Newbie question - Can using osc probe x 10 with dmm for check HV CRT

[chihaxinh]

Hi every one

When repair OSC analog , i need checking HV relative with CRT ( 1.6KV ) so my DMM have max 1000V , can i using OSC Probe in mode x 10 with DMM to mesurement High Voltage ?

I done have DMM HV Probe !

Thank you so much !

[Gyro]

Sorry, but no. Don't even think about it. The 9M series resistor in the scope probe will have a much lower breakdown voltage than your meter and you will probably kill both.

It is possible to build a HV probe if you take care, particularly for relatively low voltages and low energy like a scope supply. You need to create a chain of high value resistors, used within their voltage rating, in a non-conducting tube. Something like 9 10M 1W resistors (90M total) in a plastic tube. Don't go above 10M because you will want to accurately measure their resistance values (to 'calibrate it'). Don't go above a few kV with it, and low power circuits only.

There are a few designs on the Web, be careful of insulation, particularly in a humid country like Vietnam.

P.S. You might want to use 10 x 10M resistors and use the meter on its current range instead (giving 10uA/kV). That wouldn't rely on the input resistance of your meter being exactly 10M.

[chihaxinh]

Many thank Gyro !

I want to measurement maximum 2kv . HV before multiple to 14Kv go to drirect CRT .

I searching on internet , so have 2 type DIY HV Probe , One type using 2 resistor to make resistor bridge , and other type using only 1 resistor will mixing with DMM input impedance .

What a best type for DIY HV probe ?

Thank you so much !

[Gyro]

You're welcome chihaxinh,

Unfortunately that introduces another problem  - the 2kV before the multiplier will be high frequency AC (the multiplier works as a charge pump using diodes and capacitors). I assumed that you were trying to measure the cathode bias circuit (often in the -1 - 2 kV DC range).

Trying to measure high frequency AC will cause problems, both with the parasitic capacitances of the resistor string and the frequency response of your meter.

In this case you might be forced to measure the DC voltage at the anode cap directly. This requires a much longer resistor chain to avoid voltage breakdown and reduce the load current. It is clearly possible (as it is achieved by commercial products and internet designs) but does require more care in construction and insulation... and more resistors.

In terms of the type, my preference would be for the single resistor (chain) type. Having the 2 resistor (divider) type means that you are losing some current through the bottom resistor, which increases the measuring load... and you still have to compensate it for the meter input resistance anyway.

As I indicated in my 'P.S.' using the meter in current mode is a good option, you no longer have to worry about meter input resistance (it is very low), and you are directly measuring the current through the resistor chain (in uA). It is both accurate and reminds you how much you are loading the voltage source. This is how the self-contained EHT testers with the moving coil meters (used to) work.

Sorry if I misled you a little regarding ease of construction, but 2kV high frequency AC is a very different problem to 2kV DC.

[chihaxinh]

Many thank Gyro

I'ts my mistake ! So i review schematics it's correct 2kV AC before multiple to 14Kv . And Vk it's -1975V ,a and Vp1 -625V , ok so we can bypass for mearsurement AC High Voltage.

Let's me diy DC HV probe in next time , i need repair focus function first !

Many thank you so much again !

when not sure i don't know ask someone so i ask in forum , Sory for take forum resource

[Gyro]

Hi chihaxinh,

From the circuit it looks as if everything is taken from the same transformer and the regulation (via IC1) is taken from Vk. That means that if Vk is correct then your EHT will be correct to if there were a fault in the multiplier then you would get some indication... probably a nasty smell.

This means that you can still use a 'low kV' probe to check that Vk (DC) is correct.

Please don't apologise for asking questions. If nobody asks questions then nobody gets to see the answers and maybe learn something. Even worse, they may do something dangerous without asking first. 

Chris


P.S. For your focus problem, it is worth checking the values of all the high value resistors in the focus circuit because they often drift with age.

[David Hess]

I agree with Gyro, a x10 probe will not work because the probe itself will breakdown somewhere above about 600 DC volts.

Some x100 probes are good to 2000 volts or higher and could be used but you have to do it right.  The probe expects the 1 megohm input resistance of an oscilloscope and your digital multimeter probably has a 10 megohm input resistance.  That is easy enough to solve by placing a 1.11 megohm resistor across the digital multimeter input when using the oscilloscope probe.

Gyro's suggestion of 9 x 10M resistors in series with your multimeter is probably easier than buying a x100 probe and 1.11 megohm resistor unless you already have the x100 probe.  I picked up a Fluke 80k-6 inexpensively on Ebay for making these kinds of measurements.

[chihaxinh]

Many thanks Gyro and David Hess

9x10M Ohm it's correct for Osc probe , but it's correct for DMM ?

I check resistor Voltage Divider need 10x 10Meg series with 10Meg input impedance DMM . You can see picture, if correct i go to market and buy resistor in weekend .

Very helpful if running well with DMM .

[David Hess]

It is 9 x 10 megohm resistors in series for the multimeter and 9 x 1 megohm resistors in series for a 1 megohm oscilloscope input which is what a x10 probe does but with 1 x 9 megohm resistor.

