Hi Trevor.
Thanks for the tip re iCloud. Will look into it.
I think with your resistor problem it might come down to footprint.
In the Schematic editor select a resistor type and size. Don’t worry about value at this stage.
Make sure it has a valid footprint associated, usually displayed on the right hand side (symbol and footprint). If blank that particular resistor type has not got a footprint, don’t use. look further until you find a suitable component that HAS a footprint attached.
The value (0R or whatever) can be modified at any time in the properties box without affecting the PCB.
You can update the PCB editor with these changes if any have been made.
The version I have just downloaded for Mac is 9.0.6.
I have not used it in anger yet, just tried out a couple of previous schematics. I will have a play when I fell a bit better. Have been feeling somewhat poorly for the last couple of weeks.
Cheers Bob
Bob, I have an R with a footprint - that’s not the issue. The problem bites when trying to draw a PCB trace between the two connection points. KiCad won’t let me… and I have yet to understand why. It may be some oddity in my power circuit… I will try with a different circuit later today, eg for what I want to do with my multiple I2C QWIC connectors. Although I have yet todraw that out, I am quite sure the SDA and SCL tracks will require crossover… given I have effectively a single-side board.
Yes, at a pinch I could run a via to the bottom side, as I did for my first instance, but I don’t think that is going to be viable with 2 lines to run to multiple connectors.
Cheers
EDIT: Update… my test circuit with multiple QWIC connectors and a couple zero ohm Rs worked just fine! Conclusion: there must be something froopy with my actual project circuit. Sigh…
Hi Trevor
I have not had the time to experiment but I am not sure KiCad will allow you to place a track on the PCB which is not on the schematic. But I have just had an idea.
Try adding the wire you need to the schematic editor and let the PCB be modified accordingly. This may work.
Cheers Bob
Hey Bob…
my email has been out of action for a few days… didn’t see this util just now. Email now working again…
Did you read my edit? It’s all good on my circuit crossing SDA and SCL, using 0-ohm Rs, everything works as expected.
The only remaining issue is the original one… where I needed to cross a track on the part of the PCB with the devices/connections for power/LiPo management, in that area of my schematic I used some Power flags … and that seems to have caused the issue.
So… in summary: in a “normal” usage scenario, I have no issue using 0-ohm Rs to cross tracks. But… add a couple power flags in KiCad… no banana. I will do some more testing of this, as it seems to be a bug. I don’t believe I have committed any sins in my circuit… ERC says it’s all hunky dory.!
Cheers,
Hi Trevor
I am by no means an expert with KiCad. Only really touch it when I want to post a part schematic or something. Maybe once every 3 or so weeks.
BUT, I think there are power flags and power flags, ie; different types and usages. Would it be possible you are using the wrong type. I don’t seem to have a lot ofttimes lately so you might have a look.
Cheers Bob
Nuts.
I found the problem… ME !! Nothing to do with power flags, just plain old bad circuit design. Had my zero ohm R on the wrong line. OMG…
Good news is I learned how to put a link on the backside of the board, carving an island out of the copper ground fill area. Who knows, one day that might actually be a thing I need…
I have managed to simplify my connection to the LCD1602… just use another QWIIC connector!.
(Almost) final PCB layout attached…
I am a little concerned that the test points I added for I2C SDA/SCL, being a couple header pins, might actually be working as small capacitors - so wonder if I should just leave the headers out. See my 3D pic. Any thoughts?
Finally… got my PCB fabbed… having changed almost all componets to SMD. I got a stencil (which is huge) with the boards… and having never done this before, am looking for advice on how to proceed.
Obviously, I will do all the SMD stuff before affixing the THT parts. It’s the use of the stencil that I am really asking about. All tips from experienced makers appreciated!
I have not had a great deal of experience myself here but what stencil are we talking about.
The one for the screen print should not be a problem as it looks like the screen printing has already been done.
The other stencil that comes to mind is one to apply solder paste to the surface mount pads. This could be a bit tricky but if great care is taken id should be doable. I have not done this so some guidance from a more experienced person would be a good thing.
So come on all you modern PCB experienced guys out there. Help is needed.
The one criticism I could make is the thickness of the positive rail track. Particularly that long run across to the pressure sensor connector. OK so if the current in this track is probably low making the resistance no problem but the inductance could be quite high with the thin track. Making these supply rail tracks as thick as practical will reduce this stray inductance and minimise any strange happenings due to spurious oscillations.
It might all be OK. I think this 5V for the sensor could be supplied from the Pico. But is does not hurt to keep all of the 5V circuitry down to as low impedance as possible. ie; Fat tracks.
If you do get any funnies which could be due to spurious oscillation the first thing I would do would be to connect a thicker wire across these connections where the thin tracks are and see if this cures or improves the situation.
I wonder how the thin 5V tracks weren’t picked up by some design rules or other check. Even by the board fabricator or did you do the board yourself. maybe the final design rules check is not the holy grail after all. It is pretty good though.
