help-with-hw-design

[fervent lance]

The DW01A is already monitoring together with the TP4056 the battery voltage and is supposed to cut it off from GND once the voltage drops too low.

ahh wait, but a voltage supervisor IC could make sure the device doesn't try to turn on if the voltage at the booster entry is too low, right?

[distant raven]

Correct

[fervent lance]

soo question is, what specs are important here?

[distant raven]

Usually threshold voltage and rise time

I think 30usec is common but you can find slower ones if the regulator has a slower rise time

[fervent lance]

not sure if this simulation is accurate enough, but... are -42uA on the battery + perfectly fine? If I would add another schottky diode here, the voltage from the battery would drop noticable

oh... wait... once I set 3v as the battery voltage, this would increase to -201mA... that definitely is a problem... I don't get it... am I using the wrong mosfet here?

[distant raven]

Yeah

Looks like an N-Ch mosfet

[fervent lance]

they are both P channel mosfets

the ap40p05 is a P channel enhancement mosfet

[supple pollen]

It looks like the MOSFET body diode is conducting, you may want to turn it around

[distant raven]

Ah yeah orientation and morning brain not serving me well here lol

[fervent lance]

switching source and drain? I thought source should always be where the electricity is supposed to come from and drain where it is supposed to go to

[unreal flax]

That would make way too much sense...

[fervent lance]

in other words, that's not the case?^^

ok... now it's only nA in the simulator... that sounds better ^^

I also want to add a speaker to my LED project, for playing simple sound effects together with LED animations, based on readings of the 3 axis gyroscope I added. But I wonder if this can cause performance / memory issues with WLED (the software I use for running the LED animations) on the ESP32.
https://circuitdigest.com/microcontroller-projects/esp32-based-audio-player

[limpid nest]

Would I be breaking any hard and fast rules by doing this? I want to move the !N_OE pins around so they are all next to each other, and the NA pins so they are all next to each other, and the NY pins so that they are more spaced out (!N_OE and NA pins will also be spaced out). This will make my layout prettier and easier.

[unreal flax]

That should be fine. Schematic symbols have only vague guidelines about pin placements.

[limpid nest]

That's what I thought, thanks.

[fervent lance]

I can also use a regular resistor instead of a variable one here, right? As long as I don't mind the volume being fixed?
And how could I make the volume controllable via software? are there resistors which can be controlled via a microcontroller?

[limpid nest]

Yeah fixed should be fine. And yes there are! Adafruit sells 2, one is linear and one is logarithmic.

They are called digipots adafruit's are I2C.

[fervent lance]

https://datasheet.lcsc.com/lcsc/1912111437_Maxim-Integrated-MAX5407EKA-T_C185834.pdf
That one would also do it?

[limpid nest]

Do you just need 10k adjustable? It might be easiest to pick the exact chip that adafruit uses, so you don't have to write your own library. If you don't mind writing your own library, then that one might serve? looks like you have only have up to +6 volts on it (I'm not sure what the specs of the adafruit chip are off hand)

When picking a FET to switch 12V, I should pick a VDSS greater than 12 right? like at least 16V?

Actually I'm doing reverse voltage protection but I think similar principles apply

[fervent lance]

I am still considering if I even need it to be adjustable... maybe a fixed volume is easier/cheaper to implement for now.

think I will just experiment with placing a regular resistor for controlling the loudness and then see how it goes

[limpid nest]

I think that ease of adjustment order is going to be mechanical pot > digipot >> fixed resistor

good luck!

But digipot probably gives you better accuracy

[limpid nest]

I'm trying to pick a resettable fuse. I think I want my max current to be 2.5A. I see there are two parameters: holding current and trip current, with a deadband between holding and trip current. Should I pick a fuse with a holding current equal to 2.5A and a trip current such that the deadband is as small as possible?

Additionally, Time to Trip, does a 10s TtT really mean that this fuse will conduct X Amps of current for 10 whole seconds before tripping?

[supple pollen]

That depends on what your parameters are. Do you want to allow a brief overload, or a small overload for a little while, but cut off quickly on a large overload, do you want to respond to surges, or allow them, do you need to allow for temperature independence?

[limpid nest]

Hmm yeah, it's complicated

I am confident that the raspi people did something that works for my needs, so I've been trying to copy them, but that data isn't really available.

[spice turtle]

It sounds to me like Efuses are much like regular fuses or Circuit Breakers that have a B,C, or D curve. They will let so much current through for x amount of time, usually the end electronics can handle so much, but I would say pick the shortest time needed

But TBH resettable fuses arent my area of expertise, I would search TI to see what kind of documents they have

[fervent lance]

I am going nuts... cannot figure out how to adapt my PCB design to flash the ESP without pressing any reset button... any suggestions? I remember the ESP32 D1 Mini board can do it. But cannot find any official schematic...

[unreal flax]

Typically it uses some extra serial-port control signals (CTS / DTR or whatever) with some transistors to programmatically toggle the boot and reset pins. All of the Espressif reference schematics will show how to do that in the canonical way.

[fervent lance]

I already implemented this part and yet it still doesn't work :/ I have searched for schematics of the esp32 d1 Mini, since I remember that esp32 board can be flashed without pressing any buttons. But all the schematics I can find look a bit different. Some have the RST pin of the CP2102 connected to nothing, others connected to a resistor, capacitor and voltage/gnd. Some have resistors at DTR, RTS, TXD and RXD of the cp2102, others don't...

[supple pollen]

How do you initiate the reset on this mini board?

[fervent lance]

the cp2102 should do that, based on my understanding

[supple pollen]

Then it's probably RTS or DTR based like your diagram.

[fervent lance]

but what parts are important for that then?

I don't really get why I cannot get it to work

[supple pollen]

When you say you cannot get it to work: what have you done? What was the expected result? What was the actual result?

That's a funky little circuit. If RTS is low and DTR is high, Q3 will turn on and pull RESET low. If DTR is low and RTS is high, Q2 will turn on and pull IO0 low.

[fervent lance]

one mistake I did was I had the pins for the auto reset wrongly connected. So I fixed that with some wires. It might not look pretty, but should be connected correctly now. But once I connect the board to my Mac and visit install.wled.me for installing WLED and select the board, it cannot upload the software but asks me to press the boot button and try it again (and yes, then the upload works fine).

I found another schematic with this nice part that replaces the four previous components. Also check the result table next to it.

[supple pollen]

That truth table doesn't match the diagram, but I suspect the diagram isn't quite correct

[fervent lance]

what do you mean? the formatting? or whats inside doesn't make much sense?

[supple pollen]

You have posted 3 different diagrams. The first one (collector to RESET) seems sensible, if a little odd. The second one (emitter to RESET) may not work. The third one (with the 6-pin chip) shows DTR connected to both bases and RTS connected to both emitters, so both transistors would turn on for DTR high and RTS low, which may not be what you want, and doesn't match the truth table in the same diagram.

The truth table does seem to match the first (collector to RESET) diagram you posted, and by rearranging the signals slightly, you could make the chip do that, but the connections in the chip diagram don't match that behavior.

If you were to switch pins 4 and 5 (in other words, connect pin 4 to DTR and pin 5 to RTS), I think the chip would implement that truth table and the behavior of the first (collector to RESET) diagram you posted.

[fervent lance]

oh yeah you are right, RTS and DTR have been switched, my mistake (when trying to recreate an existing schematic)

[supple pollen]

Granted, an IMH3A is a bigger, easier to solder package, but it has a slightly different pinout

[fervent lance]

but besides that part of the schematic, what else matters to implement autoreset?
here for example they used a bunch of resistors on the USB to serial chip (U2)
https://i.imgur.com/EpE4dGj.jpg

and also connected RST of the USB to serial chip. But on other schematics of the ESP32 d1 mini I found they don't do that

and I cannot find an official one for this board https://www.az-delivery.de/en/products/esp32-d1-mini (which I used in the past and know it works there)

I already started considering if it makes sense to search for some of those boards at home (should still have them somewhere), inspect them with my macro camera and try to reverse engineer that part of the board

[distant raven]

Ah that’s an 8266 which is different than the ESP32

[fervent lance]

no, they are both esp32

[distant raven]

ah wait, read 8266 compatibility

[fervent lance]

where does it say that?

[distant raven]

"COMPATIBLE
It has a similar pin layout to the conventional D1 Mini ESP8266-12F and is therefore pin-compatible with most accessories for it."

at least based on the datasheet, the boot/reset strapping pins need to be toggled to achieve programability

so I think you can go with the plan you mentioned above

toggling the pins with the USB Serial

[fervent lance]

and the reason why it didn't work out when it tested fixing my pcb with some wires?... maybe bad soldering connections?

guess you are right... I should just adapt my pcb design, order it and the parts again and then see if it works... fixing pcbs by using wires and so on is really annoying

[supple pollen]

Annoying, but a big time and money saver in many cases. I'll normally hack up my boards to fix as many things as I can and ideally get them working properly, so I know I understand all the fixes required before I respin it. https://www.sparkfun.com/tutorials/109 Pull quote: "Don't immediately throw up your hands and layout a new board. Instead, make sure you get every bit of the functionality of your board working, by any means necessary, and then make all the revisions. Otherwise, you'll constantly spin your wheels with PCB revisions."

[fervent lance]

yeah that was also my approach so far

don't worry

[supple pollen]

"anyone can layout a PCB, but it takes a true magician to get a bad PCB working. Scratch, cut, splice, and otherwise modify your PCB until you get it working."

[fervent lance]

:D:D

[supple pollen]

We've all been there

[fervent lance]

^^

[heavy jasper]

That said - there is a point that my colleagues and I call “critical jank” where any change/rework you make has a 50/50 chance of helping vs. making something worse just because of the fact that you’re doing somewhat hacky reworks. That’s where it’s helpful to order more than one board on your first order, so you can rework your way to critical jank on your first board, then take just the reworks you know you need and apply them as cleanly as you can to your second board, then start bringup again from there.

