help-with-hw-design

[distant raven]

That feeling when you’re trying to program a board that you put the programming points too close together where you can’t use pogo pins and the pads pull off the board

[fast tundra]

Squeeeeeeeeeeeeeeeeeeze

[carmine scarab]

Oooh! Whatcha makin'?

[fast tundra]

It's a Eurorack module

SAMD51 brain

[carmine scarab]

Cool!

[heavy jasper]

Are you going to be hand-soldering it, or reflow?

(if the latter, can I make my usual appeal for a type-C header? there are some nice vertical ones!)

It might also save you some room

[fast tundra]

USB B is still super common in audio applications and it's very mechanically sound.

& what do you mean? There's plenty of room! I didn't even have to put anything on the backside! 😂

& I can squeeze at least one more quad op amp on the front lol

[heavy jasper]

I guess where the failure mode is crunch instead of cycle count, I can see that (and again, consistency with the field tends to be more important than any other concern). I will say that there are some nice staked vertical parts these days that I've had good experiences with.

[fast tundra]

Yeah, I might check those out when I eventually make a module with USB host.

[misty escarp]

I've noticed that and wondered if it's the chonkiness that keeps it around

[heavy jasper]

For dedicated fields, I think there's also an inertia of not wanting to replace all the cabling

esp. in cases where the modules are meant to be fairly compatible, and folks may have already run and nicely managed wires

[misty escarp]

makes sense.

[heavy jasper]

@tough matrix A basic scientific article that goes into the core problem (fretting): http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.29.8048&rep=rep1&type=pdf Fig 6 shows the difference between tin-tin vs tin-gold. This study is more concerned with full connector cycles, but this fretting also occurs when connectors are used in high-vibration or high-thermal-cycle (which leads to small-scale scraping due to thermal expansion/contraction) environments.

In the case of what's being discussed here, it's probably not a significant issue, it was more a warning of "given you have a tin connector, there's not much point in buying the more expensive gold crimps" but it's a good phenomenon to know about. I'll leave you with the cheekily named 'Tin Commandments' from AMP (now part of TE Connectivity): https://www.ramoem.com/uploads/4/4/0/7/44075859/tin_commandments.pdf

[bright thistle]

Question 🖐️
Since connectors are plated and not full gold, wouldn't it mean the problem is just postponed to when the layer wears out?

[elder peak]

...and I guess this is something you should fret about?

Also, non musicians forget that in a band, sometimes there's a drummer. Beefy connectors for the win

[heavy jasper]

Correct, and you can get different plating thicknesses depending on how many cycles/how long of a life you need.

at the expense of, well... dollars

[bright thistle]

You suggested the title of the research, didn't you?

[heavy jasper]

I will say here that I'm just... ahem... scratching the surface of this topic and am by no means an expert. This is something that mechanical engineers and materials scientists at connector companies and failure analysis labs can spend a lifetime working on. But it's a good thing to keep in the back of your head and also gives some appreciation for e.g. USB-C being rated for 10,000 cycles over a variety of non-ideal mating conditions.

[elder peak]

Yah, without some serious study you are just... scraping by.

[bright thistle]

orz

[misty escarp]

problems like that can really put a crimp in your operations

[fast tundra]

This board is absurd

[misty escarp]

If jackson pollock were a robot

[fast tundra]

My brain hurt

[misty escarp]

I'm real impressed, how many layers?

[fast tundra]

Two.

[misty escarp]

hot dang

[fast tundra]

Last board was 4, but since the power distribution on this one is a lil easier I was able to do two.

My one and only four layer design

[misty escarp]

wowza

hey it says trans rights!

🏳️‍⚧️

[carmine scarab]

It also says no pinch... I wonder if that's something one should otherwise beware of with other boards?

[fast tundra]

😅 it's the esd sensitive symbol

[distant raven]

i have many 4 layer board designs

but whether they were necessary is another question 🙂

[elder peak]

Trans rights are inherent in synths. Wendy Carlos and all.

[fast tundra]

I may actually make this a four layer board when I send it to production

Since my CM doesn't charge any extra for it, might as well take advantage of it and have a contiguous ground plane.

[elder peak]

....be a well grounded individual in at least one way.

[misty escarp]

before I started meditating, ground planes were the only way I was grounded

[fast tundra]

I'm always floating

[carmine scarab]

I guess extra layers are nice to have in a pinch?

[distant raven]

Continuous ground plane you say?

It’s almost entirely continuous

[elder peak]

PCB routing is really meditative, which I guess is why we're always talking about our ground planes.

[fast tundra]

Hehe it's not meditative at all for me.

[distant raven]

Top side is slightly stressful

Also not remembering SWD test points on this design was my last minute face palm

Thankfully it’s TQFP which means I can at least solder so wire to it to program

[elder peak]

Yah, like, I am making things which amuse me and obsessing over them and it's generally relaxing for me.

I think if I was trying to do a hardwaremaking business I might think differently.

[distant raven]

i only find it frustrating when i spend time on a design, reworking it, and stuff.. only to find a tiny tiny bug 3 minutes after the board goes into production

[tough matrix]

@heavy jasper thanks a lot !

[bright thistle]

Revisioned the board so that one could use the TIMER pin for the Adafruit1375, if so inclined.
Also, made made the LED-enable pin available as well

just realised I should put a warning near the LED connector as well, since there's a control jumper as well

[misty escarp]

looks great!

[fast tundra]

Unless you need all that space, you could definitely make it smaller and save a few bucks when ordering from OSHPark

[bright thistle]

Ok, that should do it 🙂
now every pin is mapped on the board, allowing for off-the-shelf cables if needed 🙂
If I don't manage to fit the pinouts (maybe on the back?), I'll write to refer to the project homepage

[misty escarp]

is it being mounted on anything?

[bright thistle]

I won't be using OSHPark (EU maker), but AISLER instead (lower price, but no purple mask 😦 )

[bright thistle]

supposed to be fitted inside an HiFi lookalike casing... D'oh! Forgot the mounting holes!

finished size is 109.22mm x 48.26mm

[misty escarp]

good thing you're not using OSH! That would be expensive!

[bright thistle]

OSH quotes for USD40.9, AISLER €27.94

it does seem to scale better with size... does OSH round to nearest full inch?

[misty escarp]

I'm not sure

[elder peak]

You could always osh them

[bright thistle]

btw, it's either proto-fab, or CNC routing

[bright thistle]

yes and no.
after €22, I'd also have to pay VAT 22%
after €150 I'd pay both VAT 22% and duties

[fast tundra]

I think osh rounds to the nearest square inch

[fervent lance]

Have you quoted with PCBway? JLC PCB?

[bright thistle]

should I use grounded mounting holes?

[fast tundra]

Oof that reminds me I'm waiting on customs to tell me how much my import duty is gonna be on my last batch of PCBs 😬

[fast tundra]

It depends. What are you mounting it to?

[bright thistle]

see why I'd rather use an EU fab? :3
also, .ch ones have minimum 5pcs order

[bright thistle]

hifi chassis. may be wood or metal, may be using plastic or metal standoffs

[fast tundra]

It probably won't hurt, then

[bright thistle]

ok, padded hole created and connected to ground plane

since they're mainly to keep the board in place, i guess two of them are more than enough

[fast tundra]

IME if you can do at least four, do at least four.

[bright thistle]

I don't think I could do any better without either sacrifying back silkscreen space, or redesigning all the nets 🙂

[fast tundra]

That's good enough for rock n roll

[bright thistle]

back silkscreen done

that poor lonely VDD trace XD

[heavy jasper]

You might do connector pinouts in 4-1 order, or flip all of your silk 180

(it'd be nice to get the text also in the order the connector is in)

[fast tundra]

Honestly it's perfect

[tough matrix]

when you write "cut the jumper on the sensor...", do you mean "cut the trace"?

[fast tundra]

Context?

[bright thistle]

yes, adafruit product page describes it as a jumper. part 1375 and 1374

it's a jumper pad

[tough matrix]

@fast tundra i was referring to silkscreen text on ozone's board

[fast tundra]

ah I see

[bright thistle]

you mean pins described left-to-right, top-to-bottom?

if so, sure can do 🙂
I thought pointing pin1 would be enough 🙂

[heavy jasper]

In theory yes, and in a more cramped layout it would be more than enough, but since you've got the space, you may as well make everything line up as it would be so it's an even quicker reference if you're going through with probes

(but this is the most nit of nits)

[bright thistle]

would it help if I put a "bracket" on the back where the connectors are placed?

[heavy jasper]

How do you mean? Are you looking at improving mechanical robustness, or something else?

[heavy jasper]

I think having a matching designator there would be helpful, but I don't see much use for the half-rectangle.

[distant raven]

modifying the original FTDI board i made for this breadboarduino project i started but never did anything with

it's going to be a FTDI bitbang board.

i don't really want to pay for a junk one on Amazon, and the Adafruit one is a bit too much for me right now. so this will do. I've got the parts for it. just need the board. 🙂

[fast tundra]

Neat

[distant raven]

basically trying to bitbang jtag over rs232 which i'm pretty sure what the Lattice Semi FTDI cables do anyway

if this works, then i will simply just rework my Lattice Wing featherwing to have a ft232r on it with a usb port and that'll be ready for the shop 🙂

interesting enough, the SPI Driver FTDI chip isn't supported by Diamond Programmer so that's a bummer.

[distant raven]

This might be one of the best boards I’ve done in terms of layout

[fast tundra]

very smol, very tidy

[carmine scarab]

That's one thing I noticed about that board @distant raven , the top (what I consider being "top", where the USB connector is) is really tidily laid out and balanced from a visual perspective.

[silk lark]

I'm triggered by tx and rx leds not being inline with the pwr and user leds

[distant raven]

@silk lark i contemplated a version with a neopixel and stemma qt connector that got rid of the rx/tx leds

[silk lark]

and also the corresponding resistors on the other side

[distant raven]

lol

[silk lark]

and those two caps are not symmetrical

and the diode

or is that a fuse?

but I have severe OCD, so ignore me

[distant raven]

xD

hahaha no fuse just a schottky diode

[jagged relic]

No idea what I'm doing 'discord' - had to layout a PCB

[unreal flax]

So far it looks fairly reasonable, though you might want to double-check the overall cable-plug footprint of the TC2030 to make sure it won't hit the JTAG header in making contact with the board.

[fast tundra]

The tc2030 has more than enough room in that layout

We put ours in the strangest places lol

[elder peak]

The TC2030 tattoo doesn't count, tho.

[fast tundra]

Haha. I've been meaning to get an electronics tattoo but I haven't decided on one yet.

[bright thistle]

Go Classic, get a 555

[ember laurel]

how terrible would it be to NOT put decoupling caps right behind a BGA MCU of 480MHz, and rather route them out to the side of the chip?

I'd prefer not to do double sided SMD, my whole design is one-sided so far, and I'm only doing this change due to chip shortage of LQFP chips

[unreal flax]

I'm not a high-speed expert, but that wouldn't immediately worry me too much.