Cheers Bob
The signal tracks are 0.2 mm … the power tracks are 0.4 mm - that includes everything related to the LiPo battery charge circuit, which maxed out on my breadboard setup at around 130 mA. chatGPT claimed my track width was more than fine for this current. We’ll see!
Yes, it is the solder paste mask I am looking for tips on. My best guess is that I should fix the PCB to the stencil with masking tape or whatever so it doesn’t move… then wipe the paste on with a scraper (otherwise known as a credit card, I think…)
Really looking forward to getting this baby built… just need to get a 14-pin DIP socket (I mistakenly ordered 16 pin)… I could use it with one row of pins pushed out, but figure a decent socket (ie not Chinese rubbish) is probably worth a little extra delay.
Bob - thanks for sticking with me on this. I well remember our initial interactions, when you scolded me for putting a diode in my CT detect circuit! The project has grown and morphed a huge amount since then… I’m most grateful to you (mainly) and a few others who have helped this old dog learn some new tricks!
Edit: PS I had the board made by JLCPCB… no way I would attempt this myself!
@Robert93820 OK… I haven’t even built this - but already I can see a possible improvement (dang…)
I2C0… presented to 4 x QWIC connectors, in which I needed the 0R things to cross tracks. It has just occurred to me… the resulting track length of SDA and SCL are different. Should I have added some squiggles to the SDA tracks to even things up? Or am I milking the proverbial mouse, given the I2C frequency is not that high…?
I did say that I didn’t see track resistance as being an issue. (refer “current” in the above statement). My bit of concern is with keeping the inductance down on the 5V circuits. They will probably be fine, just be aware of any potential problem possibly due to this..
Yes I think that how it is basically done. I have not done this myself so someone could chip in here with some instruction possibly ?? I think it would be absolutely imperative that the mask does not move or you are likely to have solder paste everywhere.
Cheers Bob
Hi Trevor
In some circles the frequencies involved with I2C would be considered almost DC they are so low.
Your difference in track length would be an extremely small fraction of a pico second so I would not worry about it
I know I2C devices are usually daisy chained but I don’t know about a star configuration as you look like you intend. Should be OK. Just be careful about the number of pull up resistors connected or you might load down the lines too much as they would all be effectively in parallel.
Cheers Bob
A v2 or v3 is not really a bad thing as you are learning.
e.g. Build the V1 then run tests fix anything that is not working, re-run tests. then build the changes into the v2.and repeat.
I started out etching my own single layer PCB, then 2dual layer PCB, but now I tend to only do that if a need it asap, else its just a much better result to get it made. Coming from the etch, you removed copper, so tracks tended to just we wider (less copper to remove) as well as the photo resist can easily have issues it tracks are too small. So now I tend to just leave tracks wider unless I need them to be thinner. When looking at current support on the tracks keep in mind your copper thickness as well as track width.
But always good to see a design come to life and the made PCBs always look nice.
As to the solder mask template, while I have not done it myself, there are lots of YouTube on how people have done it, which seems to make sense.
I have a solder paste dispenser that uses compressed air and a foot peddle. Not as exact as the stensal, but I got used to the “1 tap form small pads, 2 taps for… and so on”. Note, if this is the first time you will be using solder paste, keep in mind it does not go as far as you think, lots of flux holding the balls of solder. So you can run a bead of solder down over the pads for an IC, and when the solder activates it will pull into the join (the solder mask helps this process),
I have some cheap Chinese SMD ovens. You setup the solder profile you want, place the board and parts on the tray and press go. it heats up slow enough to avoid any rapid thermal shock, activates the flux for a period of time, then heats up to flow the solder, then a gradual cool down. this works well for a whole smd only board.
Keep in mind you can also do by hand if you need, but doing a board at once allows the components to auto align as the solder pulls the component and pad toward each other.
@Robert93820 thanks. My star topolgy derives from: a) existing b/board setup uses a Piico splitter.. and b) only one device (my Piico Transceiver) has both in & out I2C connectors All the displays only have a single I2C connection… so, it kinda made sense to do it this way. I guess star also minimises the maximum cable length…
I will get to build this thing in the next few days… and we will see. @Michael99645 Thanks Mike… I don’t have an oven, so far my (limited) experience with SMD has been using a rework hot air gun, and applying paste with a bit on the end of a toothpick! Straight out of the syringe put way too much paste on a pad… so I improvised!
I’ll wait and see if any SMD experts comment… if not, I’ll just try taping the PCB in place and having a crack.
Bob’s covered the important part on the signalling side — at I²C speeds on a small PCB you’re not in transmission line territory. The bigger real-world limit tends to be total bus capacitance rather than trace length matching.
On the SMD side — toothpick paste application is basically a rite of passage. If you’re doing fine-pitch ICs, a cheap stainless stencil is honestly worth it. Even a manually aligned one with tape hinges improves consistency massively.
And yes — build it. Worst case you get Version 2 with better ideas baked in. That’s how most of us got here.
