#help-with-hw-design
1 messages · Page 57 of 1
Nothing
Ahem, do you have a >NAME label in the part symbol?
That's curious. I don't think it should permit a blank part name. What does the name show up as when you try to edit it with the NAME tool?
ooh one thing, if I rclick and hit properties, I see the name properly e.g. JP2 JP3
let me see about the above
same thing, of course. Oof I'm sleepy
So it looks like something is weird about the >NAME label. Like it's in white text, or placed off the screen, or behind something else, or marked as non-visible, or the layer is disabled, or...
hmmmm
Or it's been added in the library but the library hasn't been updated into the schematic?
should it show up roughly where I put it in the symbol? I did update the library in the schematic but I did "update all" I'll see what happens if I update just the part
no dice
Yeah, it should show up exactly there.
werid
weird
I may have to go to the official forums
hopefully my handle is available
any idea what they mean by "reposition attributes" and "smash" I don't recall smash from EAGLE but I'm no expert
https://forums.autodesk.com/t5/eagle-forum/gt-name-and-gt-value-does-not-appear/td-p/8747245
I have added >NAME and >VALUE belated to a device in my library, which is often used in my schematic. I have updated the library within the schematic editor, but these new items do NOT appear, after I have entered any values. Adding it as a new device in the schematic, everything is normal and these...
The SMASH tool splits apart sub-elements in a part, allowing you to do things like reposition the name label on a part-by-part basis to make the silkscreen tidy.
ahhh ok
Hm, I tried moving the label around, in case that's what they meant by "reposition attributes". No luck though
It's not super duper critical but my schematic looks less professional this way
That's quite weird. You might change the label to just "FOO" or something to make sure it shows up as text in the schematic, independent of the part-name substitution.
Bizarre...
yuuup
@unreal flax I found the right google terms. Seems like this is just plain a bug
using the command reposition opens up a tool and you click the part and it fixes it
🤦♂️
Should I try to pack my board in more? The top half where the FETs are cant really move...I want to try and make provisions for a heatsink if possible (thats the white bar you see on the right), but the board feels empty mostly with all the random stuff on the bottom (like the buttons and speed pot)
Also I realize my RL is in all my file names but idc. I dont have much to hide from my fellow makers
It does make for a nice space for silkscreen, so you don't have to pack it tight. I would personally rearrange the not-super-heat-generating components to be more organized together, and leave a clearer space for whatever silkscreen I would want to put.
I like how eagle marks QFN packages as having clearance issues when I run DRC. 🤔🤦♂️ 😅
OK, I know this may only vaguely qualify as on-topic for this channel, but I've got a mfg side question.
On something like a USB A plug, that has the mechanical through-hole solder tabs.... can that be done in reflow, or is it a selective solder and/or manual only thing?
Didn't try type A, but microUSB connectors with through-hole tabs certainly can be done in reflow
E.g. this one: https://www.digikey.com/en/products/detail/amphenol-icc-fci/10118194-0001LF/2785389?s=N4IgTCBcDaIIwAY5wBxwJwBYC0C9wBkAxEAXQF8g
Order today, ships today. 10118194-0001LF – USB - micro B USB 2.0 Receptacle Connector 5 Position Surface Mount, Right Angle; Through Hole from Amphenol ICC (FCI). Pricing and Availability on millions of electronic components from Digi-Key Electronics.
Thanks @tough matrix. The connector I'm investigating looks to have similar-ish sized tabs, so we'll go with that assessment until proven oherwise. (No, I don't have the actual thing. I'm investigating a failure by remote 😦 )
anyone willing to review a Fusion360 library I'm making? I'm not done yet but just checking
EAGLE/F360 questions. How can I move the numbers 1-13 to a better position? They don't have anchors in the schematic. Additionally are the extra P$N from the footprint and how can I remove them?
For the second question, I think the pin symbol properties will have a checkbox for showing the pad numbers or not.
For the first question, I'm not sure if pin names can be "smashed" and moved around. One option is to turn off the display of the pin names and place extra labels manually. I'd tend to just suggest leaving the pin numbers where they are and making the connector box bigger to enclose them, though.
Cool thanks a lot I'll try those both
for caps, a voltage rating of 2x my applied voltage is sufficient, correct?
ceramic caps
Yup
Honestly, 6.3V are typically enough for 5V or less
But it never hurts to have more overhead
ok I see a bunch of 16V ones available in the capacitance I need
If that’s what’s available and 5V is your max, those will work great
It depends somewhat if you're using capacitors with a large voltage coefficient, and if you're depending on the value not changing.
For example if you're using Y5V dielectric 6.3V capacitors at 5V and you need the capacitance to be near the rated value, you may be disappointed.
I'm using 0.1uF 16V X7R caps to decouple 5V inputs to ICs and a 1uF cap of the same specs to do the same to my 5V power input
That will be totally fine.
kool
thx
does the 1 uF cap for my power input count as a decoupling cap or as a filter cap?
It's pretty much both, they're different ways of expressing the same general idea.
Yeah, I agree.
Do you have bulk capacitance on your input?
input to the PSU? Or to the board the PSU powers?
That's what my 1 uF was meant to be
unrelated, but how would folks go about marking a pin on a schematic as NC? I don't see a simple way of doing that in my scenario where I'm using the same IC twice but have several pins NC on only one of the two devices
I just don't hook anything to unused pins (hopefully outputs: leaving inputs floating can cause issues)
I typically drop a little "NC" label on the pin in the schematic to document my design intent.
Not to my knowledge. Other EDA tools like Altium do have a way to explicitly annotate a pin as unconnected, but I don't know of a way to do that in Eagle.
Mk thanks. That's what I gathered from googling
Naming them "NC" would tend to connect them all together to an "NC" net, which is not what you want to do.
Yeah
I have a probably silly question. I'm using one of these (https://www.digikey.com/en/products/detail/bourns-inc/4613X-101-103LF/3741136) and I made a part that looks like the attached image. Any signals I want to pull up need to be routed directly into pins 2-13, correct?
pin one is to my supply voltage
Yes, that's correct. Note that the schematic symbol does not make the function of this part obvious, though. If I were making it myself, I might have put pin 1 on the other side, labeled it "COM" or something, and labeled the remaining pins as "R[n]", etc.
Or, as bonus points, actually drew out the internal connections of the part with little resistor symbols inside the box.
Interesting. I made this one. I can't remember how I changed the pin names from P$N to just N
It would just be the NAME tool in the part schematic editor, I think.
Closer to how you'd do it?
Gotta figure out how to do resistor symbols in here
Iguess I can just use lines 12 times
Yeah. I mean, this is your design, so do what makes sense to you, of course.
I want it to look nice
This looks a little better:
don't like the placement of the numbers but what can you do
Cool, yeah, much more obvious what the part does.
they don't super look like resistors but they don't look like inductors either so that's good
@unreal flax Decided to use 2 6-channel resistor networks for routing reasons. This one had a lib for it. Much more pleasing drawing too
Then there's our nutty resistor network...
You can't tell but my face is doing the thing the bad guy's does at the end of Raiders
If you're curious, here's the physical part...
It reminds me somewhat of the gorgy mosaics you see in mosques.
On the upside, I got two patents out of it.
Heh, yeah, it does kinda look like a spiderweb.
You can see how adafruit does it, e.g.: https://www.adafruit.com/product/2652
basically the outer edge of the mounting pad is aligned with the edge of the board
Why didn't I just think to check adafruit boards.
Any idea if they can be opened in f360? I'm away from my PC for dinner or I'd try it. Buying a 🌯
Where is something like this used?
The basic notion is like an analog serial number.
Like you read the resistance and the unique value gives you your code?
Right. With the 8 ports, there are a lot of combinations I can use. I hook 'em up to GPIO pins, so I can ground some pins, connect some others to Vcc, and read the resulting voltage on another. With 8 pins, that's 15456 different combinations that can give useful readings.
Neat
I was originally asked to design a compact ohmmeter for the task, but then came up with the zero extra parts approach that gave many more available readings.