The strongest version of this being to just release a second rev of PCB with your known changes and start again from there.

[supple pollen]

Been there too...

[distant raven]

hehe...

[fervent lance]

yeah can relate... ^^
oh and another important question, since my LED / PCB project is battery powered - how do I read the battery charging status? can I connect the TP4056 to a GPIO pin for that? or can i just connect +BATT to any GPIO pin from my ESP32?

[supple pollen]

You'd need some signal conditioning as your battery voltage may be high enough to damage a GPIO (especially during charging). You could connect the "charge" output of the charge control chip to a GPIO, but it's open-drain, so you'll need a pull-up (either the GPIO built-in one, if it offers one, or an external resistor).

To actually monitor the state of charge would require a little more.

[limpid nest]

Me to Autodesk: This feature doesn't work

Autodesk: Have you tried using the feature, it's really how it's supposed to work.

Me:....

[supple pollen]

I had a couple of rounds of that with sheet metal fabrication in Fusion 360. However, watching a video by another company and doing some experimentation finally yielded the results I was looking for.

[limpid nest]

If anyone was considering switching from pure EAGLE to Fusion360, wait a while. They completely broke the library management workflow.

[supple pollen]

Nope, I'll use my expensive EAGLE license for as long as I can, then I plan to switch to KiCAD

[limpid nest]

I'm considering a switch to KiCad when life slows down and I have the time.

[distant raven]

I’ve contemplated KiCAD as well

I love eagle and fusion, I’m coasting on an edu license since I’m not selling anything and it hasn’t kicked me out yet lol

[limpid nest]

Yeah I needed a commercial CAD program that did EDA and 3D well. Fusion is supposed to be that, but they keep adding ingredients to the meal and making it harder and harder to stomach.

The 3D CAD is "fine", but they have borked the EDA pretty well.

For example....a part I just added isn't showing up on my dang board.

[distant raven]

Strange

Eagle works for me for now

[limpid nest]

Oh of course! It created a copy of my board, dissociated it from my schematic, kept the name the same, and didn't tell me!

[distant raven]

Fusion360 works pretty good for 3ad CAD for me

Though sometimes it acts stupid

[supple pollen]

I use Eagle (and DesignWorks Lite) for PCB design, and Fusion 360 (and OpenSCAD, and pencil and paper) for 3D design.

[limpid nest]

I prefer Inventor but I can't afford it

[supple pollen]

I considered Altium (for PCB design) and Solidworks (for 3D design), but they're DOS-only, which is an automatic fail for me.

[distant raven]

I 3D print stuff so Fusion360 kind of is perfect for what I do. Inventor is cool though

For a while I used a tool called DataCAD for doing architectural design in high school

[limpid nest]

Fusion has some really weird aspects to it for 3D. It has a more active online community outside of the forums though

[distant raven]

Yeah

Fusion is nice though that I can open up the 3D model of an eagle project natively without really any issues

[limpid nest]

Yeah it has a lot going for it. I just wish they'd stop helpfully breaking it

[distant raven]

Same

I haven’t updated in a while

[limpid nest]

The "joking" conversation that started this thread is actually really, really close to what happened.

[supple pollen]

Its bent sheet metal support is actually fairly good (once you find a site that explains the process intelligibly)

[distant raven]

Ohh looks good

[limpid nest]

I've heard the CAM is very good too

[distant raven]

I’m just designing brackets like this for RGB matrix PCB I have

[supple pollen]

Once I had the design dialed in, I had Fusion flatten it out and produce a DXF I could hand off to a laser cutting service to fabricate.

[distant raven]

Oh sweet, I plan to eventually get to that design level

For now this is the complexity I’m at

[limpid nest]

Just to whine about a program for moment longer, I wish fusion360 3D cad had more unit flexibility in it. Doing things explicitly in mil is nice sometimes.

Yes, I can multiply/divide by 1000, but it doesn't seem like it would be too onerous to add

[distant raven]

Yeah, I wish there was more flexibility like there is in Eagle

[slow plover]

in my experience, F360 felt like they told a dev team "make Great Value™️ Inventor from scratch"

it was slow as an old dog on my system (RTX 2070 super and i5-9600k at 4.9 GHz) even with relatively simple parts, and its file management (at least in the free version) was worse than nothing

frustrating, to say the least, considering its previously decent reputation

I was expecting a serviceable application with some limited features compared to Inventor

[limpid nest]

Yeah I may see if I can just afford inventor when my license expires, depending on how easy it is to move files between the two

Inventor and kiCad

[fervent lance]

a friend suggested connecting the battery via a voltage divider to one of the Analog to Digital converter GPIO pins. And maybe another GPIO pin then connected to the "charging" output of the charge control chip?

[supple pollen]

Yes, that will allow you to read the battery voltage.

[craggy plank]

I help teach EAGLE at college and the transition to their new Fusion 360 version has been very slow and painful. The new library management has been a nightmare. Really wish they'd just fix bugs with the original version and not make the setup process that much more complicated.

[limpid nest]

Oof, see above where I was complaining about it. It's really bad.

Like I paid 500 dollars for this software, I'd expect them to not release updates that completely break it.

[supple pollen]

I paid $1400 for a full professional license. Shortly thereafter, they moved to a subscription model and no longer maintain the older version I have. This is why I'll use the licensed version for as long as I can (basically until I no longer have a computer that can still run it).

[unreal flax]

Yeah, annoying. At least for the time being, they still let you run the standalone Linux pro version with a F360 license.

[distant raven]

I’m awaiting the day my edu license for eagle stops working. Until then, I’m gonna keep trucking.

[supple pollen]

I didn't know about that, do you have a pointer to the details?

Granted, I only have the free F360 license, and the offer may well not apply to that.

[unreal flax]

You'd get the free Eagle version with that, I think.

[supple pollen]

That makes sense, but the free version is insufficient to my needs.

[unreal flax]

Yup, same here. I lurk for Autodesk's typical 30%-off sale in January and grab a F360 license when those happen, heh heh.

[supple pollen]

One big attraction to KiCAD is that the development team is likely to pay attention to my wishlist (especially if it arrives as a PR for inclusion into the build)

[limpid nest]

I'm interested in how well kiCad x Inventor would work

I will have to find a good inventor community, that's the other hurdle.

KiCad would make it easier to do FOS projects

More accessible ones at least

[supple pollen]

I am interested in keeping my work FOS-friendly. Inventor isn't really a factor for me.

[limpid nest]

I've used it a bit on a student license. I liked it a lot. There's a lot to like about fusion, but they bungle so much of the cloud part of it that it just ruins the experience for me.

They get a TON of cloud stuff right, but the stuff that they get wrong is just awful

[supple pollen]

The cloud stuff I don't mind, but they have some truly strange UI choices that seem to make things unnecessarily tedious.

[limpid nest]

I don't want to make a library party for a resettable fuse if I can easily just use another lib part. The fuse is a PTC, 1612 package. Anything particularly gross about just using a 1612 1812 resistor package here?

Hmm I forgot my resistor packages, can't find that

[supple pollen]

Do you mean creating a part (re)using an existing footprint, or using a resistor part with a "this is really a PTC" string in the value field, or something else?

[limpid nest]

The second one. Sorry. I'm all over the place. I'm just making a part, it's going to take about the same amount of time as finding another part that will work

[supple pollen]

Using a resistor part for a different component is kinda gross, but even though you said "the second one" (which is that), you also said "I'm making a part", which would be the first one, which isn't gross (I re-use footprints all the time)

[limpid nest]

Oh. Again, all over the place. The second thing you said, using a resistor part.

Can I use the dimensions from the recommended footprint of another 1612 part? The particular one I need doesn't have a footprint rec.

I decided just making my own is better practice

[supple pollen]

Making your own is indeed good practice

[limpid nest]

I always have to relearn that

What do you think about re-purposing the recommended footprint from another mfg?

[supple pollen]

As I said, I do that a lot.

[limpid nest]

Sorry, wasn't fully online yesterday. Thank you

Is there a generally accepted method for picking the radius of a fillet on a board? The board has 2.5mm holes approximately 5mm from the corner, at a 45 degree angle. 5mm radius? 2.5mm? 5 seems a bit steep, 2.5 seems nice but I don't know if it's "optimal"

[supple pollen]

I'll sometimes make them concentric with mounting holes, but since it sounds like your mounting holes are offset, your best bet is just to choose something that looks good to you.

[limpid nest]

Ok yeah that's where I was leaning, just wanted to make sure there wasn't some "you must do this" standard I would be ignoring

thx

[limpid nest]

Do folks have a preferred way to strongly connect solidcore wire to a board?

I've seen some latch-down contraptions that I like, but I can't recall what they are called. You put the wire in and then flip a plastic latch down, and it clamps on the wire.

[broken python]

Terminal connectors I think is the generic term, Phoenix style often are pluggable with screws, speakers often use a specific style sized for larger wire gauges, and I found this list of a bunch of other styles:
https://www.metz-connect.com/home/products/u-contact/pc-board-terminal-blocks/spring-clamp-terminal-blocks.7e.en.html

[elder peak]

These are slightly cheaper and easier to find than the METZ ones: https://www.digikey.com/en/products/detail/cui-devices/TBL009V-254-04GY-2GY/10442221 or https://www.digikey.com/en/products/detail/cui-devices/TBL004-508-02BE-2GY/10442162

[supple pollen]

Wago 255-404 are another option, there are lots of varieties out there

[limpid nest]

Thanks folks

[amber torrent]

Thanks 👍

[elder peak]

You might say that wire terminals that don't require a screwdriver are a hot-button issue in this crowd.

[tough matrix]

Talking about hot button.
Got a chance to use high-quality thermal camera to test my motor controller design, at 3.5A load on one of the channels.

