Suggested PCB fab?

[adamcritchley]

I did a run with v3 of the board with 1oz inner and outer layers but I'm having issues with the velocity touch buttons. If the velocity is hardcoded then it works with a single magnet but not accurate enough to keep the entire car on the track. I'm thinking it may have been the fab I used.

Do you have a recommended PCB fab?

[mcbridejc]

Hi! I used JLCPCB for mine, but I don't really think it should matter much where you get it built.

Capacitive touch

I'm not sure what sort of problems you're seeing, but I wonder if the calibration I did might not work well enough for your boards. Although I hoped it was primarily a function of the trace geometry and that it would be consistent across batches, I don't know if it is (it seemed to be consistent within my batch of boards, at least). I determined the calibration by pushing the raw capacitance measurements out the serial port, and capturing them while sliding a finger across the sliders. Then I just computed scale factors to "normalize" them, so the maximum values were consistent. If you're up for some data collection, and able to read to the serial port on the debug connector, I could send you a firmware for it.

Keeping cars on the track

There could be a lot going on here; and there are always some operating conditions where a car will fly off the track -- e.g. if you go too fast. Is what you're seeing that the cars move but go straight off the track when it hits a curve? The main batch-to-batch variation I would expect to matter is trace resistance. Since the software just controls duty cycle and it can't measure the actual current, if the resistance changes so will the currents. Are you able to measure the trace resistances? On mine, the phases are about 5 ohms, and the guard rails are 0.8 ohms, and this is codified here (note the 1.4 ohm extra is accounting for the resistance of the FETs).

[adamcritchley]

We used JLPCB as well. I can absolutely collect those slider measurements. How would I get that firmware from you?

The cars kind of slowly and gently meander off the side of the track. A single magnet works pretty well but not the entire car. I'll reply with those trace resistance measurements tomorrow. Thanks!

[mcbridejc]

Any chance the car magnets are installed backwards? They all need to be aligned, and also there is a "right" polarity based on the gaurd rail current direction. The other thing to check: validate the guard rail current. Find the capacitor across the guard traces, get the resistance on it while powered off and then measure the voltage across it while powered on. Oh, and since it sounds like you modified the software to bypass the touch control, I guess just make sure that you modified it in such a way that the guard rail is still turned on.

…

On Sun, Jul 16, 2023, 2:15 PM Adam Critchley ***@***.***> wrote: We used JLPCB as well. I can absolutely collect those slider measurements. How would I get that firmware from you? The cars kind of slowly and gently meander off the side of the track. A single magnet works pretty well but not the entire car. I'll reply with those trace resistance measurements tomorrow. Thanks! — Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABFQFIPT6QI7DUFDWAVNCE3XQRKYPANCNFSM6AAAAAA2MBP5JU> . You are receiving this because you commented.Message ID: ***@***.***>

[adamringer]

Are the magnets all supposed to be installed with the same polarity, or is one side supposed to have opposite polarity from the other side? Right now we have them installed all with the same polarity.

[mcbridejc]

All with the same polarity, and it cant be arbitrary. One polarity is wrong and if backwards it will move but the guard rails will not contain the car, but rather will be unstable and tend to push it off the track. Switching the direction of the guard rail current has the same effect as flipping the magnets around.

…

On Sun, Jul 16, 2023, 2:59 PM adamringer ***@***.***> wrote: Are the magnets all supposed to be installed with the same polarity, or is one side supposed to have opposite polarity from the other side? Right now we have them installed all with the same polarity. — Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABFQFIOIVFFI6HEF7VYXUX3XQRP4TANCNFSM6AAAAAA2MBP5JU> . You are receiving this because you commented.Message ID: ***@***.***>

[adamcritchley]

Thanks for the measurement directions! I'll have them to you shortly.

[adamcritchley]

Is C10 the guard rail capacitor? I get 5.5 ohms powered off. And voltage is 2.5 to 2.9 powered on. For C9, I get 0.6 ohms and 0.5v