[ember laurel]

vias of course add some impedance as well

and there is the board thickness of 1.6mm

this chip is 10x10mm, and traces would most likely be 6mm or so, for the decoupling caps

and 0.2mm wide

I haven't soldered any UFBGA-176 manually myself - is this feasible just with a hot air station and a microscope?

(for prototyping I'd like soldering myself)

[unreal flax]

Yes, likely feasible, though spring for ENIG on your board.

[ember laurel]

Yeah I always do ENIG. I think I’ll give this a go today. Will route some vias and pads down in bottom layer as well, and compare noise with caps placed there.

[minor axle]

Hey does anyone know a good way to import or convert kicad projects into eagle?

[distant raven]

I’ve never tried LOL

[minor axle]

Yeah, I've seen a lot about converting from eagle to kicad but next to nothing about the other way around

[heavy jasper]

@ember laurel Depending on what power planes you have, you might not even bother running specific traces out for power and ground, just drop with a via directly to the plane and then vias again at the cap.

[signal topaz]

capacitors placed on the same side as the BGA will actually work better, but only if stackup is designed properly

[ember laurel]

I managed to score the LQFP chips I needed, so fortunately no BGA redesign is required

[ember laurel]

it all comes down to the impedance of your path to the caps, no?

[distant raven]

This is usually more true than anything but multilayer boards can actually make the via path higher impedance which can be bad

[ember laurel]

yeah I know that

in any case, I ordered a new proto board with LQFP again, and I just placed an order for 525 MCU's. woop!

Farnell today suddenly got a delivery of my chips (STM32H750VBT6) so I placed a quick order to cover my production needs for the coming months

I have 59 unique components, and about 280 total components on the board - but only this MCU has proved troublesome to find in stock.

[distant raven]

Nice nice!

[tough matrix]

@ember laurel you could hoard some chips and then resell at a premium!
(just kidding...)

[ember laurel]

considering some quotes I got from China, I think some companies are doing just that.

[heavy jasper]

You've now entered.... the broker zone.

[ember laurel]

scalpers

[distant raven]

I’m definitely getting tired of scalpers

[rain remnant]

intel seems to get things to states and europe for they chips so it might change some of this

[distant raven]

A bunch of sellers are cracking down on scalpers and bot buyers

[ember laurel]

does anyone have any really good explanations as to why there are so many chip shortages at the moment?

[elder peak]

Fires.

[silk lark]

just-in-time manufacturing and really narrow margins, to optimize costs

[elder peak]

Especially at this point, a lot of folks have moved from just-in-time to dragon's-hard.

[silk lark]

the fires wouldn't matter as much if there was reasonable stock and multiple manufacturers

[elder peak]

yah

[silk lark]

but both the equipment and the masters are very expensive, so they tend to only have one

I wonder if the masters were destroyed in those fire — that could possibly mean an end of a line

[tough matrix]

I can only repeat what I read in the news - some say that the chip foundries reduced production anticipating low demand for cars caused by pandemic. Well, the demand proved to be stronger than anticipated, so now they need to bring production back up, and can't do it fast enough.
also there is severe draught in Taiwan: https://www.bloomberg.com/news/articles/2021-03-24/taiwan-raises-red-alert-over-water-cuts-supply-for-chipmakers

[distant raven]

TSMC is at full capacity

They don’t even have fall back capacity either

The demand for consumer electronics too like cell phones, smart TVs, laptops, gaming consoles, have all soaked up fab demand

That and shipping delays have also hindered key areas of the supply chain for making chips

LiPo charging devices have increased in demand to

Just a month ago there was probably 3-5K pieces of the LiPo charger ic I use and it’s back ordered till April

[rain remnant]

i put to offtopic one thing too

[distant raven]

@rain remnant your comment is still on topic for this channel. Talking about sourcing issues is very relevant to PCB design

[rain remnant]

yeah then the contaienr ship stuck in suez canal not helping the parts

[rain remnant]

as containers from asia to europe in there stuck so china cant even use them to US so some are stuck, so we have to wait and see when parts get back to normal again what comes to price

[frigid seal]

Slightly concerning note from digikey.

Normally I'm the one who forgets which part is which.

[distant raven]

😂

[fast tundra]

Everything was fine until the auto manufacturers attacked

[elder peak]

I mean, I'm an avid cyclist. Car hate is easy for me.

[distant raven]

100 days past and my brother and I found a new parts supplier. They maybe overwhelmed, but we think they can help us make our deadline

Avatar the Last Airbender references for the win

[fervent lance]

I cycle a lot - is a challenge not to get smooshed.

[onyx copper]

Does anyone willing or know anyone willing to create a custom pcb design for the feather 32u4 bluefruit similar to this?

also does anyone know approximately how much a design like this will cost?

[silk lark]

why 32u4

also, looks like it should already work with any feathers?

[onyx copper]

I'm not sure, its the one that most of these designs seem to use

the main reason was because it is small enough to fit in a keyboard case

[silk lark]

all feathers are almost the same size

[onyx copper]

oh i just like that one because of the way you can use a battery with it easily

i don't really know this very well

[silk lark]

all feathers have that battery socket and charging

what code did you plan to run on it? did you want to write it yourself?

[onyx copper]

yeah i was planning to write it based on code other people have made for similar projects

[silk lark]

do you have an example at hand?

[onyx copper]

i have a link to a project someone made https://medium.com/@friggeri/the-new-blanck-keyboard-c7563c4f9fa

[silk lark]

I had designed the circuit planning to use the Feather 32u4 Bluefruit LE as the micro-controller, but between then and assembly time Adafruit released the Feather nRF52 Bluefruit LE which has a compatible pinout and has only one chip on board instead of having one dedicated to computation and the other one to Bluetooth LE, so I swapped that at the last minute and rewrote my firmware to support the new API.

[onyx copper]

i also wanted to use the 32u4 because it is supported with qmk firmware, im not sure about the other feathers

[silk lark]

does the qmk firmware support the blufruit module though?

[onyx copper]

yes

https://beta.docs.qmk.fm/using-qmk/hardware-features/feature_bluetooth

their website says this specific board is supported

[silk lark]

cool

you know, I would still try to wire it on the breadboard and test if it works before you start designing a PCB

without are the buttons, of course

just one, to test

[onyx copper]

oh, ok im not really experienced with this

[silk lark]

that's even more reason to try it first with a breadboard

[onyx copper]

yeah

[silk lark]

I made a bunch of keyboards using the samd21 and circuitpython, and there is the KMK keyboard firmware for circuitpython for the NRF52 chip

[onyx copper]

cool, did you hand wire them?

[silk lark]

no, I used low-profile switches and pcb: https://hackaday.io/project/174095-dorsch-40k-keyboard https://hackaday.io/project/174917-dorsch-48k-keyboard https://hackaday.io/project/175043-flatreus-keyboard https://hackaday.io/project/175712-kamina-keyboard

by the way, if you are going ortholinear, I really recommend making a split keyboard, otherwise your wrists will hurt after longer use

[onyx copper]

do you know anything like that for a tkl layout?

[silk lark]

like this? https://hackaday.io/project/167912-flounder-keyboard

[onyx copper]

i was thinking of a layout that is basically a full sized keybaord with the numpad chopped off

[silk lark]

right, that's getting expensive fast, with all those keys

[distant raven]

That’s why custom keyboards are pricy 🙂

[silk lark]

unless they are small

[distant raven]

Yeah

Even then still pricy

[silk lark]

I really like the Alice layouts

I mean, if I still used staggered layouts

[distant raven]

The staggered split layout you did was pretty sweet

[silk lark]

it wasn't staggered

it was ortho alice

[distant raven]

I wasn’t sure the right term

[silk lark]

if you mean https://hackaday.io/project/174982-turbot-keyboard

[distant raven]

The keys were offset/staggered looking

No the smaller one

[distant raven]

This one

[silk lark]

the flatreus?

but they are staggered vertically

a lot of people like the Atreus layout

[distant raven]

Keyboards are just a very interesting projects and one day I’m going to make one

For now... I need to do the impossible thing of clearing out project backlog 😅

[silk lark]

I recommend leaving a room for pluggable modules, so that you can keep experimenting without having to rebuild the whole keyboard

[distant raven]

Thinking about designing a RGB light kit to sell on my shop

Then maybe I’ll grab some of those reverse mounted sk6812

[silk lark]

@onyx copper have you seen https://makerdiary.com/products/m60-mechanical-keyboard-pcba ?

but that one is also a 60% not a tkl

[onyx copper]

yeah i really wanted to keep arrow keys

[silk lark]

I'm worried that a feather will not have enough pins for so many keys

especially when some of the pins are needed for the ble module

[onyx copper]

yeah

i am looking for another controller that can do 25 pins

for the keyboard

is there a way to make a wireless keyboard using a controller like the teensy 2.0?

or another version of the teensy controllers

[silk lark]

you will probably want something like the itsybitsy nrf52

https://www.adafruit.com/product/4481

but that's only 21 gpio

you could use a shift register or some other gpio expander

but of course then you would need to write the code for it yourself

[onyx copper]

if i only use 21 gpio is there code already available for it?

[silk lark]

well, kmk will work on it

you will only need to write the layout definition

actually the feather nrf52840 express also has 21 gpio pins, plus that battery charging circuit

so either will work

[silk lark]

pogo pins?

[distant raven]

Yeah

[frigid seal]

Me today ^

[distant raven]

Basically 🙂

Not perfect but it’ll probably program a few hundred boards 🙂

Need to make the clamp taller though

[bright thistle]

Icymi, system76 makes a OSHW keyboard

[carmine scarab]

Nice! There quite a few OSHW makers out there, including the folks who make the keyboard Scott uses - KeyboardIO

[distant raven]

Yeah

How long do you think this LED will last 🙂

[carmine scarab]

Seems fairly low power, not too bright or hot, so... at least until I send this response

[distant raven]

😅

[carmine scarab]

Probably a good long while I would guess

[distant raven]

I’m guessing these Innolux LEDs need a smidge more current than the 10k resistor is providing

But this would be the ideal low power ADC 🙂

[carmine scarab]

(asking for technical clarity, not pedanticity) - is it that the resistor is providing, or that it's allowing through?

[silk lark]

@carmine scarab providing of course. If you remove the resistor, no current flows.

[distant raven]

Current limiting resistor

[distant raven]

Not true

It will pull infinite current

[silk lark]

not true, air conducts too

[distant raven]

Which is why is dies 🙂

[silk lark]

and not infinite if you just disconnect and remove the resistor, without any connection

[distant raven]

Electrical theory would say LEDs will pull an unlimited amount of current because it’s a current driven device. Hence why we use “current limiting resistors” if you removed the resistor it would be an open circuit so 0 amps yes

[carmine scarab]

That would be a safe LED since it would be air-gapped 😄

[distant raven]

But if it was straight power to the LED, you deal with runaway current until the LED becomes a quick blow fuse

[silk lark]

who said straight power

[carmine scarab]

I think that @silk lark means that, given the EXACT layout you have not, but unsolder/remove the resistor, and leaving the "nothing" in its place

[silk lark]

besides, your power source has internal resistance too

[distant raven]

It’s true, LDO limits to 500mA or about 100 times more than this LED is tested for 🙂

[carmine scarab]

The difference in inferance between "nothing" = "not a single thing" and "nothing" = "no component" can be tricky!