10 resistors in series yield this input current:

Iin = Vin / (10 * R)

Output voltage is input current through one resistor:

Vout = Iin x R

Substituting Iin:

Vout = Vin * ( R / (10 x R) )
Vout = Vin * (1 / 10)
Vout = Vin / 10

[chihaxinh]

Ah i deeper understand , hehe . 9 x10 M + 10 Meg input impedance = 10 resister series and we mesurement voltage drop in one resistor   

Sory for my concern !   

[tggzzz]

I've just bought two things you might like to have, but no, you can't have them

The 40kV meter is conceptually simple: a precision FET amplifier measures the current through a 30Gohm (30000Mohm) resistor and shows the result on a conventional analogue meter. The tricks are in the mechanical construction, i.e. probe handle to keep 40kV away from the operator, ditto flexible probe wire, and preventing breakdown in or around the resistor. The resistor is potted in a block that's about 2cm * 4cm * 6cm.

The 12kV scope probe is also conceptually simple (1000:1 divider plus frequency response compensation), but again the mechanical construction is vitally important. The manuals at http://w140.com/tekwiki/wiki/P6013 have exploded diagrams showing the internal construction.

[Gyro]

Ah i deeper understand , hehe . 9 x10 M + 10 Meg input impedance = 10 resister series and we mesurement voltage drop in one resistor   

Sory for my concern !   

Yes, you've got it. 

The 9 x 10M resistors + the meter input resistance will give you a divide by 10 divider, so your meter will display 200V for 2kV input.

If you want to get best accuracy, then measure the actual resistance of each 10M resistor (they might be +/-5% tolerance or more). Then add all your resistance measurements together to see how close it is to 90M total. For example, if the actual total is 89M then your meter display will be 1% high. With up to 5% tolerance resistors you  probably don't need to bother, but if you can only find 10% tolerance ones at the market then becomes more important that you calculate the actual value.

P.S. A trap that many people fall into... Don't hold the 10M resistor leads to the probe tips with your fingers when measuring them. Your body resistance will make the reading wrong. 

[tggzzz]

The 9 x 10M resistors + the meter input resistance will give you a divide by 10 divider, so your meter will display 200V for 2kV input.

Have you considered that the meter's input resistance might not be a well-controlled value?
Have you considered 10*10Mohm + 100kohm, and measuring the voltage across the 100kohm resistor with a "10Mohm" meter. 1kV -> 1V, which is suitable for most DMMs.

[joeqsmith]

Short clip showing an attenuator I made to work with my meter. 

[Gyro]

The 9 x 10M resistors + the meter input resistance will give you a divide by 10 divider, so your meter will display 200V for 2kV input.

Have you considered that the meter's input resistance might not be a well-controlled value?
Have you considered 10*10Mohm + 100kohm, and measuring the voltage across the 100kohm resistor with a "10Mohm" meter. 1kV -> 1V, which is suitable for most DMMs.

That's why my other suggestion (1st and 3rd replies) was to use the lowest current range and measure the current through 10 x 10M resistors... To make meter resistance an insignificant factor. Assuming a 200uA lowest range 2kV would give 20.0uA (3 significant digits on a 3 1/2 digit DMM).

You're right 10x 10M + 100k forms a 1000:1 divider. Paralleling a 10M input resistance meter across the 100k drops it to 99k which would cause 2kV input would display 1.978V on the meter.

Hopefully the OP has the meter instruction leaflet and it specifies input resistance and tolerance.

Now we've probably really confused him. 

[joeqsmith]

Best thing for DC would be to just find a used one in good condition or get a new one.  I'm too cheep for that and attempted to make my own wideband probe.  I use it with both the meters and the scope.   There were 5 parts to the video showing the progress.   These two show the schematics and then the final probe in operation.   May give you some idea what is involved.   

[Gyro]

@joeqsmith:
The problem is that due to the OP's location, he may not have ready access to used HV probes, he is looking for a quick, cheap and easy way of allowing his meter to read 2kV (Edit: DC now... Vk).

@tggzzz:
What about the following (assuming the 9 x 10M series resistor approach).
- Find a reasonably stable approx 10 volt source.
- Measure it accurately with the DMM on its 10V range.
- Add the probe and measure the resulting divided by 10 voltage.

This is relying on the DMM having the same input resistance on the 100V, 10V and 1V ranges, but not that unreasonable an assumption.

[tggzzz]

@joeqsmith:
The problem is that due to the OP's location, he may not have ready access to used HV probes, he is looking for a quick, cheap and easy way of allowing his meter to read 2kV (Edit: DC now... Vk).

@tggzzz:
What about the following (assuming the 9 x 10M series resistor approach).
- Find a reasonably stable approx 10 volt source.
- Measure it accurately with the DMM on its 10V range.
- Add the probe and measure the resulting divided by 10 voltage.

This is relying on the DMM having the same input resistance on the 100V, 10V and 1V ranges, but not that unreasonable an assumption.

The uncertainty about that assumption is unnecessary and - so far as I can see - offers no significant advantages.

