DIY Milliohmmeter - how to solve grounding issue?

[Avelino Sampaio]

Hi

I found another batch of fake mosfets. Now it was the IRF3205.
 
1 - IRF3205PBF - (datasheet RDS = 8mR) - OK
2 - IRF3205PBF - (datasheet RDS = 8mR) - FAKE (RDS=27mR)

[Avelino Sampaio]

I went to measure some transformers and got a scare with the totally wrong readings. INA143 did not like the presence of the inducers at all. The addition of a 100nF capacitor in the driver line resolved the issue.

1- wrong reading
2- wrong reading

After inserting the 100nF capacitor:

3- correct reading
4- correct reading
5- correct reading

[Kleinstein]

The revised circuit looks OK (though still a bit confusing) for the voltage measurement side and seems to work.

The detection side should not care very much about inductance. If needed it is relatively easy to slow a down with some capacitors (e.g. 1-100 nF range) in parallel to the 100 K.

The current source can be tricky with a highly inductive load and this is more tricky to fix. A possible fix could be an RC sereies combination in parallel to the DUT to dampen a possible resonance. The rather complicated current source makes it hard to get a good guess. on the suitable valaue. Also just a series resistor at the current source side can help.

[johansen]

I had relatively good success building a micro ohm meter using a 7$ 1kilogram scale.

What you do is remove the strain gauge and run the 4 leads out to kelvin clips... the positive and negative from the battery (one of which isn't directly from the battery but is rather switched on through the circuit board) are sent through 2 resistors. they don't have to be the same value each, but you want approximately the same value so that the amplifier is biased half way between the 3 volt battery and ground.

i recall something like 115 ohms or something for full scale being .01 ohms.

anyhow the way it works is that strain gauge amplifiers reliably amplify the .1uV difference between the bridge when the load is .1 grams... 1 to 3mv is frequently full scale.

when you use a strain gauge as a microohm meter, you aren't directly reading ohms but rather are reading the ratio between the known resistors sending current through the kelvin clips and the unknown resistance.

you have to clip the leads to one terminal of the unknown resistor, tare the scale, then clip the leads to the other end. i have found that 2% accuracy is relatively easy to achieve. the resistors used to load the unknown resistance need to have low thermal temperature coefficients, especially if you want to send 100ma through the unknown resistance with only two 1/4watt resistors . also you may want to just directly connect them to the battery rather than relying on the mosfet on the circuit board, which has a resistance that is probably highly temperature variable.

[Avelino Sampaio]

The current source can be tricky with a highly inductive load and this is more tricky to fix. A possible fix could be an RC sereies combination in parallel to the DUT to dampen a possible resonance. The rather complicated current source makes it hard to get a good guess. on the suitable valaue. Also just a series resistor at the current source side can help.

I continued performing various tests with inductors and everything is working as expected, after I added the 100nF capacitor.

It was a great challenge and satisfaction for me to obtain the results achieved, using only 50mA as a current source. The entire circuit consumes 87mA.

The percentage of accuracy and precision achieved was far beyond my expectations.

Kleinstein, thank you so much for being here sharing your knowledge.

Here's the updated schema.

[Avelino Sampaio]

Hi

Project finished and thank you very much for the support of the participants.