[bright thistle]

Me, but daisy chain up to rev 25 in 30'

[ember laurel]

using PCB traces as fuses is always a good idea.

[bright thistle]

Heyooo!

[silk lark]

well, it would work

and it's cheap!

[bright thistle]

And, you'd get to learn new way for swearing in Chinese from the fab!

[silk lark]

as long as it's within the specs...

[bright thistle]

Plot twist: they'd rather place a fuse holder and give you a box of fuses on the house 🤣

I was wondering... Except for cost, is there any reason for not using 1 way dip switches, especially for hobby components, instead of jumper traces?

I think it might improve reusability, since not everybody might be skilled enough to solder/desolder the pads. Plus, they might wear/detach the more you change the state of the jumper

[tough matrix]

@bright thistle cost is the main one
also, there might not be enough room for it if your board is densely populated
jumper traces can be placed on the bottom

[elder peak]

I mean, you can always just use a piece of header and a jumper that way.

DIP switches are only rated for mA levels of current, you can do better with a header and jumper

[distant raven]

sliming the RP Sapling down

bottom side

ditched USB-C because i like my big honkin' buttons and it's just a mess trying to do everythin g+ USB-c

[fervent lance]

big honkin' buttons are good!

[distant raven]

they are buttery smooth

[fervent lance]

That pico BOOTSEL button is pretty nice.

[distant raven]

slightly pricier but the action is worth the price i think

[fervent lance]

My boss told me early desktop calculators used reed switches (and I suppose permanent magnets that move as you press buttons)

He redid a core memory by hand at least once. ;)

[distant raven]

hehe

[carmine scarab]

We all remember that famous song by Solder Wicks-a-lot "I like big buttons and I cannot lie"

[distant raven]

Haha 🙂

[distant raven]

Okay ordered the RP sapling revision

Can’t wait 🥳

[spark sun]

any recommendations for easily generating pretty-looking 2D images from gerber files for eg. documentation? maybe just something as simple as setting good color schemes in gerbview or something. the ones you posted above @distant raven look quite nice, by default most gerber viewers seem to show a bunch of garish colors and jaggy aliasing

[distant raven]

I use the manufacturing view that Eagle provides

[spark sun]

ah gotcha 👍 i'm using KiCad. Maybe i'll work on a little workflow/script for exporting SVGs from Kicad gerbview/pcbnew and prettifying them a bit

[spark sun]

Honestly the JLCPCB web-based Gerber viewer is the closest I’ve seen to what I want, but it’s v. aliased/jaggy. Wonder if there’s an easy way to upscale it to antialias.

Here’s the new version of my PCB badge I just ordered in their viewer 🙂

[distant raven]

That’s pretty sweet looking

[spark sun]

Thanks 😊 now just have to endure 7-20 days of waiting 😅

[carmine scarab]

That looks really nice @spark sun. Hope you'll post pics when the 7-20 days come to pass!

[heavy jasper]

@spark sun Maybe take a look at Altium's web viewer: https://www.altium.com/viewer/

[spark sun]

That does look nice, will give it a try, thanks!

[spark sun]

tried the Altium viewer, seems like it could be good for documentation, but like most others, it's more engineering-focused and makes it hard/impossible to use "pretty colors" for rendering.
Discovered that the analogous tool on PCBWay's site is actually quite good and lets you export SVGs of all layers + combined top/bottom views! https://www.pcbway.com/project/OnlineGerberViewer.html

[heavy jasper]

Ah, yes, I'm used to the actual Altium underlying tool which lets you configure the view type (colors/transparency/etc)

[vapid grove]

@distant raven are ya a mechanical keeb fan?

Like, the feel of buttons is important for you

[distant raven]

It’s a texture and response thing .

[distant raven]

slowly working on the new standalone Icy Tree Feather

[south gazelle]

Why not simply go to the 3D view of KiCad, disable orthographic projection, and export in PNG. Otherwise, if you only want the layers, you can also select plot in Kicad and then, instead of selecting a gerber file as output, you can export each layer as an SVG and then recombine them in a tool like Inkscape, change colors, etc.

[spark sun]

Ooh I’ll try that too (2D-ified 3D view in KiCAD) and compare, thanks. Yeah I know how to export the SVG layers, was hoping to have an fairly quick/repeatable process without messing with layer colors etc every time. But yeah exporting layers + writing a little script for compositing/recoloring would probably be the most customizable option

[radiant wolf]

@distant raven what microcontroller will it have? And what is the second chip on it?

[distant raven]

It won’t have a microcontroller, yes a FT232H

And a lattice ICE5LP4K

[radiant wolf]

An fpga on a feather sounds really interesting, will you be selling them on your store?

[distant raven]

Yeah, hopefully by May

My speed of development would be faster but it takes money so I’m staggering releases so I can afford to make things with enough scale to give good pricing

[radiant wolf]

Do you have an estimate on the price it will be?

[distant raven]

Well.. probably $19.95

The board provides a great base for doing more FPGAs from Lattice

[elder peak]

I'm betting it'll be a gate new toy for a lot of people.

[distant raven]

I’m just hoping people latch on to it

[elder peak]

Yeah, so far FPGA boards seem to suffer from a lot of gatekeeping.

[distant raven]

And a lot of people flip flopping on their usefulness

[spark sun]

i'm just not sure if I need a microcontroller AND a FPGA **OR ** if I just need a microcontroller AND NOT an FPGA 😛

[radiant wolf]

$20 sounds like a great price, there is another fpga in a feather formfactor that someone made but its over $100 which isn't worth it for me. For $20 I will get one once you release them provided I can get my hand on one in Australia. I have been wanting to get into fpgas for a while now and I really like the feather formfactor so this seems like a great product.

[unreal flax]

It's more expensive, but the QuickFeather is also worth looking at. That's using the EOS S3 chip from QuickLogic, which is a combination Cortex M4 and FPGA.

[radiant wolf]

The m4 core does seem nice, looks like it has micropython support as well which is good

[distant raven]

I’m working on some MCU+ Feathers as well. Starting with a M0 and Lattice ICE5LP4K

[carmine scarab]

So Youtube just popped this up for me...
https://www.youtube.com/watch?v=J4q59KwA_Vk

[distant raven]

Slightly pricier version of OshPark but red solder mask

Kind of cool

[bright thistle]

Now we know who sources sparkfun

[distant raven]

Lol

Royal Circuits

[tough matrix]

Royal color is purple!

[distant raven]

Indeed 🙂

Thankfully Royal also does a lot of other colors

It is interesting though how Purple was the color of royalty

Mostly because purple dye was from indigo which was a rare commodity

[fervent lance]

It's because it says 'purple mountains majesty' I think. /rimshot

[tough matrix]

IIRC, it is not indigo - indigo is a plant-based dye, and purple was made from some mollusk found in Mediterranean

[distant raven]

Huh interesting

[tough matrix]

https://en.wikipedia.org/wiki/Tyrian_purple

[distant raven]

A snail of all things

[distant raven]

You think people shelled out a lot of money for those dye?

[elder peak]

HEY that's my job.

A long long time ago in the mists of history, SparkFun launched one of the first PCB batchers.

They decided to focus solely on the bits industry and sold their batch PCB service to OSH Park.

[distant raven]

And then they started A la carte

[carmine scarab]

Did they get board with that side of business? 😄

[distant raven]

not even done yet lol

I still need to add the eeprom for the ft232h

bottom is still pretty clean though

[carmine scarab]

I've spotted a most EGREGIOUS error. That is, unless you'd rather still be living in 2020? 😛

[distant raven]

didn't you know? it's still March 2020

still 2 weeks to flatten the curve 😛

[carmine scarab]

March 382nd?

lol

[distant raven]

Just like the east sale on my shop runs till April 4th 2020

😉

only then will covid and march 2020 be over

[carmine scarab]

*eyes calendar*
Gonna need a longer calendar!

[stable pollen]

Hello
I am a software developer.
If i make a project on one of the development board ,
And i would like to make it be mass manufactured.
With little electronic experiment can I design the pcb for it to be producted

Or is there an automated solution for pcb preparing from development board

[south gazelle]

You have created a design with a dev board, some wires and peripherals, and you want to transform it into a PCB, to produce hundred units?

[stable pollen]

Yes i am planning to create design

Or use a ready board with a cam and speaker module

For example a raspberry pi with a cam and speaker

But any development board is ok for me provider that it can convertible to mass production solution easyly

[south gazelle]

Using "classic" Raspberry Pis for mass production is not very realistic, because it's hard to get them in volume. You can however use a "compute module" (see https://www.raspberrypi.org/products/compute-module-4/?variant=raspberry-pi-cm4001000).
But in any case, that's just part of the problem. If you want to build a product at scale from a prototype, you will either need to do it yourself, or pay someone to do it for you. To do it yourself, you could learn a tool like KiCad or Eagle, and then design a PCB, get a few samples made, and test your design. Iterate until it's good. Once you're happy with your design, you can get it made in bigger volumes in China (e.g. PCBWay is popular with makers). Paying somebody is quicker of course. Personally, I would advise you to try at least doing it yourself first, because you will learn a lot, and you will be in a better position to understand potential problems in the future even if you ultimately outsource things. Moreover, PCB design is fun!

[silk lark]

sparkfun has a service for this -- where you build something with their components, and they turn it into a product, but it's rather expensive

https://alc.sparkfun.com/

[south gazelle]

Yes, but the choices are limited. No raspberry pi for example.

[silk lark]

@south gazelle the choices will always be limited, the important thing is that they cover the use cases

[stable pollen]

Thank you Deshipu
I will consider it as a last resort it is little expensive if i need to make some changes few times after first design i will have to pay too much

[stable pollen]

Thank you Alain,
I think i will pay somebody for it and trying first myself is a good idea.
By the way, for my purpose, which development board should i chose?

[silk lark]

it's hard to say without knowing your project

[south gazelle]

Well I don't know your project well enough to give you an informed opinion. I just understand that you will use a camera and a speaker. That could cover a lot of things, including machine learning or image recognition. If your background is software engineering, designing a first prototype with a Raspberry Pi is probably not a bad start, because you can use python and scripting in a Linux environment. Otherwise, you could look micro-controller solutions like OpenMV, or the ESP32-CAM projects. I'm sure other people might have even better suggestions.

[stable pollen]

I am thinking to build a device which captures the text from book and later read it
I will use tensorflow lite as a software

[tough matrix]

Rpi could probably work. But instead of writing your own tensorflow model, why not use one of several existing open-source OCR implementations?

[stable pollen]

I am new in this field
trying to learn tensorflow
As you suggest ocr
Then i use pre open source ocr
Thank you very much

[pearl tapir]

"bottom is still pretty clean though" Room for more parts.