IC case temp: 118C; PCB temp: 90C front (near the chip); 102C back (under the chip)

no overheating, but I'd feel safer with some heatsinks

[tough matrix]

I used a power 3 Ohm resistor as load, and it was dissipating around 35-40W, so the resistor got really hot - even though I mounted it on a copper pipe. But I was not worried about the resistor, though

[thorn prism]

Hey folx, I'm looking for a good PCB milling machine for rapid protos. Any suggestions?

[distant raven]

Bantom Tools sells them

Pricy but good

You might find an OtherMill on eBay for $1000-$1500

[limpid nest]

These are board-mountable right? https://media.digikey.com/pdf/Data Sheets/DFRobot PDFs/FIT0586_Web.pdf

either male or female?

[unreal flax]

Those look like just cable-end connectors, where bare wires would be soldered into the half-pipe cups.

[limpid nest]

hmm mk

I want to interface a nice IP rated power supply with a board, and it has a male of these on it.

[unreal flax]

Maybe https://www.gobilda.com/right-angle-pcb-mount-xt30-connector-mh-fc-2-pack/

[limpid nest]

I need female unfortunately. But we're in the same places https://www.andymark.com/products/xt30-2-pin-female-pcb-connector

[elder peak]

I got the impression that Bantom Tools wants to light their customers on fire lately tho.

[limpid nest]

Bantom?

[unreal flax]

Bantam.

[limpid nest]

Oh different conversation

If a hole for a press fit part specifies a #5 drill, is that #5 gauge? I.e .2055"?

oops wrong channel

eh, people here will know

[steel valve]

What are improvements I can make to this routing

There now it should be a bit more readable

The only part of it that is a mechanical constraint is the horizontal diodes above the ic in the middle
And the location of the on off switch

[limpid nest]

The question I usually ask is: How much current is each trace expected to handle?

[steel valve]

The USB is a USB C

So 5V is going through it

Then the big trace on the bottom is a lipo

It turns into a small trace on the other side of the switch because idk what to do

Cause a .254mm trace would not fit there

How do I calculate the proper trace width

[limpid nest]

There are trace width calculators, one moment.

Here's one I use regularly https://www.4pcb.com/trace-width-calculator.html I usually size up by at least 10%

[steel valve]

How exactly do I use it

[limpid nest]

These are the two most critical parameters

[steel valve]

How do I tell what thickness I need to put in

Is it just from what my fab can do

[limpid nest]

Indeed

[steel valve]

I use jlc

[limpid nest]

2 Oz/Ft^2 is common. 1 also isn't unheard of.

[steel valve]

checkin

[limpid nest]

I think they use 2 unless you specify

[steel valve]

yeah

2

So how do I tell how many amps I am gonna use

[limpid nest]

That depends on what's downstream of your voltage source.

[steel valve]

WDYM by downstream

[limpid nest]

bad term for between your high voltage and your ground reference.

[steel valve]

How do i calculate that

[limpid nest]

It depends on your components you're using. Do you have a schematic we could look at?

[steel valve]

Yes sir one second

This is the schematic so far

[limpid nest]

What is the chip with the two VDDs?

[steel valve]

A MCP73831

For charging the lipo

And supposedly the VDD and VBAT nets should be shared

https://www.mouser.com/datasheet/2/268/20001984g-846362.pdf

using the dfn package

[limpid nest]

So you're going to want to note this https://ww1.microchip.com/downloads/en/DeviceDoc/MCP73831-Family-Data-Sheet-DS20001984H.pdf and basically copy their capacitor setup.

The device may not work properly otherwise

[steel valve]

That is what I am using in the schematic

It just looks weird cause the dfn package drawing

[limpid nest]

They use 4.7 uFs. Am I just missing it?

[steel valve]

Oh thats just a placeholder

[limpid nest]

Ah OK.

So it looks like your charge current is up to 500 mA

So you'll need traces that can handle 500 mA at a bare minimum.

[steel valve]

Yeah, I am using a 110mah lipo

[limpid nest]

Should charge quickly at least!

[steel valve]

The charge current on the mcp is programmable

[limpid nest]

yeah, up to 500 mA

[steel valve]

Oh so it doesn't matter what the charge current is set to

[limpid nest]

No it does.

I'm saying your traces for the charge circuit need to be able to handle 500 mA

And your power traces to the device need to be able to handle something a bit more than that. Are you planning on powering this from a USB port on a computer, or from an adapter?

[steel valve]

A usb port on a computer

[limpid nest]

Ah OK

Then you won't be able to get the full 500 mA out of your device.

[steel valve]

Yeah

Plus I don't want to get the full 500mA

Because that would destroy my battery

[limpid nest]

Still worth considering making the traces comport with that, just so you're safe

[steel valve]

http://cdn.sparkfun.com/datasheets/Prototyping/spe-00-DTP401525-110mah-en-1.0ver.pdf

It has 1C Maximum charge

So assuming I wanted to have a meh battery span on my device it would be 110mA max charge current

[tough matrix]

actually most fab houses - including JLCPCB - use 1oz copper by default. 2oz is much more expensive

[limpid nest]

you want to charge it in 1 hour?

[steel valve]

An hour and a half

[limpid nest]

Ohhhh I was worried I'd mixed that up!

[steel valve]

An hour and a half is the intended amount of time I want to spend charging it

[limpid nest]

ok

so even less than 110 mA charge current

[steel valve]

Yeah

Back

[limpid nest]

Welcome back 🙂

[steel valve]

Ok so pretty much I am tryna charge it in 1.5H

But dont understand the equation of the rprog

How would I get the resistor value that I need to use from this formula

[limpid nest]

Do you know your desired current?

I calculated 73

110mAh/1.5hrs

ohh I see

Let me pull up the datasheet

Yes, OK I understand now. So I calculated the current you want the charger to produce as 73.33333 repeating mA

[steel valve]

Yeah

[limpid nest]

So we plug that in for I_Reg, and move R_Prog and I_Reg around.

We get 1000V/73.3333mA = 13.636 kOhms

then we can look for a resistor near that value

[steel valve]

So 13.6kOhms

[limpid nest]

roughly.

If you can find exactly that, great! Otherwise you'll have to find something close

[steel valve]

Yeah

So now that we know the charge current, how do I get the amps out of it for the trace width formula

[limpid nest]

So that's gonna depend on each trace.

Your overall power bus should have more than 73 mA, since the charging circuit is downregulating from 5V. That's going to burn some power. You can look at the datasheet for the chip you're using and get values for its power consumption. What's the chip on the right of your schematic?

I think 250 mA would probably be safe?

[steel valve]

MIC5365-2.0YMT-TZ

https://ww1.microchip.com/downloads/en/DeviceDoc/MIC5365-6-High-Performance-Single-150mA-LDO-DS20006605A.pdf

That's its datasheet

It's a LDO

[limpid nest]

Shoot, I gotta run, hopefully someone else can help. You're not far from an answer. I suspect that a value like 250 mA is going to give you a safe cushion, but I don't want to say for sure.

[steel valve]

Once you come back if no one else has a answer, ping me and I'll be on all day

[limpid nest]

mk!

[steel valve]

But to reply... The MIC5365 has a max 150mA output current, and the current it is downregulating is variable

Pretty much...

USB C (5V) -> MCP73831 (Charge Lipo) -> 3.7V LiPo -> MIC5365 (2V Regulated)

And then the 2V goes to the MCU, and Sensors, which draw barely anything

15 Potentiometer-type Sensors, and a Sparkfun Artemis module.

And for reference, this is the chip datasheet.
https://cdn.sparkfun.com/assets/1/5/c/6/7/Apollo3-Blue-MCU-Datasheet_v0_15_0.pdf

[peak rose]

This is probably the wrong question for this channel ~ but I needed some general help in designing a shield for Arduino Mega and I'm using some parts from Pololu. Does anyone have EagleCAD libraries for Pololu Parts?

[steel valve]

Assuming it is running at 1.8V, High Performance Mode, it is running 41μA/MHz

Unless I am wrong w the datasheet

And the sensor power draw is negligible (and also partially unknown to me due to one of the sensors lacking a datasheet)

[steel valve]

So, I just fixed the schematic to make it more readable. I also put the current voltage on each section.

[tough matrix]

what Pololu parts are you using?

[steel valve]

@limpid nest How do I know what amps are coming from the USB C

[limpid nest]

That depends on the "load", your circuit. From what I've seen, you should be safely under the 500 mA cap

[steel valve]

Yeah

[limpid nest]

In DC, current is equal to voltage/resistance.

[steel valve]

I have been able to put in the 73mA charging current

And then the up to 20mA in a nightmare scenario the rest of the project current

[limpid nest]

Yeah so we can conservatively say 150 mA to be safe.

[steel valve]

Ok, cool, and 1 oz/ft^2 is the unit that jlc uses

So, 0.127mm is what should be used

[limpid nest]

That's the minimum

[steel valve]

Since that is the lowest trace jlc can do

[limpid nest]

I wouldn't go that low

I don't go lower than 10 mil myself

When you solder, with teeny traces, it's easy to have the copper delaminate under heat

[steel valve]

What do I do in situations where I can't do 10mil

[limpid nest]

Ah

[steel valve]

Like in my original trace that I showed

[limpid nest]

10 mil = 10 milli-inches

[steel valve]

I use 10mil

Except for the LDO

Which has DRC issues when I put in 10mil onto those tiny pads

[limpid nest]

Hmmm

[steel valve]

But some of that issue is probably due to easyeda

Cause the trace is rounded

[limpid nest]

Is the yellow a silkscreen?

[steel valve]

And snaps to the pad

Yes

[limpid nest]

Hmm

[steel valve]

It's a silkscreen since these are lcsc footprints

[limpid nest]

Ahhh ok

What error does it give you?

[steel valve]

clearance pad to trace

[limpid nest]

Hmmmmm

Weird

Because that trace looks totally legal to me in the CAD I use.

