pi, Computer Module 4 + RP2040
plus a few other peripheral
tough matrix
You got a lot of plans 🙂
tough matrix
meanwhile, I found a great brushed motor driver IC, Toshiba TB67H451FNG. 4.5-50v, 3A continuous, easy to solder HSOP8, and costs under $1.50. It will replace my previous favorite, TI DRV8871
heavy jasper
They'll be a ton of software work, but having a sense of your application, you could be a good candidate for some Qorvo parts
(ARM MCU, drivers, and FETs integrated together, most commonly these days for BLDC motor control for industrial and hand tool use)
Along with integrated charge-pump or inductor-based switching converter for the core rails to be powered directly from supply Vin
Though many of the super-cute integrated solutions these days will be targeting brushless motors. (if anything they're meant to enable brush motor replacement with brushless by building in some local brains)
tough matrix
thanks! took a look at it... it does seem interesting, but will indeed be a lot of work. maybe a little later
heavy jasper
@digital moat Congratulations, you would have done very well on Car Talk's 'stump the chumps.' I believe I have searched through the active catalog of all major mezzanine connector manufacturers (Samtec, Hirose, Amphenol, Molex, and then more generally on Digikey and Google Images) and have not managed to find this part, or its product series. If you have a set of calipers, it'd be super helpful to measure from center-to-center of the pins of each side, and from one side to the other.
And/or some more info about the product that this was found in
But in short: if you need debug access, you might be best off soldering some small jumper wires to the pads instead.
digital moat
@heavy jasper thank you so much for your time. I too did lots of digging, but to no avail. My guess at the moment is that it is indeed some sort of stacking header - having tried to interact with Devlin who makes the actual device I'm guessing they use this as the basis for lots of their designs. that unfortunately leaves me back trying to use what should be a serial interface via the external db9, but without it responding at all. i'll have to see what the pins actually connect to - it is entirely possible that, given these were connected to larger 'integrated trading systems' they are carrying power in some weird way via the db9. thanks again for your help - and it made me smile to remember Car Talk!
heavy jasper
Given it's a leaded package, if your soldering is sufficiently good you might just solder JTAG/SWD directly to those pins
Given at least those pins are very clearly documented (if they left them enabled)
distant raven
You got a lot of plans 🙂
Lots of ideas I want to make happen hehe 😉
digital moat
Given at least those pins are very clearly documented (if they left them enabled...
apologies @heavy jasper you mean solder to the chip traces themselves? not sure my 50 year old hands can stay that steady 😆
median stirrup
unreal pollen
One thing I've noticed about the through holes in the feather proto, is that the header pin holes are a smaller diameter from the rest of the board. I can't find any references though of where the larger through hole came from. I'm assuming the only way I could get a strip of those larger through holes connected is to create my own component and have it match the size of holes that adafruit is using there?
The larger through holes do have trace lines to connect some of them but they are not preset in the schematic, nor are there any parts listed for them, which is why I'm confused as to where they came from.
tough matrix
The larger holes in proto-shield are done as vias
Vias can be added to the board in any place , connected to any signal, and they do not correspond to any component in the schematics
and yes, they are larger than necessary, in my opinion
unreal pollen
I was also wondering about that. I think it's because they are individual, so you could maybe chuck two wires in there if you needed to? I kind of prefer the smaller ones tbh.
tough matrix
for standard breakaway headers, I used holes with drill diameter 1.016 mm
which is same as 40mils
this is also waht SparkFun uses in their library
for most through-hole components (resistors, capacitors, etc) you can use even smaller holes
unreal pollen
I have not been able to get two Vias to route together, it doesn't accept that as valid.
Thats why I thought I had to have them on the schematic, to allow for the connection to be valid.
tough matrix
what software are you using?
unreal pollen
Eagle
tough matrix
then it is easy. In board editor, click on "Info" tool ("i")
then click on a via
you will get a dialog like this
unreal pollen
yeah that's how I was editing the size
tough matrix
now find there SIgnal->name and edit
make the name of the signal the same for two vias
unreal pollen
dude
tough matrix
then Eagle will allow you to connect them by a trace
unreal pollen
That's amazing, I've been futzing with that for so long.
tough matrix
Been there 🙂
unreal pollen
Thank you so much
tough matrix
no problem
tough matrix
yeah, I benefited a lot from advice I got here from others - now I am passing it on
unreal pollen
baby steps
this was my first goal though for working with Eagle and making something that I wanted, which needed to be an easy target.
distant raven
Not bad 😃
elder peak
So, I guess you are learning via vias.
heavy jasper
Or learning through vias, perhaps. (Blind no longer)
pale current
I'd like to get more into PCB design and have a slightly above beginners knowledge of electronics. Any thoughts on a good project to work on as a "first real project", even if it is almost just copying another project out there, just for the practice?
distant raven
@pale current you can start by designing simple breakout boards
I’ve learned a ton just by making i2c breakouts
Like for the at42qt1070 capacitive touch sensor, or an ADC board I should be getting soon in the mail
Or you can do what I did and make your own Arduino Nano
pale current
<@!228322543507865601> you can start by designing simple breakout boards
Thanks, breakouts are a neat idea. Just pick a chip from digikey and go.
distant raven
Yeah, some chips have better application notes than others
pale current
I had thought of just almost copying some open design even though its "a waste" but just sometimes doing it myself really helps
distant raven
Texas Instruments is usually great in terms of electrical applications
pale current
Yeah, some chips have better application notes than others
Having read lots of these to write drivers I fully agree!
distant raven
Arduino publishes their Nano schematic
It’s very well detailed and easy to understand, and provides a great jumping point for learning to design microcontroller boards
pale current
Oh thanks. yeah that's the type of thing I was thinking of. Something complicated enough that there is some challenge but also where there is sorta a reference to refer to
pale current
Yeah, next thing ya know you’re launching an electronics company
Who would do that? 😛