[distant raven]

Make the top as packed as possible for highest level of assembly difficulty 😉

[pearl tapir]

Almost any contract manufacturer would rather have a very packed top side rather than double-sided assembly.
However, you could add some optional user installed parts to the bottom similar to the QT Py. I was thinking additional I/O connectors. I2C or SPI with lots of CS.

[distant raven]

Yeah, I might add a Stemma connector on the bottom since this lattice FPGA has two hardened I2C cores

[twilit mango]

I read that as "abandoned I2C cores" on the first pass and was a bit confused.

[distant raven]

Haha 🙂

[carmine scarab]

I really hate when my eyes and my brain read different things!

[twilit mango]

At least sometimes it's entertaining.

[carmine scarab]

Very true

[bright thistle]

Please do not abandon your I2C cores when you leave for vacation

-- People for the Ethical Treatment of Electronics

[distant raven]

Smöl boards

Hopefully it works as intended lol

[distant raven]

Smoke test check

Detected by diamond programmer check

[distant raven]

Dang might not work

[tough matrix]

you found what was wrong with it?

[distant raven]

Well, it won’t work with Diamond Programmer because it won’t bit bang jtag. OpenOCD can bit bang jtag using the FT232R

[silk lark]

another try at a flat pewpew

[distant raven]

Oh cool

[jagged relic]

Done and tested https://oshpark.com/shared_projects/Arco2mZq

[bright thistle]

So, my PCB requires 3v3 in order to make the outputs Pi-compatible.

Should I trust the user, and let it provide correct power, or should I slap a uUSB socket and a voltage regulator?

[distant raven]

I always suggest accounting for the most novice users, as reasonably as possible

[bright thistle]

Regulator it is then, thank you :)

It might make it incompatible with SBCs using 5v, but better safe than sorry :)

[distant raven]

You can use Level Shifters if you need both 3.3V and 5V logic

[bright thistle]

Or use a jumper for setting voltage between 5v and 3v3

Default: no jumper, pay attention 😈

Btw, decided to drop the 1375 (toggle caps), and only use 1374 (momentary caps), much less of a hassle to route and versatile to use.
Also, shaping it so that, with proper headers, one could choose to go either board-to-wire, or board-to-board

[still sphinx]

i'm making a footprint for a part with wire leads (the regular pins are out of stock). since the wires go out a bit, should i just eyeball the size?

[distant raven]

I would match the footprint given that you may want to use it with headers down the road

[still sphinx]

ok

[still sphinx]

oh also in KiCAD the DPDT switch is broken up into two symbols, which i somehow need to assign as one footprint

[spark sun]

if you place both symbols & then open the "Symbol Fields" table view, you should just see one row in there for both & you can assign a single footprint

[still sphinx]

oh thanks

[bright thistle]

how would one implement power from µusb correctly? 🙂

[tough matrix]

@bright thistle add an LDO

and, of course, stabilizing caps

what current do you need to provide?

[bright thistle]

lemme do some math, but not much, it's 5 capacitive buttons and their LEDs 🙂

[bright thistle]

weighing in the worst values from adafruit's and sparkfun's versions of the sensors, i'd say ~210mA

[tough matrix]

Then I'd take this LDO: https://www.digikey.com/en/products/detail/diodes-incorporated/AP2112K-3-3TRG1/4470746

[bright thistle]

this for converting 5→3.3 right?

[tough matrix]

yes

[bright thistle]

ok, and I just tap from 5V/GND from usb? no mode resistors?

[tough matrix]

no resistors
but add capacitors
check the "typical application circuit" in datasheet

[bright thistle]

thank you ❤️

nice, it's know to kicad 😄

[tough matrix]

I'd expect so - this is one of the most common LDOs.
Adafruit uses it on all their boards

[bright thistle]

I'm learning as I go 🙂

[tough matrix]

another common choice is AMS1117

but if you just search digikey, it gives you dozens of LDOs with 3.3V fixed output voltage. Honestly, I am not sure how one chooses among them (other than checking a couple of basic parameters such as current limit and voltage drop), so I usually copy what others are using - in this case, AP2112 and AMS1117 🙂

[bright thistle]

I take I just tie Ven to Vin :3

[tough matrix]

yes
if you ever need to turn off the regulator - e.g. to conserve battery - you just bring En to GND. For normal operation, tie En to Vin

[bright thistle]

no need in my case, but good to know 🙂

[tough matrix]

what the ...
AP2112K-3.3 is out of stock everywhere!
Mouser says "201,000 Expected 11/8/2021"
this is crazy
I knew microcontrollers such as STM32 were hard to get, but LDOs? this is not rocket science, they do not require high-tech chip foundry

[bright thistle]

they do require smart sand, tho, and I guess it's being all hijacked to {C|G}PU foundries

[tough matrix]

ok, maybe you do need to check AZ1117
or AMS1117, basically same thing
hope it is in stock

[bright thistle]

SOIC8 and SOT895 are available, tho

they're short only of SOT255

[tough matrix]

oh, ok.
SOIC8 would work, obviously
Or https://www.digikey.com/en/products/detail/diodes-incorporated/AZ1117IH-3-3TRG1/5699672

[bright thistle]

what's wrong with sot89-5? 😂

[tough matrix]

nothing 🙂

[bright thistle]

I'll go with that, less pins to mistake 😄

[ember laurel]

AZ1117 is good

[bright thistle]

do I connect usb shield to ground, or should I just "nah" ?

[ember laurel]

depends on what rejection frequency you want

[bright thistle]

It just carries signal from capacitive buttons, no audio or RF involved in the board 🙂

[ember laurel]

I meant the regulator 🙂

[bright thistle]

oh 🙂

[ember laurel]

probably you are regulating something, like a buck switched supply or so

[bright thistle]

just dropping from 5v to 3.3v, so not to fry Pi's GPIO in 🙂

[ember laurel]

hmm what?

regulator for dropping a logic voltage?

[bright thistle]

eh... no 😄
the board just aggregates a bunch of those https://www.adafruit.com/product/1374
they have the "habit" to take any Vin 1.8-5V, and copy it on Vout upon touch

I want to use them as buttons for a Pi, but I can't feed its GPIO 5v

[ember laurel]

ok, so you want to power those with 3.3V

[bright thistle]

exactly, by regulating µUSB, so to minimize user error

[ember laurel]

yes ok, I see

[bright thistle]

I might just slap a "3.3v" silk near a 2pin connector, but

1. yet another cable assembly to take into account
2. people don't read 😂

btw, so not to exclude future/alternative usage, I'll be putting a jumper for choosing 5/3.3v as well, in case somebody wants to use it with an arduino instead 🙂

uh? does the az1117 output from a pad and the tab at once?

yep, it is 🙂

[ember laurel]

middle pin goes through to pad, yes

[bright thistle]

uhm... I see mixed opinions about connecting usb shield to gnd... but I'm not carrying any signal, so I guess it's ok to leave it ungrounded?

ok, looking at other breakout boards, there's no apparent connection

aaand, sparkfun and adafruit have different opinions as well "-_-

[distant raven]

If all else fails for regulators, the TLV75533 (might have it wrong) is a great choice

3.3V 500mA

It’s my go to

[bright thistle]

uhmmm... typical application mentions ceramic caps, but shows polarized in diagram... ?_?

[distant raven]

Ceramics are fine

So long as they are X5R or X7R

[bright thistle]

00:40... *sigh*, night is long, let's do this

[distant raven]

Central European Time?

[bright thistle]

aye

*Summer Time as of now 😉

[fast tundra]

I like the TLV76733

[distant raven]

Oh that’s a good one too

[fast tundra]

but for 5v -> 3.3v the TLV755 is a better choice

[distant raven]

Most of the TI regulators are great because they all mostly have low quiescent

[fast tundra]

Thermal resistance 100.2 C/W 😱

[distant raven]

Yeah, super stable

[fast tundra]

that's pretty high lol. Don't draw 500mA from that thing unless you've laid down some serious copper heatsinking

[distant raven]

I use pours for ground

Plenty of thermal decoupling

[fast tundra]

here's some heatsinking on a TLV767 board

[distant raven]

The SOT223 for the 7533 isn’t terribly bad thermal wise

[fast tundra]

yeah

[distant raven]

I don’t think I’ve run more than 400mA through it though

[fast tundra]

exposed pads are chef's kiss though.

[distant raven]

If I have to got more than 400mA I’ll opt for a switching regulator

[fast tundra]

yeah my rule of thumb is anything more than ~200mA

but I also generally have to deal with 12v -> 5v/3.3v instead of 5v -> 3.3v, so I got a lot more watts to deal with.

[distant raven]

I found some sweet tiny footprint 3.3V/500mA super low quiescent 86% efficiency dc:dc regulators

The 5V/500mA isn’t too bad either

Takes 17V makes input

[fast tundra]

nice

I have a murata part that I like.

hold on

[distant raven]

https://www.oakdev.tech/store/p3/tps62173-acorn-5V-buck.html#/

[fast tundra]

https://www.murata.com/products/productdata/8807037992990/oki-78sr.pdf?1583754815000

aw it's so cute

[distant raven]

I think quiescent on it is like 25uA

Sorry 17uA average lol

I didn’t realize how nice it was until one desk of Ladyada where she was looking at boost/buck regulators and talking about how good they were for low power operations

That murata part is nice

Flexible output depending on your needs

[bright thistle]

amazing how I always forget I have two layers available for routing signals, and keep to go all around the board

[still sphinx]

I cant find the design rules section in kicad cos all the tutorials are in an older version

also should i increase the clearance for HV

[heavy jasper]

I can't help with the first part since I don't use Kicad, but for the second part, yes, and there are a variety of UL/IPC calculators.

[bright thistle]

Top right of the PCB drawing area, there's a drop down with "Track: ..."
Press it, last entry should bring you to the design rules window

[pearl tapir]

@still sphinx If your voltage gets as high as, say 120VAC and you want to meet UL, it is typically easier to slot the board between the traces to reduce the chance of arcing across the surface.

[unkempt remnant]

Anyone ever had success milling flex PCBs, or am I stuck with etching?

Sticking them down to a flat surface miiiight work, but they're so thin...

[bright thistle]

Oshpark has a flex service

Also, I know you might have been told many times before, but please be careful with the chemicals involved in etching, and dispose of the leftovers correctly :)

[unkempt remnant]

yep. for jewelry I've actually just gone with electrolytic etching (saltwater and a random wall wart) and it's worked fine; I'll probably start with that

I have quit ea bit of prototyping to do so I'd rather not have to go back and forth with oshpark for each iteration

at least I don't need super fine pitch for this particular part of the project, although there's some stuff I'm going to be working on later that I would like to use QFN components on

[bright thistle]

This will probably end in lulz being ensued, but did you tried using conductive pens? :D

Or 3d printing with metal filament?

[distant raven]

Got some much needed parts in. Time to solder up 4 panels of LiPo power packs

Might make it a solid 30 for stock and do 5 panels

[distant raven]

Nevermind I forgot capacitors LOL

[silk lark]

a pro micro footprint with a samd21, for refreshing old projects

[unreal flax]

I'm not sure what the layer colors indicate, but it's suspicious that the exposed pad is different from the others. No solder paste on it?