Woooo
only goof is I tied the LDO enable to VUSB and not 5V so the battery could turn it on as well
but this is why we do prototypes lol
I don't see the frames or similar library that I got used to using in EAGLE in Fusion360. Does anyone know where that went?
Looking to add a titleblock to my design
I found it. You have to enable it in the library manager.
Weird
yeah lots of stuff about Fusion 360 EDA is just odd
tl;dr: not only is your resistor network absurd, it's patently absurd.
What was the application of this network?
I replied above: #help-with-hw-design message
Ah okay, very cool
Wheeeee destroying board files to make Fritzing objects work properly. Good thing eagle2fritzing uses a copy of the board file....
I'm very sleepy and tend to second guess myself when so. Does this dimension seem like the distance between the center of the header and the center of a mounting hole?
I'm 99.9% sure it is but wanted to check
I want to say yes but can I have the full drawing view?
it is centred on both hole and header
That is correct then
what an odd way to measure though
usually you would do from the edge to center
ie like the 3.5mm
yup that's my experience as well, which is why I asked
chained dimensions are a thing but I think IIRC that I was taught to avoid them
Its pretty much useless. I mean the IO connector is there, but you would generally give from the left edge, then to the right edge of that connector OR you give the dimension of the connector it self.
🐑
opps
I meant 🤷
My favorite band is The Mountain Goats so I'll take the ram happily
lol
what are folks
oops
what are folks' thoughts on this placement of a barrel jack connector?
Let me get dims
It's ~6mm of overhang
You'll want to check the spec for the copper to board-edge distance for the leftmost hole, but the overhang itself shouldn't be very constrained. It might also be worth checking the connector drawing to see whether it has any bottom-surface features that would benefit from having a board underneath them for mechanical stability.
I've accidentally bought barrel jacks with that little nubbin on them before, let me see if this one has it
It doesn't seem to have any protrusions
Good callout though. I ended up moving it in slightly so less overhangs
lesson I seem to refuse to learn: Laying something out nice and neat in a schematic does not mean that your actual pads will be arranged in a helpful manner
@worldly schooner no rush but a while ago we discussed a pre-AC/DC converter filter. I can't find it in the log. Do you recall which one? I can dig deeper too, just doing several things at once
Google emc-20, I think?
Ahhh right thanks
I can't find just the ic for the adafruit AW9523 anywhere. anyone know why?
I guess I should have qualified that anywhere wasn't really anywhere. Thanks!
I've never ordered from them but they seem legit
Ah it's an assortment of vendors
I'm not sure the manufacturer has any direct sales (some do), but here's their site: https://www.awinic.com/en/index/pageview/catid/19/id/15.html
Breathing LED driverFounded in June 2008, Shanghai Awinic Technology Co., Ltd. is a high-tech company focusing on high-quality and high-performance IC design of analog, Digital-analog mixed signal, and RF for mobile phone, AI, IoT, automotive electronics, wearables and consumer electronics areas.
So is it sinking current for leds?
Yes, looks that way.
Kinda?
I expect it uses mosfets switching them on/off with high frequency, plus possibly a cap to smooth the current
I was thinks fets too
I'm guessing it's more like running them in the linear region.
But more with varying the gate voltage to vary rds on
That's what "running them in their linear region" means, yes.
Ahhh ok thanks. I'm still learning fets
Is it possible to have a custom board that runs off of a wall adapter (e.g. a barrel jack) but then has USB exposed (e.g. microusb) for programming in such a way power lines don't fight each other?
Would just connecting data and ground from the USB port work?
yes, that would work
of course it means that it won't work without the power supply
Which is what I want
I don't want anyone trying to power it off a limited USB supply
I don't think what I'm thinking of uses more than 500 mA but I don't want a low ceiling like that
Can anyone recommend an off brand crimp tool for criming 2 pin molex connectors?
I want to make cables that can connect to a wire to board housing but don't want to spend hundreds
which ones?
problem is, molex makes hundreds of connectors...
That's true. Let me do more research. Best friend asked me what I want for Christmas and the answer is always a new tool
https://www.molex.com/molex/products/family/minilock?parentKey=wire_to_board_connectors
The 2 conductor version should work fine.
Mini-Lock Wire-to-Board/Wire-to-Wire 2.50mm pitch system is a versatile and proven product with a broad range of features that include TPA, positive-lock and glow wire to meet varying industry needs
I could do a 12 conductor version I guess.... I do need 12 connections...
engineer is a well respected https://www.adafruit.com/product/350?gclid=CjwKCAiAtdGNBhAmEiwAWxGcUq-4JcNEbhRhMjkwQuvpgNTxtORI-6y6z_3Mu9oPu7L9g3P8kq0ioxoCexwQAvD_BwE
or some crimper from IWISS
That says 1-1.9mm is that referring to the pitch?
Or the actual size of the part you crimp down?
Their site implies that this one can be used for 2.5mm pitch
diameter of contact
Ah ok so I just need to figure out the diameter of the contact I want to use.
Google website of manufacturer, they have a table of all types of connectors and recommended crimpers for each
Engineer Co? Or molex?
Cool I was just on their site, it mentions at least one molex type so I'll just see if I can't use that molex
I've generally heard negative things about JLCPCB's assembly service. Has anyone used them and can give a positive or negative review? I can use another fab house, of course, but I'm familiar with JLC. I want to get assembly because I don't fancy hand soldering any WS2812s
I used them many times and was generally happy with their service
Ok cool. Thanks. I just heard some bad things (can't remember where) and didn't want to let that influence my decision too much
Can they do for example WS2812s on two sides of a board?
Ahh ok so I'll have to do some assembly on my own. Probably will just hand solder them all then.
how many boards do you need?
Undetermined. Probably not more than the minimum of 5
But I need RGB LEDs on both sides of the board
How does 2 sided assembly work? Do they have a heat proof jig that holds the bottom parts in place?
no, afaik, you do two rounds of reflow
Do you just hope that the solder paste doesn't melt and drop bottom components off?
first solder one side, then turn over and populate the other and put it in the oven
and bottom ones are held by surface tension of solder
but I have no first hand experience. Maybe others will comment
in your case, for small quantities, you could just use hot air reflow
It might be worth it to make a test board that's just a few rows of neopixels and try hand soldering them first and see if they light up
Yeah I need a reflow gun
talk to @distant raven
yeah, hot air or a hot plate for neopixel
Only issue is I want the board to be bigger than my hotplate. So hot air it is.
Ok that was my first instinct but wasn't sure how valid it was as a technique
So hot plate for one side and hot air for the other?
it's pretty valid
Yay, nay?
if it's two sided design, one side will have to be hot air'd
That's what I thought
i would do the neopixel side on the hotplate
Both sides have neopixels
I'm making a Christmas tree
Want the neopixels to look like lights
oh i see
yeah, i suppose just do the best you can
if you had a reflow oven, use higher temp paste on one side then low temp on the other
I may have to use hot air for the whole thing.
I'm planning a reflow oven build but it won't be ready in time for this.
They make samd51s with fewer pins, correct? I'm working on a design that doesn't need nearly the number of possible pins.
Well there don't seem to be any in stock anyways.
Are they just not findable right now?
But yeah, availability is zero
Yeah that seems tough to solder
Hot plate I guess
Any idea how many neopixels I can drive with 16 MB flash?
Well I got the last two of these https://www.digikey.com/en/products/detail/raspberry-pi/SC0914-7/14306009?s=N4IgTCBcDa4GwBYAcBaAygYQAwE4CMCAFAOwCUGAKigHIAiIAugL5A
That's kind of apples and oranges. Neopixels are usually driven from RAM, not flash, unless you intend to have a completely static pattern.
Ahh ok.
Let me make sure I didn't get the wrong thing, to go along with SC0914(7) there's also a SC0914(13)
Anyone know the difference? They seem identical
I think it's just the reel size. TR7 is 500/reel, and TR13 is 3400/reel.
Ahhh perfect
May as well buy the other two remaining
(Sorry to anyone who wanted one)
All's fair in love, war, and chip shortages...
Yeah I can make at least one of my upcoming work projects work with these and maybe another project
Hi!
