Please help me design a kelvin / 4 point meter circuit

Thread Starter


Joined Mar 29, 2011
I'd like to design a meter with the following requirements:

Microcontroller: Atmel ATMEGA / similar
Object to test: expected to be between .03 ohm and 4 ohm
Interface to object: Proprietary connector with v+ and ground
Output: oled display
Budget: $350 in components and boards / unlimited labor (mine)

I have been asked to do very accurate and precise ohm meter that is capable of properly measuring resistances down to .03 ohms. I understand how challenging this is, and if it's not possible for my budget, I'll head back with a decline on the request!

Based on my understanding of all of this, I'm going to need:
A high quality constant current driver capable of driving a very specific current
A high quality reference voltage for the ADC
A high quality ADC dedicated to measuring the voltage drop across the subject for the voltmeter portion of the measurement ( I was thinking of the LTC2400:
As good a reference resister as I can afford for doing current measurements. This will allow me to calibrate the device based on the output of the constant current driver.

What I need is a basic block diagram of how I'd set this up so that I'm not making a dumb mistake. It SEEMS to me that I'd have a microcontroller connected to the LTC2400 over SPI that would first check for NC / high resistance if the subject isn't plugged in. If plugged in, it would queue up the CC driver and then test and save the current for later arithmetic. Then, it would test the voltage across the subject, do the math and display the output.

This very likely is far harder (at least in terms of getting a precise and accurate result) than I am thinking, and am more than happy to get responses that indicate why this is so. I suspect I'm missing a second ADC for testing the voltage across the subject (in addition to the one being used to test the current).

I sincerely appreciate any help anyone can give, and thank you in advance for making it this far!


Joined Jun 19, 2012
What is missing is the level of accuracy you need- and the level of current you can subject your unknown R to.
You need a precision current source and a differential amplifier with high common mode rejection to measure the voltage across the R
Probably the most difficult part is designing the actual physical connection method- lot's of accuracy can be lost there if you are not careful.

Thread Starter


Joined Mar 29, 2011
Well thought out. I had been thinking of using a 1A constant current driver from ON SEMI as I had assumed that so long as I knew what the current WAS (as opposed to what I intended it to be), I wouldn't need to go too crazy with the reference current source. Given that I can subject the R to up to 5A, I won't be measuring VERY low voltages, though the amplifier is of course a solid idea and will make it far more workable.

The connection point is not as much of an issue as you might think ONLY because this is a safety situation where the connection to the meter will ALWAS be better than the field application, and what we're looking for here is to ensure that the R is not BELOW the expected value. Given that it will always be higher in the field due to a poorer connection, it will realistically not be an issue.

If there's something wrong with my thinking in regards to not needing a high precision current source in as much as ANY current can be measured, then the voltage across R, and so long as we know what the current was at that point, we can do the math, please let me know!

Thank you so much for taking the time to respond!

Thread Starter


Joined Mar 29, 2011
Looks like someone wants to measure vaporizer coil resistance.
You didn't hear me say this, but you nailed it. The guy that asked me is trying to get ahead of the market. If you've got something in the works, I'll head back with a "can't do it" and get out of your way! It's not going to be a money maker for me in any case :)

Just let me know when you finish so I can buy one for myself!