It’s Portland, they will just lounge in the street if I look confident enough in what I’m doing. Probably apologize for interrupting. 
4 Likes
If you wear hi-viz and carry a clipboard no one would even question you.
5 Likes
Critical detail.
I used this trick to roam the halls of Intel to waste time in my youths
7 Likes
hardhat also maybe? well actually probably a helmet counts
2 Likes
I foresee a Craigslist post in the future
3 Likes
This was my thought as well. The 60D/100D has a set of 3 mosfets and an IC for the anti-spark system that turns it on and off. The 100S doesn’t have that.
It could also be affecting the MCU based on the bootloop and setting corruption comments. The 60D/100D uses a 100 pin version of the STM32, whereas most VESCs and the 100S use the standard 64 pin version.
You could bring the ESC to that spot and monitor voltages on the anti-spark IC, the anti-spark mosfet gates, or the 3.3v for the MCU to see if anything pops out… though it’s kind of funny to imagine Andrew sitting in the middle of the road with an ESC, battery and oscilloscope, yelling at people that he’s doing science
15 Likes
It’s Portland, there’s folks on unicycles and flaming bag pipes
14 Likes
This is exactly what I foresee in my future.
Truly though, I think people would go around me and let me be. Car culture in Portland is super relaxed.
I just need a labcoat to sell it from afar
13 Likes
That’s a good point, I don’t know but I think there are at least a couple of possible scenarios…
-
The magnetic field from the loop/coil induces current flow in wires and/or pcb traces which can charge FET gates and even small caps. Normally I would expect pulldown resistors on the FET gates which would prevent this type of turn on but perhaps not (as a cost saving measure).
-
RFI from the loop/coil circuitry (radiated by the loop(coil itself) gets rectified by the input protection diodes of the pins for any chips and creates voltage spikes or perhaps a varying DC voltage. This could create all kinds of havoc.
I’d recommend bringing a box/foil for testing RF shielding (non-ferromagnetic) and another box/foil for magnetic shielding (mu-metal or equivalent) to see which might have an effect. This would block directly radiated energy but anything coming in on cables would have to be dealt with differently.
This is an interesting one! I agree, we have some wonderfully brilliant people here and I’m really looking forward to seeing how this goes.
7 Likes
Fantastic suggestions & hypothesis!
1 Like
5v from CAN does the same thing, not just feedback from motors
1 Like
Yes, but where is the 5v coming from?
Not quite the same mechanism we’re discussing.
That’s supposed to be special? I thought unicycles and flaming bag pipes were the standard.
The 5v comes from the other esc in a 4wd setup.
I’m not an engineer’s arsehole, just pointing out that an esc (only unity and stormcore that i know of) can be tricked into turning on or off via this method.
On the unity it needed a resistor on the 5v line for it to be able to turn the board off properly. If it didn’t have the resistor, when you tried to turn the board off, it stayed in a weird state where it seemed the 5v was leaching back and forth between both escs and powering them back on
5 Likes
This is another bug, very common on stormcores. Suspected to be due to data corruption on the flash memory. Connecting with vesc tool and reflashing the bootloader (which does a swd flash erase) fixes this issue, although I don’t think anyone knows why really. I recommend any stormcore user does this as the very first step when programming their ESC.
Anyway, that’s unrelated to the issue this thread is about.
4 Likes
I’ve seen this behavior on 4wd setups before- but I don’t think it’s related to what we were observing during CarvePDX.
I don’t think any of the impacted boards were running FWD.
5v circuit being impacted by the inductance based sensors feels like a further stretch from the roll-to-start/antispark latching circuit hypothesis. 5v comes from the DRV, which it will only provide it if the VBAT input is above the minimum threshold and the ESC is running.
3 Likes
A loop antenna and a portable oscilloscope should be able to pick up near-field effects (more on the magnetic than electro- side of things. IIRC Dave from EEVBlog did a video less than a year ago showing how loop antennas/probes can be used to sleuth near-field EM on circuit boards, and the concept is the same. Even showed how to make your own loop antenna for cheap from a piece of coax.
It’s important that due to the relative sizes and positions of these inductive loops, they must be treated and considered as near-field interactions, which are usually very small and localized (in consumer electronics, it’s generally within 10cm or less). The sheer size of these loops (designed to interact with entire cars) means that the “near field” is pretty damn large, while still being very localized - it’s not radiating much energy at all to anything outside its local sphere of influence, but that sphere of influence is big enough to stop a whole posse of esk8ers dead. Think of it like a really giant Qi phone charging pad - a Tech Deck is pretty close in scale.
3 Likes