N.B. some DVMs have input resistances that vary with time, since they are bootstrapped to minimise the input current, and it takes time for the bootstrap to settle. However, that is unlikely to be relevant to the OP's problem.

[Gyro]

N.B. some DVMs have input resistances that vary with time, since they are bootstrapped to minimise the input current, and it takes time for the bootstrap to settle. However, that is unlikely to be relevant to the OP's problem.

Sure, input bias drift is an obvious source of error for a high impedance bench DMMs with bootstrapped inputs (like my old Datrons for instance) but definitely not relevant for a simple handheld DMM like the OP has. His thread title is "Newbie question - Can...". He's looking for simple advice. If we're going to that level then voltage coefficient of the resistors is probably a more relevant error source (that we can't do anything about) we don't need to be worrying him with such esoterics.

Sorry if that came out as a rant, but I think we (as a group) have a tendency to go off into long technical discussions that don't match the specific needs of the questioner.

The OP probably needs, what?, 2-5% measurement accuracy for checking his analogue scope supplies? It's a question of whether a 10:1 divider using the 10M input impedance of his DMM or a 1000:1 divider. He will need to measure the values of the resistors for both to achieve even that level of accuracy though. I have to say, I'm now leaning towards the 1000:1 even though it's a bit more complex to build than a simple series probe, however I think both will do the job. I guess he can test them both with a lower voltage source, either 10V in 1V out or 10V in 10mV out (input impedance on the bottom range is sometimes different to the others on low-end meters though).


@chihaxinh: You should probably purchase at least 10 x 10M resistors (get some spares) and make sure you have some 100ks too. Also see what information the instructions for your DMM have for the input resistance. (There's no point in building something more complicated than you need).

[tggzzz]

When selecting resistors, be very careful that the breakdown voltage is appropriate. Not much use having a 10Mohm resistor with a breakdown voltage of 100v!

Once one resistor breaks down, more stress is put on the others in the chain - making them more likely to break down. And that's a way 2kV could propagate to a DMM and your fingers.

IMHO, 5% (~120V) would be undesirably inaccurate in this application. 1% (24V)would be acceptable, and it is readily achievable.

[Gyro]

That's why I specified 1W resistors in Reply #1 

The OP has the service manual, It will tell him the required measurement accuracy.

[tggzzz]

1W may or may not be sufficient. A quick look at Farnell indicates there are 10Mohm 1W 200V resistors, which is inadequate for the OP's requirement.

If you need to withstand voltage, specify the voltage! Anything else either indicates or engenders dangerous misunderstandings.

[Gyro]

Fair point tggzzz, I've never come across a 1W resistor rated that low.

EDIT: What on earth did you find that was that low? Farnell is down for maintenance at the moment but looking at RS, even bog standard 0.25W TE connectivity carbon film are 200V limiting element voltage (400V overload) and 0.5W carbon / 0.6W metal film are 350V element limiting (700V overload). The lowest 1W I found were also 350V (700V overload). That's in line with my expectations.  Are you talking smd or something? I'm really curious about what you found that was that bad? I'd like to know so that I can avoid it!

[tggzzz]

Fair point tggzzz, I've never come across a 1W resistor rated that low.

EDIT: What on earth did you find that was that low? Farnell is down for maintenance at the moment but looking at RS, even bog standard 0.25W TE connectivity carbon film are 200V limiting element voltage (400V overload) and 0.5W carbon / 0.6W metal film are 350V element limiting (700V overload). The lowest 1W I found were also 350V (700V overload). That's in line with my expectations.  Are you talking smd or something? I'm really curious about what you found that was that bad? I'd like to know so that I can avoid it!

Low breakdown voltage => small => SMD. Yes, I realise you wouldn't be so silly as to use one, but ... 😃

[joeqsmith]

@joeqsmith:
The problem is that due to the OP's location, he may not have ready access to used HV probes, he is looking for a quick, cheap and easy way of allowing his meter to read 2kV (Edit: DC now... Vk).

@tggzzz:
What about the following (assuming the 9 x 10M series resistor approach).
- Find a reasonably stable approx 10 volt source.
- Measure it accurately with the DMM on its 10V range.
- Add the probe and measure the resulting divided by 10 voltage.

This is relying on the DMM having the same input resistance on the 100V, 10V and 1V ranges, but not that unreasonable an assumption.

Reason for the video showing the attenuator.   

[Gyro]

Low breakdown voltage => small => SMD. Yes, I realise you wouldn't be so silly as to use one, but ... 😃

   Not Again! Why cite it as a serious objection then! Even if the OP used a string of 10 1/4W bog standard carbon film resistors it would just about manage 2kV (maximum resistor rating) and 4kV Overload / dielectric breakdown. With the chain of 1W wire ended resistors there's no way he's going to suffer breakdown as long as he puts them in a straight line in an insulating tube as I already advised him. Also remember that he's working on a low energy circuit too. He's not going to get to the market and be able to start quizzing sellers on datasheet voltage rating, wattage is all they will know.

We REALLY should be trying to support the OP to achieve what he wants to do, within his available resources, rather than engaging in pointless point scoring and pi**ing contests! Why don't YOU start addressing him directly with helpful suggestions rather than picking holes in my attempt to help him. 


