[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: High voltage probe, odd NST measurements



Original poster: "Gerry  Reynolds" <gerryreynolds@xxxxxxxxxxxxx>

Hi Ed,

OK I'm beginning to see your concern, I went back to your original post and interpreted the stray capacitance of 0.26pf as being the stray per resister meaning that with 10 resisters in series the effective capacitance would be .026pf. I also calculated the correctly compensated probe (with a 10pf load across the 1Meg resister) and found that the stray per 1Gohm resister needs to be 0.01pf. So if your .26pf stray is for each of the 10 1G resisters, then the probe would be overcompensated by a factor of 26. Resisters for proper compensation are of the HV type that are tall and skinny that also reduces the parisitic capacitance between the resister ends. With the stray capacitance included, there are two dividers: a resistive and a capacitive divider. The resulting voltage (the affect from both dividers) requires superposition to calculate. Even with 0.26pf of stray per resister, the resisters dominate impedance wise by a factor of 10 (0.5% error). I did a spice with 0.026 pf across the 10Gohm aggregate and at 60 Hz, the effect was negligible. At higher frequencies, the capacitance becomes dominate and the error goes up.

If the 0.26pf is the aggregate stray of 10 resisters, then (as you say) the error would be 41%

I dont know what the real stray capacitance of resisters are, but if the resisters used have stray that is too high, one should be able to minimize the effect by increasing the load capacitance to bring the capacitance divider ratio back to 10000:1

Gerry R
Original poster: Ed Phillips <evp@xxxxxxxxxxx>

"Original poster: "Gerry  Reynolds" <gerryreynolds@xxxxxxxxxxxxx>

Hi Ed,

Yes, over compensation will cause high readings.  One might calculate
how much capacitance it takes across each resister to properly
compensate.  I'm guessing this is an order of magnitude larger than
the strays and may be hard to accidently do.

Gerry R."

If I remember the original design correctly it uses a 10G ohm (10^10 ohm) series multiplier. At 60 Hz an 0.265 pF capacitor would have that reactance and cause ~ 41% high error. That's not much capacitance at all!

Ed