Noob pcb+assembly ordering question:
Why my pbc component preview looks incorrect - component placement looks like it trying to place stuff on bottom layer where i have nothing but copper pour? The pcb house is jlcpcb.
this is how it detect gerber files:
this is how it detects assembly:
and this is on final page before saving order into cart:
you can see there is no almost no traces here
after closer look i no longer sure this isnt a bottom layer at all - this looks like mix of 2 layers
I think i read that JLC is sometimes wonky but I've never used any assembly services
how do you look at middle layers of pcb in jlcpcb?
their viewer has only top and bottom layer tab
I'm not sure. It's been a while since I did a 4 layer with them. I think it just showed up but I don't remember specifically
JLCPCB assembly service orders are always checked by a human before putting to production, so many problems will be caught at that stage. On the other hand, they only have a very short time to spend on each board, so they can miss some problems...
44$ for two boards - not that high level of risk
(including shipping)
ordered 💸
lets see what they would say
They've reached out to me numerous times about my art projects being super weird PCBs.
Also their visual representations of your CAD are imperfect.
Hoping for some advice on how to route my 2 layer screw terminal (pictured). I have 6 unique signals and 6 power lines (each signal has a power line associated with it). Would folks do all power signals on one level and associated signal on the other level? Or alternate left to right across the top/bottom?
also what's the best way to make a really nice and neat library part like this? I want to make a part that's just a 2x20 with 4 holes in the exact positions I want so that I don't have to position my holes in the actual PCB anymore.
Okay so I’m really scratching my head here with this transimpedance op amp design. I’ve used this before in a similar application but with higher supply voltage without issue.
It’s from an OSH VLC transceiver that I’m adapting to another project which I can’t really share details
When I prove the feedback side of C19, I get the expected variation in voltage when the input to my photodiode is changed (more light/less light)
But opposite of C19 where it’s pulled up ~1.65 voltages by the R17 and R15 voltage divider is where I am losing the signal
I wouldn’t think using 3.3V vs 5V would make such a difference but at this point I am a little miffed as to what’s going on
What modulation frequency are you using to test with?
Ah. Well, the capacitor is going to block the DC response, so you really should test with an AC-coupled signal.
Sure, but logging the analog input to the Micro I should see more than just 1mV differential when logging at 60Hz+
weird Fusion360 EDA issue. I'm making a library part, but when changing coordinates, it's rejecting my pressing of the - key.
I have it logging ever ms
But it only varies 1mV when I should be seeing a 150-300mV change
This photodiode is sensitive enough to pick up the 60Hz noise of household lighting. I don’t even see that
I don't have a good intuition for what the cutoff frequency is for the path out to the ADC, but I'd assume they would have set it fairly high for typical communication applications.
If I do this same measurement with my 5V application I can measure the response of 250mV into the ADC
I thought maybe it was the supply voltage being too low but acceptable range is 3.1-5.5V
I’m wondering if I need a weaker voltage divider
I tried to see if I changed the bias resistor for the inverting input would change my outcomes but that was a dead end
Im wondering if I need to just use the 5V design with off chip ADC
Cutoff frequency is 4.24MHz
0.5Msps should be able to handle that but maybe not
Apparently it's a known bug fyi
Won't accept a negative unless there's a number in there already
We’re shooting for fairly low data rates but perhaps I need to re-evaluate my design
I'd send you my old 10Mbps design for reference, but alas it's all behind NDA, heh heh.
Lol
All good. This op amp can support 4GHz GBW which is super nice considering it’s $13
A few hundred Kbps should be a breeze
4.244MHz cutoff is supposed to support up to 400kbps so some fine tuning (changing feedback resistor) should easily bump to 1Mbps
Anyway, I think maybe I just need to go back to a standalone ADC
I'd run the power to one connector/level and the signal to the other: that would likely be the easiest to route. But it's hardly the only way you could do it.
Best? You can just knock it out manually, or you can write a program to generate the library part automatically, or write a script to do the same. Each is best in different circumstances.
Hmmm thanks. I'm doing it manually for now. I'd like to learn to write scripts to do this.
Maybe when things calm down I can learn that
Is it possible to add holes to a footprint in Fusion360?
just plain holes, not pads
I can do a PTH with drill=diameter I suppose
I got it
cmd line hole
There should be a separate hole tool as well, rather than the via tool.
It doesn't show up in my toolbar, or I'm too sleepy to see it, but running just hole worked great
any idea why my pins on my schematic all have G$1.# on them? I changed all the names to just be #. That is reflected in my pad names correctly.
The G$1 is the placeholder name for the schematic symbol. A library device can have multiple symbols, like a quad op-amp chip where there are 4 individual amplifier symbols.
Ahhhh ok. That makes sense. Thanks
(I think G stands for "gate".)
nice
That's what I was thinking as well. Thanks for the input
Update on pcb - they accepted it without asking me any more info
only corrected some placement - but what they corrected is beyond my comparison skill
How hard is it to solder an SD card slot onto a board? I'm thinking it's easier to just buy the spi breakout from adafruit but that feels a little like cheating
Uh? I cannot use /shrug?
I guess not
I’ve been trying to get some of these for months
Found 30 on LCSC and snagged them
Nice looks at roughly 70 RP2040 left from 500 piece reel
To be fair, I sold over half the reel to makers in need
Would these marks on the footprint correspond to the dots on the physical device do you think?
Highly doubt it.
It’s not rotationally symmetric, is it?
How do you mean? It's def not axi symmetric
Seems like it’s only possible to put it in one way.
Ah good pt
e.g. the orientation markings tend to be most useful for parts that you could rotate some angle and it’d still fit
I can’t quite tell based on the drawing - it’s possible this one still is, but then also it doesn’t much matter, except to see which direction the interlocking features face if you want to solder in just one, then interlock and solder another one later (which seems a bit obscure)
ah ok. So any ideas what the two markings on the footprint mean?
I think they mark the "front" of the device. Kinda confusing tbh but from the part drawing it looks like the wider section is in the back
I can't seem to get this stemma QT footprint closer than this to the edge of the board. Anyone have an idea as to why?
The red box is tKeepout, is that hte issue?
It shouldn't be. I suspect it's how close the copper pad is to the edge. You might check your DRC settings to see what that clearance is. Maybe 1mm?
ahh ok.
I'll look at some adafruit boards and see how close their stemma QT connectors go
Theirs are pretty darn close to the edge. Thanks @unreal flax, I'll examine my DRU. Is there any danger in changing the settings?
Just depends on what your PCB fabrication house can support.
ah ok
I'll see what JLC can do. What's the relevant metric called? Like i can describe it but it probably has some technical name
Looks like they call it "trace to outline" clearance. Only 0.2mm, so you can scoot things a lot closer if you need to.
They may have an Eagle spec file that you can just download.
I've had a really hard time getting files like that from JLC. 5% of the time it just downloads but 95% of the time it just opens a webpage with a text file in it
also any idea what the metric would be called in Fusion360?
Looks like it's called "Copper/Dimension" in the Distance tab.
ahhhh! I should have realized that's what that diagram meant.
It was confusing because the default board color in the actual CAD program is black, but here it's white!
Thanks!
is 8 mil small enough for traces between low power ICs?
e.g between a GPIO and the gate of a MOSFET?
or even 6 mil?
Yes, that should be fine for low-current signals.
Whatever your PCB house is happy with.
Also should be fine. If you're curious, there are various trace-width calculator tools online which will tell you how much heating you'd expect from different widths with different currents.
Yeah I wasn't sure of the current in my smol traces or I'd have gone to those. I confirmed that 10 mil is enough for 20 mA
I'm considering the CircuitBrains Deluxe board because it's the only way to get a SAMD51 and it simplifies my design work. Since the Deluxe has QSPI Flash on it, does that mean that the SPI bus is busy and I can't use it for an I2S chip?
afaict, the SAMD51 only has one SPI bus. But I'm reading about the I2S functionality
For that chip, they have a dedicated QSPI peripheral, plus up to 8 SERCOM peripherals that can be configured as regular SPI, plus a separate I2S dual-channel peripheral, it looks like.
interesting. Was that in the datasheet and how did you know which table to go to?