@joeqsmith: That's a really instructive video that takes a serious and methodical approach to the problem, thanks. 

[chihaxinh]

Many thank for many advice

I think in easy question allway make hard answer like "Why does it rain? ? " ! Hehehe

So if you have any advice , i can learn and i can make or do it in next time ( for collect resistor, capacitor, isulation material ...etc ) .

You know, in my country it's very hard to buy good brand resistor, capacitor ...etc like vishay, bc phillip , epcos....

So when i ask and take answer , i need 1 2 week to collect item.

Yes, simply question : I really want to quick way check HV with DMM with simple material . For quick check HV Board CRT .

Many thank for your support !

PS: But any conversation about mearsurement technical , i allway hope to see

[tggzzz]

Low breakdown voltage => small => SMD. Yes, I realise you wouldn't be so silly as to use one, but ... 😃

   Not Again! Why cite it as a serious objection then! Even if the OP used a string of 10 1/4W bog standard carbon film resistors it would just about manage 2kV (maximum resistor rating) and 4kV Overload / dielectric breakdown. With the chain of 1W wire ended resistors there's no way he's going to suffer breakdown as long as he puts them in a straight line in an insulating tube as I already advised him. Also remember that he's working on a low energy circuit too. He's not going to get to the market and be able to start quizzing sellers on datasheet voltage rating, wattage is all they will know.

We REALLY should be trying to support the OP to achieve what he wants to do, within his available resources, rather than engaging in pointless point scoring and pi**ing contests! Why don't YOU start addressing him directly with helpful suggestions rather than picking holes in my attempt to help him. 

You have, responsibly, admitted you gave the OP incorrect advice and guidance (a watts rating instead of volts rating).
The OP is a newbie and might not appreciate that - and might blindly follow that guidance.
Dangerous voltages are involved; that's the whole reason for this thread.

Hence the points I made are necessary to prevent the OP (or anyone else reading the thread) endangering themselves by following your deficient guidance. Given that objective, I wonder how many people would regard that as "pointless point scoring"?

[Gyro]

Many thank for many advice

I think in easy question allway make hard answer like "Why does it rain? ? " ! Hehehe

So if you have any advice , i can learn and i can make or do it in next time ( for collect resistor, capacitor, isulation material ...etc ) .

You know, in my country it's very hard to buy good brand resistor, capacitor ...etc like vishay, bc phillip , epcos....

So when i ask and take answer , i need 1 2 week to collect item.

Yes, simply question : I really want to quick way check HV with DMM with simple material . For quick check HV Board CRT .

Many thank for your support !

PS: But any conversation about mearsurement technical , i allway hope to see

That's good  chihaxinh, I was worried that we might have scared you away!

It sounds as if tggzzz is best equipped to give you your 'shopping list - for the resistors at least, hopefully he will provide that to you shortly. 

Regarding the other things:

-an insulating plastic tube capable of taking the diameter of the string of resistors.
- some heatshrink sleeving,. High voltage test lead wire if you can get it.
- 4mm plugs to fit your meter (you don't want the connection between the probe and meter coming undone).
- Also a secure clip to make your ground connection (again you don't want the ground connection comming undone as your meter will float up to 2kV!).
- Anything else that tggzzz can think of.

[Gyro]

You have, responsibly, admitted you gave the OP incorrect advice and guidance (a watts rating instead of volts rating).
The OP is a newbie and might not appreciate that - and might blindly follow that guidance.
Dangerous voltages are involved; that's the whole reason for this thread.

Hence the points I made are necessary to prevent the OP (or anyone else reading the thread) endangering themselves by following your deficient guidance. Given that objective, I wonder how many people would regard that as "pointless point scoring"?

We'll I think others can judge the merits of the 'discussion' from now on!

In the meantime the OP has come back and clarified his needs and available resources again (thankfully). It's is time for you as a fellow super-contributor to 'Put up' and give him suitable and practical advice, relevant to his situation. Please do not let him down!

[tggzzz]

In the meantime the OP has come back and clarified his needs and available resources again (thankfully). It's is time for you as a fellow super-contributor to 'Put up' and give him suitable and practical advice, relevant to his situation. Please do not let him down!

Sigh.

I already have - and you have replied to it and thought it was correct useful advice! So clearly you were aware of it, and have forgotten it.

I don't think I can help you further at this point.

[Gyro]

I don't understand - the OP needs concrete advice on what resistors he should buy for a safe implementation at his location. You clearly have an opinion. Please provide appropriate advice to him!

[joeqsmith]

@joeqsmith: That's a really instructive video that takes a serious and methodical approach to the problem, thanks. 

Anyone working with potentially lethal circuits should.

A few things that may not have been clear from the video are effects of loading the circuit you are trying to measure.  The unit I made does have a metal shield internal to the box to help minimize noise.   It's fairly low impedance so noise was not much of an issue.   The higher bandwidth probe I showed is much higher impedance and required better shielding to get the performance I was looking for.   

One other video that may be worth watching.   Rather than using a 10X oscilloscope probe, maybe a low cost 100X probe would work.  Then again....   In this video, I tested a cheap HV probe I bought from Amazon. 