I'm entirely unfamiliar with Easy EDA I'm afraid

[steel valve]

This is my design rule on easyeda

(The capabilities of jlcpcb)

[limpid nest]

I wouldn't use their minimum capabilities btw.

Like it's probably fine, but I'd hedge your bets and add a little overhead.

Well I guess scratch that. For DRC it's appropriate

[steel valve]

I just have them set to that for drc

[limpid nest]

Yeah good call

[steel valve]

I could try and disable trace snapping

To see if I can get that 10mil in the edge

Since it snaps to the center of the pad

Yep that solved it

EasyEDA was adding a extra diagonal bit into the pad

[limpid nest]

Oh great+

!

[steel valve]

For this section, is it better to have a via on both ends

Or is having the via go between diodes diagonally ok

[limpid nest]

For GND?

[steel valve]

Yep

GND

[limpid nest]

Should be fine how you have it

[steel valve]

Ok cool

Once I turn the traces to the correct size

Any changes to component placement

My main concern is the LDO

Since those typically should be far away from components

[limpid nest]

What's it for?

[steel valve]

Linear Dropout Regulator

[limpid nest]

I know that

[steel valve]

It takes the LiPo voltage and makes it 2V

[limpid nest]

Hm, for what purpose?

[steel valve]

Oh you mean the purpose of the pcb

[limpid nest]

Of the LDO on the PCB

[steel valve]

Oh

To keep the project at its lowest possible voltage

It saves energy from my understanding

Running the Sparkfun Artemis at 2V, and 15 potentiometers at 2V uses less energy

[limpid nest]

Ah ok. So this seems like a hat board, is that right?

[steel valve]

As in a expansion board?

If so, yes I believe so

Pretty much, it houses a Sparkfun Artemis RF Module

This is the whole board so far, making pin changes to it

Since on those two red circles will be two 8 Channel Analog multiplexers

[limpid nest]

Ahh I see

Neat project

[steel valve]

Pretty much, the artemis has 10 ADC, it supports using external ADCS via QWIIC but that isn't easily done with the parts shortage RN

And I wouldn't have any benefit from doing so

I am putting in 15 different potentiometers on it, which all require a voltage divider

The thing is, there just is not enough space for 15 resistors

Plus the module already does not have enough ADC

So I am using two mux

On the output of the mux is a voltage dividing resistor

[limpid nest]

Neat!

[steel valve]

I've just gotta find a mux that is a small enough package

And also in stock on mouser

Cause so many of the ones that are a package that would fit have <1000 of them in stock

[limpid nest]

Are you making a ton of these?

[steel valve]

I am not, but I don't wanna have times where I make a couple and the part is just oos

Cause this is for two devices that send stuff to each other

So that means I would be using four for one project

[limpid nest]

Ahh ok. Yeah.

I assume you looked at resistor arrays?

[steel valve]

And lead times suck nowadays

Yeah

[limpid nest]

Yeah it's ruff getting parts

[steel valve]

But even then I have to use a mux or external adc

[limpid nest]

Ah gotcha

[steel valve]

Because even if I use resistor arrays I need 7 more analog channels

So I am just putting the 15 main pots on two 8 channel mux

But any mux that fits the project size has <1000 of them in stock usually

And I am reflow soldering with a hotplate so I would like to do this in one layer

[limpid nest]

Ah ok. Well, hmm, 1000 is IME not a bad buffer

These days you have to be a bit more risk tolerant

[steel valve]

Yeah, plus I can easily swap it out for another one in the same series

Since I only have to change two parts (the ldo and mux)

Just to support a diff voltage since going down to 1.8V is just overkill in the first place

Considering the Apollo3 (chip of the artemis) consumes practically nothing

https://cdn.sparkfun.com/assets/1/5/c/6/7/Apollo3-Blue-MCU-Datasheet_v0_15_0.pdf
From my understanding, this thing is gonna consume like maybe 5mA max when running TX, 96MHZ, useless stuff disabled.

The purpose of the Artemis is to take the values of the sensors, and send it to the other pcb for it to be processed as fast as possible

[limpid nest]

Nice!

Oh weird messages loaded out of order

[steel valve]

Well, just found a mux that is 1.8V, fits the size criteria, and has 10k in stock, nice. it's pricey (97c each) but works

(Originally i was using 2v for the ldo since the mux was 2v originally but when I moved to using one layer part placement the nicer ldos are 1.8V)

Oh yeah I forgot since the conversation shifted from the original question, how far should a LDO be from the rest of the parts?

Since it's converting a LiPo voltage to 1.8V*

[unreal flax]

No particular distance requirements. An LDO doesn't have any switching circuits, so it's electrically pretty quiet.

[steel valve]

Oh I am saying for heat

Don't these things get toasty

[unreal flax]

You'd want to do the math on that based on the current you expect to draw. The LDO datasheet will have some numbers to convert the watts it dissipates into a temperature rise in the chip package.

[steel valve]

Ok

https://www.mouser.com/datasheet/2/268/mic5365-1082359.pdf
So, PD = (4.2V - 2V) * 150mA
(Output current is always 150mA apparently)
PD = 0.330W
0.330W = (125C-TA)/(250C/W)

Am I doing the formula it wants right or is this the wrong part of the datasheet

[unreal flax]

150mA is the maximum rated current of the LDO. Does your circuit actually consume that much all the time?

[steel valve]

No

It would be shocking for it to consume more than 10mA

[unreal flax]

Then that's the number you should use for the heat dissipation.

[steel valve]

Ohh ok

I was getting confused cause I thought it was saying no matter what the output current is 150mA

So:

PD = (4.2V - 1.8V) * 10mA
PD = 0.024W
0.024W = (125C-T)/(250C/W)
0.024W = (125C-T)(250C/W)

[unreal flax]

So it looks like you're looking at about a 6C temperature rise for the junction (silicon) temperature. The chip package should warm up even less than that.

[steel valve]

So pretty much it's negligible

[unreal flax]

Yep, yay for low-power design. 😁

[steel valve]

Ok, cool.

https://www.ti.com/lit/ds/symlink/tmux1208.pdf?HQS=dis-mous-null-mousermode-dsf-pf-null-wwe&DCM=yes&ref_url=https%3A%2F%2Fwww.mouser.com%2F&distId=26
How exactly do drain pins work, it says they can be an input or output, but what does a drain pin do

Is it supposed to be the general output pin?

I am intending to use this as a multiplexer, with a 8:1 conversion

[unreal flax]

The multiplexer is bidirectional, so you can choose one of the source input pins to be the drain output, or you can send the drain input to be one of the source outputs.

[steel valve]

Ok so I think I get it, I was just confused by the TMUX1209 in the same datasheet that has two drains

This is pretty normal then, cool.

Oh yeah, forgot to ask. How much bigger should the gray area of a through hole be in comparison to its hole

Like I have a 0.5mm diameter hole, how big should the gray area around it be

[unreal flax]

That'll typically be a specification of your board house, depending on their fabrication equipment. You want the copper ring around the nominal hole position to be large enough to account for any inaccuracies in where they position the drill versus where the center of the circle is. It'll typically be called an "annular ring" spec.

[steel valve]

Oh ok that's what its called

So for stock 1oz copper, 0.13mm ring is the minimum

Or is it 0.3mm cause its a through hole

[unreal flax]

The 0.3mm is the minimum hole size.

[steel valve]

Ok just making sure, cool

So then it would be a .8mm pth, with a .5mm diameter hole

Or wait, you mean 0.3mm as in the minimum drill hole diameter

and 0.13mm is the gray ring

[unreal flax]

Yes. So for a 0.5mm hole, you'd want more than a (0.5 + 2 * 0.13) = 0.76mm diameter pad around it.

[steel valve]

Ok cool

[unreal flax]

I'm not familiar with this tool's UI, but I presume that's doing the right thing. Note that while you CAN aim for exactly your board house's limits, adding a little margin is also not a terrible idea, unless you're really optimizing the heck out of the design.

[steel valve]

Yeah, I am gonna be using the .8mm circle instead

So that I don't have weird issues

I am just tryna minimize the pinhole space since I am optimizing my pcb design a bit more since I can probably fit this in two layers and in that case there is a lot of unused space, and I am going a couple mm wider than what I originally planned to so I am tryna lower the width to be what i originally wanted to fit in some spare things I want to add

(current pcb pinholes)

[unreal flax]

Nice. What's that funky 16-pin chip towards the bottom left?

[steel valve]

That's one of the two mux that will be used

The one I asked about how the drain pin works cause I misread the datasheet like a dodo

[unreal flax]

Gotcha.

[steel valve]

Pretty much, there are 15 potentiometers used

And only ten analog pins on the module

Also, the pots all require a voltage divider

So I am using two 8-channel mux, and each mux has on its output one voltage divider.

Since that should work

And in case the module of choice goes OOS out of nowhere due to the parts shortage it would be easy to swap out for a different one

Since a big focus is making it so that its very unlikely that things will go oos to the point where there isn't anything similar enough to make the project

[unreal flax]

Yep, that's a super-annoying design consideration these days...

[steel valve]

Plus it's only 90 cents extra to do the two mux thing in the first place so why not

Which would mean, this is (generally) the final parts layout of the pcb aside from me shrinking the sides

Excluding the extra decoupling caps I would need to add for both mux, and one for the artemis. And also the two resistors that the mux need

[marble edge]

hi, this seems like a good place to ask: I'm involved in a makerspace and we're looking at acquiring a reflow oven. Does anyone have any recommendations for particular products that are budget-friendly? Doesn't have to be the cheapest option available; but we're not really looking to spend over $500.

There have been attempts to build DIY solutions in the past, but at this point it seems preferable to buy something that has already been tested and assembled (no one really has the bandwidth for a 12 hr project)

Seems like the PCB design channel might have some good experience to plumb

[distant raven]

Well, $500 would probably get you a drawer oven which would likely need some modifications to operate consistency

[distant raven]

Can I send you a dm?

