Measuring Low Ohm with micro-volt resolution precisely

Using Arduino General Electronics
1

691×229 57.1 KB

Hello, I am trying to make a 4 probe resistivity meter. My setup is something like this pic. I am supplying the specimen material with a high voltage variable DC source (50V to 400V) and trying to make the measurement VD.

Please check the sample design attached. I am trying to measure the voltage drop across R1 where R1 is typically very low (micro/milli ohm range). Everything above the black line is system and below the black line is the meter. The loads are maximum 1K. Using 22bit MCP3551 and 16bit ADS1115 ADCs for measurements.

My idea is to use divider networks and a gain stage to adjust the high voltage DC to adc levels. Use filters to remove the unwanted noise etc. Then measure the value using the adc and arduino.
Not sure if this design idea is suitable to measure this low voltage drop across R1. Also need help with the "filter buffer gain amplifier" stage.

Calculations:
Least count for the 22bit adc at 5V reference = 5/(2^22-1) = 1.2uV

Why this peculiar setup? Why can't I use a fixed current source?
Because I am also measuring the IP (induced polarization) of the material for which the higher the voltage the better. In IP we measure the decay of voltage when the supply is turned off. So basically when the device is charging the material I want to measure the resistivity and when the supply is turned off, I want to measure the voltage decay across R1

Sample Design.jpg
Sample Design.jpg611×538 70.8 KB

2

Not sure if this design idea is suitable to measure this low voltage drop across R1.

Nope.

The more interesting issue is the effect of applying 50-400 V across a sample with resistance in the micro/milliOhm range. Take a moment to ponder the current that will pass through it. Hint: use Ohm's Law.

3

Yep. And needing to measure microVolt resolution for a voltage like 50 Volt is kind of not necessary. Because the relative error would be relatively tiny.

And JRem has a good point. Better calculate what sort of current and power dissipation to expect.

4

Perhaps I'm missing something but the schematic shows 2 1k resistors in the loop. Meaning at 400V the current would be ~ 200 ma. And each 1K will dissipate 40 watts.

My question is the OP might do better if they defined the basic goal (accuracy sample range etc).

My guess is the sample current would be "known" if the power supply voltage were known. However resistors capable of dissipating 40 watts are not likely to be terribly stable.

Issues I see are:

  1. Getting meaningful data from the 2 lsb's of the 22 bit daq is near impossible. Even with long integration time.

  2. When measuring µV one must be careful of the materials being uses else a termocouple will be formed causing temperature related error.

Jon

5

He had another thread. He wants to measure the resistance of a conductor under load. He was talking about measuring nanovolts. All the added 'stuff' just adds more error.

But Ohm's law. You can measure voltage and current to get the resistance. Can he build a volt meter that will accurately read in the nanovolts at 400VDC and read it with an Arduino? Does he have a current source that is that accurate? I have my doubts.

6

No doubts here, he can't.

7

jremington:
Nope.

The more interesting issue is the effect of applying 50-400 V across a sample with resistance in the micro/milliOhm range. Take a moment to ponder the current that will pass through it. Hint: use Ohm's Law.

There is load on both side of the resistor, please see the diagram.

JohnRob:
Perhaps I'm missing something but the schematic shows 2 1k resistors in the loop. Meaning at 400V the current would be ~ 200 ma. And each 1K will dissipate 40 watts.

My question is the OP might do better if they defined the basic goal (accuracy sample range etc).

My guess is the sample current would be "known" if the power supply voltage were known. However resistors capable of dissipating 40 watts are not likely to be terribly stable.

Issues I see are:

  1. Getting meaningful data from the 2 lsb's of the 22 bit daq is near impossible. Even with long integration time.

  2. When measuring µV one must be careful of the materials being uses else a termocouple will be formed causing temperature related error.

Jon

Thanks for your reply Jon. And thanks for noticing the loads on the diagram, unlike others. Just consider the loads and other things above the Black line as fixed in a particular scene, but the supply and loads can change form test to test.

That is why I want to measure V across R1, and current in the loop at the same time to find out R1. I need to design the measurement part that can accurately measure the drop across R1. If you need a range for R1, it will be 10^-5 to 2 Ohm. Accuracy that I am looking for is <1%.
I am aware of the issues you have mentioned. I am trying to use good ppm/C resistors and will take care in pcb designing and soldering as well. Lets just start with something first.

outofoptions:
He had another thread. He wants to measure the resistance of a conductor under load. He was talking about measuring nanovolts. All the added 'stuff' just adds more error.