[Gyro]

That was an entertaining test to destruction and teardown, that internal construction was shocking, I wonder if there was any science in the specific length of insulating film wrapped inside the shield to obtain the right capacitance or just winged it.

@Joe, I'm still looking for support in advising the OP on resistor choice for his situation. Just to reiterate, he's in Vietnam and has limited access to decent components. His measurement need is 2kV DC (scope internals). My best is 9 or 10 (depending on attenuator style) 10Meg 1W axial resistors in a long series chain. The way I figure it is that minimum element rating for resistors of that size will be in the order of 350V with dielectric breakdown occurring at around 700V (say 500V accounting for high humidity). That should give an in-rating limit of about 3.5kV (for 10 resistors) and breakdown at 5kV. Obviously a higher safety margin would be preferable, but roughly x2 is probably all that can practically be achieved. My best advice to the OP would be to pick the largest physical body size available, with closer tolerance being an indicator of better quality. The whole assembly would have to be mounted inside a plastic tube and preferably overall heatshrinked too. Can you think of anything better / safer given the non-avaliability of an off-the-shelf product?

[David Hess]

I did not comment about the breakdown voltage of the 10 megohm resistors simply because I assumed 1 watt values were suggested for this very reason and the original poster is limited in what he has access to.

As far as accuracy, just measuring the resistors may not be enough.  As high a calibration voltage as available should be used because the effect of voltage coefficient of resistance on accuracy can be considerable.  For best accuracy when calibration is not available, it may be better to use an external shut resistor and high input impedance voltmeter so that at least some of the voltage coefficient of resistance cancels out.  Voltage derating the resistors helps minimize the voltage coefficient of resistance as well.

Have you considered that the meter's input resistance might not be a well-controlled value?
Have you considered 10*10Mohm + 100kohm, and measuring the voltage across the 100kohm resistor with a "10Mohm" meter. 1kV -> 1V, which is suitable for most DMMs.

I am not sure what kind of junk DMMs you are used to but all of mine have an input resistance tolerance of better than 1%.

You cannot always directly measure the input resistance of a DMM.  Modern DMMs may have problems doing this because of charge pumping from their analog to digital converters.  Old DMMs which use external input buffers lacking chopper stabilization before their analog to digital converter do not suffer from this issue.

[joeqsmith]

That was an entertaining test to destruction and teardown, that internal construction was shocking, I wonder if there was any science in the specific length of insulating film wrapped inside the shield to obtain the right capacitance or just winged it.

@Joe, I'm still looking for support in advising the OP on resistor choice for his situation. Just to reiterate, he's in Vietnam and has limited access to decent components. His measurement need is 2kV DC (scope internals). My best is 9 or 10 (depending on attenuator style) 10Meg 1W axial resistors in a long series chain. The way I figure it is that minimum element rating for resistors of that size will be in the order of 350V with dielectric breakdown occurring at around 700V (say 500V accounting for high humidity). That should give an in-rating limit of about 3.5kV (for 10 resistors) and breakdown at 5kV. Obviously a higher safety margin would be preferable, but roughly x2 is probably all that can practically be achieved. My best advice to the OP would be to pick the largest physical body size available, with closer tolerance being an indicator of better quality. The whole assembly would have to be mounted inside a plastic tube and preferably overall heatshrinked too. Can you think of anything better / safer given the non-avaliability of an off-the-shelf product?

Maybe you could post some clear pictures detailing your construction techniques to help guide them along.   The high voltage parts will show the life derate curves in the data sheets.  Something to consider.   

Yea, that 100X probe is not something I would recommend even though it was not very expensive.  I gave up and rolled my own.   I use them mostly for 2KV and under. 

[tggzzz]

As far as accuracy, just measuring the resistors may not be enough.  As high a calibration voltage as available should be used because the effect of voltage coefficient of resistance on accuracy can be considerable.  ... Voltage derating the resistors helps minimize the voltage coefficient of resistance as well.

Good point; I should have mentioned that.

Have you considered that the meter's input resistance might not be a well-controlled value?
Have you considered 10*10Mohm + 100kohm, and measuring the voltage across the 100kohm resistor with a "10Mohm" meter. 1kV -> 1V, which is suitable for most DMMs.

I am not sure what kind of junk DMMs you are used to but all of mine have an input resistance tolerance of better than 1%.

I use Flukes, and have never made measurements where the tolerance of the input resistance is important. Hence I've never looked at that specification.

I have no idea of the OP's equipment, and hence merely noted that the tolerance should be considered before relying on it.

You cannot always directly measure the input resistance of a DMM.  Modern DMMs may have problems doing this because of charge pumping from their analog to digital converters.  Old DMMs which use external input buffers lacking chopper stabilization before their analog to digital converter do not suffer from this issue.

Just so; ISTR I mentioned that point.

[Gyro]

Maybe you could post some clear pictures detailing your construction techniques to help guide them along.   The high voltage parts will show the life derate curves in the data sheets.  Something to consider.   

