I tried to do lot of stress testing with assembled m52 PnP speeduino using real components, but I really can't get the traces to burn or even hot. Also even with big spark gap and the coils/spark plugs close to ecu, I couldn't cause any noise to speeduino inputs. I probed around the ecu with scope and all is clean. So there shouldn't be any ignition related noise with this design when assembled correctly.
I had the setup running high RPM for hours and tried different dwell settings. With 1-3ms dwell there is basically nothing happening the ecu stays cold. And no wonder, because for example 6000 RPM and 2ms dwell was only 18% duty for the coils. But then with higher dwells the duty for the coils gets high really fast because it's wasted spark. Like here, this is over 50% duty for the coils:
Leaving the system running like that for couple of hours warms up the ecu case quite a bit, so in hot engine bay in e36 for example it could be possible to overheat the thing
So high dwell (4ms), hot engine bay and constant high RPM could maybe cause the traces to burn.But there shouldn't be any reason to run dwell that high. I for example run 1.5ms dwell in my m52 turbo and that makes 560hp at the crank using e85. Then if the coil(s) are shorted, it seems that it's really possible to burn the traces. It doesn't require much to get those smoking. Lastly even with good condition coils, the traces (and coil) gets destroyed immediately if the IGBT body touches the case. Like for example in this case when there was manufacturing fault in the IGBT which did lead the IGBT body to touch the case through insulation:
But well in any case. I would recommend using reasonable dwell settings with this PnP ecu to avoid possible problems. I changed the dwell from 2ms to 1.5ms in base tune, because that should work in most cases.
It also seems that even the 4.7nF cap for C25 isn't small enough for 60-2 trigger wheel. The speeduino 0.4.3 design uses originally 10nF cap, and when calculated with the 1k resistor in the line that should be plenty enough for 60-2 trigger wheel. It works nicely with simulated signal from speedystim for example. But with real trigger wheel and hall sensor, you don't get that nice and even square wave that you get with simulator. And in reality, because there is 1k pull-up to 5v, the voltage to high state goes through 2k ohms worth of resistors. Which makes the signal rise time way too slow and there will be problems from 6000 RPM upwards. So that's why I changed to 4.7nF cap originally. But with high RPM and the trigger wheel vibrating slightly, the pulses from the teeth can still get short enough to cause problems. So after testing with real trigger wheel and hall sensor, I ended up using 1nF cap for C25. That gives nice and robust signal for the mega. I did change that for the BOM files.