[marble edge]

yep, actually I just beat you to the punch 🙂

[limpid nest]

I have a suggestion as well if you don't mind!

[marble edge]

please!

[limpid nest]

DM OK?

[marble edge]

yep that's fine

let me know if you can't dm me, i may have messed with my privacy settings and disabled it for non-friends

[limpid nest]

ok!

[limpid nest]

If I have a surface polygon of Y width, sized to allow enough current to pass through, and then I add a bunch of stitching vias to an internal plane that's bringing current elsewhere, do the stitching vias greatly impact the first polygon's ability to transmit current? I would think they would be bad for it

[distant raven]

Technically it would increase surface area if done right

[limpid nest]

hmmmm

[distant raven]

Or in general it would. But you would be creating parallel paths I imagine

[limpid nest]

Hmm that makes sense I guess now that I think about it.

[elder peak]

Sometimes it's a good thing if your PCB design is unparalleled.

[limpid nest]

heh

[steel valve]

What stackup should a four layer pcb where all of the components are on the top use?

[limpid nest]

Hmm, I can give you the numbers I use for my CAD program. I assume they mean the same thing

Just give me a moment, I learned of a critical security bug in my machine through google news instead of through my manufacturer, so I just finished rebooting after an update.

[steel valve]

\

I am at the phase where I am routing the mux

Which is the most complicated part

And am just making sure my stackup is fine before I go all in

Signal Layer (Components)
2V Plane
GND Plane
Signal Layer (No Components)

[limpid nest]

I just lost wifi >.< and fusion is cloud based

[steel valve]

F

[limpid nest]

I usually see/read it as signal/GND/Pwr/signal

Oh I see what you meant

I can't recall why people do it S/G/P/S but they do

[steel valve]

Also, is having complicated routing on one layer part projects normal

[limpid nest]

Yeah

Pretty normal

[steel valve]

Ok cool just making sure that I am not an idiot with these like billion vias

[limpid nest]

I almost always redo it multiple times until it's how I want it

[steel valve]

This routing I am gonna redo once or so to make it look a bit prettier

But the routing is pretty much the same every time

Since I have to put the sensors in order to be read easier from the software-end

[limpid nest]

Yeah there are a finite number of solutions that approach optimal

[worldly schooner]

Heh, if you want to see messy routing, you should see my 2-layer RP2040 keyboard...

It's so bad I'm struggling to figure out what my ground pour is going to look like...

[limpid nest]

I'm lazy, I just do 4 layer for everything. Even when I could probably do 2 layer

[steel valve]

Yeah, in my case I am just stuck doing 4 layer unless I want to have a single component on the bottom of the pcb

Being the mux

C

[worldly schooner]

Hmm, seeing the way some of those are routed, I wonder if the routing is cleaner if you rotate the whole chip 180 degrees...

[steel valve]

Ima check actually 1 sec

[worldly schooner]

Ooh, the pin order gets reversed, if that's an issue at all.

[steel valve]

Nah

I am just tryna connect the sensors in order

So Sensor One connects to Input One (or Eight)

Sensor Two Connects to Input Two (Or Seven)

[worldly schooner]

Well, there's nothing wrong with using more vias, but you can check out your options and see which one you prefer.

I'd save a copy of what you have just in case.

[steel valve]

I still am not sure on what the simplest method of doing what I want in this case is

Pretty much, the mcu has 10 analog inputs

I need 17

Also, 15 of the 17 analog sensors use voltage dividers

I am using two mux, with a voltage divider on the output of each for the sensors

Which is really funky to route

No matter what I am gonna have to use a mux most likely since most external adc are oos or super expensive and out of the scope of the project

[worldly schooner]

You pretty much have the right idea, an analog MUX is generally the choice there.

[steel valve]

Also, 10 of the sensors are a flexible potentiometer, the other five are normal potentiometers

So two of the flexible potentiometers would end up in the other mux

To make it fit

But then that might screw with the voltage dividing stuff

[worldly schooner]

If you only need to poll the sensors at a low frequency, a general-purpose analog MUX should do the trick.

[steel valve]

Since I assume the two sensor types use different valued resistors for the voltage divider stuff

[worldly schooner]

You're using 2 8-to-1 MUXes?

[steel valve]

Yeah

I am using mouser parts

[worldly schooner]

If you have a voltage divider and want to isolate the input impedance, you can use a buffer amplifier on the output of the mux?

What kind of voltage divider do you need on each pin?

[steel valve]

Typical Bend Sensor Voltage Divider ^

10 bend sensors are used

https://www.mouser.com/datasheet/2/18/1/12_PT10v03-1499201.pdf

Five of these trimmer potentiometers are used

Which use this recommended voltage divider circuit

[worldly schooner]

The other five "normal" potentiometers, are they being monitored with a similar voltage divider?

[steel valve]

They are all intended to use a similar voltage divider

[worldly schooner]

Because a pot inherently has its own voltage divider.

A flex sensor does not.

[steel valve]

Wait so a potentiometer does not need the other half of the voltage divider

(Flex Sensor Diagram)

A pot has both resistors?

[worldly schooner]

Well, the pot has two terminals for VCC and GND, and the analog voltage is read from the wiper pin.

[limpid nest]

Yeah a pot is a voltage divider

[worldly schooner]

The flex sensor needs the other resistor, the pot technically does not.

[steel valve]

Ok, so I gotta do this a bit differently then it seems

[worldly schooner]

If you have 10 ADCs, you can measure the pots on 5 of them, and put the other 10 flex sensors across two MUXes on two ADCs.

[steel valve]

So, the pots go to 5 ADC, the two extra sensors go to 2 ADC, and the ten flex sensors go to 2 ADC via two mux.

[worldly schooner]

Yup, unless you have different frequency requirements across the sensors.

[steel valve]

Yeah all the bend sensors are the same

If only there was a 6:1 mux on mouser, that would make this a bit more even

(Routing-Wise)

[worldly schooner]

MUXes only come in 2^n circuits.

[steel valve]

Six bend sensors come from the right, four come from the left

At least mouser has a 1x4:1 that's from texas instruments (they make the other mux I use)

[unreal flax]

I mean, you can just use 6 channels of an 8:1 mux if you want a 6:1 variant.

[steel valve]

Oh, it's just for routing purposes

Already onto routing that section and just choosing the simplest path

Then all I need to do is add in pins for programming this.
What is the best method of putting arduino bootloader and code onto this module?
https://www.sparkfun.com/products/15484

[proven mason]

Anyone help me out by pointing me towards the part I need to include in my project to provide through-hole footprint for headers on my PCB in KiCad? I've been through the entire library, and found nothing

[worldly schooner]

Pick a board like https://www.sparkfun.com/products/15444 with the module on it and use its schematic as a reference? (click the Documents tab)

[worldly schooner]

This?

Should be entire categories for Connector_PinHeaders and Connector_PinSockets in various pitches.

[steel valve]

Yeah, they use a 12 pin jtag header, I just don't think I am gonna be able to fit it in

[worldly schooner]

They also have a USB port with a CH340 chip?

[steel valve]

That's doable. My question would be, can it program it without a bootloader or anything being on the module?

The module comes completely blank

[worldly schooner]

https://learn.sparkfun.com/tutorials/designing-with-the-sparkfun-artemis?_ga=2.185550249.1824767937.1650499806-989614518.1636901821 might be a good place to start. It talks about the factory bootloader and the necessary circuitry to get it working.

[steel valve]

Sorry for the late response. I've already looked into that, they come with a preprogrammed sparkfun bootloader, but how is it programmed with arduino code aside from using a CH340

Do I just add in pins for a serial programmer? Or is there like another way of programming it aside from that

[worldly schooner]

There's a factory bootloader in every module. The section titled "Factory Bootloader" should be what ships onboard every module if I understood it correctly. That should tell you how that works...

Actually, I think the modules might come with both bootloaders preinstalled. You should be able to program with Arduino right away with the Artemis boards loaded in your IDE.

[steel valve]

Yeah they do

That is my question, how do I program it with arduino

Like what pins do I need to add

I could put a CH340 IC onto the pcb but then I gotta order from alibaba or lcsc to get one to the USA

[worldly schooner]

I'd reference an existing development board schematic for that. The USB connector and serial converter circuits should provide that.

I think you can adapt a CP210x as well, but I haven't looked into the connections too deeply. They should be fairly similar IIRC.

@steel valve If you need detailed help with the Artemis module, the SparkFun forums might be a better place to ask.

[steel valve]

I'm probably just not understanding the terminology

So I am gonna try and re-explain

The Sparkfun Artemis Module has a bootloader, how do I upload arduino sketches to the module without using a ch340 on-pcb

[worldly schooner]

https://www.adafruit.com/product/5335 or some other USB-UART programmer? FTDI?

[steel valve]

So I should use an external one and just put the pins for the programmer onto my pcb to simplify things

[worldly schooner]

If you don't have any means of adding a serial converter or prefer an external programmer? Absolutely valid.

[steel valve]

Yeah, that's what I was trying to ask originally if that was a good idea, sorry for being confusing lol

Pretty much I have the option of adding a programmer such as a CH340 or CP2102, but would have to place a component or two on the bottom, which would use up more vertical space on top of being annoying to assemble since the top is soldered with a hotplate

[proven mason]

It seems whatever libraries shipped with my KiCad v6.0.4 do not have that category or those parts

[untold moat]

Anyone know if a jtag connect leg touches a trace of a pcb on the other side if issues will occur
Image
I will be adding a spot for one of these for ftdi programming
But the area it will connect to is directly under the main mcu module
Those gray pins will touch the back of the top pcb
Will issues occur if the traces of the top pcb touch the gray part of that tag connect

https://media.discordapp.net/attachments/622118305930674192/966577475410874438/unknown.png

[supple pollen]

These are stacked PCBs in close proximity?