In terms of construction, I was simply thinking 10 (or 9) resistors soldered end to end and enclosed in a plastic tube. The actual detail would depend on what the OP has available, if flexible then the ends could be constricted by cable ties, otherwise retained by epoxy or other means. On reflection I would avoid heatshrink, my picoammeter experiments have shown it to be relatively conductive and I don't know how this would scale with voltage. Termination at each end would be flexible leads so that the OP could solder to the Vk test point rather than using it handheld (less risk of direct contact). For the voltage divider approach I guess this would simply be a matter of hardwiring the 100k resistor across the DMM input jacks.

In terms of life derating curves, I very much doubt that the OP will have access to relevant datasheets, he has already said that he doesn't have access to quality branded parts. We are very much depending on physical size (creepage path length) as an indication of voltage capability. This sounds like pretty much a one-time test anyway.

As far as accuracy, just measuring the resistors may not be enough.  As high a calibration voltage as available should be used because the effect of voltage coefficient of resistance on accuracy can be considerable.  ... Voltage derating the resistors helps minimize the voltage coefficient of resistance as well.

Good point; I should have mentioned that.

I did mention it, but only in terms of something that we don't have control over as it depends on the resistors that the OP has access to. I don't see a way around this as he will not have access to a higher calibration voltage.

Have you considered that the meter's input resistance might not be a well-controlled value?
Have you considered 10*10Mohm + 100kohm, and measuring the voltage across the 100kohm resistor with a "10Mohm" meter. 1kV -> 1V, which is suitable for most DMMs.

I am not sure what kind of junk DMMs you are used to but all of mine have an input resistance tolerance of better than 1%.

I use Flukes, and have never made measurements where the tolerance of the input resistance is important. Hence I've never looked at that specification.

I have no idea of the OP's equipment, and hence merely noted that the tolerance should be considered before relying on it.

This is why I still favour measuring the current through the resistor chain rather than a voltage divider (either external divider or using the DMM input resistance as the lower part of the divider). If the OP uses his lowest current range, assuming that it is low enough (200uA) then the meter impedance ceases to be an issue. It is simply measuring current through a pre-measured resistor chain. Additionally there is no added complication of an external 'bottom resistor'. Admittedly it does limit resolution as 2kV will come out as 20.0uA and still doesn't address the voltage coefficient, but I don't know how we can, other than recommending buying physically large metal film.


It seems that we are mainly dependent on details of the DMM at this point...

@chihaxinh: Can you please give us some details of your DMM? Does it have a manufacturer and model number that we might be able to find outside your country? Do the instructions specify the input resistance on voltage ranges (eg. 10M ohm +/-1%)? What is it's lowest current range?

[chihaxinh]

Hi Gyro

I using Agilent 34401a - Hioki 3238 - Sanwa PC5000a - So with Agilent i have 10M input impedance in 100mv to 100V range - 1M in > 100V
Hioki 3238 same Agilent
Sanwa fix in 10Meg input impedance

I alway interesting method measurement like tube , solid state curve , Inductor , Capacitor measurement ...etc

@Tggzz
So i hope  tggzzz draw clearly method you want show me ( using jfet or opamp for reading current of 30Meg resistor )

[Gyro]

Ah, you have access to rather better test equipment than I had realized your description of component availability and Newbie question tread title  . In fact you have rather more accurate test equipment (the Agilent) than I have! 

It seems that there are more measurement options available than I had thought. It sounds as if you can very easily calibrate a resistor chain (with decent components?) and accurately measure the current through them.

[joeqsmith]

I using Agilent 34401a - Hioki 3238 - Sanwa PC5000a

I am surprised you can get your hands on some decent meters but not a probe or resistors.   

Sounds like this is what is being recommended. 

[chihaxinh]

Ah, you have access to rather better test equipment than I had realized your description of component availability and Newbie question tread title  . In fact you have rather more accurate test equipment (the Agilent) than I have! 

It seems that there are more measurement options available than I had thought. It sounds as if you can very easily calibrate a resistor chain (with decent components?) and accurately measure the current through them.

Hi Gyro
I'm not graduation major  Electrical and Electronic Engineering , my job it's IT system admin . So Electronic it's my hobby , every thing i learn by reading ebook, forums...etc==> i'm newbie

In time , I like simple method for quick test ( dont care about accurate ) . So if i learn best method , high acurrate i will buy some part and buy in next time.

You know

I am surprised you can get your hands on some decent meters but not a probe or resistors.   

Sounds like this is what is being recommended. 

Yes, i buy all from one factory bankrupt . If you buy new , it's very very expensive in VN .
In VN can buy from Mouser, Digikey but take 15-20 day for shipping. Some time when need repair equiment , i dont have time for wait delivery !

[tggzzz]

@Tggzz
So i hope  tggzzz draw clearly method you want show me ( using jfet or opamp for reading current of 30Meg resistor )

Some of the issues are mechanical, i.e. ensuring adequate materials and construction techniques. Some of the issues relate to component characteristics. While I can mention important characteristics, there is no point in my specifying components, since I would almost certainly specify a part number that you could not obtain.

