Creating high voltage components is challenging because of the parasitic capacitances to the surroundings. The electrical field at 10KV requires careful isolation of any conducting surface of about 1”; furthermore, any sharp niches can trigger an arc between the component and the surrounding lower potential items.
Goal
I needed to measure high voltages around 30KV; so I undertook the task to create a “reasonably accurate” voltage divider. The goal was to divide the voltage by a factor of 1000 and still present an input impedance of less than 1M to the lower part of the divider.
Accuracy
As it turns out, the accuracy of divider does not depend only on the tolerance of the individual resistors – the parasitic capacitances to ground play a much more important role than the resistors themselves, as shown in the figure on the right. One way to mask-out the parasitic capacitors is to add much larger capacitors linearly with the resistor.
Construction
I used 100 pieces of 1206 chip resistors of 10M each. As these resistors are stringed together, a total impedance of 1,000 MOhm or 1 TOhm is possible. But how to you solder 100 chip resistors when each one of them is about a 0.6mm in width? The answer: line them up using a razor blade – use scotch tape to hold them temporarily until a tiny drop of solder flows between them! Stringing 15 of them at a time was not that bad – it took about 2min per segment. Then, after enough segments were made, I used 1/4” of 30AWG wire to solder the segments together – this way the overall string can maintain its flexibility for thermal expansion and mechanical stress. I was about to solder capacitors in parallel with the resistors, but at the end I thought that he capacitance of the soldered sides will be just enough to reduce the parasitic effect.
After all resistors were soldered together, I used an one foot long glass tube of 1/8” of diameter to place them. To ensure that the resistors live in a non-humid environment, I used the heating gun to overhead the glass tube – while hot, I dipped the two ends in silicon glue which was sucked in as the inside air was cooling down…
Testing
I used a HV generator of 5kV which could create positive or negative DC voltages – amazing enough the accuracy of the divider was excellent: when I applied 3,735 volts, the meter was reading 3.72 volts, i.e. an error or 0.4%. This was hard to believe since, all components: the DVM, generator, and the divider are not more precise than 1%.
Finally, I used a HV generator, called “the ionizer” to test the voltage rating – I would only go to 30kV; so far, so good.
Enjoy!
Your are absolutely right for DC. One of my reasons to make this divider was to capture the discharge voltage from a fully charged TV tube. The discharge rate was expected to be 1000 Volts per nanosecond and I needed a “balanced divider” with high frequency response. At the end, my experiment did not work, despite my FET probes of the oscilloscope — the bandwidth was too small for any level of HF measurement. Well… next time…
Regards
I don’t understand why parasitic capacitances would be a problem for DC measurements. Under DC conditions, capacitors don’t source or sink any net current, they just stabilize the voltages that are already there. To first order, you can just ignore them completely.
For AC, of course, they would be a serious problem.
I recommend to buy from Mouser the
MOX94021007 (1GOhm),
MOX94021006 (100MOhm),
MOX94021005 (10MOhm),
all are rated 45kV of 1% tolerance. They are about $20 a piece.
The MOX750 is 10GOhms, also 1% for 7.5kV for about $10. To make a TeraOhm, you will need to string 100 of them.
The MOX1125 is 100GOhms, 1%, 7.5kV for about $40. To make a TeraOhm, you will need to string 10 of them.
(I also sent to you a private email on this matter)
we want to purchase decade of 1M /1G/1T range from, 1 G ohm to 10 T ohm standard resitors woking at 1000/5000 volts for calibration of high resitance meters capable of measuring upto 10 Tera ohm at 5000 V……..pl contact us
(Copy of Win’s comments from his email)
1G and 10G resistors are fairly easy to come by. But you can buy
Expensive, special order, and a long wait.
1T-ohm resistors from Ohmite (was Victoreen), e.g., their Maxi-Mox
thick film MOX-5-12 or MOX-5-13 types, rated at 50kV. These are
five inches long.
Mouser has 10 pieces of a standard 3″ long, 1G-ohm, 10-watt,
45kV resistor on back-order, MOX94021007FTE, only $19 each.
They stock the MOX-750231007, a modest 1G part, and the 10G
MOX-750231008, both 1″ long, but only rated at 5kV. Sigh.
Ohmite also makes ¼” 3kV chip resistors to 50G, but Mouser only
stocks low-voltage 10G versions in 0805 and 1206. Expensive.
The highest-value low-cost SMT types I’ve found with distributor
stock are the Vishay Dale 200M and 470M parts, in 0603 and 0805,
very cheap but only rated to 75 and 100 volts. Mouser will order
KOA Spear HV732ATTD476J parts, 47M in 0805, rated at 400V.
100 of those would make a 5G 40kV wand resistor for about $20.
At that rate the MOX94021007FTE doesn’t look so bad at $19,
allready assembled! if you can live with 1G instead of 5G. Hey,
maybe you prefer 1G for faster discharge?
Have you explored electrometers? These can measure high voltage
with capacitive rather than resistive dividers. I have an old Keithley
probe for doing this. You can make your own with the appropriate
electrodes and shields, and fA-sensitivity op-amps, like the LMC6041.
I’m hoping to find time to write about that when we do chapter 4x.
Speaking of high-value resistors, to 10^12 ohms = 1T, one way I’ve
found to get them is in old Keithley electrometers, cheap on eBay.