But Ohm's law. You can measure voltage and current to get the resistance. Can he build a volt meter that will accurately read in the nanovolts at 400VDC and read it with an Arduino? Does he have a current source that is that accurate? I have my doubts.

I actually hijacked that thread instead of posting my own - i don't need the nano volt measurement, micro will just do. And I don't think I need an accurate current source, I can use a good Vref for the ADC, and when finding R1 =I*R1/I = VR1/I , I gets cancelled out?

Grumpy_Mike:
No doubts here, he can't.

Of course I can't, at-least not without help. And that is why I have posted the problem here.

Edit: To make it simple, This is just like a micro-ohm meter, but without any constant current source. We need to determine the current on the spot as well as the voltage drop to determine the R1. I can easily determine the current with <1% accuracy. So in this post I am only focusing on measuring VR1. I hope I have made my query clear enough. Thanks for looking into it.

8

OP ..... just show a complete diagram ...... one that shows where the voltage is applied across.

At the moment, your diagram is incomplete, so no room for coherent discussion about this system you want to build.

You did mention something about 50 to 400 Volt across a 'low' resistance. You need to show it.... show where the source is applied....across which points? And what sort of 'low' resistance are you taking about --- approximate resistance.

9

There are two diagrams, that's the confusion.

I don't understand the use of 1000 ohm resistors either side of a low resistance sample - is
that a necessary feature of the appratus, or just to make things difficult. Why not use a current
source directly?

10

Thanks Mark! I see now. The file attached in the opening post is different from the one displayed visually. Thanks for mentioning this.

Totally agree with you as well. Having those much larger resistors in series with the relatively low resistance sample won't help to develop a large voltage across the sample. That's if the OP wants to develop a relatively large voltage across the sample itself (even if it destroys the sample).

11

Southpark:
That's if the OP wants to develop a relatively large voltage across the sample itself (even if it destroys the sample).

As far as I know the sample is a conductor so ideally no voltage drop. That's the range of drop he is looking to measure. The embedded diagram is actually what he is trying to do. The one you have to download is more a proposed method.

12

MarkT:
There are two diagrams, that's the confusion.

I don't understand the use of 1000 ohm resistors either side of a low resistance sample - is
that a necessary feature of the appratus, or just to make things difficult. Why not use a current
source directly?

The first diagram is just to explain how resistivity testing works. The one that you have to download on the other hand shows what I am trying to build here. The "HIGH LOADS" on both sides are necessary features of the apparatus. Try searching for resistivity and IP meters for reference. They are mostly used for soil test etc. But I am going to use it for measuring any resistivity of 'any' material. The reason I can't use current source is in my 1st post (italicized).
Edit: Can I use a current source if the "1k loads" change every time I test it??

Southpark:
Thanks Mark! I see now. The file attached in the opening post is different from the one displayed visually. Thanks for mentioning this.

Totally agree with you as well. Having those much larger resistors in series with the relatively low resistance sample won't help to develop a large voltage across the sample. That's if the OP wants to develop a relatively large voltage across the sample itself (even if it destroys the sample).

Yes, Having those larger resistance at both sides makes it challenging. There is not much voltage to be developed over the sample and that is the reason to measure micro voltages with precision.

outofoptions:
As far as I know the sample is a conductor so ideally no voltage drop. That's the range of drop he is looking to measure. The embedded diagram is actually what he is trying to do. The one you have to download is more a proposed method.

Even if the resistance is 10^-5 there will be a voltage drop, I just need to measure it.

13

A small differential voltage over a voltage that could be floating 100V or more is really hard. Is it possible
to guarantee good symmetry in the voltage source to keep the sample close to ground? If you can get
that common mode voltage down to 5 to 10V or so you can use an instrumentation amp. And that'll
probably have to be an auto-zeroing instrumentation amp.

How long is the voltage applied for? Can those 1000 ohm loads survive for long enough to get a
good static reading with a dual-slope integrating ADC for instance?