[silk lark]

yeah, grey is paste

[light ermine]

Assuming the brown thing on the right is just a logo on the silkscreen, the corner radius on the inside is too high.

[silk lark]

not sure what you mean, the corners are not rounded on this image (later I did make them rounded a little)

[unkempt remnant]

Wait. I am a dumb. I can't electrolytically etch copper completely off of a flex pcb, it will stop being conductive

on the other hand, since this circuit is so simple and I'm going to be potting it in silicone anyway, I could very well just cut it out of adhesive copper sheet/tape and drop it on a sheet of kapton

I have a vinyl cutter, but it would probably be doable by hand too

[silk lark]

use the regular etching stuff as with other pcbs

[unkempt remnant]

yeah I was just attempting to be clever, but hten I realized it was stupid

[silk lark]

it is doable, with a conductive brush setup

best effects when the brush bristles are carbon fiber

[unkempt remnant]

I am having a dumb. How do these right-angle LEDs sit on the PCB? Every other right angle LED has the lens fcing the direction the light is supposed to go, but the pads on this package are in the wrong place for that. https://www.digikey.com/en/products/detail/qt-brightek-qtb/QBLP617-IW/4814678

[unkempt remnant]

tbh it sounds like I should just grab some ferric chloride 😛

[vapid grove]

Yay @silk lark continues his low-cost series of boards!

If QMK supports samd21, then I believe that would be the best pro micro replacement

For custom keyboards

By the way, I have found a very interesting arduino clone

It is called “vanillin”

Look at these holes lol

This is a Russian uno clone

And it is based on atmega328p

[supple pollen]

Interesting

[elder peak]

When the designer's buddy asked them to poke holes in the design, I don't think that's what he meant.

[silk lark]

@vapid grove ZMK supports SAMD21

[unkempt remnant]

Hmm. I have a heat press, and I figured it would be exactly the thing for toner transfer, but so far it has been disappointingly ineffective. Could be my printer, though.

the more pressure the better, right?

[silk lark]

also print it on shiny magazine paper

and disable toner saving

[unkempt remnant]

I was using transparencies. I definitely could use more toner, though

I'll see if I can change that setting somewhere

[bright thistle]

There's also another way: get a flatbed printer (or modify an existing one like you'd do with cd/card printing)

https://youtu.be/JPva4kMxmpg

[distant raven]

Hey @fast tundra if a data sheet recommends using a crystal with 50ohm ESR but the footprint you want only has crystals with ESR rated at 150ohm, is it okay to use the higher ESR?

Should I add a parallel resistor across the inputs to bring ESR closer to 50ohm?

[fast tundra]

I don't know off the top of my head.

Microchip has an app note about osc selection

[distant raven]

RP Sapling. QT Py for reference

[distant raven]

Ah wait, so looking at the data sheet the 50ohm ESR and 1K resistor are chosen to make things more stable. So choosing a 150ohm ESR crystal and a 900ish Ohm resistor should yield similar results

[unreal citrus]

I'm hoping to get a sanity check on my robot arm driver + peripherals schematic. This is my first robotics/serious electronics project so it's very possible I'm missing something simple. VCC is 20v DC from a high quality power supply.

The schematic is composed of 4 main parts:

• Bottommost circuit: a 12v converter and pair of MOSFETs for controlling a 12v LED strip via the RPI's hardware PWM. The LED strip has a set of white (W) and warm white (WW) LEDs to control colour temperature, hence the two MOSFETs/PWM lines. The LED's +12v is fused for 2A which is in line with their 18W max draw
• Second from bottom: a 5v converter for powering the RPI (via USB) and a fan header for cooling. There is no possibility that the fan could be rotated to feed voltage back into the circuit
• Second from top: a pair of TMC2208 chips. Each connects to various GPIO lines to control their respective motors. Additionally, VCC (20v) is fed to VMOT and the RPI's 3v3 is fed to VIO. These then run to headers that will connect to the motor itself. Like the fan, there is no possibility that the steppers could freewheel to feed voltage back
• Topmost: a simple header to connect the reed switch used for homing

Questions:

• The RPI's ground is connected to the common ground. Is this needed/wanted?
• Does the RPI PWM circuit to control the LEDs look correct?
• Should I add any diodes between the MOSFETs and the RPI? Or otherwise add any (very simple) protection circuit?
• Does this pass the sniff test? Are there any glaring issues around power conversion, etc?

Thank you!

[spark sun]

They go sideways on the board like this, the pads don’t sit flat. They could prob be soldered facing up too but I think this is the expected usage. Ignore the crookedness 😅

[unkempt remnant]

Huh ok. The pads just... Aren't on the bottom? That's super weird

I wonder why?

[broken zenith]

I'm working on a robot controller PCB. This is my second PCB, made in kicad. How's it looking?

[supple pollen]

It looks like you have more room for clearance if you move your traces and diagonals around some

[spark sun]

It is a bit weird. Guess it's a mix of 1) having more pad surface area (lying that way, if they were on the "bottom" only they'd be very tiny pads) and 2) structural stability, the LED is less prone to getting knocked loose when the solder wicks up the sides like that

plus this way there are 2 mounting options (ie. could mount facing up)

[unkempt remnant]

I guess it's nice because it gives you three different mounting options so you don't have to plan as much around the polarity of the diode... but still

it seems like tehy could add side pads in addition, but oh well

(insert "both is good" gif here)

[spark sun]

yeah depends on the led, i've seen some where the pads do wrap around to the side more

[unreal citrus]

Sorry for the bump, but would anyone mind sanity checking this? I'm eager to order some PCBs haha

[fast tundra]

It was fun fitting all of this stuff on the board

@unreal citrus I'm taking a look now

[unreal citrus]

Thank you! I've noticed that I had the LED MOSFETS hooked up incorrectly, so I've since swapped the GPIO and W-/WW- pins on the MOSFET

[fast tundra]

Wanna send an updated schematic?

[unreal citrus]

Gladly

[fast tundra]

Since this is around motors and such, I'd recommend adding a filter before the two switchers. The datasheet has a recommendation (both switchers can share this same filter, but be aware of current)

[unreal citrus]

I'll look into that! Seems pretty simple

[fast tundra]

you need bypass capacitors for your motor drivers

and a lot more, hold on

Are you using the module version of the TMC2208? https://www.digikey.com/en/products/detail/trinamic-motion-control-gmbh/TMC2208-SILENTSTEPSTICK/6873626

Okay, if you're using the module version you're fine

I would still consider adding some bulk capacitance here, since there seems to be a lot of current draw.

(post regulation)

general thing: use power symbols for VCC/GND/+5v/+12v/etc because all of these looping wire connections are hard to read.

[spark sun]

^this (capacitors) was my gut feeling also, but didn't feel qualified to answer since I haven't ever built a similar circuit. glad to see it confirmed. nice to see you on Hackster the other day @fast tundra , great interview 🙂

[fast tundra]

Aw, thanks. 🙂

[unreal citrus]

@fast tundra thank you very much for the input! I'll implement those changes

[fast tundra]

no problem!

Good luck!

[spark sun]

PSA: the Digikey kicad footprint called "SOIC-8_W5.3mm" is not, in fact, 5.3mm wide. Double check your footprints means measure them, not just check that the name matches your package. Don't be like me and order $100 worth of PCBs with the wrong footprint 😅

[light ermine]

Ouch. I guess you'll be bending pins for a while 🙂

[south gazelle]

Does anybody know how the PCB keyboard of the Arturia MicroFreak works? (https://www.arturia.com/products/hardware-synths/microfreak/overview )?
It's based on a PCB and some form of capacitive touch. What surprises me is that is seems to be pressure sensitive, and actually quite good at that. All capacitive touch PCB interfaces I've seen were typically binary (on/off) and do not offer any gradient of touch/pressure sensitivity. Any idea what chip/technology they are using to drive this?

[unreal flax]

Capacitive sensors are generally inherently analog, since they're measuring capacitance... a digital output would just be thresholding the underlying value. There's some correlation between the reading and the surface area of the fingertip that's touching the electrode, so you can get a rough pressure measurement that way.

[south gazelle]

@unreal flax Makes sense. Do you know of an capacitive touch IC that provide an analog reading?

[unreal flax]

I'd generally expect any of them with a register-level (like I2C) interface, as opposed to just digital button outputs, to provide that. For example, the Cypress CapSense MBR3 seems to.

[elder peak]

I got burned by the DigiKey KiCAD repo

[distant raven]

At42qt1070 is a great way to go

[south gazelle]

@distant raven OK. I'm looking at the datasheet of the At42qt1070 right now. I assume that what they call "Key Signal" is some kind of analog level that would be proportional to the amount of contact the finger has with the PCB.

[distant raven]

Yeah

Each channel data corresponds to 2 bytes of data

Easily tunable

I’m working to launch a breakout for it this month

[south gazelle]

Interesting!

[unkempt remnant]

Getting good solder joints on these is unbelievably frustrating. Usually, you blob some solder on the pads and surface tension saves the day. NOT THIS TIME

Do you have any secrets? Or is my solder paste just old

[carmine scarab]

Maybe you shouldn't touch that part... or add a heatsink? 😄

[supple pollen]

My usual approach is "moar flux"

[carmine scarab]

Careful with that around capacitors!

[supple pollen]

Built one

[unkempt remnant]

Oh. Ohhhhh. The recommended pad size for these right angle LEDs is HUGE. I see "0603", I use an 0603 footprint, is that so wrong

I guess to give you more area to glob solder onto...

Now, if you flip it so the lens faces up, an 0603 footprint works just fine

[unkempt remnant]

I just came across these via grommet things. How are you supposed to set them? I have plenty of wrong tools that will probably do the job, but I'm curious what the right tool is.

[supple pollen]

https://www.mcmaster.com/grommet-setters/

[unkempt remnant]

the hole diameters on those grommet setters start at 3/16"

I have plenty of grommet setters in that range

but down in the 1.5mm range? not so much

[supple pollen]

Something like this? https://www.ebay.com/itm/273601184461

[elder peak]

Wow, madbodger, your explanation for this was especially.... riveting.

[haughty wolf]

not gonna lie, all this rivet talk has me fixed

[unreal citrus]

I've never worked with inductors before and I'm getting mixed answers on a question I have.

Per @Stargirl#6666 's advice I'm adding a filtering circuit before a DC-DC converter. The recommended inductor is a RLS-126, but digikey Canada doesn't have any in stock. Is it a problem if I use a "beefier" inductor such as the RLS-186?

Description for RLS-126: FIXED IND 12UH 800MA 420MOHM SMD

Description for RLS-186: FIXED IND 18UH 1.89A 100MOHM SMD

[distant raven]

I’d say yes because it changes the fundamentals

And depending on the DCDC converter, you could accidentally add an inductor that makes it self resonant which is something you should avoid at all costs.

I’d say order from DigiKey USA or try mouser or Arrow Electronic

Alternatively you can search for one with the same or similar values

[supple pollen]

For a filter, it's probably fine. For a resonant component in a switching converter, details are more critical.

