Design Review - AC Milliohm Meter

[Shay]

I'd like to get some feedback about my 9V battery powered AC milliohm meter design, with the goal of measuring 0-1Ω with 0.01Ω (10mΩ) accuracy (or better) with 4 wire Kelvin sensing.
schematic:


Working princple:
U1.1 forms a 100 mA current source, that goes to the H bridge to convert the DC to AC. In theory, this is done to avoid any kind of noise and only measure and worry about the pure ac part; correct me if I'm wrong.
U1.2 is the square wave osilicator set to 40 Hz to alternate the voltage to the H bridge mosfets. It is also used to power a charge pump to generate a negative rail for the AD620 instrumentation amplifier to amplify the ac measured voltage across the resistor by 10 and output it to the DMM.
I also have a low battery voltage indicator.

I'd like to get some feedback about the design — things to improve, consider, or add. Any cool features that are useful and easy to add are also welcome.

[moffy]

You do realise that the body diode of Q2 negates the FET control altogether? You have a 25 ohm resistor and diodes in series connected to VCC. If you want to use P channel FET, swap the drain and source so that the body diode is blocking VCC.

[mawyatt]

The 2N7000 is not a low RDSon NMOS FET (5~10ohms max), selected a NMOS FET with a lower value, this should yield better results.

Synchronous sampling/detection will provide better noise immunity and allow higher gain for lower valued DUT, however this is more complicated than what you've shown, which is not such.

Best,

[Jay_Diddy_B]

Hi Shay,

Here is a quick model of the circuit in LTspice:



The low-pass filter C1, R2 and R3 in the model, C7, R17 and R18 on your schematic, the -3dB point is 80Hz. This is too low if you want to pass the 50Hz signal.

The model can be simplified further to:



Look at this for an example of synchronous detection, as recommended by mawyatt.

Link: https://www.eevblog.com/forum/projects/esr-meter-adapter-design-and-construction/


Regards,
Jay_Diddy_B

[Doctorandus_P]

It looks like you will have shoot trough between the "A" and "B" FET's.
The boxes around your circuit sections also make your schematic more difficult to both read and maintain. I'm wondering, how many times have you re-drawn those boxes?

I'm not sure about the rest of your schematic. Your opamps look like square black boxes, I don't even look at such schematics.

Reading though the other comments, there are also already plenty of other flaws detected.

It also looks like you are overthinking it. It's easy to make a 100mA current source or sink with just an LM317 (or LM7805 or similar) and a resistor. With 100mA you also already have 1mV over a 10mOhm resistor and simple DMM's usually have a resolution of 0.1mV, so there is no need to amplify your signal at all.
With an external power supply, a simple current source/sink and a separate DMM you also already have a 4 wire measurement that eliminates wire resistance (Or you can explicitly measure that too of course). And you can of course also check your current source with the current range of your DMM to verify if it still works.

Some years ago I read a CircuitCellar (#314) article about a milliohm meter, which was quite cleverly built. It's quite simple with a small uC, an ADC1115, a display and a bunch of passives. I attached the schematic, but the rest of the article is worth reading too. They claim a resolution in the micro ohm range, which seems reasonable as the 1115 can measure quite low voltages.

[jonpaul]

No need for this fancy AC design

Gold plated probes, clips, and ordinary DC technique , regulated By source >>current terminals
DC millivolt or micro volt with instrument opamp, eg gain 1k, 10k.

Our Keysight 34465a is DC easily measures fractions of mOhm

J

[precaud]

Get a HP 4328A

[jonpaul]

easy DIY

Lab PS>>>series R 1% >>Rtest

DVM 4.5..6.5 digit or DC millivolt meter

done.

The AC bridge idea is good in noisy industrial environment, noise, DC offset is reduced but needs synchronisation detector,

That technology is used for microohm applications like strain guages.

Jon

[Jay_Diddy_B]

No need for this fancy AC design

Gold plated probes, clips, and ordinary DC technique , regulated By source >>current terminals
DC millivolt or micro volt with instrument opamp, eg gain 1k, 10k.

Our Keysight 34465a is DC easily measures fractions of mOhm

J

I agree, the AC designs are need for measuring very low value of resistors to cancel offsets and thermal EMFs.

A design which is a nice balance between complexity and performance is found in Maxim's application note AN-106.

Link: http://www.wentztech.com/radio/resources/Projects/AN106.pdf



This can be simplified if you only need one range.

You can replace U1 with the TL431.

The op-amp should have a low offset voltage and work with the inputs near the negative rail.
You may need to add a capacitor, try 10nF, between the output of the op-amp and the op-amps inverting input, if you substitute the op-amp. This capacitor is to stabilize the current source.

This circuit is the current source. You measure the voltage across the resistor with a DMM.

It looks like you will have shoot trough between the "A" and "B" FET's.

snip ....

shoot-thru is okay in this application. The bridge is fed with a current source (except for C1 which would get discharged).

Regards,
Jay_Diddy_B

[Gyro]

I've been using that design for the past 15 20 years or so and have yet to need to change the batteries.

If you use a dual opamp, you can use the second one as a comparator to monitor the gate drive voltage and light an LED to indicate when the resistance is too high for the current source compliance range.


P.S. I used an LM385, OP295 and BUZ10 - basically what I had to hand.

[moffy]

A design which is a nice balance between complexity and performance is found in Maxim's application note AN-106.

Link: http://www.wentztech.com/radio/resources/Projects/AN106.pdf




Regards,
Jay_Diddy_B

That is a pretty cool design.