Pre v0.3x hw changes

[martinroger]

• DNPs updated based on minbuild
• Removal of unused Vsense circuits from board
• Internal harness schematic
• Pull-up on boot line to avoid wake-ups in download mode
• Insertion of 40106 inverter Schmitt triggers, double pass before 4050 with protection diodes to 12V_PROTECTED (Should resolve ITF - Finalize denoising circuit on pulse buffer #58 )

[martinroger]

1. 40106 used as a wave shaper with the 4050. Inputs protected to stay within VDD+0.5V with RC stage and BAT54SW diodes :

2. RPM and WaterT channels swapped around to simplify layout

3. DNP load emulators to minimum build

4. Excluded two VSENSE from board because they are not used. Option to use one of the ADS inputs as a 3V3 reference voltage for comparison.

5. Added pull up and made capacitor optional on Boot button circuit to avoid accidental startups in "Download mode" when capacitor is fully discharged.

Layout for 40106 added near the 4050

[Alex-XNR]

Something I've noticed/thought about on the original pulse buffer circuit is that both input and output signals are capacitively loaded quite heavily with the 100nFs. This may cause ringing and slower slew rates (and put stress on the line driver).
For simplicity sake, it could be that by just adding 100ohm resistors in series with the input and output signals without any capacitance (or very small <1nF to make an RC network) would help out here. Space for parallel resistors (100k maybe) could also help to terminate/load the line and prevent ringing.

Perhaps worth allowing for both the proposed Schmitt-trigger circuit as well as a more simple option, to rule it out?

[martinroger]

Something I've noticed/thought about on the original pulse buffer circuit is that both input and output signals are capacitively loaded quite heavily with the 100nFs. This may cause ringing and slower slew rates (and put stress on the line driver).

Yeah, I think the 100nF input + output strategy was some poor mistake on my side. For board number 2 (the "min build), I actually built it without the output caps and with 22pF input caps only, and it is the board that was used to get the CAN logs and that shows the rare-but-not-absent occasional pulse "panic". I still need to try and trace the odometer to see if some of the pulse surges were missed by the fast metrics signals but still integrated in the pulse counter.

I don't have logs of the original "100nF" board 1 on an actual drive because I didn't receive the displays in time for that. I should try and remedy this in the coming weeks, pending the punishing repair of the water pump...

For simplicity sake, it could be that by just adding 100ohm resistors in series with the input and output signals without any capacitance (or very small <1nF to make an RC network) would help out here. Space for parallel resistors (100k maybe) could also help to terminate/load the line and prevent ringing.

Where would you put the parallel resistors ?

Perhaps worth allowing for both the proposed Schmitt-trigger circuit as well as a more simple option, to rule it out?

I can definitely try to add some room for the pads. It's getting a bit crowded but not unfeasible. 100Ohm resistors as close as possible to the 4050, I imagine ?

[martinroger]

Ok, it's not pretty but it should be functional :

When using solder bridges JP13-15, C65-67 can be used for a small extra capacitance with a 0-Ohm on R143-145. In those cases, BAT54SW and the 40106 should be left unpopulated.
Also 100k added for each input to ground, and an inline 100R (I still need to assign a component to it, but for now 0603 as usual).

PCB is getting crowded and not so pretty, but I guess given this is not yet production release, we can do some de-populating once there is something that works fine.