[untold moat]

Yeah

I have an Artemis RF Module on my Breakout PCB

The Breakout PCB has a area for jtag connect programming

That area happens to be directly under the RF Module

[peak rose]

I'm using their High Power Motor Driver https://www.pololu.com/product/2517
Using 3 to interface with 6 Motors

[untold moat]

Also this is what it would look like on the pcb (no rf module attached since I have no 3D model of it)

But there would be a Sparkfun Artemis RF Module on the top layer

[unreal flax]

That shouldn't be a problem electrically, but it may be a problem physically. The TagConnect alignment pins ordinarily poke completely through the board and come out the other side, so you wouldn't be able to fully insert the connector. It's not obvious whether the spring pins would make good contact with the pads with just a partial insertion.

[tough matrix]

I do not think Pololu publishes schematics or board designs of their boards. Maximum they do is publish dimensions and position of holes - so if you want to recreate their board, you are out of luck

[steel valve]

https://www.sparkfun.com/products/15484
https://cdn.sparkfun.com/assets/9/a/7/1/c/ArtemisModule_Pinout.pdf
https://cdn.sparkfun.com/assets/learn_tutorials/9/0/9/Apollo3_Pad_Mapping.pdf
The sparkfun artemis module has 10 analog channels
Why does it show that there are way more on the pinout?

[untold moat]

Oh, so the tag connect pins go through the other side. Ok, what are other solderless ways of programming a pcb?

Or is the best option to just add some programmer pads on the bottom

And quickly solder and desolder wires to it when needed

[untold moat]

I have the same question. The pinout datasheet is confusing.

[unreal flax]

It looks like they may be counting some of the ADC-associated pins, like the trigger inputs, as "analog" even if there are only 10 actual channels.

[untold moat]

How do I tell which ones actually link up to the channels

[unreal flax]

The source of truth is the chip datasheet. You'd work backwards from which module pins are connected to which chip pins.

[untold moat]

That's gonna be annoying, F.

[unreal flax]

Heh heh, welcome to embedded engineering, where The Details Matter™.

[untold moat]

Yeah cause even this is confusing on what it is about

https://cdn.sparkfun.com/assets/learn_tutorials/9/0/9/Apollo3_Pad_Mapping.pdf

Which is the pad mapping to mcu pins

[unreal flax]

That document looks like it's the MCU pin table, so you'd need to look at the module documentation. The numbers on the Artemis doc you linked above would correspond to these pin numbers.

[untold moat]

So that's what it means

I go off of the pad number

From the function mapping chart

And see where it is located on the module visual chart

Ok that makes sense it just clicked on what it meant thank you.

I have another question, I plan on using an external programmer for my pcb due to space constraints. I plan on using a USB to TTL programmer, most of them use 3V3 or 5V pins that are supposed to be used to power whatever it is programming, how do I make it so that the current doesn't flow where it should not?

I would have my project be turned off (no battery powering it) to program it via the programmer from my understanding, and would have the 3.3V (or 5V) connect to the MCU, how would I prevent that from flowing into other parts of the PCB?

To explain with a diagram

Both 1.8V and 3V3 would be in the same net

And 1.8V would not be used while 3V3 is used

Wouldn't that make the 3V3 flow into the 1.8V area

And the 3V3 flowing into the 1.8V area would cause problems, wouldn't it

Since that 1.8V area powers a bunch of sensors and goes back to a ldo

[untold moat]

When should a capacitor be used aside from when a datasheet requires one
I have a charging, linear dropout circuit that has a decoupling cap going into the charger, out of the charger to the lipo a decoupling cap is there, and from the lipo to the linear dropout there is a decoupling cap, and out of the linear dropout to 2V which the project runs on is a decoupling capacitor

https://images-ext-1.discordapp.net/external/Td2PnzapwtyDqal3LHfyLKCAQ9fF78Mg3_sSW26OdFs/https/media.discordapp.net/attachments/622118305930674192/966885868155596850/unknown.png

[supple pollen]

It's actually somewhat difficult to identify where capacitors are, in fact, needed (the answer usually boils down to "it depends"), and lack-of-capacitor problems tend to be difficult things like it fails once in a while during hard-to-identify conditions, so the usual approach is "capacitors are cheap, they can't hurt, just sprinkle them around"

[untold moat]

So pretty much just use them before an IC, in a long trace, etc to save the chance of blowing a component or something.

[supple pollen]

That's a good summing-up.

[untold moat]

Yeah, better safe than sorry. Ok, just was making sure that I was doing this right.

Then to what my actual question is for my situation. I have 2V going into an IC, which has an Enable pin, these two pins are pretty far away, would it be better to do some complicated routing to make it work

Or would it be better to use two decoupling caps

[supple pollen]

I wouldn't use a capacitor on the enable pin, but I will often put a capacitor near a voltage source and another near whatever is using the power, if they're not close to each other.

[steel valve]

What is CTX/CTS and DTR/RTS for an external FTDI Programmer?
Trying to see if it is necessary for programming a sparkfun artemis module

[supple pollen]

Those are originally intended as flow control signals to throttle data transfer, but these days, they're often used as auxiliary signals to perform functions like resetting a processor to enter the bootloader to upload code. I don't know offhand if the Artemis uses this functionality, however.

[distant raven]

But just remember the output capacitance range of the onboard LDO too, too much capacitance can lead to long rises. I think this is where if you’re running a lot of peripheral on one board you have multiple supplies. But maybe if things are far enough away the inductance of the trace makes up for things

And this is actually a pretty interesting topic too because I’m designing an IoT board and I’ve got a 1000uF electrolytic capacitors on the 5V line for power buffering neopixel strips. If nothing is attached neopixel wise, the system starts up, the capacitor charges dropping the input voltage to the DC-DC converter causing the system to restart as the capacitor comes up lol

I could probably fix this by adding a supervisor or something but neopixel will be connected more often than not

Hey @supple pollen have you ever used an On-Off switch IC for doing power on/off on a board?

More specifically that fancy board you share every now and again

More or less looking for suggestions because all the options I’m seeing are $5+ a single

[supple pollen]

I've usually just used regulators with an enable lead, but I was looking at (mis)using some USB power switch ICs to control/monitor power to an IC under test in an IC tester. There are also a couple of AdaFruit boards with power switching ICs, I don't know if you had looked at those chips or something else.

[distant raven]

I haven’t

I thought of just putting a switch in line with the barrel connector

[supple pollen]

Ah, the direct approach. That works too.

[distant raven]

Though the power I/O switch IC seems like such a clean approach but at a step price for such a simple function

[fervent lance]

What's the problem that's being addressed using a power switch inline with a 2.1 mm barrel plug's cable?

(ex. I have an 8051 target board running off a current-limiting bench supply; in order to avoid wearing out the 2.1 mm plug jack on the target, I've taken to unplugging the banana plug at the PSU)

[distant raven]

Just on and off power control

[fervent lance]

Why does it need to be powered OFF?

[distant raven]

For if I’m out of town and don’t want it on

It’s primary function is running neopixels

[fervent lance]

When I had a desktop PC that would automatically dial up an ISP internet connection on a modem, it'd send email I could read while travelling .. and I could pickup its exact dynamic IP address that way.

Then ssh in.

When it didn't work I'd have someone cycle the power switch on the UPS that powered the PC.

I wanted a way to do it by remote but couldn't think of one. ;)

[supple pollen]

Now I'm wondering if X10 would work through a UPS

[fervent lance]

I'm thinking a relay that takes no energy to hold the current state.

[supple pollen]

Latching relays are good for that

[fervent lance]

Then just build up from there, to 1000 watts 120 VAC if you like.

[distant raven]

Anyway, just would be cool to include an I/O controller but it’s a bit steep of a price lol

[fervent lance]

@supple pollen Yeah I'm still looking at the USD $7-ish adafruit relay board for that. ;)

just haven't gotten a round tuit yet

The new Honeywell thermostat I put in this year makes a loud click when it cycles on or off.

I can hear it from 12 ft away, easily.

I think it runs on two AAA alkaline cells for a year.

[steel valve]

Yeah, I have no idea either which is why I was asking, cause the artemis has crappy documentation.

Currently looking into the PCBS they've made to see if they utilize those pins

[supple pollen]

I bought one myself, drawn in by the "get blinking in 5 minutes". Several hours later, in extreme frustration, I put it away and never picked it up again.

[steel valve]

I mainly want it for the low power consumption

Like this thing consumes so little

[supple pollen]

I ended up opting for the TI MSP430FR series, which offer not only amazingly low power consumption, but non-volatile memory so even if the power runs out completely, it can pick up where it left off when it's powered again.

[steel valve]

Are there Premade RF Modules of the TI MSP430FR series?

[supple pollen]

Not really. There are some low-power RF capable ones in the SensorTag range, and some NFC enabled (and optionally, powered) variants.

[steel valve]

Yeah, I am using the Artemis for RF capabilities

Cause it's bluetooth 5, and consumes pretty much nothing

It's like 3mA when transmitting +4dBm if I remember right

Compared to the only other alternative module I know of, the NRF51822

Which is 16mA +4dBm

And Bluetooth 4 (oof)

Adafruit has the NRF52832 module which has a close enough power consumption, but they're almost always OOS

Like RN it is OOS on Adafruit, Mouser, Digikey, and Amazon

[supple pollen]

Maybe look at the TI CC2650. Bluetooth and runs for over a year on a coin cell.

[steel valve]

Has anyone done premade modules of that chip yet?

[supple pollen]

Yeah, that's the "SensorTag" reference design I mentioned earlier

[steel valve]

Is it sold on mouser?