[distant raven]

Accompanying capacitors will help it not resonate which is usually what happens but even then inductor is dependent upon dcdc frequency and required filtering capacitors no?

[heavy jasper]

So, short version: You want the output impedance of your DC-DC input filter to be much less than the input impedance of your DC-DC converter to avoid causing loop gain problems and instability.

Technically there's a lot of math to that input impedance that depends on circuit type but a simplified minimum (worst-case, kind of rule of thumb) can be obtained via a variant of ohm's law: Vin^2 / (input power) == Vin^2 / (efficiency * output voltage * output current).

The trouble comes when you have an input filter with super low resistance inductors and capacitors, because then at the resonant point, the output impedance can spike massively.

(however when not at that point, a lower series R of the inductor actually helps keep the output impedance low! Mostly this problem comes from super-low-resistance capacitors like ceramics).

It's possible that your filter as it stands may have no problems, but it would take some analysis. If you want to be extra-sure, you can do something like this:

where Rd is a damping resistor that provides a bounded maximum output impedance of the filter (L | C | R ) when the L || C are resonanting to form an open-circuit. An easy way to determine values here that's pretty much mathematically optimal: Rd = sqrt(L/C) and Cd = 4*C

Often this is done with a ceramic cap (or several) for the normal C to give low-ESR high-frequency performance, and then a big electrolytic for Cd since its series resistance is actually a benefit.

Lots of gory details here: https://www.ti.com/lit/an/snva538/snva538.pdf?ts=1618099371816 and here: http://ecee.colorado.edu/~ecen5807/course_material/Ch10slides.pdf for those interested in diving into the math in more detail

[unreal citrus]

Thank you very much for this!

[unkempt remnant]

Vinyl cutter flex pcb, copper tape on kapton tape

I made an attempt with thinner traces, but either they got pulled up while cutting or I couldn't consistently get them to transfer

[unkempt remnant]

anyway, this is going to be potted in silicone, so I'm not super concerned about fragility as long as it lasts long enough for me to pot it

...although I might track down some kapton film instead of tape. otherwise there will be dog hair in it

kapton tape was a convenient form factor, though, since I didn't have to cut pieces off a sheet and then end up with a funny-shaped sheet that's hard to use efficiently

[distant raven]

This is a pretty cool project

[unkempt remnant]

this project is nuts

it can accurately be subtitled "Inadvisable Uses of LEDs"

so far there are at least three switching power supplies, two different wireless technologies, and... lemme count 'em... 8 40W LEDs, 4 15W LEDs, and a couple dozen little surface mount guys

not that I'm going to be running the 40W LEDs at full power

next time I think I'll just use the hot air gun instead of the oven; since I needed big chonky traces and don't even have real pads, the LEDs tend to slide out of place and then I have to rework them anyway

hah, I could try cutting another piece of kapton tape for a soldermask

[elder peak]

Subtitle: If your light-up cosplay doesn't have you worried about asploding, you need to add more lights.

[unkempt remnant]

💡

[tough matrix]

Today I Learned
that there is such a thing as solderable PCB captive nut, for attaching standoffs:
https://www.adafruit.com/product/4207

why didn't they tell me this earlier?

[twilit mango]

We all got together and decided we should let you find out on your own.

[silk lark]

@tough matrix there is also a bolt like that

[drowsy drift]

hey @crisp crystal I've been loving your recent PCB art videos, thank you! What DPI do you normally use for the bitmaps when importing into Eagle? And do you use the hoary import-bmp.ulp or do have a better way?

[crisp crystal]

@drowsy drift 1200 dpi, which is probably excessive but nobody really has an answer. Ladyada does the Eagle import, and yeah, I think uses the import-bmp.ulp script.

[drowsy drift]

cool thanks! I'm usually at 600 dpi but have been loving your clean silks so I think I may bump it up

[tough matrix]

anyone has experience with edge connectors, aka "gold fingers"?

I need a custom pcb to be plugged into the micro:bit slot.

Advice i had seen online suggests that I need to order "hard gold" plating on the edge contacts if I want it to survive many insertions.

problem is, hard gold plating is quite expensive, so it doesn't make sense for small run. So my question is: if i do usual ENIG instead, how many insertions can i expect it to survive? 50 would be plenty for me, 5 would be problematic

e.g., does adafruit use hard gold on their trinkey?

https://www.adafruit.com/product/4870

[elder peak]

I can see how this would be a golden opportunity for you to ask.

[tough matrix]

i tried modifying this design, but can't make it work...

[distant raven]

I believe you just extend a pour over the edge to do gold fingers

[tough matrix]

The real question, though, is about the hard gold plating - is it really necessary for hobbyist products?

[pearl tapir]

@tough matrix I believe that the automotive world does not use gold plated connectors due to expense. The last time I read something along those lines, I think the words "gas tight" were how they prevented corrosion. Note that in the automotive world, they expect very few removal and re-insertion cycles so that may not fit your design goals.

[tough matrix]

I do not have a choice - I need a board that would work with an existing product, and that product has a slot for plugging in micro:bit, so I must use same edge connectors as microbit has

[carmine scarab]

You can check out the design on the Clue board, since it's got the same (or compatible if not the same) connector as the micro:Bit
Product page: https://www.adafruit.com/product/4500
Main learn page: https://learn.adafruit.com/adafruit-clue/
CAD, Eagle files, etc: https://learn.adafruit.com/adafruit-clue/downloads

[tough matrix]

@carmine scarab yes, I certainly will.
But the Eagle files have no info about manufacturing process - same gerber file can be used with HASL, ENIG, or hard gold finish

[carmine scarab]

I see. That's definitely something that Ladyada or someone else who's in the engineering bits could answer. I suspect that the finish isn't hard gold on those, since it's aimed at the hobbyist, not to mention the cost difference it might make

[distant raven]

It’s likely ENIG

[tough matrix]

PCBway quoted me price of $45 for 5 copies of board with ENIG finish, and over $200 for proper hard gold plating of gold fingers

that's quite a difference

[haughty wolf]

jlcpcb tends to be a bit cheaper

20 bucks for 5 boards with immersion gold, as long as the board is under 10 x 10 cm

[tough matrix]

yes
But they do not offer hard gold at all

[haughty wolf]

i bet you like it hard ( ͡° ͜ʖ ͡°) oh ok

[heavy jasper]

Hard gold is definitely the most durable solution, but for your application ENIG will almost certainly be fine.

And the automotive world spent a lot of time away from gold-plated connectors due to expense, but is moving back to them as the signals running around the car get lower-voltage and higher-speed.

And veng is absolutely correct that if you look at any automotive connector, the number of rated cycles will be 10-50. Many (but not all) are able to do many more cycles physically, but there’s not the push to do the validation of additional cycles since it doesn’t matter for the intended use.

The big advantage of automotive connector series (IMO) are that they generally have good positive latches and terminal retention (so-called TPA for terminal position assurance), and also tend to be available in a variety of keys since the companies really care about mistake-proofing the manufacturing.

[tough matrix]

Thanks a lot! this is what I wanted to hear 🙂

[heavy jasper]

Even more so, for what you’re doing, you’re probably not counting the milliohms of contact resistance.

[tough matrix]

no, all I need is regular digital IO pins and I2C - either would work fine with some resistance

[elder peak]

No resistance to some resistance.

[minor axle]

Does anyone know if I can use the scl and sda pins (PA13 and PA12) on the Feather M4 to connect to usb D+ and D-?

[silk lark]

you can't

[minor axle]

Are there any pins other than the ones already being used for D+ and D- that you could use?

[silk lark]

no, you have to use PA24 and PA25

there is only one USB peripheral

[minor axle]

Ah, ok. I’m assuming that’s a hardware thing right? Like I couldn’t just change the firmware a bit?

[silk lark]

they are broken out on the bottom of the board as test points

[minor axle]

Ok

[silk lark]

yes, it's a hardware thing

the test points are under the usb socket

so you can solder wires to them and use your own usb socket

or cable

[minor axle]

Ok. The plan is to put this in a keyboard that might go to group buy so that probably wouldn’t work

Thanks anyway! You definitely saved me a lot of time adding that to the pcb and ordering it

[unkempt remnant]

I just realized something. From what hte chipquik guy said about modifying a reflow profile for an aluminum substrate, the reflow profile is targeted at hte board temperature, not the oven temperature. I wonder if I should change my existing reflow profiles where I just plugged in the datasheet temp without measuring the board temp

Because even for FR-1, if I plug the reflow profile from the datasheet directly into my reflow oven profile, the temperature lags behind fairly significantly

like, it's only barely hitting the reflow temp according to the thermocouple I taped to my board (of course, this assumes that the thermocouple isn't lying)

I suppose as long as my existing profile works, and my thermocouple is consistently wrong, I just have to modify the aluminum profile so that the temperatures end up the same as I measured for FR-1

[pearl tapir]

@unkempt remnant The chipquik guy is right. The profile temperature is for the solder and it is best measured with a thermocouple on a pad on the board. Tape the thermocouple lead down to the board and then thermally bond the end of the thermocouple to an unused pad of copper area with heat sink grease.
Get a thermocouple that you can rely on. One possibility is put it in boiling water and see if it reads 100C or 212F.
If you want professional results, you are going to need a controlled soldering profile. It's possible to watch the solder paste and as soon as it is liquidus, turn off the heat but that is subjective and not truly repeatable. It's probably okay for prototypes but not production.

[unkempt remnant]

My goal here is to be able to successfully reflow prototypes on aluminum substrates, rather than FR-X. Seems like there's room for improvement even on my FR profiles too, though, if the board temperature is the important part.

Sounds like I have the right idea - I have the htermocouple taped to the board wi dab of heat sink paste

So fancy

He said that adjusting for aluminum is a matter of lengthening each phase however long is necessary to reach the target temp, which makes sense

But then I realized I didn't have a baseline for FR-1, and I could hardly expect the board temp to be the same as measured by the reflow controller

[unkempt remnant]

Okay, I think the right answer here is to switch the roles of the oven controller thermocouple and the separate thermocouple for this test. Let the controller do its PID thing with the temperature measurement I actually care about, and record the air temp so I can set up a profile with the corresponding air temp targets (that way I don't have to attach the thermocouple to the board every time)

now, is the controller thermocouple actually long enough...

[pearl tapir]

@unkempt remnant An approach I've used is to have a sacrificial board of each type Al, FR-4, single double sided, multilayer, etc. with the thermocouple permanently attached. Then put the type board I intend to reflow into a corner of the oven with the thermocouple plugged into the PID controller and the actual target board in the oven next to the sacrificial board. The temperature of the two will be close enough so the temperature profile will match.
If you have or get some higher temp solder than what you normally use, you can solder the thermocouple to the sacrificial board to save fooling around with Kapton tape. I find after a couple of cycles, it pulls loose. Thermocouples are dirt cheap.
The sacrificial board can be anything reasonably close from a failed design. Assuming you have any failed designs. I have many.