It would be unwise of me to quickly tell a beginner how to do something potentially hazardous, unless I was there to check what they actually did. I would probably miss out a detail that is obvious to me but not to you. I don't know you, don't know how much you do and don't know, and don't know which bits of of any instructions might become "lost in translation". There's an old engineering saying: it is impossible to make something beginner-proof, since beginners are so damn ingenious

[Gyro]

I using Agilent 34401a - Hioki 3238 - Sanwa PC5000a

I am surprised you can get your hands on some decent meters but not a probe or resistors.   

Sounds like this is what is being recommended. 

Pretty much, yes. The resistors shown in the video look like 1/4W metal film, so an element limiting voltage of about 250V (though higher breakdown). I was suggesting physically larger ones, both for higher voltage capability and hopefully better voltage coefficient. The ptfe tape wrap is a good idea, I've found that some heatshrink is conductive at the nA level at relatively low voltages (carbon black pigment content?). I would have liked to have seen some strain relief, but for short term use it is probably ok. The deformation of the heatshrink would hold the resistor chain intact and plenty of layers shown. Oh, I hate seeing people heatshrinking with a flame.

Looking at the Sanwa PC5000a pdf specifications (nice 50000 count meter + optional extra digit)...
- They confirm that the DC voltage input resistance is 10M ohms on all ranges but no tolerance stated. Accuracy 0.05% rdg+2 digits on 500V range (1kV range is 0.1% rdg.)
- It has a 500uA current range, resistance burden 100R (negligible in this context), resolution 10nA, 0.15% rdg+20 digits.

Given the resistor quality and voltage coefficient limitations of the resistor chain, I guess in practice it could be used either on the voltage range (9 resistor chain) or the current range (10 resistor chain).

I can't think of any particular benefit, given the OP's stated requirements and component sourcing, of building a complete external divider (with 100k) and risking the high impedance Agilent or Hioki. Not until he wants to do a proper permanent setup anyway - it's not worth it until you use decent precision resistors and build a proper enclosure (Edit: and some protection).

[David Hess]

This is why I still favour measuring the current through the resistor chain rather than a voltage divider (either external divider or using the DMM input resistance as the lower part of the divider). If the OP uses his lowest current range, assuming that it is low enough (200uA) then the meter impedance ceases to be an issue. It is simply measuring current through a pre-measured resistor chain. Additionally there is no added complication of an external 'bottom resistor'. Admittedly it does limit resolution as 2kV will come out as 20.0uA and still doesn't address the voltage coefficient, but I don't know how we can, other than recommending buying physically large metal film.

If I wanted an accurate measurement and had no high voltage calibration source and inadequate specifications for the resistors, then what I would do is use a high input resistance DMM (electrometer) or buffer with an external series/shunt resistor divider.  This way the voltage coefficient of resistance for the resistive divider using stacked identical resistors should cancel out to some extent; each resistor has the same voltage across it.

If you buy resistors intended for high voltage applications, they will have a specified voltage coefficient of resistance and this allows you to voltage derate them for the accuracy you require.

[Gyro]

Sure David, I 100% agree, but unfortunately above 20V the high impedance voltmeters (the Agilent and Hioki) will switch in their 10M ohm resistive dividers and no longer be high impedance. That forces a change to a 100:1 divider to get to (barely scrape) the 20V limit. In that case he would have different resistor values (and maybe types) for the 10M and 100k resistors, with unequal voltage distribution. The only way to do 100:1 by stacking identical resistors to get equal voltage distribution would have to be a chain of 100 1M resistors (as far as I can see anyway).

If you buy resistors intended for high voltage applications, they will have a specified voltage coefficient of resistance and this allows you to voltage derate them for the accuracy you require.

Again, agreed but the OP doesn't seem to have time or ability to get properly specified resistors at the moment... a mistake in my view.

Chris.

Edit: I have to admit, I am beginning to lose a little bit of my sympathy (probably an unworthy of me ) since learning of the OPs high end test equipment - I'd be tempted to sell either the Agilent or the Hioki to fund the purchase of other useful gear. Even that Sanwa is no cheapee, in fact it looks like a gem of a hand-held, accurate, feature packed and with a very high resolution option.

[tggzzz]

Sure David, I 100% agree, but unfortunately above 20V the high impedance voltmeters (the Agilent and Hioki) will switch in their 10M ohm resistive dividers and no longer be high impedance. That forces a change to a 100:1 divider to get to (barely scrape) the 20V limit. In that case he would have different resistor values (and maybe types) for the 10M and 100k resistors, with unequal voltage distribution. The only way to do 100:1 by stacking identical resistors to get equal voltage distribution would have to be a chain of 100 1M resistors (as far as I can see anyway).

If you have a 1000:1 ratio (10*10M+1*100k), then there would only be a negligable 2V across the 100k. There would be no requirement to match the voltage coefficient of resistance; only that of the 10M resistors need be assessed. And I'd measure that voltage with a 3.5 digit handheld DMM.

But personally I'd be more concerned about leakage/arcing/breakdown, particularly w.r.t. mechanical construction.

[Gyro]

If you have a 1000:1 ratio (10*10M+1*100k), then there would only be a negligable 2V across the 100k. There would be no requirement to match the voltage coefficient of resistance; only that of the 10M resistors need be assessed.

