Hi,

Attached is a work in progress circuit diagram for en EDM spark generator. It contains some notes to self and tentative part numbers and components values. Please ignore at you discretion.

The idea is that the buck converter will be forced into current limiting and provide pulsed current (~1A to 10A) into spark gap E1. The expected gap voltage is in the order of 1V. Experimentation will be needed to verify this. At a minimum I'd like to detect a short circuit at E1. Ideally I'd like to monitor non-short circuit operating conditions as well, for example to detect if the 100V pulse from the flyback caused a breakdown or not.

To sense a proxy for the gap voltage I added R4, R5, U5. If this even works at all it would probably give very poor resolution. But instead of explaining why I think this won't work I'd like to ask how you would do it.

 

To accurately monitor and record a spark gap voltage you need two channels of amplification with very different gains . Prior to the arc striking, a channel able to accurately scale the voltage to the measuring/display/recording system. As the arc strikes and maintains you need a channel with instant overdrive recovery for the much lower arc voltage, because an arc has a very low resistance that seems to decrease as the current rises.

I am guessing that the TS actually wants to measure the arc voltage rather than to just sense that an arc has started. If that is not the case please let us know.

 

As already stated, at a minimum I'd like to detect a short circuit at E1. Ideally I'd like to monitor non-short circuit operating conditions as well, i.e. measure the arc voltage. Detecting a short and an arc can be done in another way, but sensing the voltage would be nice.

Regarding the two different gains, the gap voltage range of interest is estimated to be in the 0V - 5V range. The rest can be clamped. Perhaps the attached circuit will do that.

 

The circuit shown has no signal or output reference shown. Are we supposed to assume that the reference ids the ground symbol??
Certainly that will produce very noisy output voltage. AND I doubt that the op-amp will last very long when the arc voltage supply is active.

 

This looks like this might work. V3 is a -3V3 regulator that supplies the MCU.

 

not sure what the diodes are for and why are they connected like that. D2 is reverse biased so V3 has no function.
D1 and D3 are in series and create short circuit for V2. My bad, did not pay attention to DC offset for sine source

 

not sure what the diodes are for and why are they connected like that. D2 is reverse biased so V3 has no function.

V3 is there to power the micro. Then D2 is there to clamp V_sense, going to the micro ADC input, so that it doesn't destroy the micro. Perhaps I don't really understand what you're getting at.

 

Given the way arcs function, the voltage rises until the material between the "points" breaks down and the arc strikes. Sensing a shorted circuit is simple because in that instance the voltage between the two points never rises above the voltage at which the arc would strike. In fact, the voltage across the short circuit between "points" never rises beyond some very low voltage caused by the current thru the connection resistance.
So the logic to sense a short circuit will only need to compare the measured voltage with the anticipated voltage.

 

The challenge of measuring the voltage ONLY across an arc is primarily one of not measuring other voltages. A suitable differential isolation amplifier connected directly to the arcing elements should be fairly close, BUT it will also need an adequately isolated power supply. The good news is that such isolation is not a new requirement at all and so there is a fair amount of information available, and even a few reliable isolation products are available.