Some noob high voltage divider questions

For Short Term Learning Discussions ONLY. This area is for CURSORY questions and connecting with other users ONLY. ALL technical contributions need to be made in the appropriate forums and NOT HERE. All posts are temporary and will be deleted within weeks or months. You should have already search the extensive FAQs in each of the forums before posting here as your question may already be answered.
Post Reply
Thomas Henderson
Posts: 8
Joined: Tue May 02, 2017 6:56 pm
Real name: Thomas Henderson

Some noob high voltage divider questions

Post by Thomas Henderson » Wed Feb 13, 2019 7:37 pm

I think it is finally time to crack into my x-ray transformer, and build a nice HV supply, as it is slowly crushing the table i have it sitting on. I have been reading up on the HV divider FAQ, and I think I have most of my parts sourced. I want to build a divider that could in theory read 100KV, and give me a 0-100v readout to an eBay sourced 0-100v led display. After some "back of the napkin" math I proudly ordered 10x 100meg 2watt 15KV 5% resistors, and a 1meg .5w 1%... my math showed me that this should give me my 1-100v scale, and not blow anything up.
After ordering everything, I started thinking... what about the resistance of the meter itself? Turns out some of these meters are speced for like 50k ohms... which in my mind throws things way off if in parallel with my 1meg resistor. What is the best way to counter this? can I add a resistor in series with the meter? should I up the 1meg to something higher, so I have a little room to play with my mess of series and parallel resistors? I feel like this could be one of those annoying text book problems unfolding in real life.
HV divider.png
HV divider.png (4.72 KiB) Viewed 362 times

User avatar
Dennis P Brown
Posts: 1654
Joined: Sun May 20, 2012 2:46 pm
Real name: Dennis P Brown
Location: Glen Arm, MD

Re: Some noob high voltage divider questions

Post by Dennis P Brown » Wed Feb 13, 2019 9:19 pm

Since your chain of 100 m-ohm resisters are only 5% (so each can be +/- 250 k-ohms), these could either cancel (unlikely) or all add up the wrong way (also, not likely.) More likely, you will have a significant error of at least a few 100 k-ohms in the chain. In any case, these might create a bigger issue than if the meter is as high a resistance as you think it could be (and it might be a lot less.) Rather, I'd suggest building the device and testing it before deciding if the meter issue is the dominate problem. It might actually increase the systems accuracy.

In any case, if you decide to measure voltages above 40 kV, I'd suggest an oil filled glass rod for the first resister chain. I built a 100 kV supply and discovered that those voltages can be both very unpredictable and surprising in finding paths to ground; really, this is true when voltages get above 50 kV, so one needs to be very careful to insulate the device. There's a reason high voltage probes tend to be long and have multiple shield plates between the handle and probe tip

Dan Knapp
Posts: 223
Joined: Wed Aug 06, 2008 12:34 pm
Real name: Dan Knapp

Re: Some noob high voltage divider questions

Post by Dan Knapp » Thu Feb 14, 2019 2:52 am

Use one of the Harbor Freight red freebie meters (input impedance 1 Meg.), and the bottom of your chain will be a net half Meg. Another use for the very handy red freebies! I just used one to build a ground loop detector (for circuit see https://authors.library.caltech.edu/8499/1/BELrsi07.pdf).

User avatar
Rich Feldman
Posts: 1058
Joined: Mon Dec 21, 2009 11:59 pm
Real name: Rich Feldman
Location: Santa Clara County, CA, USA

Re: Some noob high voltage divider questions

Post by Rich Feldman » Thu Feb 14, 2019 5:59 am

Thomas,
As I learned & described in recent NST Power thread, your divider design should BEGIN with the panel meter parameters.

If your 100 volt LED panel meter really has 50 K input resistance, it would draw 2 whole milliamps at full scale, and dissipate 200 mW in its internal divider. Leads me to doubt the numbers. Ebay sellers of cheap stuff like that often get details mixed up.

I used a nominal 30 volt panel meter that's supposed to have 100 K input resistance, for full scale current of 300 microamps. Actual measured resistance was a bit higher. It depended significantly on whether the panel meter was powered up, and on which direction the ohmeter was connected. I added an adjustable resistor in parallel with meter, to get desired sensitivity with existing HV resistor string. Calibrated it using a known HV source, which was much smalller than full scale range, but still a big improvement over the individual component tolerances stacked up.

Now your excellent 100 kV, 1 gigaohm resistor string delivers only 100 microamps (and dissipates 10 watts) at full scale. Would be a good fit to a common analog panel meter with 100 uA full scale range -- the voltage drop in meter coil would be unimportant.
Could also be an excellent fit to an inexpensive or free multimeter with 100 volt range and 1 megohm input resistance. What's the meter input resistance when switched off, or as you turn the knob between OFF and 100 volt range? If HV is on, the meter would have to eat 100 microamps with 100 kV behind it, no matter how the switches are set.

To make the instrument much less sensitive to meter impedance (and generally safer), why not include a built-in low side resistor and aim for a higher divide ratio? Like 1 M for 2000:1 ratio, as Dan suggested. Or approx 110 K, which in parallel with 1 M multimeter gives you 10,000:1 division ratio. 100 kV would be indicated as 10.0 volts, and the voltage would be only a little bit higher when meter is disconnected or switched off.