Yeah, I was just looking at the initial "features" section of the datasheet.
ahh ok
I see it now. Thanks
seems like this is a good board. I have been pressuring my boss to buy a few. I think I'll point out that if we don't use them, I will buy them from work because I want them lol
Although you'll want to check the particular chip variant against the Configuration table on page 17, as not all D51 variants have I2S, it seems.
let me check then
looks like the J19 does have I2S
good looking out
if I don't want it to be possible to power my device via USB, but I do want some indication that USB is connected, can I just connect the 5V line from the USB port directly to an indicator LED?
Yes, if "directly" also includes a current-limiting resistor in series.
yes of course haha, I just left that off
And then as long as I have common ground between all my DC components, D- and D+ will behave?
Not quite sure what you're asking there, can you clarify "behave"?
Like I can transfer data over USB even if I'm neglecting to use the power from USB for anything but an LED. As long as I provide external power to the chip
Yep, totally. From USB's perspective, it has no idea if that trickle of power for the LED is actually enough to drive your MCU or not.
how can I make covered (not tented) vias in F360? My friend suggested it and I can't find the setting
I'm not sure whether that's something which is done in F360 or done at the fabricator in their setup. The copper in the Gerber file already extends over the via normally, I think, so it's just a matter of telling them that the vias needs to be filled and covered.
"Via in pad" is the usual terminology for that technique.
(Assuming that I'm interpreting your question correctly.)
I think what my friend meant was mask fully covering the via?
I can ask again
Definitely not via in pad
Oh, mask, not copper? I think that would be in the DRC settings. There's one for the via-mask clearance, probably.
Yeah, seems to be in the "masks" tab. You can set a limit such that through-holes smaller than some size don't have a mask clearance at all, while larger ones do.
Would I make both of these 0?
No, that didn't seem to do anything
Limit is the field you’re looking for
Just adjust it until the via are tented in manufacturing view
Ahhhhh nice thanks
Pretty proud of this board. Sure it's physically bigger than my last one, but that's just so I can use the mounting holes. But I didn't autoroute a bit of it!
And I used less space on this one than my last one, I think
Or at least used it more effectively
hmm @distant raven I took it all the way down to 5 mil in increments of 5, and it doesn't seem to affect the manufacturing view.
no luck. here's an example of what I see. Doesn't seem different than before.
Huh
Holes are .35mm
can you get custom female headers cut? I'm real bad at cutting them myself
Yes, Digi-Key will do that as a "value-added item" when you order a breakaway header of a particular size.
i'm working on a low power project
and wanted to add battery voltage sensing, and was reminded of that using a voltage divider between vbat and ground puts a steady drain on the battery.
doing some searching, a mosfet disconnecting vbat from the vmon divider would allow the micro to turn sensing on and off.
i then stumbled upon this schematic from the Nordic Thing, and have some questions about whats going on
its using both a p and n channel mosfet to control voltage to the vmon divider
why two and not just one mosfet?
Can this stepper motor driver (https://www.adafruit.com/product/2448) be used with other steppers? And if so how can I determine compatibility?
Spin two DC motors, step one bi-polar or uni-polar stepper, or fire off two solenoids with 1.2A per channel using the TB6612. These are perhaps better known as "the drivers in our ...
@snow jackalYou could technically use one P-FET, and wire the gate of the PFET directly to your microcontroller if you knew these 2 things:
- your inputs were 4.2V tolerant, and didn't leak enough that they'd overwhelm a 1M shunt resistor and pull the line low. (having the second FET means that you can specifically purchase an ultra-low-leakage option)
- you could set the output to open-drain configuration (effectively mimicking the N-FET shown here) - if you only had push-pull available, depending on your high-side output voltage and the VLi-Ion, you could either never turn the battery sensing off when your MCU is powered (say you had a really low voltage MCU - 1.8V - it could only drive 1.8V to the P-FET gate, the source would be up to 4.2V, so the Vgs would be sufficient to turn the FET on).
Also a word to the wise if you're going to do this: check the input impedance of the ADCs you're going to use resistors this huge, and add some cap. This voltage divider has a combined thevenin impedance of 160K - so imagine the "correct ADC voltage" with a 160K resistor in series with it heading into your ADC. If your ADC has, say, a 100K static input impedance, the actual read ADC value would read out as only 38% of what the correct reading would be. Generally the actual input current of the ADCs will be spiky, so a cap can buffer that out to have vagely correct readings. But even more so: if you're adding the FET switching circuitry anyway, you can use way smaller values of resistors (which will make the effective source impedance less) without facing the static leakage penalty.
can the 2 circled parts be combined into one?
Likely yes, though it'd be nice to have the cap near the input of the LDO instead of near the connector. Also FYI -it seems like you've double-named the top net VBUS and +5V by placing two power port indicators.
i actually combined a qt py and a powerboost1000c and they have the 5v traces named different things
find-replace is your friend here
The net can't have 2 names at once in the final design, so best option is your ERC complains loudly at you, worst option is half the things just don't get connected
i technically already did in a previous version, but i wanted to trim things up and kinda started over. let me see if i can remember where i placed it
Not a huge deal - just something to get sorted out before you head into layout (and worth putting a TODO on your schematic as a reminder to yourself if you aren't going to fix it immediately otherwise it's way too easy to forget - ask me how I know)
yeah, its right in the original
here are the 2 main components in my original layout, all labeled correctly
i figured that since the usbc and the voltage shifter(not sure if thats the right name) share the same 5v, cap, gnd layout, i could eliminate a capacitor
This seems like it'd be fine; if anything I'd shift the cap over towards the LDO so it's also not getting in the way of the track currently snaking through.
remind me what ldo is please
Low-dropout regulator (the 5V -> 3.3V linear regulator)
ok. so if i remove one capacitor, i should put the remaining one near the ldo
im moving the ldo closer to the connector so it shouldnt be an issue
Might want to double the capacitance of the remaining one, no?
Meh - just 10uF total is probably fine, especially if you're choosing a good cap.
Something with 2x+ rated voltage?
That's part of it; I'm spoiled and also don't spec less than X7R dielectric if I can avoid it.
I do the same, with other people's money. But I don't design anything that gets made at scale
More so in this context - if you need a bulk cap for larger power smoothing on the 5V rail, it's probably worth specifying a separate bulk cap specifically for that purpose
ah thanks for the writeup!
that answers a bit on the other oddity to the design of the 1M resistors
for my circuit i'm using 3.3v logic and nimh cells, i was trying to figure out the reasoning in their design
So - I think you meant to use a P-channel MOSFET here, but you used an N-channel MOSFET
so it will never be able to truly turn off (note the body diode going from 3.3 to R3)
whoops
i take it i want to pull the gate up on the p-channel then to have it default to closed?
Correct - you want the default to be 0V Vgs, so the FET is off.
Then by pulling the gate low, you get -3.3V Vgs, which turns the FET on (as long as you have a logic-level FET)
Yep! Worth noting - your current through the 10K now completely overwhelms the current in your sense divider when you have it turned on
Depending on your total duty cycle - this might still be acceptable for your application, but if so I'd advise lowering the resistances of your divider (see above about ADC input impedance)
i hadn't considered either ADC impedance
or the affect that pull up resistors being might be having on the current draw when being driven low by the mcu, and the change in voltage that might have on the divider
thanks!
whats decent software to plan a pcb free?
nothing super complicated, but not terrible
i've been using KiCad to much success
has a bit of learning curve though
most of the pcb software has some builtin part lists
common "footprints" of standardized part sizes
perfect
kicad is a multistep process
schematic with symbols -> assign part footprints to symbols -> layout and path circuits on pcb
looks like you can open the kicad pcb editor and just place footprints and draw circuits
but that might not be a good idea depending on the complexity of the design
If you decide you want to spend money, EAGLE is what adafruit uses. Autodesk is slowly merging eagle into Fusion360. Fusion360 is great because you can do solid modeling as well. You can make a 3d model of your PCB with components on it and add it to your assembly
The only things I don't really like about F360 is the automatic rat's nest and the cloud of it all. I liked to backup my EAGLE projects with Github
For fab houses, if you want to buy American, OSHPark is a great choice. If you don't care to have the quality of OSH or don't care about buying American, PCBWay and JLCPCB are great. Not an exhaustive list.
what?