Doesn't that just bring us back to the problem of the unknown voltage coefficient of the 10M resistors, which David was trying to compensate by using a chain of identical resistors? Without that if becomes pretty academic whether you put the bottom 100k (with its own tolerance) in, or just measure the current through the chain (which is what the 100k is doing, acting as a shunt is doing anyway). It doesn't sound as if the OP actually has anything as humble as a 3.5 digit DMM.

[tggzzz]

If you have a 1000:1 ratio (10*10M+1*100k), then there would only be a negligable 2V across the 100k. There would be no requirement to match the voltage coefficient of resistance; only that of the 10M resistors need be assessed.

Doesn't that just bring us back to the problem of the unknown voltage coefficient of the 10M resistors, which David was trying to compensate by using a chain of identical resistors? Without that if becomes pretty academic whether you put the bottom 100k (with its own tolerance) in, or just measure the current through the chain (which is what the 100k is doing, acting as a shunt is doing anyway). It doesn't sound as if the OP actually has anything as humble as a 3.5 digit DMM.

The keyword in my statement, which I even emphasised is "match".  For some reason you chose to omit that in your quote, and might not have even read it before quoting.

Apart from that, your "Doesn't that just bring us back..." statement merely reiterates what is in my statement; therefore I don't see the point of your question.

[Gyro]

Huh?  Actually I just cut and pasted your text into a quote box, it wasn't worth hitting quote button on your reply as it was the next one up - I've no idea why the italicize didn't go with it, maybe it doesn't if you just paste into reply window unless you <ctrl>I a word, you really think I'd waste my time of retyping the stuff you're writing?  .

I didn't say anything about the 100k resistor having to match (about time you started reading!). All I said was that it had it's own tolerance, not VOLTAGE tolerance just tolerance - 1%, 5% whatever! (EDIT: It's an extra source of error).

What I was saying was that whatever is there or not there, it doesn't help us with the fundamental issue of the voltage coefficient of the 10M resistor chain - the only suggestion that does is David's (using identical resistors for the divider and high impedance measurement of the bottom one. As I pointed out this good idea unfortunately fails because the Agilent and Hioki do not maintain their high impedance input up to a high enough voltage before cutting in their 10M resistive dividers.

I don't know why you are so keen to keep putting in the 100k resistor. has it not occurred to you that as soon as you introduce the quote "3.5 digit handheld DMM" with its 10M input resistance across The 100K you immediately add a 1% error to the divider Just like that or didn't you do the sums? 

The Sanwa meter is quite capable of reading the current through a chain of 10 x 10M resistors with 10nA resolution and sufficient accuracy to match any 10M resistors is likely to buy! No need for a 100k resistor, no need for the extra connections to include it, no need to consider it's tolerance (%!). Just a simple chain of connections.

Now I'm sure David understands this, I'm sure Joe understands this, I'm even pretty sure that the OP understands this.

@tggzzz at this point you are either being deliberately obtuse, have some troll type need to keep this thread circling or some desperate self gratification need to have the last word? However I'm finding the Return on Invested effort becoming distinctly negative.

[tggzzz]

@tggzzz at this point you are either being deliberately obtuse, have some troll type need to keep this thread circling or some desperate self gratification need to have the last word? However I'm finding the Return on Invested effort becoming distinctly negative.

Pot, kettle, black.

I believe my approach is easier more predictable than yours. I doubt we will agree on that.

[Gyro]

We certainly won't! You haven't addressed the tolerance contribution of the 100k resistor or the error introduced by paralleling the meter across It. Or the additional complexity involved in physically implementing it.

[chihaxinh]

I really dont want make a war in topic !  Like a said i really want to learn from basic to complex method to measurement hv. 

So if everybody have good method pls advice me ! 

[Gyro]

@chihaxinh (and tggzzz) I didn't mean (or want) to participate in a war.

My advice is to go with 10 x 10M ohm (the best quality, biggest / highest voltage ones you can get, not less than 1W metal film package size). Connect them in a very well insulated chain, as demonstrated in that video. PTFE (sleeve or tape) is a good method of reducing leakage caused by the heatshrink. Don't forget to accurately measure each resistor so that you can add them and get a more accurate resistance value for the whole chain.

Then use your Sanwa multimeter in Current mode. It has very good low current resolution and its internal 100 ohm current shunt will contribute one thousand times less error than an external 100k one.

[Edit: Actually the thousand times less error contribution probably isn't correct, however the 100 ohm shunt will be more accurate than an external resistor and the meter will have been calibrated to measure the voltage drop across its exact value. This removes one source of error, leaving only the accuracy spec of the meter and the accuracy of the 10 x 10M resistor chain.]

If you go to the time / trouble of getting some high quality (close tolerance, low voltage coefficient) high voltage 10M resistors and mounting them more permanently then you have the makings of a very good high voltage probe for general use.

I hope this helps. 

P.S. This is my opinion of course. You need to take the responsibility for building carefully and insulating sufficiently (if in doubt, add more) and using it responsibly. I guess that if you are already poking around inside the scope then you have already taken that decision.