And if so, link it ima check it out

Ok so apparently the artemis does use one of those functions

It uses RTS

This is a CH340E schematic for the Sparkfun Artemis Nano PCB

And RTS goes to a bootloader reset circuit

Wait I need a circuit for doing a bootloader reset on this module? OOF

Maybe this is just for the RST button on the nano pcb which is why it is more complicated?

Cause I just want to be able to program my thing with an external ftdi programmer

Ok so from my understanding, the RTS is still optional according to the documentation

https://learn.sparkfun.com/tutorials/designing-with-the-sparkfun-artemis?_ga=2.120244175.632158558.1650658281-1346487478.1650658281

RTS is just for the bootloader reset circuit, which is used for more secure programming

Am I correct on this?

[supple pollen]

More convenient (you can have the software reset and load the board for you instead of having to press the button at the proper moment), not necessarily any more secure.

[steel valve]

Wait press what button at the proper moment

The program button?

[supple pollen]

The reset button.

[steel valve]

Oh I don't even have a reset button on my pcb

Should I add one?

I only have a on/off switch

What's the best way to go at it so I can just connect the FTDI jumpers and load the arduino sketch

Also, there isn't a BOOT/1 on the schematic for the bootloader reset circuit

https://cdn.sparkfun.com/assets/5/5/1/6/3/RedBoard-Artemis-Nano.pdf

I am confused, curse you sparkfun

[steel valve]

What exactly made it frustrating to work with

[supple pollen]

Apparently the design has changed some, and you need a particular combination of bootloader, programmer, compiler, build suite, libraries, etc. that all works together. But backdating things in phases never did come up with a working combination.

Oddly, mine does have a reset switch

[steel valve]

Yeah, I just added the bootloader circuit to it for programming via an external FTDI programmer

And plan on programming via arduino

So it (should) be more simple to program it?

[gray lance]

Question: If a board house has a spec of 6mil for trace width, is that their maximum or their minimum?

[heavy jasper]

Minimum

[gray lance]

Oh whoops nevermind....... I was mixing mil and mm.

Im good at reading.

I was thinking 6mm was far too large to be their minimum

[steel valve]

I made a PCB in Eagle and it is not converting to EasyEDA and says an error occured when opening the BRD. How do I fix this? I am using eagle 9.6.2 and newest version of easyeda

[tough matrix]

i have very little experience with EasyEDA, but may i ask why you need to convert it to easyeda?

[tough matrix]

hi all:
anyone has experience with PCBs with thicker copper (2oz)?
I am thinking of using 2oz, 4layer pcb for new revision of motor driver - to help dissipate heat. What worries me is that same board also has RP2040, which is a QFN with 0.4mm pitch, and it seems to be at the limit of what JLCPCB can do for 2oz copper (they require minimal track width and minimal spacing of 0.2mm for 2oz).

Do you think it will be ok, or I should avoid it?

[supple pollen]

I suspect it will be fine. Other approaches you could consider are another PCB manufacturer, or putting the finer pitch parts on a daughterboard.

[signal topaz]

What exactly are you trying to cool on this board? Thicker copper is not always the answer. Maybe it will be better or even cheaper with a heatsink?

[tough matrix]

motor driver
it needs to dissipate up to 5W

[signal topaz]

Is it a single integrated circuit?

[tough matrix]

I am adding heatsink too

yes

[signal topaz]

Can you share the part number?

[tough matrix]

tle9201sg

rdson=200mOhm (high+low)

at 5A

[signal topaz]

Try to optimize heatsink, planes and thermal vias. In my opinion it is not always worth going from 1oz to 2oz.

[tough matrix]

the current design is here: motor-controller.rtfd.io

and it can sustain 4.5A with heatsink

I'd like to increase that to 5A

[signal topaz]

Ok. It looks nice. But I think you could improve by having copper planes on both sides of the ICs. And spam more vias. Of course I do not exactly the routing, so I am not sure how feasible that is 🙂

[tough matrix]

thanks!

[signal topaz]

And how does you actual metal heat sink looks like?

[signal topaz]

I am bored so I actually opened gerber files and I see that there is no GND fill on internal layers under the drivers? I suggest to add it. It will help as well

[tough matrix]

never thought about that... will do for next version

[signal topaz]

and you can also unmask metal under on heatsink areas, including bottom side of board. It helps only a little bit, but it is basically for free

[tough matrix]

as for heatsink, I used 14*14mm heatsink

[signal topaz]

Ok... Still it only sink heat from one side of the ICs.... But there is still some optimization of layout possible in my opinion. SO maybe it will be fine.

part of the heatsink goes over some signal lines. Is there some thermal compound/pad between heatsink and board?

Again - a lot of small improvements are possible. You can unmask copper under heatsinks. Signal lines should not be allowed there, it is easy to make a shortcircuit and also can be a EMC nightmare if traces couple to heatsink which is not e.g. grounded.

[tough matrix]

sorry, had to leave - only got back now.
There is indeed a pad under the heatsink
I was thinking: would it be better to place the IC on the bottom of the board, directly under the heatsink? what do you think? @signal topaz

or in the current configuration, put heat sink on the bottom, under the IC

[signal topaz]

No idea. Maybe. I think you still have a room for improvement on current approach. Maybe two smaller heat sinks on both sides of IC would be better. Or even something that goes over the package.

[signal topaz]

You need to consider how the board is placed or installed. Heat generally goes up.

[tough matrix]

yeah, that's why I didn't put the heatsink on the bottom
But I tested it with thermal camera (without heatsinks)

and found that the area on the bottom under IC is hotter than the heatsink area on top next to IC

meaning that thermal resistance from IC directly to opposite side of PCB is lower than to the side

which got me thinkign about placing the IC on the bottom and the heat sink, directly on top

[signal topaz]

That makes sense. Depending also on amount of vias. You can still improve the spread a bit with copper on inner layers.

Also, not sure if you consider like mass production with dual reflow, but this IC looks quite heavy.

[tough matrix]

Ok, thanks!
No, at teh moment I have no plans for mass production. I had JLCPCB assemble everything other than through-hole and the driver ICs; those I added manually

[signal topaz]

Have fun. 😉

[limpid nest]

do they make resistors that are meant to go into pin headers?

[unreal flax]

Not really "meant to", but the beefier resistors generally will have thicker leads on them, so they'll probably work okay.

[limpid nest]

mk, I've considered that. But when there just aren't parts to do what you want to do? Maybe it's not the best thing to be doing

[unreal flax]

I tend to agree. But people will try to bodge all sorts of stuff on this server, so I try not to judge, heh heh.

[limpid nest]

Yeah I can make what I want work.

[heavy jasper]

The lesser-jank variant can also be to solder an appropriately sized surface mount resistor across two adjacent pins of a male pin header, then plug that unit in.

Advantage being that the resistor doesn’t extend out past the footprint of the header, so it’s less likely to get knocked into and break the legs.

[limpid nest]

Hmmm

I'm trying to balance some semi-disparate needs.

[limpid nest]

For a resistor, is a lower value temperature coefficient more precise than a higher (e.g. +/- 20 ppm/degC vs 100) ?

[elder peak]

Other thing I do is put a pin recepticle that's about the right size for a resistor in the board itself.

[limpid nest]

I was thinking of doing that but that has downsides as well

[unreal flax]

Yes. Smaller values means it's more stable over temperature, since it changes resistance by a smaller percentage per degree C.

[limpid nest]

Kool thx

[steel valve]

If a trace goes through the pad of a resistor to get to another diode, will issues occur.
(I have a programming pin on a bootloader reset circuit, DTR, that needs to go to a resistor and a capacitor, can it just go through both of their pads?)

[steel valve]

Can through hole components ground holes be used to move ground traces between layers

Like having a through hole that a wire connects to getting ground from the ground plane, and then some traces come off of it to go onto a nearby smt component

[supple pollen]

Both of those are valid, but in high speed design (which may not apply to you), you have to be more careful about things like ground continuity

[limpid nest]

I'm confused by this, re: capacitors. Generally larger capacitance is worse at decoupling, no? Like 500 uF isn't going to have a particularly stunning high frequency response.

Additionally, if you're dealing with a DC motor, you are going to have reverse voltage spikes, which I would think is generally not good for large caps, which tend to be polarized items.

[tough matrix]

for an ideal capacitor, its impedance is inversely proportional to capacitance, so ideal large capacitor would be very good at bypassing high-frequency signal

the problem is, real-life high capacitance capacitors are very much not ideal

[limpid nest]

Yeah, but real large caps are not good at it ha

[tough matrix]

because they are typically polarized AL capacitors and thus have high ESR

[limpid nest]

Yeah

[tough matrix]

normally one puts large caps (like 220 uF) on power input lines

and, in addition, also put smaller caps with lower ESR

[limpid nest]

Yeah I was thinking something similar, but what about the reverse voltage spikes for a polarized cap?

[tough matrix]

normally between power in and the motor you have a motor driver (or FETs+snubber diodes)

[limpid nest]

this is decoupling a servo motor

[worldly schooner]

Those reverse voltage spikes are the inductive spikes mentioned at the bottom of that note, the diode is responsible for safely discharging those.

[limpid nest]

Wouldn't they still affect the cap?

since it would be in parallel?

[tough matrix]

I do not think it will be a problem.
And I never seen snubber diodes used with servos.
A servo contains a motor and a motor controller, which - I'd expect - has the FETs and snubber diodes already built in. But I do not know for certain

https://learn.sparkfun.com/tutorials/hobby-servo-tutorial/all

this is the internals of a servo

[limpid nest]

It's for the in219, the parts are like 1.25 total so it's not an issue including them.

[tough matrix]

what servos are you using? voltage/current?

[limpid nest]

Uncertain yet. Probably driving at 7.5V, the cables can't do more that like 920 mA so I'm specing the whole design around 1A motor current.

[tough matrix]

I'd put a 100uF, 16V or higher AL cap for bulk capacitance, plus - if you really want - a 10uF smd cap.