To have the board made
Interesting.
i needed to buy baby oil or something too
I'm mildly confused.
bro me too!
Are you gonna etch your own board on copper?
Im a python programmer
uh
i think im gonna use a flashlight on a piece of copper
and then nail polish off the ink
Interesting. What are you gonna use kicad for then?
gimme a sec, ill find the video again
apparently I cant swear here
it looked very clean
i hope to use kicad to generate the image that I can use to print onto paper
Learn to fabricate Single Sided PCBs at home! Prototype PCB fabrication is one of the essential skills you must learn in electronics. Instead of using breadboards and perfboards, custom PCBs would make any project smaller and more compact!
PRESENSITIZED PCBs:
Also known as photopositive PCBs or photoresist PCBs, are regular PCB copper clads, l...
Ahh. Tbh this sounds like a pretty convoluted and difficult way to get a PCB but I def hope it works and you have fun
Id love to hear your feedback
I'm no expert, not even close, but if you're just having fun I have no qualms
If you were designing a product for work I'd advise against this method
getting simple 2 layer PCBs printed is pretty cheap, probably less cost than the supplies.
but the DIY approach will be very fun!
It's the way a friend of mine (in his 70s) did it. Except they had to draw the traces. No CAD
If you open up an old enough electronic device you'll see hand drawn traces.
We're spoiled by CAD and library services.
i've seen them! traces drawn that way can be very organic looking
Hope it goes well! It's a cool method
If you don't paint your nails, maybe start so you can get practice with nail polish application lol
if I only have a single sided pcb.... how do i hop over another line?
drill through it and use a lead?
Yes, it's called a "jumper".
If it's worth anything, your ADC doesn't have power or data connections.
AISLER is a good middle ground for what I've seen.
- cheaper on bigger boards (compared to OSHP)
- allows to pubblish boards
- fabs in the US as well, otherwise in DE
- supports Kicad out-of-the-box (allows to donate to them at checkout as well).
- pricing is "open source" as well (https://aisler.net/help/getting-started/pricing)
- their site on github (https://github.com/AislerHQ).
they only offer two or quad layer, ENIG or HAL, with no exotic features like flex PCBs, but I think for prototyping is good enough
(errata: US is the country of origin for their 2-layer ENIG boards, not the fab location 😅 )
@worldly schooner I couldnt find a device that has the same layout of my lcd driver, it's an 12c line
What device?
I was just using something in place to show that yes, those pins are being used
standby
just needed something to emulate the J1 connector
which of these footprints is better? (same device, found in different libraries)
I expect it is QFN?
I'd go with the second one
LGA16
left one is copied from adafruit
I tried the one on the right recently and had a poor success rate, wondering if adafruit had a good reason to think the shorter pads are better
Also @cedar salmon I've generally had it recommended to me not to power an LED from a gpio pin. It will work it just isn't the best practice.
Oh looks like you fixed it
how do i enable it otherwise?
Mosfet
Is this board 3.3V only?
Expand your project possibilities, with the Adafruit AW9523 GPIO Expander and LED Driver Breakout - a cute and powerful I2C expander with a lot of tricks up its sleeve.GPIO expanders work ...
I want to drive my LEDs at 5V, is that possible with this board?
Somehow I thought there wasn't a learn guide. I don't have my glasses on (working nights so this is my morning)
Thank you.
what LED's need 5v?
I can level shift the i2c signals.
since it includes constant-current driver, it will provide exactly enough voltage to each LED to maintain required current - so it doesn't really matter if supply voltage is 5v or 3.3v
Ones that are 30+ feet from the board. I just don't want to have to worry about voltage drop in the lines.
I'd hate to spend a bunch of time designing the board and then realize I have too long of cables. I don't think it's an issue but I'd rather avoid it entirely.
Thanks @tough matrix @unique patio
Does it have boost circuitry in it?
no, as far as I see, so indeed, for long runs of the cable, 3.3v might just not be enough to produce sufficient current
you are right
Thx
It may be the difference between a footprint for hand-soldering (where you'd want the extended pads to be able to get an iron on them) versus reflow.
A friend of mine who is consulting for us reviewed my PCB and he's really against my using a giant single signal power plane (+5V). In one instance he's against it because it means there isn't a "direct" connection between my decoupling caps and my ICs. However, electrically it doesn't seem like there's a difference between a short trace and a short travel across a polygon. Am I wrong?
Well, is the power plane on the same layer as your components?
I’d argue that it’s okay to have your power planes as inner layers if you have 4+ layers
I would just use traces of its 2 layers. With the exception of powering an array of LEDs since a mostly continuous plane will deliver current better than a trace.
You still would want to have some of your decoupling capacitors adjacent to your ICs
and then via out from the inner plane to the outer layer and a small trace that runs to your decoupling cap and the component
(at least for the ~0.1uF class decoupling caps - any bulk higher-ESR caps are fine to put further away)
Ah I have only 2 layers
Next revision will be 4 but I have to get this one done soon
I know this has got to be possible, and I'm super over using the group tool and clicking dozens of traces. There has to be a way to globally change all traces under a certain width to a new width, in Fusion360. Anyone have a tip?
A plane is a better (lower resistance and inductance) connection than a trace, it also adds useful distributed capacitance.
hmm that's opposite of the advice I've gotten elsewhere. I'm kind of confused as to who to believe. I've already re-worked my board to use traces to nearby planes for my decoupling caps. We'll see who's right if it doesn't work.
A lot of people don't really comprehend the basics. These folks work in industrial lightning protection, for things (like radio transmitter towers) that expect to be struck by lightning while in operation, and are intended to keep operating. They have to get their physics right or equipment and people can be damaged. They have some good explanations on why flat wide conductors are good low impedance routes. Sure, you aren't dealing with lightning, but the same sorts of considerations apply to fast spikes and decoupling capacitors. https://www.polyphaser.com/News/DownloadFile?downloadGuid=3676caa6-581e-40ed-b25a-590bad89ec80
Thanks I'll give that a read.
Hm, I must be missing it, but I didn't see anything in the white paper that directly relates to the decoupling cap thing.
It was my intuition that a broad connection using a plane is best but I have had a number of EEs tell me that that's incorrect.
I know plenty of people with EEs that couldn’t design a PCB
These are PCB designers by trade
Here is what was recommended to me. This is a cap connected to ground and the nearby 5V plane.
Well, doing something professionally doesn’t always mean you’re following best practices
with 20 mil traces
What’s the thing to the right of the cap?
a 5V plane
Huh…
I guess the idea is that there should be 1 and only 1 path from the plane to the terminal of the cap and then to the power pin of the IC
I would generally agree that would be the case even with a solid plane
that was my intuition as well but I didn't have the extensive electrical education these people had
There’s only one entry to the 5V end of the cap
nor practical experience
That’s the 5V pad
The more critical thing here is closeness to the IC it’s decoupling
yeah everything is <1mm away. Basically as close as I can get it and still be able to solder
Then the plane should be fine
power distribution network
Thats what I guessed, thx. I get why technical articles are written for a certain audience but the phantom initialisms can be annoying
What is the best resource to get started with PCB design?
Have experience in CAD
The Art of Electronics is well recommended. I have purchased it but haven't read it yet. It may not cover PCBs but it will help you with components
a battery mounting board with hookups for a volt meter (with momentary micro switch,) toggle switch (love clickies,) and a 2.1 barrel port. designed to fit in an altoids tin. with a barrel jack to allegator clip, you can quickly plug it into any 3.7v project
how does it look?
so, does anyone have experience with aisler for pcb fab? their boards are cheap, 16usd for 3 budget boards
not me.
But it seems comparable in price to OshPark, which a lot of people - myslef included - have used a lot.
same pcb on osh was 33usd
sorry. 37
gerber labs cant even upload the gerber files to tell me how much it is
as for the board design:
- I would use thicker traces for power. You have lots of room to spare - why not make the trace from positive pin of barrel connector 40mils?