If your 100V panel meter actually comes in with resistance of 100 K or more, here's a circuit modification that is less disruptive to your design.
Skip one or two resistors in your HV chain, making the total about 900 M or 800 M. Then put a fixed resistor in parallel with your panel meter, bringing the low side R value down to 90 K or 80 K for 10,000:1 ratio. The external resistor would limit low-side voltage in case the panel meter failed or its input connection broke.
Mike echo oscar whisky! I repeat! Mike echo oscar whisky, how do you copy? Over.

Rex Allers
Posts: 274
Joined: Sun Dec 30, 2012 8:39 am
Real name:
Location: San Jose CA

Re: Some noob high voltage divider questions

Post by Rex Allers » Thu Feb 14, 2019 11:25 am

Thomas,

You are on the right track but here's my input. The meter input impedance you mentioned is a big factor.

The 10 100M resistors you want to use sound like they should to be up to the task but if you really want to measure up around 100 KV, the layout and connections will be important. Some kind of zig-zag long layout and smooth, rounded connections to avoid arc-over.

I have a divider board I found on ebay a few years back. I think it was used for measuring around 100 KV. I've used it so far to measure 50 KV, with only air around it, and had no problems.

Here's a pic of that board.
HV divider.jpg
It is almost 20" long. Each of the orangish-yellow rectangles is a 200 Mohm 1% resistor. I think 20kV rated and 2W. So the total dropping string is 2 Gohm (2E9 ohm). I chose to make it 0.1 V out for 1 KV in, or divide by 10000. There is also a 500pF 20KV capacitor in parallel with each of the 200M resistors. Not required but should make the output faster for quick transient changes on input HV.

Now on the resistive divider principles...
Here's a simple diagram of a resistive divider.
simp_divider.png
In that, Rd is the main divider string resistance, Rm is the lower resistance across which the output voltage is generated.

In your message you wanted a divide factor of 1000, as...
D = Vin / Vout = 100000 / 100 = 1000

Your circuit implies that
D = 1000 = Rd / Rm = 1E9 / 1E6 -- But that's not quite right.

In terms of R, the divide factor is Rm vs the Total resistance
D = Rtotal / Rm = (Rd + Rm) / Rm
So the actual divide ratio you presented is 1001. That is: D = (1E9 + 1E6) / 1E6

For design, if the givens are the divide factor and the big resistance, then we can get
Rm = Rd / (D-1)

So for divide by 1000,
Rm = 1E9 / 999 = 1001001 , so only about 1K more than 1 Meg.

Probably not enough to worry about, especially with 5% tolerance in Rd but it is the correct calculation.
Just wanted to point out the correct ratio is not D = Rd / Rm , but one more than that

Now for the measurement the actual resistance of Rm must include the parallel resistance of the voltmeter input resistance in parallel with any resistor placed in the Rm position.

If the meter resistance was actually 50 Kohm as you showed, then the effective Rm would be
1 M || 50k = 47.6 K

Way off. D ~= 21000
Hint, for gut feel, with two resistors in parallel, the combined resistance has to be less than the smaller one.

For this, a meter with 10M input would be better. Many cheap meters have that (note: I'd put them into a fixed range rather than auto) but aren't nice if you want an easy, default panel meter.

Maybe others can share how they've dealt with having a nice meter in this kind of a divider configuration.

An op-amp across Rm maybe could help, high impedance across Rm, low impedance out. But that's a whole other discussion for those who are capable.

---
One last thing on how I calibrated that divider board I showed. Goal: adjust out any tolerances and deal with any math errors or other unexpected factors.

The Rd in this case was 2E9 ohms. I used a meter with 10M input. I wanted 1 KV in to give 0.1 V on the meter (D = 10000). This means 100 KV in would give 10 V measured. I calculated I needed an effective Rm of a bit more than 200K. I used the following circuit for the "Rm".
Cal circuit.png
The pot in the 3 resistors of "Rm" and with the 10M meter lets you set the output to the (fairly) exact ratio desired. To do this, you want a rather accurate and moderately high voltage to apply to the HV input of the divider. I have a good quality multimeter that measures up to 1 KV. I also had an HV DC supply that could make a voltage in the 900 to 1000 V range.

With my good meter measuring Vin, I set the voltage to something like 990 V. This meter was left connected so any loading didn't change.

The meter used for the divider HV measurement was then calibrated by adjusting the "Rm" pot for an indication 0.099 V.

Especially with 5% resistors, this may not work. If the range of the pot is not enough to get the desired output measurement indication, you can change one or both of the fixed resistors shown to adjust the needed Rm into the range of the pot.

The scale on this (now HV) meter was probably set to a range of 2 V for good accuracy on mV readings for calibration, but it was previously tested that the input impedance was 10 Mohm for the lowest range and also for the one that covers full expected 10V.

How do you know the meter's input impedance? If you have two DMM's you can verify the input impedance of one by connecting their inputs together and putting the other in resistance mode. Then switch the measured one through the DC volt ranges to be tested.

So, -- summary -- if you can find a way to apply a moderately high voltage to do the kind of calibration described, you should be pretty good for higher voltages.

---
Hope that wasn't TLDR and had enough info to make sense. I'm open to questions or corrections.
Rex Allers

Thomas Henderson
Posts: 8
Joined: Tue May 02, 2017 6:56 pm
Real name: Thomas Henderson

Re: Some noob high voltage divider questions

Post by Thomas Henderson » Sun Feb 17, 2019 7:36 pm

Thank you all so much for the replies!! You have helped shine a little more light into my divider, and given me some amazing information. It looks like I am going to have to wait until my china display comes in, and get some real world measurements on it. I may go ahead and just eat the $2, and find another one that sports a 10meg resistance, so that it will arrive around the same time. I will update when supplies come in.

Post Reply