[limpid nest]

I don't think a low ESR low value cap is needed really

TFW you decide to just go with the free capacitor footprint. Those are mm.

Ah I see, it's the axial version. Crisis averted

[limpid nest]

@distant raven wondering if you experienced this in that qt-py footprint. I have a trace going to the pad, but it says it's not connected to the via and I can't actually connect it with a trace.

[unreal flax]

What happens if you try to route that airwire?

[limpid nest]

doesn't let me

[unreal flax]

Can you clarify? It beeps, the program crashes, it starts routing but the trace won't get any closer than 1mm away, what?

[limpid nest]

It just doesn't route

[unreal flax]

Restart the app?

[limpid nest]

I'll try that

Now before it was not making a distinction between my board and the background, but it was all black. Now it's all grey >.<

[steel valve]

Yeah, the MCU runs at 48/96MHZ, I went and did a bunch of stuff to remove vias on my pcb

So like three gnd pad traces on the pcb come from a ground through hole

That's intended for a wire for a sensor

Was able to remove like over ten vias on the pcb

By removing the redundant ground vias

Since from my understanding, less vias = better ground plane

[distant raven]

Click the pad with the wire. It’s likely your clearance is too large for it to reach normally

Typical distance from the edge is 40mil, you can change it to 6 mil and you should be able to attach the wire no problem

[limpid nest]

I'm using a 16 mil trace

Can I change that setting in DRC?

[limpid nest]

What do you mean by "click the pad with the wire"? Click the pad using the wire? I think I tried that.

[distant raven]

Yes

Let me see what it is exactly

[limpid nest]

I realized there's another issue with this footprint so I'm just making my own

It uses Dimension to draw lines, which messes up the rest of the board.

[distant raven]

Shouldnt be an issue with using dimensions because it should render as a cutout

@limpid nest this is how is renders for me

as for rounding to the pad, you should be able to change this to 10mil or smaller and have no issues rounding to the pad

copper/dimension that is

[limpid nest]

Thanks @distant raven I'll check these out

[supple pollen]

What you want is to avoid gaps in the ground plane. Vias are useful for low-inductance connections (many high speed designs will use "via stitching" with a bunch of vias to get the best connection).

[limpid nest]

How do you determine how much current a diode can handle passing through it?

[worldly schooner]

Continuous current and peak forward current should both be in its data sheet?

[limpid nest]

hmmm

must not be seeing it

[distant raven]

Link to data sheet?

[limpid nest]

one sec

https://www.smc-diodes.com/propdf/S1A-S1M N0560 REV.B.pdf I'm not convinced this is the diode for me anyways

[distant raven]

Ah yeah that’s a rectifying diode

Looks like 30A is the max current rectified

What’s the application?

[limpid nest]

just blocking current from flowing backwards

5V

but it needs to be able to pass >1A

Where did you see the max current?

[distant raven]

Ah okay

You want a Schottky diode

[limpid nest]

that's what I was thinking

[distant raven]

I found this part with the Digi-Key mobile app. https://www.digikey.com/en/products/detail/PMEG4010CEJ,115/1301023

Something like this would work

[limpid nest]

ha, I'm looking at the same line

different part

The only hard part is making the schottky symbol!

[distant raven]

Yeah lol

Have you tried ultra librarian

[limpid nest]

Yeah they didn't have this part

It's fine, I need the practice

[distant raven]

SnapEDA?

Lol

[limpid nest]

haha I'm OK with doing it myself

[fervent lance]

I checked the pins didn't find one. Would like an overview of how you move from breadboard to PCB with specific things that I can't fathom, like if you use resistors do you ship them do the company or you list them and hope they have it ? What about sensors and the microcontroller you send it to them too? How do you manage the fact with a breadboard you can only use DIP whereas you can use everything with a PCB so obviously there is a lot of circuits you can't go from breadboard to PCB for ? Do peoples make theorical circuits in some software and test all of it in that software ?

[spice turtle]

1A continuous or 1A for a few ms?

Because you can go by the 30A rating as long as your reverse signal doesnt exceed that time.

[limpid nest]

1A for a few ms

I found one that's comfortably in the right range and is cheap and is easy to solder so I'm calling it good

[distant raven]

FPGA board

[steel valve]

Yeah my ground plane has no gaps in it aside from vias

Aside from the edge through holes

Which is just normal

[supple pollen]

That should be fine then

[proper anvil]

some PCB fabrication/assembly places are integrated with a parts supplier (JLCPCB/LCSC) so you upload your design and BOM and they go next door to their warehouse and have all the parts. Some places support sourcing the parts themselves, or from you, but im not so familiar with those. Same for all components, really. With a breadboard you only use DIP, you can definitely use DIP in your circuit boards, but (probably most) many of your components come in various packages, sizes, pin configurations, etc. so you have many to choose from when designing a PCB. Through-hole, surface mount, SOP, DIP, etc. So when converting from breadboard to PCB you have to decide if you want to use DIP or something else.

So there's not really a circuit that you can build on a breadboard that you cant build on a PCB. For cost/convenience/size reasons you'll often want to switch from the big 1/8w throughhole resistors to tiny SMD resistors, big fat dip-18 microcontrollers to a much smaller surface mount package, etc, but if you want to use your components on hand you can certainly design for and use them.

also note there's a distinction between ordering bare PCBs that you assemble yourself, which depending on complexity may be cheaper/simpler, or you can have them manufacture and assemble the whole board which sounds like what you're expecting

[fervent lance]

yeah but if I can't physically test say a QFN like I can a DIP how do I actually know that it will work and that I have the correct resistors etc from it besides making a theorical circuit from the datasheet ?

[worldly schooner]

There are some simulators available for certain subsets of components, but ultimately, there isn't one strict process to move to PCB. Some people prototype with perfboard circuits, others can make do with breakout modules on breadboards. Regardless, most people will find themselves having to make a couple of modifications to a PCB before it works as expected, meaning more often than not, a PCB will have to go through at least one respin to cover the mistakes made in the first PCB.

[fervent lance]

it seems from my research one would have to use something like eagle/fusion 360 to make the pcb design at 80$/month then use solidworks to simulate it at 150$ish/month

[worldly schooner]

Free PCB design software exists for hobbyists. Perhaps look into KiCad or EasyEDA?

[limpid nest]

KiCad is like Pro grade

It's no joke

[fervent lance]

I didn't find anything free to simulate it if you don't have the gear to prototype the pcb with SMD chips etc

[worldly schooner]

Simulators are less common, but LTspice is free from Analog Devices. https://www.analog.com/en/design-center/design-tools-and-calculators/ltspice-simulator.html

[proper anvil]

In addition to Hem's input on breakout boards, keep in mind the goal is that the IC behaves the same no matter what package (other than thermal dissipation and the like) so if it works on breadboard, as long as you make the connection right, the hope is that it would also work on a PCB with an equivalent circuit

[limpid nest]

There can be wrinkles there, like the existence on a PCB of a suitable ground plane that isn't easy to have on a breadboard.

[tough matrix]

another simulator (limited but still useful) is wokwi: https://wokwi.com/

[worldly schooner]

https://www.falstad.com/circuit/ is also useful for basic circuits, but doesn't offer many IC options...

[limpid nest]

What's the next level up for wire-to-board solutions when you want to deliver power and go beyond screw terminals?

[distant raven]

Molex or JST connectors

[limpid nest]

ok thx, that's what I was thinking but I can't get a crimp so I've been looking for solder cup solutions

there aren't many

[distant raven]

You can order prefabricated cables with molex and jst connectors too which is nice

Adafruit sells some

[limpid nest]

Yeah I was considering doing that

[distant raven]

Depending on your current needs, you could start at a JST-PH and XH

Those are rated for 1A and 2A respectively

[limpid nest]

I need more than those can deliver unfortunately

[distant raven]

How much do you need?

5-10A?

[limpid nest]

I mean, in practice? Probably not near this, but yes, 5-10 A

with factor of safety

[distant raven]

Maybe these? I found this part with the Digi-Key mobile app. https://www.digikey.com/en/products/detail/0430450219/1635012

8A rating

[limpid nest]

Those are nice. I was looking at Molex pre-made, they have a lot of options.

[distant raven]

Those are just the board connector

[limpid nest]

yeah

you know this, but they sell a bunch of pre-crimped wires

Can't get custom colors but I'll have to live with that.

[distant raven]

But it’s like the standard for motherboard power which is cool to think how much power goes through those

[limpid nest]

yeah that's a ton of power!

[distant raven]

As for color, you could probably put like washy tape to label

[limpid nest]

Yeah and board-level labels

that plus keyed connectors is pretty solid

[distant raven]

Yeah, 100%

So many good options

[limpid nest]

I also own paint, I can paint the cables heheh

[distant raven]

Hehe

[limpid nest]

Oh that's RIGHT, I forgot. Molex don't provide any useful CAD formats in their "electronics" CAD gen section

It's all like Autocad which sure, fine, if you're an architect or electrician making PCBs, sure you need that

Not like the right tool for the job

[distant raven]

Yeah, the part generator is kind of disappointing

[limpid nest]

I can make my own, it's just a bit annoying

Must not be a lot of people buy tiny amounts of pre made cables heheh

[distant raven]

Probably not in the volume we expect

Probably mostly computer manufacturers

[tough matrix]

JST VH or XT30

you can buy cable assemblies/pigtails for either of them on ebay, aliexpress, amazon...

e.g. https://www.amazon.com/uxcell-3-96mm-Female-Connector-Header/dp/B07FM5J96B

(they do not use words "JST VH" for legal reason, but these are JST VH clones)

[limpid nest]

I have had bad experiences with XT30 but I'll keep it in mind, thanks

[tough matrix]

try jst vh then. I like it a lot

[limpid nest]

Anyone know a good place to find a JST-SH 4 pin board level connector for EAGLE/Fusion? Adafruit.lbr doesn't seem to have it.