- usually the footprint for barrel connector has slots rather than round holes - the "pins" are commonly flat contacts
is there a reason why you use round holes in this case?
a, the closest i could find, b im using a breadboard barrel port
in eagle?
yes
all the traces or just those to the connector
I am not using Eagle, but I am sure there shoudl be slotted footprints for barrel terminals somewhere there
these are the only 2 and only one of them is through hole
all traces that carry main current to/from battery
(but not necessary e.g. for traces to voltmeter)
I'd get footprint eslewhere then. E.g. https://www.snapeda.com/parts/PJ-002A/CUI Devices/view-part/?ref=digikey
Download schematic symbols, PCB footprints, 3D Models, pinout & datasheet for the PJ-002A by CUI Devices. Exports to OrCAD, Allegro, Altium, PADS, Eagle, KiCad & Pulsonix.
would a fill work better than a trace on this board?
probably
also: in two places, your traces are too close to another through-hole pad
at the switch SJ1 and SJ2
you have room, move it farther away
@tough matrixhows that
apparently the distance between all traces and pads is 6mil
i did not set that up. that is default in eagle
6mil should be ok with most fab houses, but there is still (small) chance of a short. given that you have lots of space, I'd set the separation distance to bet at least 8 mil, better 10 mil.
6/6 is usually what I’ll do
I’ve done 5/5 with OshPark successfully many times
And I usually do 10mil holes with 5mil ring
It’s the trick for smöl designs
So I'm looking at frequency vs impedance plots for the decoupling caps I'm considering. I'm working in the 100 kHz range. All the caps I've looked at have really high impedance (relatively at least) in that range. Yet, people are telling me it doesn't matter??
100kHz noise, or 100kHz signal?
Signal
No idea about noise
But if the noise is lower frequency that's even higher impedance
The noise frequency is generally the switching frequency of any switching converters that power your device.
That's the range your decoupling caps ideally have lower impedance.
You usually don't put a decoupling cap directly on a signal line, so that shouldn't matter too much?
Capacitors generally have lower impedance at higher frequency. At 100kHz, a 1000pF capacitor will have about 1600Ω of impedance. I don't know if that's "really high" to you, or you're looking at even smaller capacitors.
I was wrong about the freq of the signal I'm decoupling. I'm decoupling the frequency of the PSU not the SPI bus
Hmm. If it's an LT1170 based PSU, that frequency would be around 100kHz. 🙂
Okay… so I’m looking at the various versions of the MCP73831T in the SOT23-5 package. And I’m trying to figure out why there is none of one variant vs the other
I'm currently trying to figure out how to get an old TV to do what I want but my next step is to figure out the freq of the switching supply
For instance, the MCP73831T-ACI/OT has 34k in stock on Digi-Key
But the ATI/OT variant has 0
The difference in them isn’t all that much
Based on this table here you wouldn’t really tell the difference between using one over the other
One might not be as great as power consumption when it’s not charging but that seems to be the biggest difference
7.5 vs 20 is almost a factor of 3, but I don't know what those terms are, offhand
Ah, charge termination current ratio, in percent. Huh.
So ideally for the average product this won’t matter that much
Obviously the AT variant has a better ratio but if you need to continue to make a product with the part, the AC variant would be a perfectly acceptable replacement
I wonder if those are basically the same dice, but binned by performance, and they just didn't get many of that bin in the last few runs.
Hence my wonderment
I might just snag a bunch
Not a bad idea, all the grades seem like they'd be fairdinkum chips
34k in stock at DigiKey vs waiting 2 weeks to get the same but over priced for LCSC of the AT variety
They are charging $0.40 over singleton value
And $0.04 over multiples
I’m also wondering if LCSC ordered all of the TI chips
If they're seeing an ongoing demand for them, it would be a smart move on their part.
Because there are some TI regulators I can only get on LCSC but they have 2022/2023 lead times elsewhere
My go to 3.3V 500mA LDO I’ve only been able to get from LCSC
which one is that?
Are LDOs hard to get atm?
3.3V ones, yes
my favorite LDO, AZ1117, is widely available: https://lcsc.com/product-detail/Linear-Voltage-Regulators-LDO_Diodes-Incorporated-AZ1117H-3-3TRG1_C110474.html
Diodes Incorporated Diodes Incorporated AZ1117H-3.3TRG1 US$0.1303
LCSC electronic components online Power Management ICs Linear Voltage Regulators (LDO)
- leaded datasheet+inventory and pricing
I found this part with the Digi-Key mobile app. https://www.digikey.com/product-detail/en/RT9080-33GJ5/1028-1509-1-ND/6161657
Order today, ships today. RT9080-33GJ5 – Linear Voltage Regulator IC Positive Fixed 1 Output 600mA TSOT-23-5 from Richtek USA Inc.. Pricing and Availability on millions of electronic components from Digi-Key Electronics.
it is larger than most, but also provides up to 800mA
I don't design anything to be produced at scale, so I don't mind if something is a little more expensive
Wow the digikey mobile app actually works for you?
Yeah
it simply does not function on my galaxy S21 Ultra
The iOS app is pretty stable
Looked it up, it's 65kHz.... how can I possibly source a suitable bypass cap for this supply with such a low frequency? Don't I want as low of an impedance for my bypass cap as possible?
Honestly, 0.1uF is a very good value for filtering in terms of response/impedance.
hmm but the impedance is at or near it's highest in that range..
You could use anywhere from 0.1uF to 1uF and be perfectly fine
Maybe I'm giving too much thought to this
I was planning on having 0.1 and 1 uF caps
Problem is, the more capacitance, the slower response of the power line
With power filtering, you balance speed and impedance which is why filtering ranges between 0.1uF and 1uF.
ah ok
Some designs will use 4.7uF and 10uF to decouple lower frequencies and smooth power up
huh it seems like 10 uF is better from an impedance standpoint. I assume it leads to lower rise time in the PSU?
Pulled this image from a stack exchange answer
Correct
Which might not be desirable for some microcontrollers and devices
oops I'm in re-routing eternal punishment now
I'll just use 0.1 uFs and do more reading about the issue.
Lol
0.1uF is just a very common and good value to use
If you look across lots of designs, it just works really well and is a good balance
what do you think about the issue of, on a two layer board, connecting your decoupling caps using the power pour vs connecting them to the power pin with a trace from the power pour to the cap then to the IC power pin? my friend is insistent that the latter is best practice and demands I find a source saying otherwise. Notably he hasn't provided a source besides him.
I haven't been able to find a source saying either way is best. More copper between the cap terminal and the IC pin seems like it would just lower the impedance of the path.
Oh I totally misread the plot. Is the black one both 10 uF tantalum and .1 uF ceramic?
no, I didn't read it wrong
In terms of the power plane question, power plane delivery is actually much better
Traces are current limited and offer much less EMI shielding/decoupling than solid pours.
this is my argument but he has a master's degree and I don't
On 4+ layer boards especially, power planes are basically vital for power delivery
That's a pretty high frequency, easy to filter.
I don't have a master's degree either, but you're right and he's wrong.
The way I see it, and this is a view from wiring houses. It’s better to have more capacity than you need and never use it, than not have enough and need it.
Madbodger has made a few great points too about impedance and EMI
yeah I'll just do my thing and Live, Laugh, Love.
ah mine is 2 layer, does that matter? I notice adafruit doesn't use big pours super often
I'm looking to put one of these (https://www.cui.com/product/resource/emc-20.pdf) before one of these(https://www.cui.com/product/resource/psk-s25d-t.pdf) but I'm confused about one aspect. What do I do with the ground pin on the PSK?
just leave it floating?
or do I need to loop it back to the ground pin on the EMC-20?
Thank you for watching. More information about me: https://feranec.com/wiki/Robert_Feranec
Would you like to support me? It's simple:
- Bitcoin donation: bc1qy3pq0mthrhw9gtdxh5qv0jcc4wxes4e32ckd39
- Ethereum donation: 0xf8C68847b7fC6e904Bc948991580f8a8A5DD6b1D
- Sign up for my Hardware design and PCB Layout online courses: https://academy.fedev...