[unkempt remnant]

Ohhhh that's a good idea! I'm mostly using low temp solder, so that won't be a problem. I'd just need to set things up so I can switch the thermocouples around.

If only this kind of thing was rated for reflow temps https://www.ebay.com/itm/224116016975

Right now the thermocouple is hard-wired to the controller

[pearl tapir]

@unkempt remnant Why do you want the plug and socket to be rated for reflow temps? Only the actual cable needs to be in the oven with the plug and socket at the PID controller.

[unkempt remnant]

Not sure how I'd have access to switch them out, otherwise

Wait, if I could run the external thermocouple in from the outside to calibrate, I should be able to do the same with these. Not as ideal as a more heat-sealed port (the build guide for this kit has you seal all the holes with thermal silicone), but the insulating door gasket has enough squish

[unkempt remnant]

Or run them in through a sealed hole and have a switch on the outside

[narrow ruin]

hi all!

does a crimping tool for JST XH (one of these here https://www.qiannipicture.com/pic/imgsgroup1/M00/36/71/E52724939499C68C878007BDCDBEBA9B99876EBE14B4CAC425D4BD6FADE614CC5353F513.jpg) also work for JST SH 1.0mm (as used by Qwiic connectors)?

hmm, and also, i can't find the little black sockets that are used on Adafruit breakouts on eBay searching for something like "JST SH crimp set" ... could somebody help me find one of those boxes that are very commonly available for the bigger XH system?

[nocturne wasp]

I doubt it. The JST XH series has a 2.54mm pitch, and the clips are much bigger than a 1.0mm would be

The jaw size really matters.

Are you doing a lot of them?

[narrow ruin]

hmm i'm planning on using them for everything that's not THT in the future

so probably it'd be a good investment

but first would have to find one of those sets

curious they're so ubiquitous for XH but nonexistant for SH ... all you can find is readymade cables with sockets which don't fit the footprint of those used by Sparkfun and Adafruit

[nocturne wasp]

This one may work: ~~https://www.pololu.com/product/1929~~

This one claims specific compatibility: https://hobbyking.com/en_us/hobbyking-jst-sh-connector-crimping-tool.html

[narrow ruin]

ah, those yellow ones are also available on ebay

what i'm having trouble finding is the sockets and housings used by Qwiic

[nocturne wasp]

Oop, actually ignore that pololu; I misread mm^2 as mm

Here's a socket: https://www.sparkfun.com/products/16766

The housing appears to be a: SHR-04V-S-B

https://www.digikey.com/en/products/detail/jst-sales-america-inc/SHR-04V-S-B/759868

Ugh that was a slog...

[narrow ruin]

i'm sorry @nocturne wasp 😦

thank you for sifting through the interwebs for me

should have had the idea of going elsewhere but ebay, myself

[nocturne wasp]

Sparkfun links to the family's datasheet: https://www.jst-mfg.com/product/pdf/eng/eSH.pdf?5ec58813d214e
From here: https://www.sparkfun.com/products/16766

[narrow ruin]

the datasheet even has the footprint i found on EasyEDA ... same one!

[nocturne wasp]

On the second page, there's a housing chart--I think 4 circuits

Oh, the flange is different than I thought--it's not the polarity strip, but a shoulder to hit the jack

So just SHR-04V-S will also work

[narrow ruin]

ah and the header is SM06B-SRSS-TB, there we go

just seems one has to settle for Ivory ... black must be a special deal Adafruit and Sparkfun got at volume 😛

sorry, that's ofc the 6-pin ... SM04B-SRSS-TB for the 4-pin, Qwiic-compatible

and it seems if you're in the US, those $0.50 for one of the headers is actually a very good price 😦

(but then i'm not even in the US, soooo...)

@nocturne wasp these look compatible but much cheaper: https://www.tme.eu/de/en/details/nx1001-06smr/raster-signal-connectors-1-00mm/joint-tech/a1001wr-s-6p/

[tough matrix]

minor correction: JST XH has 2.5 mm pitch, not 2.54

[distant raven]

super tight spacing on this one lol

most of the components belong to the FT232h

just need to remove the labels and do a final sweep of connections

beautiful

[heavy jasper]

Unless the cad model is not entirely accurate, looks like the JST connector impinges on the crystal, and possibly the passive to the right?

[distant raven]

The outline is slightly bigger than the actual connector

Tight packing a board is tough

Especially only using one side for components

[silk lark]

I think that tight-packing both sides would be harder

routing-wise

[distant raven]

You’re right about that

[elder peak]

Let's not pick sides here.

[tough matrix]

sooner or later, they will allow 4-layer boards with components on all 4 layers 🙂

[distant raven]

I’m going to make a ridiculously unnecessary RP2040 stamp that makes it BGA

[elder peak]

That's one way to make sure that your design skills are on the ball.

[unkempt remnant]

Aww yeah. Flex boards v2: on normal kapton film with a kapton tape "solder mask" (it's really more to protect the traces - too much trouble to line up actual pads)

[distant raven]

The RP 2040 stamp that no body asked for

need to shift the BGA footprint slightly over to be perfectly even

[tough matrix]

Ballsy

[distant raven]

It’s one of those, there isn’t really a market for it. It’s just a cursed stamp lol

[haughty wolf]

is putting subcircuits on a bga pcb a thing that people do? i cant find much about it online

its a supercool™️ idea but i dont see it being much use in the real world lmao

[heavy jasper]

That’s in fact how most BGA parts are made

[unkempt remnant]

how the heck do you solder them? the solder does not want to stick AT ALL to the thermocouple 😦

[heavy jasper]

The BGA base itself is a very dense PCB, often with some decoupling capacitors on it, all running to a much smaller die that’s often its own much smaller and denser flip-chip package-on-silicon part (or is wire bonded). The package PCB is super dense, a bunch of layers, often blind/buried vias.

[unkempt remnant]

hopefully I haven't destroyed the thermocouple by this point

[unkempt remnant]

huh, this indicates that a combination of thermal tape and kapton tape does an OK job https://kicthermal.com/technology-information/conference-papers/261-a-comparison-of-methods-for-attaching-thermocouples-to-printed-circuit-boards-for-thermal-profiling-3/

although it might require that specific thermal tape - I expect stickiness is important and varies between brands

[fast tundra]

We still showing dense layouts? 😎

[elder peak]

I guess you could say that there's always room, @fast tundra

[fast tundra]

Hey if I didn't have to put anything on the back, that's a win!

It's also a 2-layer design. 😭

[compact lotus]

Hello <@&741361254035095602>

I am working on designing an OSHW PCB

what are the best practices or tips you can share on how to document a OSHW PCB design

I already included the board specs, BOM, and KiCAD files what other information or files should I also include in the git repo ?

thanks

[unreal flax]

I'm always grateful for a nice PDF schematic in the repo.

[heavy jasper]

I also like to see the final manufacturing files you send to your manufacturer, e.g. gerbers/drill/ODB++/any drawings, along with an exported step file if your tool supports it with nice 3D bodies.

Also any design collateral that you produced and isn’t already documented on the schematic (things like calculating out power dissipation in voltage regulators, tolerance of resistors, and the like)

And then subtle but important in the git repo: an appropriate LICENSE file for whatever hardware license you’re using.

[compact lotus]

good thanks

this also a guide from Open Source Ecology

https://docs.google.com/spreadsheets/u/0/d/1sgfOTDR9UB7rPGRN3iA1H0WNOakNV3WXTdQfPTJrs-Y/edit#gid=2

[tough matrix]

I am thinking of doing a "learn to SMT solder" badge kit for kids. (Yes I know there is already about 100 such kits, I need themed design for a summer camp).

Should i put a 555 or an ATtiny to control LED blinking? I know both are possible; any reason to choose one or the other?

[crude ocean]

555 ?would have more parts to solder (resistors/capacitors, oh my!), and has some analog hackability
ATTiny ?might be more expensive, but can be coded a little bit

are you more inclined to teach analog or digital is my question/thought

[tough matrix]

it is supposed to be just a 1 hr event, to show kids who have never seen solder paste how it works. I will not have time to properly explain operations of a 555 nor programming - but can show them enough to get them interested

price is really not an issue at the moment - I am fine spending $5 per kit

[crude ocean]

I suppose the ATTiny can be set up to blink more than one light in some vaguely interesting manner

is pondering all the mission creep ideas apparently.

I personally seem to favor the 555, as it opens discussion of various analog circuitry elements. But more opinions are probably useful here.

I imagine witnessing blink rate change when shorting elements with their fingers being somehow more engaging.

[radiant wolf]

If you want it as simple as possible to solder use the attiny. If you want a variety of things to solder go 555. If you want to have multiple leds go attiny.

[tough matrix]

i will probably want at least 2 LEDs - two eyes

[crude ocean]

in theory the 555 should be able to drive alternating LEDs

[radiant wolf]

All this is presuming the attiny doesn’t require any extra circuitry to function, I can’t remember what they require.

[crude ocean]

power, and a program installation (and maybe more, I'm no expert there!)

?seems to have an internal oscillator, cool

might have a wider operating voltage range than I expected

[tough matrix]

yes, looks like it can be used with lipo without a voltage regulator

just some capacitors

[crude ocean]

depending on power draw I'd ponder coin cells (or some other alkalines), lipo might be too hard to charge at home, and too volatile around silliness (also the prices tend to make me clutch my chest)

[tough matrix]

coin cells, sure

[radiant wolf]

Exposed lithium ion batteries and kids that have never soldered before seems like a bad combination

[tough matrix]

but typical 2032 coin cell is lithium, not alcaline

[crude ocean]

yeah, but all the others, AA, AAA, 9v, are

[tough matrix]

i want a wearable badge, so ... 2032 is the easiest choice

[crude ocean]

casing for AA/AAA does get absurd

[serene ermine]

Hello folks, I'm hoping someone has a simple solution for the problem I'm having.

First of all, I'm using Eagle to design a PCB and wanted to have some text print on it in exposed copper (Purely as an aesthetic choice)

From the research I have done thus far, everyone suggests that I place my text on the "tStop" layer. This however does not have the desired output.

Other ideas are to put it onto the top layer, but I'm going to assume this is bad practice.

I've had these boards manufactured and the text is there, but the copper is not exposed. (I've attached an image of the manufactured boards)

Here's an image from the manufacturing tab, as you can see the text isn't being recognised as a copper layer.

[silk lark]

is that another 32u4?

[serene ermine]

Yeah, pro micro

I seem to have some success by placing the text in both the Top layer, and tStop layer. However not sure If this is how It's meant to be done

[silk lark]

oh, I see what you did

normally you would have a ground fill on the all of unused parts of the board

so you would have copper under that text

but you are telling them to remove all copper except for the traces

that's why your text is just exposed fr4

[serene ermine]

Ah right, so should I be adding a ground fill?

Is this something I should do with every board?