14

MarkT:
A small differential voltage over a voltage that could be floating 100V or more is really hard. Is it possible
to guarantee good symmetry in the voltage source to keep the sample close to ground? If you can get
that common mode voltage down to 5 to 10V or so you can use an instrumentation amp. And that'll
probably have to be an auto-zeroing instrumentation amp.

How long is the voltage applied for? Can those 1000 ohm loads survive for long enough to get a
good static reading with a dual-slope integrating ADC for instance?

The voltage will be applied for long enough to take measurements, and the loads will survive.

I am just a newbie here, so can you please care to make me understand why the small voltage measurement is difficult when the supply is 400V?
I only understand that I have to bring down any voltage to the ADC level range for measurement.

What do you mean by
1/ Good symmetry in voltage source? It is a variable DC source. 50V to 400V.
2/ keeping the sample close to ground?
3/ common mode voltage down to 5 to 10V

15

firststep:
I am just a newbie here, so can you please care to make me understand why the small voltage measurement is difficult when the supply is 400V?

Ultimately, for safety reasons, you may find it hard to get suggestions here because what you don't understand can kill you at these voltages. Look up high precision volt meters and see how many digits of precision you get at what price. If you want a home brew solution you might start here.

16

In the OP's previous thread on this subject I suggested the enclosed, and also said it might be possible to hang all the measuring electronics at the +200v or so and couple readings of the adc via optos to a arduino, pc etc at ground level, but did not pursue this approach.

Not an easy problem.

Allan

Imeas2.pdf (25.8 KB)

17

firststep:
I am just a newbie here, so can you please care to make me understand why the small voltage measurement is difficult when the supply is 400V?

Is that DC voltage source "floating", or grounded somewhere.

If it's floating, you can ground one side of the specimen, and use a common A/D at ground potential.
The specimen supply will just generate -200 and +200volt related to ground.
Leo..

18

allanhurst:
Not an easy problem.

Allan

If you just count the number of significant digits involved it is pretty overwhelming. 400.000000VDC is 9 places. Add one for meter resolution brings us to 10. Also, unless he knows current (at the same instant no less) to the same precision that number is useless . So, now we need 11 places on each voltage and current to be reasonably sure you have precision to 9 places for resistance. Reducing the voltage to measure doesn't reduce the resolution needed.

OP. Dead serious here. See if you can build a simple circuit to measure 6 places without the last 2 or 3 numbers flopping around like dead fish. I think that will give you a much better idea what you are up against here. Leave the voltage problems for later. If you don't solve this one there is no point worrying about the rest.

19

outofoptions:
Ultimately, for safety reasons, you may find it hard to get suggestions here because what you don't understand can kill you at these voltages. Look up high precision volt meters and see how many digits of precision you get at what price. If you want a home brew solution you might start here.

Electronics -- Signal Analysis and Data Aquisition

Thanks for the link, I just checked it out. It may be helpful, but there is no explanation of the schematic. Anyway thanks.

I know that the high voltage can kill someone - but that's why I was using a voltage divider to reduce the voltage and measure it. My result is fluctuating around milli volt range currently. I used software averaging to get the average value which is close but still has huge errors.

Can anyone suggest a design like MarkT suggested with an instrumentation amplifier?? I may get some idea from there.

20

outofoptions:
If you just count the number of significant digits involved it is pretty overwhelming. 400.000000VDC is 9 places. Add one for meter resolution brings us to 10. Also, unless he knows current (at the same instant no less) to the same precision that number is useless . So, now we need 11 places on each voltage and current to be reasonably sure you have precision to 9 places for resistance. Reducing the voltage to measure doesn't reduce the resolution needed.

OP. Dead serious here. See if you can build a simple circuit to measure 6 places without the last 2 or 3 numbers flopping around like dead fish. I think that will give you a much better idea what you are up against here. Leave the voltage problems for later. If you don't solve this one there is no point worrying about the rest.

Yes, I did that - and the result is fluctuating a lot, used averaging in the code to get somewhat close results with huge errors. I just measured simply like this...

9096d40c80924c38ad684128e748173bff6c865d.jpg

allanhurst:
In the OP's previous thread on this subject I suggested the enclosed, and also said it might be possible to hang all the measuring electronics at the +200v or so and couple readings of the adc via optos to a arduino, pc etc at ground level, but did not pursue this approach.

Not an easy problem.

Allan

Hello Allan sir, Sorry, I could not make that to work because of the OPA192 which is not available to me.