This guy has a lot of great videos and tips on PCB design
Ground goes to ground
Is that in reference to this?
I was hoping to avoid making an AC power PCB.
Well, grounds by nature should be connected
Ah I missed this on the first pass thru the datasheet
so I'll have to make a small PCB to bring these together I guess.
You want to maintain common ground in your design to ensure you complete the circuit
that makes sense, IDK why I thought it would be otherwise.
Any safety considerations with a small board that just has like a 3 pin terminal block with only grounds connected to it?
I'm thinking ground of the AC/DC converter, Ground of the EMC and ground of the AC power cord all in a 3 position terminal block.
Use a ground block instead
Learn about StructuredGround™ UGB2/0-414-6 Universal Grounding Bars and other Grounding Busbars & Strips. Contact Panduit today.
Certainly cheaper
Your stuff is din rail mounted so I assume its going in a metal panel
You can buy them at homedepot, lowes, etc
It's not din rail mounted and it's going in a plastic enclosure so no real chassis ground
Will the enclosure have a sub plate?
It will look neater 😉 lol
It doesn't seem like it has room for that
You mount those things on din rail, mount the ground block to the panel
I mean it could float around as long as your grounds all go to one place
I already have the chassis mount version so I'm a bit locked in there.
bam!
We use those on our panels work
Not sure where to ground the block to. I guess it'll go out thru the AC power cable to earth ground.
earth ground
when power comes in, we have ground go direct to that block
then your other grounds connect the same way
interesting. I'd then stick it into the earth right?
I'll need to find a copper rod. Maybe I can weld a wire to the rod
I guess a bit of copper pipe would work.
what about shock potential with that big un-insulated ground bar?
Yes, ground rod is generally earth but you may want to read up on the requirements for that
Its more electrical design
But Ive seen 3ft poles hammered into the ground. Theres usually a screw at the end
https://www.homedepot.com/p/Southwire-5-8-in-Grounding-Rod-or-1-2-in-Rebar-Ground-Rod-Clamp-for-10-SOL-STR-2-STR-Wire-65176440/312648487
https://www.homedepot.com/p/ERICO-5-8-in-x-8-ft-Copper-Ground-Rod-615880UPC/202195738
Southwire ground rod clamps are used for connecting conductors to ground rods or rebar. This copper alloy clamp is suitable for direct burial. This clamp is specified for use with #10 SOL/STR - #2 STR
Oh you probably want this one
https://www.homedepot.com/p/Blackburn-1-2-in-Type-G-Ground-Rod-Clamp-for-10-SOL-to-2-STR-Wire-G5-B1-30/202907599
thx
what do you mean by this? More of an electrician thing?
yes
kool thanks, I'll look into it. Do you mean local code?
Or general code
local codes are stricter than the state codes, so I'll use those. Thanks again
pg 8 has the selection guide for trimming resistors. I ended up getting an 80k 1/2W resistor but I'm not sure if I should get a higher wattage. Anyone have an opinion? https://www.cui.com/product/resource/psk-s25d-t.pdf
What's your output voltage? You'd need like 200V to hit 0.5W through an 80k resistor.
output is 5V but I don't know what the voltage on the TRIM pin is
I presume it's supposed to be Vref, since that's what it would be in the absence of the trim resistor. You shouldn't have any trouble with resistor wattages... this is a low-power feedback circuit.
that's what I thought just figured I'd check thx
What do I need to google to figure out which pins on this connector are power and ground?
It looks like the pins are just a straight connection to the USB cable wires, so Googling the USB-A standard cable pinout should set you straight. As I recall the middle pair are D+/D- and VBUS/GND are at the ends, but you'll want to verify that and ensure you have the correct orientation.
yeah I don't want to put Power to Ground and vis versa Thx
why might a USB A library part have this big keepout area on it? Is it something I can safely remove? It's very inconvenient.
It may be designed for a connector that is that physical size
Hmm one sec let me get the part
does the mating part not just go inside the receptacle?
It does, but as you can see, USB-A receptacles are chonky
yeah I'm just confused by the huge keepout. I can get around it but it's odd
Like do I need to have the keepout fully on the board or can it be sticking off like it is?
Another thing. I'm making a diode library part but because of the size of the symbol, in the symbol part of the library, the names of the symbols are in the "wrong" place. It's quite misleading. Is there a way to move the name around? There must be but I never learned it.
it appears as if K is A and A is K in this configuration.
Fusion360 though it's probably the same in EAGLE
if it's possible
EAGLE has some weirdness with pin labels, I think there's a way to move them but I don't remember how (and I'm using EAGLE 7.7 which is pretty old)
Hmm, I'll see if I can google my way out of this.
But now that I see the text there I'm going to have to re-listen to the whole Dark Tower series
Looks like it's just a shortcoming of Fusion360. Nobody has answered this question from may https://forums.autodesk.com/t5/fusion-360-electronics/moving-the-name-on-a-pin-in-library-editor/td-p/9448851
It appears that it's not currently possible to move the name of a pin in the library editor. This makes it really difficult to design symbols such as a quad package op-amp because the drawings intersect the names. See here for an example: https://forums.autodesk.com/t5/eagle-forum/move-pin-name-text...
the question being the follow up at the bottom
oh I was about to say, you can totally move them, but, I realized that you're in the library editor
I played around with EAGLE, Fritzing, and KiCad a while back, just trying to understand how they worked. I think I like EAGLE best out of all of them, even though it has quirks.
Then I quickly realized I have no idea how to actually design the circuit board itself, like, with routes and layers and all this other stuff that I have no idea about. Are there any good classes or tutorials you guys can recommend?
which version of KiCad did you use?
A fairly recent one, it was like 4-5 months ago.
stable version - even now - is 5.x, but they have been working on 6.0 for a long time; it is currently 6.0 RC1
I don't have any tutorial recs but I recommend just picking a simple project you want to do and go from there. Could be as simple as mounting several adafruit breakouts onto one big board with a feather or similar.
as for eagle, sparkfun has a nice series of tutorials
the feather tripler is nice but sometimes you just want to have a bunch of sensors in one place and stringing tons of jumper wires together is not as neat
Is there a way to do simple holes in the board view of F360? I know how to do them in a library part but I don't see the same tool here
I'd prefer true holes over vias
ah there it is
It's under the Place dropdown if anyone was curious
When I do breakouts and things like that, I usually have QT/Qwiic connectors on those boards, and I tend to just 3D print a base plate with screw bosses. I don't bother with nutserts on these, I usually just use the metric screw features in the BOSL library for OpenSCAD and add threads that way. It's just for little PCB screws anyway...
I've even printed some that were stacked. 😄
You can mount them fairly close to each other, and use the super-short QT cables
You recommend I try KiCad or should I just stick to EAGLE?
i switched to kicad 6, partly because Autodesk is killing standalone Eagle and moving users instead to electronics design component of Fusion 360. Even though that is obviously based on Eagle, somehow I found it much less appealing.
If you want to do 4+layer boards you'll need to pay money for EAGLE/F360
and with how cheap 4 layer boards are, it's worth learning to do them
but everyone has to decide for themselves what you like better. My only advice is,if you are trying kicad, try the 6.0 RC1, not stable
but with F360 for $495 dollars you also get a very good solid modeling program, and can export your PCBs into 3D objects
thats $495/yr
for me it was not a question of price - i have free educational version of F360
ah nice
and i do use F360 a lot for mechanical design
I think all you can do is turn off the pin names in its properties, and then place your own labels if you want them. Not really necessary for a diode symbol anyway.
but for EDA, i use KiCad
That's what I gathered as well. They need to fix this, it's a pretty egregious oversight
can you not export 3D objects from KiCad?
I have never used KiCAD.
yes, you can
Yeah, I just don't have $500 to spend on a software I use as a hobby, like 10-15 times a year. I already don't like F360 for designing 3D models, and I don't really do mechanical stuff, so I have no need for the fancy features of F360 on that side, and like shurik said, the electronics design component in F360 was for some reason kind of like EAGLE, but way more annoying.
