I think this explains the 100V issue. I know I'm blocked and all, but the blocking implementation has been as good as the MS implementation. R51 on the V3.0 schematic is 1m ohm. That's just bad design, and will produce very high voltages. I see the LM1815 datasheet notes a 18K with a max allowed voltage of 108vp-p, so actual power would be at least half of that 108, probably closer to 1/4 of that voltage. Also this is the max the chip can handle, not the max expected VR signal. I'm going to round that to 100v for the heck of it.
The MAX chip uses two 10k's that are effectively 20k-ish. This nearly matches the suggested impedance for the LM1815.
If we assume the VR signal is a sine wave that runs at the max allowed noted 100vp-p, the RMS voltage would be .707 of 100, so 70.7v, not 100. It's not a sine, and by eyeballing the curves I've seen, I think it's much closer to .5 then it is to .7. The power at 70.7v would be (70.7^2)/20,000 = .249 watts.
I believe the max expected rms power could be as high as 1/2 of 100 so 50v not 100v, then we calc (50^2)20,000 = .125 watts. That 1/2 is an estimated voltage. I would want to check and see several wave forms before I can consider that voltage to be correct. A fairly small change in that voltage will result in a large change in power. Also we will likely run at an elevated temp, and the 1/8 watt spec is based on ideal room temp conditions, so a 1/4 watt probably isn't a bad idea, but it's also designed on maintaining that %1 or %5 ect error bracket, not that it will burn up, or cause the solder to melt, or any thing significant like that. If the elevated temp cause the 20k to become 22K I don't think that's a problem, the issue is we would want to keep it 11k and 11k. If we heat sink one and not the other there is a chance it could become 10k and 12k, which I think is still not an issue.
Can anyone with a VR and 20k resistor confirm the RMS voltage? I'd like to know how much heat we are actually looking at. I'm curios if we will want to heat sink these to help keep the case cool. As far as I can tell, the resistors on spin1 puma are fine, but it would be nice to verify that we don't need to bump them up to a 1/4 watt. Then when we do bump it up, we'll know it's just excessive.
I'd also like to verify the max voltage on the max leads. How high can we go before it arcs. Perhaps these will need conformal coating to increase that voltage barrier. Or perhaps we can assume normal voltages will be present, and use a lower RMS voltage.