[silk lark]

it's always a good idea to have a ground fill

[serene ermine]

Alright I'll look into how to do that then

[silk lark]

it improves your ground, does some shielding, and makes the pcb manufacturing process use less chemicals

because they don't have to dissolve as much copper

[serene ermine]

so basically, you would be filling everything but my traces and pads

[silk lark]

yes

[serene ermine]

That makes sense, thanks! I'll see If I can find out how to do that

[silk lark]

eagle might have a special layer for it, or I may be confusing it with something else

[serene ermine]

Thanks 🙂

[tough matrix]

@spark sun how difficult was it to solder the side-emitting LED to your Perseverance badge?
(not for an expert but for someone new to this)

[spark sun]

Definitely a bit trickier than eg. resistors but not too too bad, honestly. When I did my first one it was my first SMD project, though I've done lots of THT soldering. It took some tries to get it right but now it only takes me 30 seconds or so... I might put together a video for the Tindie kit page but here are the tips I have:

• I use 1206 size LEDs, still small but much easier to deal with than 0603s etc. (can give you a part # if interested)
• Use plenty of flux and tin your LED pads first
• I pick up the LED from the lens side with tweezers and hold it on the pads or very close to them
• Then use the iron to heat both LED pads at the same time
• Wiggle the tweezers a bit if necessary to get both LED leads in contact with the melted pads
• Then I switch my tweezer grip - let go of the pinch and just press downward on the lens side of the LED, so it doesn't move when you pull the iron away
• Pull the iron away slowly, sometimes I stroke it upwards along the back of the led to make it wick up the leads nicely
• Finally add a bit more solder on each lead for structural stability

[tough matrix]

Thanks! Will give it a try.
Wondering if it is possible to do the same with solder paste and reflow oven

[spark sun]

Yeah, I imagine so! Though they might have the tendency to "tombstone" if they are not positioned on the pads just right. I have been considering buying a hotplate so would be interested to hear your experience reflowing them.

@unkempt remnant you had a similar question that I meant to respond to and totally forgot, ^hope that might be helpful for you too

[unkempt remnant]

from the research I did yesterday, it seems like soldering k-type thermocouples to a PCB requires some special setup - extra high temps and acid flux or something 😕 so my best bet may be aluminum tape, which is apparently reasonably effective

[unkempt remnant]

I could bust out the blowtorch and silver solder but I'm not sure I can avoid frying the boards

particularly since I'm prototyping with FR-1

[quasi gust]

I'm hoping to start a PCB design using the new NeoKey Featherwing design as a base, but I don't see it on GitHub yet. Does anyone know if there is just a delay between product and GitHub release?

[distant raven]

Usually takes a little bit for everything to get up to speed on GitHub

[quasi gust]

Do you know if there is any way to get in touch with Ladyada to ask about it? I've got an idea of a design that is directly related to the NeoKey she has been working on that I think she'd be interested in. I don't see any way to contact her though (probably for good reason, as I'm sure she gets tons of messages)

[distant raven]

I would ask @twilit mango

She might have a better idea or possibly some of the other Adastaff

[twilit mango]

Ladyada definitely gets a lot of messages. 😄 As for getting it up on GitHub, that usually comes along with the guide which isn't on my radar yet, or if someone requests it. So, I will add it to my list to get the files up on GitHub tomorrow. Today is newsletter.

That's the FeatherWing that's already available in the shop, right?

I mean I'll ask Ladyada to make sure there's not a reason they're not posted, but if it's released, we're usually fine to post the EagleCAD files.

[quasi gust]

Yep! I've been working on an idea for a Neopixel capacitive touch for a while. Basically a button to shine through the LED acrylic with a glowey spot you could touch

[quasi gust]

And I was super excited to see the NeoKey, because if you remove the switch mount, and put a capsense area in it's place, it is it. done. 🙂

[twilit mango]

I'll ping you once the files are posted or to let you know a reason they are not. Should know tomorrow.

[quasi gust]

So once I get those PCB files, I'm going to try it out (I've already got the reverse-mount neopixels on order). If it works out, I'll bring it to show and tell or something!

[distant raven]

Sounds really cool!

[quasi gust]

thanks @twilit mango! I'm hoping to get an effect like this: https://www.youtube.com/watch?v=0FOg6ECdxbY

[carmine scarab]

My kids LOVE those tables! I'm not averse to them either, and have been tempted to take a peek and a poke around to see what drives it!

[quasi gust]

Yeah I have no idea how the company is actually doing it. They've been around for a while, pretty innovative! I think a reverse-mount neopixel with a capsense area should do the trick though! With LED black acrylic it should look pretty awesome (if it works!)

Ladyada has already done all the hard work. The seesaw library will work a treat, and the SAMD21 in the Crickit has PTC (a capacitive touch sensor). That will be enough to try it out with four neopixels and four capsense areas, and then if that works, on to doing a matrix for larger displays.

[elder peak]

So, tl;dr: any steps towards making one yourself have been tabled until git is updated.

[pearl tapir]

It's almost a matter of surrounding the thermocouple with solder and holding in place while the solder cools.
The soldering process should not affect the thermocouple as they are normally welded together.

[silk lark]

some fun with new LED matrices

[distant raven]

Ohhhh

Those are neat

[twilit mango]

@quasi gust https://github.com/adafruit/Adafruit-NeoKey-FeatherWing-PCB

[quasi gust]

Thank you so much!

[unkempt remnant]

I couldn't get it to wet. Like, at all. Trying to stick it into a giant blob of solder just got me a thermocouple sitting on top of a blob of solder

No matter how big the solder blob was (and won't a big solder blob be slow to respond to temp changes?)

[primal schooner]

Was thinking of designing a board with bga component. I plan on using my reflow oven to solder this component to the board and I'm assuming that I just coat the board in a flux for those pads. Is this right?

[distant raven]

Some people use flux to make sure there isn’t any bridging of pins

[primal schooner]

..but is it enough to affix the component to the board?

[pearl tapir]

I don't think it will wet unless you use a very aggressive flux. The most common thermocouples use iron and constantine wires neither of which are easily soldered.

[silk lark]

hmmm

https://hackaday.io/project/168395-keys-featherwing

mine is better, more key, and low-profile!

no hotswap, though

[distant raven]

Between solder paste and some flux you should be good

Highly recommend a stencil

[primal schooner]

Are you sure about solder paste for bga part? I've done this with smt parts, but not bga parts. It would be redundant to have solder paste with a bga part, I think. I'm just not sure, since pcb leveling would be critical here, I might be inclined to apply it. If so, is there a reduced mask size for the stencil?

[pearl tapir]

I suspect the part manufacturer has a solder profile, pad size and mask aperture for the part.

[primal schooner]

Unfortunately, I'm unable to find it. I was hopeful that someone knew if there was such a thing. That's why I'm thinking it's a flux only assembly for this part.

[distant raven]

@primal schooner there are a number of people on Twitter I follow who do a lot of BGA stuff. I’ve never seen them not use paste

[pearl tapir]

I've been involved in the contract manufacturing business and don't believe anyone in the business would ever skip paste. I don't know about how people might do it at home.
Sometimes the part manufacturer has a guideline document for all of their packages rather than put it on each data sheet.

[cerulean crest]

You are correct, if using core solder you certainly have issues trying to get the same amount of solder on each pad, which is an issue im having with a reworked board I have

[distant raven]

https://www.circuitnet.com/experts/88005.html

[cerulean crest]

Oh but is that with or without balls?

[distant raven]

The Twitter link was a bad example because it was a reball

[primal schooner]

I think the spartan 7 I'm eyeing has been......balled.........

[distant raven]

All new BGA should be preballed

It wouldn’t make sense for them not to be

[primal schooner]

yes

[primal schooner]

I know what you mean. At home, this project I'm doing, I simply don't have the equipment available that an assembly house would have. It's just me, a reflow oven, and a desire to put together a prototype on the 'cheap.'

[distant raven]

https://mobile.twitter.com/GregDavill/status/1331512144626556929

Greg (@GregDavill)

🤏

Twitter · Nov 25, 2020, 8:16 AM

Here’s a good example

Greg Davill does a lot of BGA stuff

[pearl tapir]

I would actually look at it the opposite way. Soldering a BGA without flux would require excellent process control in the amount of flux, the co-planarity of the placement and the solder profile. I think at home, buying a cheap stencil would reduce the process variations.

[primal schooner]

Thank you, everyone. This is awesome and I really appreciate.

I'll be looking for those documents. regarding aperture sizes on stencils.

My only concern is bridging......that's why I'm hesitant on extra solder for something so small.

[elder peak]

If you want to work with BGA parts, you really need to be on the ball.

[vapid lark]

so i want to add this -> https://easyeda.com/editor#id=9b05243e8b064f2ba1ce038859def97f|2ce42d0125644f428aed39da4ae8e403 to my PCB as i plan to use the MSGEQ7 with my ESP32 but here he uses all SMD

im new to this and i know TH is Through hole and D is for Dimensions or? and what does the F stand for? and what should i choose here for the RIN -> 0.01uF CR2 as i want all components to be TH? 🙂

https://cdn.discordapp.com/attachments/622118305930674192/834368919443079208/unknown.png

[pearl tapir]

@vapid lark I suspect the D with a number is the body diameter and the F is lead pitch. Tan = tantalum and Ele = electrolytic. Typically these are only packages and you specify the capacitance and working voltage yourself in the CAD. After making sure that such a part actually exists, of course.

[vapid lark]

so i made my first PSB kinda like a firstoff/prototype, is there anyone who can have a look over it (traces and stuff) to see if everything is ok? (im using easyEDA)

[spark sun]

post it here 🙂 if anyone has the right combo of free time + expertise they will chime in 🙂

[vapid lark]

i got some nice help 😄 ❤️

talk about being first time making a pcb xD i checked the pricing... 144$ xD well i learned and it's now 10$ 😄

[spark sun]

i always prefer to do it in public rooms if possible, so others can benefit if its relevant to them

yeah, wild how cheaply you can get them made these days 🙂

[primal schooner]

Has anyone done RF design?

Wondering about delay traces.......

[distant raven]

I’ve gone some RF design but not enough to tell you about delay traces unless you’re talking about length matching of routed pairs

[tough matrix]

https://twitter.com/JLCPCB/status/1386879398003941379?s=09

JLCPCB (@JLCPCB)

The SMT roadMap in 2021 ！ 💥💥

Twitter · Apr 27, 2021, 3:06 AM

[unreal flax]

That sounds really good. The more components they can support for assembly, the more business they'll attract, since that's the big tradeoff in using them right now.

[spark sun]

bigger question is, how many of those components will still be In Stock come august? 😬

[pine breach]

Hey folks! Anyone have favourite free or cheap software for writing circuit schematics and/or designing for stripboard/breadboards?

[distant raven]

a lot of people use KiCAD or FreeCAD. some use EasyEDA i think it's called?

[pine breach]

Oh nice, I didn't know KICAD did both schematics and PCB layouts, looks like it does 👌

[elder peak]

Well, you can get ESD acrylic.

Ponoko will laser cut it for you.

[rustic linden]

I use EasyEDA nowadays, but I used KiCAD in the past

[rustic linden]

does Adafruit have footprints and such for products? I bought something I wanted to mount something else to with the holes on the PCB and I'm not sure how to do that without having to draw up the entire item