@tough matrix how do the auto-tracing (routing?) compare between F360 and KiCad?
makes the evil eye sign at the auto-router
i did a project where i designed board in kicad, exported it as STP and imported in F360 together with other mechanical components.
there is no autprouting in KiCad, and i dont miss it
For a female USB-A thru hole connector, should the SHIELD pins be grounded?
they don't need to be (AFAIK)
there is "assisted routing": you drag a trace, and it automatically goes around obstacles
wait, which ones are the shield pins? The ones that clamp on the board, the big ones?
yerp
nah, only reason you solder that down is to improve stability in my knowledge of USB
so you may still want to have ... holes with conductivity ... for those
hahaha, dude, my Swedish/English brain broke
what are those holes called, the ones with copper on 'em?
holes
plated
or vias sometimes
aaah, plated that's it
I'm getting conflicting answers from StackExchange
Technically the shield pins should be connected, but opinions vary on what they should be connected to. One popular choice is a high value resistor to ground to dissipate static electricity before the contacts begin to mate.
nice, haven't seen that before
interesting. What qualifies as "high value" for this?
this is only transmitting power btw
For instance, Cypress recommends a 1Meg resistor in parallel with a 4.7nF capacitor (to absorb RF interference coming in through the shield).
is the RF interference only an issue for USB ports that will be transmitting data?
Intel recommends grounding it directly.
Kinda-sorta: RF interference can couple into the power leads (which can act as an antenna) then affect anything else connected to the same power net.
man these are all reputable sources, it's hard to pick between them
Hmm
well I guess I'll do the 1MOhm resistor + 4.7 nF cap method. Do you happen to have the source handy?
bookmarked that, thanks
thx
What wattage resistor would you use?
oh, duh
I don't see a 4.7 nF cap there, just a 1 MOhm resistor
How about these? I have filtering on my power line already
The ones in the top-right would only be necessary if you need a "USB is plugged in yes/no" logic signal.
iiinteresting. How does that work, electrically? I'm not seeing it.
Can I ignore the .1 uF cap to 3v3? This board doesn't have a 3v3 signal anywhere on it.
Yes, you can ignore it. That's more specific to the chip this app note was written for.
ahh
The 39K/62K is a voltage divider to drop 5V down to 3.3V, and the 0.1uF cap stabilizes the output.
oh my tired brain put the second 0.1 uF cap in series with the 39k resistor. I'm burnt apparently
that makes sense, but hmm, doesn't that drop the USB line voltage to 3v3? Or is this a "specific to the chip in the app note" situation?
Yes, the 3.3V chip would want a 3.3V input, so those resistors provide that from the 5V signal.
VBUS is still available for the raw 5V... the voltage divider just produces a secondary path with a 3.3V signal.
oh, isn't it odd to use a resistor divider as a regulator?
It's not a regulator, just a source of a voltage to be detected.
ohhhhhh
that makes sense
How would folks go about making this symbol in F360 EDA? I know there are convoluted ways to bring art in to F360 but I hate them more than people who have wronged me
Arc tool, line tool.
why do I keep forgetting to expand the dang menus
SMH
This'll have to do
Hm, any idea why the footprint for this part would indicate it's polarized when it isn't? https://www.digikey.com/en/products/detail/kemet/CK06BX105K/2397510?s=N4IgTCBcDaIIwAYFwLRgQZlQOQCIgF0BfIA
Order today, ships today. CK06BX105K – 1 µF ±10% 50V Ceramic Capacitor BX Radial from KEMET. Pricing and Availability on millions of electronic components from Digi-Key Electronics.
Part libraries are made by a lot of different people, so... 🤷♂️
Very likely, yeah. I tend to make a lot of my own library parts, or otherwise carefully check anything I use... 😅
I'm not super careful about checking but the one time I did, it was wrong.
i went with 10m and it looks pretty good
I've gotten away with 6 mil before but it's dicey hand soldering that, esp if you don't have a bunch of soldering experience. Real easy for the heat to lift up traces
right now, i am hand soldering, and like was said before, i got plenty of room
i just hope the volt meter isnt too big
yes. raven
Ahhh ok seemed familiar
Which part is that on the board?
the square hole in the upper left
ahh
This is what it started as
but it looks like crap an is only insulated with bandage wrappers
so i redesigned it to work with these
Nice work!
the toggle cuts power entirely to the barrel and the microswitch turns the volt meter on so its not always on and drawing power
the 6 header holes near the bottom are just to solder wires to keep the battery from moving
if my measurements are right, it will fit in an altoids tin
Getting measurements right is tough. Especially on something so small
im currently looking for other barrel plugs so i can use it in many applications
yes it is. luckily, thats part of my formal training
Nice. Can I ask what kind of training?
machining, with a furthering in cnc, as well as some draft work
anyone know where i can get pink pcbs that wont cost an arm, leg, my soul, and my firstborn?
https://www.pcbbuy.com/online_pcb_quote.html
is the only place coming up. For a 100x100 PCB, x5 its $67+$26 shipping to the US
PCB Prototype and PCB Fabrication Manufacturer,Custom PCB Board Buy Online!
PCB Buy and PCB Way do pink
PCBWay technically does purple, not pink, but the preview makes it look more like a magenta than a real royal OSHPark purple...
Oooh, PCBBuy explicitly has a Sakura Pink though
Yeah
I heard good things about production quality of PCBBuy. On the other hand, there was also a story when they posted for marketing purposes photo of one of the orders without bothering to ask permission of the client, which made him quite angry.
So pcbway is trustworthy?
for sure
I haven't personally tried them yet, but I know people that have used them and been happy
a mix of power boost and qtpy rp2040, this little board tells you (I hope) what the power level of your battery is, and shines the appropriate colors on a panel mount rgb switch. it also controls an electromag, with or without a power interrupt switch. its use? to enable switching from a physical pistol to a physical rifle for more in depth immersion into the oculus vr system (no it doesnt act as a controller, you still need the default touches.)
anything that needs to be fixed? i mage the traces as large as i could and added some space between parts. the spacing is about 6mil to10 mill or so.
It's hard to review things in this format, but I am somewhat worried about placing vias directly on some of your pads. It's sometimes a necessary evil for thermal pads under QFNs, but you've also got them on LEDs and capacitors, which could lead to some potential reflow problems.
Also not all fab houses can do that, correct?
I think OSH can but for example JLC cannot
That = via in pad
They can all "do it" if you just want the drill hole to go through the pad, which isn't great for soldering, but is okay for things like thermal pads. Fabricating a board with true "via in pad" support means that they take the extra trouble to fill and cap off the via so the pad on top of it is still smooth like it normally would be.
Ah yes. Thx
going with what @unreal flax said, I would also put vias under the thermal pads connected to ground (assuming ground is your 1st and 2nd layer as well) just so they dissipate more heat.
Move that one random via from the largest IC to closer to the pin.
EM+ and D3 look a little close
Remove the values from your PCB. Keep only names (R1,R2,C1,C2, etc etc)
Same with your pin connectors. Just call out the signals (+,-,RX,TX, etc etc). Make the names around 50mils
Normalize all your text for your name designators (I usually use 32 mils)
Im assuming you still have clean up to do, otherwise try to align all the names to one side.
These are only suggestions though 🙂 you may do things differently but this is how i try to do things
i assume by thermal pads, you are refering to the ones under the 3 main chip (P$21 and P$1). if so, all those large pads do indeed have a via in them. im not sure what one random via you are refering to. em+ and d3 have .5mm between them. should it be more? and yes i still have clean up to do in which most unnecessary values go by by.
as to the via under pad concern, it shouldnt take much to tweek that, just need to spread things out a bit more.
These are FETS but the idea is the same:
So yes, more pads means more heat transfer
and the 0.5mm looks deceiving is all 🙂 I couldnt tell from my end.
wait, Now I have to go read up on something. Maybe I should have via's under my drain pins.
This becomes a bit of a trade - yes, it does improve heat connectivity. But also - it improves heat connectivity when you're soldering, so it makes hand-soldering them (even with a hot air gun) a bit of a pain because you have to dump heat into your entire plane in order to get the pad hot enough to melt.

