DIY milliohm/volt meter kit

[robrenz]

I don't know squat but this is what I see:
You have the ability to turn the 3V on and off since the unit can measure voltage alone. The unit measures ratiometrically comparing the voltage drop of the R17 reference resistor to the voltage drop of the DUT.

The offset compensation procedure would be: turn off V3 for 200mSec measure and store sense connection voltage. Turn on V3 just long enough to get a stable reading maybe 200 mSec. measure and store the value. Algebraically subtract the voltage off value from the voltage on value. Display the calculated value and repeat the procedure continuously.

[jaxbird]

I don't know squat but this is what I see:
You have the ability to turn the 3V on and off since the unit can measure voltage alone. The unit measures ratiometrically comparing the voltage drop of the R17 reference resistor to the voltage drop of the DUT.
...

It is good thinking 

But unfortunately, switching between the two modes (Ohm and Volts) is a very manual process that requires moving 5 jumpers around on the back. My bad, I didn't cover that or post any pictures of the back side 

I was thinking a couple of tiny MOSFETs (P and N) would do the job together with a few resistors, but looking again, I don't think that is the case. Also, I also wonder if it's really worth the trouble as a ~10C change in temperature will make the meter drift about 7-10 micro Ohms (I slowly heated it up over about 20 minutes). That could of course also be compensated as the Silabs 8051 mcu contains two complementary diodes that can be internally connected to the ADC for fairly accurate die temperature measurements

[dannyf]

switching between the two modes (Ohm and Volts) is a very manual process

Yeah. Those two modes of operations are mutually exclusive - I think I might have mentioned it earlier in this thread.

[robrenz]

IMO you would only need to add a mosfet to turn the 3V on and off while the jumpers are in the resistance mode. The unit will be measuring the sense voltage in resistance mode. You just need access to the sense voltage value itself to get the two readings and be able to turn the mosfet on and off.

[jaxbird]

IMO you would only need to add a mosfet to turn the 3V on and off while the jumpers are in the resistance mode. The unit will be measuring the sense voltage in resistance mode. You just need access to the sense voltage value itself to get the two readings and be able to turn the mosfet on and off.

It would still need something to keep the reference part of the ratio approximately the same.

I was thinking something like this, with a couple of SOT23 MOSFETs and a few resistors, but I'm not convinced it would work:

[jaxbird]

Yeah. Those two modes of operations are mutually exclusive - I think I might have mentioned it earlier in this thread.

Yeah it's 2 completely different parts only sharing the mcu, display etc. it would have been a lot simpler (cheaper) without the volt meter part. No need for precision reference, chopper amp for buffering, amp for virtual ground, switch for resistor ratio etc.

But then again, the Voltmeter part is actually very nice.

[robrenz]

It would still need something to keep the reference part of the ratio approximately the same.

I don't think so. you just need to have it read the sense voltage with NO 3V then read it again with the 3V and do the math. Now that calculated sense voltage needs to go into the ratiometric calculation that is using the ref voltage that was present when the 3V was on.

[jaxbird]

It would still need something to keep the reference part of the ratio approximately the same.

I don't think so. you just need to have it read the sense voltage with NO 3V then read it again with the 3V and do the math. Now that calculated sense voltage needs to go into the ratiometric calculation that is using the ref voltage that was present when the 3V was on.

It's using the ref+/- (from my simplified schematic) as the reference for the ADC to easily get the ratio between the reference resistor and the sense+/- over DUT. So the ref+/- must be kept at approximately the same level for reasonably accurate offset readings while the source current is removed from dut, that is the task of the NMOS + 2.7R resistor.

[robrenz]

Ok, I understand now

[jaxbird]

Ok, I understand now

Nice, yeah, I wasn't providing the complete picture, too much implicit info.

But I do think sometimes when getting down to these more extreme measurements I just want to swallow the blissful ignorance pill and ignore all the potential errors, being happy with no worries

On the other hand, I do find an unhealthy attraction towards the nearly impossible. 

[dannyf]

So the ref+/- must be kept at approximately the same level for reasonably accurate offset readings while the source current is removed from dut,

Yeah - you want to maintain rougly the same common mode voltage levels.

that is the task of the NMOS + 2.7R resistor.

Those two mosfets can be just two io pins.

[jaxbird]

...
Those two mosfets can be just two io pins.

I haven't checked but I don't think the IO pins will source/sink ~100mA.

[dannyf]

yeah.it wouldn't work with high current.

[jaxbird]

yeah.it wouldn't work with high current.

Understand .. yeah, you could do an extremely simply ohm meter like that. Even with a number of reference resistors for different ranges.

[iloveelectronics]

I am finally done with the English translation of the documentation that comes with this kit. For those who have sent me PM's for pre-ordering, I have just sent you a PM, please check. If you haven't got one, please send me a PM to remind me. Thanks a lot for all your patience!

[AKM]

When will it be available to the rest of us?

[iloveelectronics]

When will it be available to the rest of us?

It is available now. Please PM me for details. I will make a post in the "Buy and Sell" section soon.

[jaxbird]

Been playing around a bit more with an implementation of thermal EMF compensation, as robrenz pointed out, is needed for more valid measurements in the micro ohm range.

I'm going to attempt a slightly modified version of my previous schematic (using 2 NMOS and a transistor instead of both P and N MOSFETs)



It should be possibly to test it, without having to permanently modify the PCB (no need to cut traces or anything, just use existing pads)

[jaxbird]

Test modification soldered in:




A bit fiddly with the little sot23 and resistors  but no cut traces.

I'll make it a bit more permanent with some glue if it works.

[dannyf]

I'm going to attempt a slightly modified version of my previous schematic

What if you wire the lower mosfet's gate to the upper mosfet's drain? you can save the npn + resistors.

[jaxbird]

I'm going to attempt a slightly modified version of my previous schematic

What if you wire the lower mosfet's gate to the upper mosfet's drain? you can save the npn + resistors.

It's a good idea, but I think the voltage will be too low on the upper mosfet's drain, down to a max of ~300mV, depending on dut.

Anyway, it works with the transistor. Only I got some headaches from the mcu pin I assumed available, as it isn't connected to anything.

It's the only "free" IO pin, the other unused pins are analog only.

I'm not experienced in silabs 8051, so it took some datasheet reading to discover why it seemed to have weak pull up and not reacting to being set, when it was configured as push pull output.

Explanation found when I stared looking at the hardware peripheral assignments. Turns out it's assigned as MISO for the hardware SPI used to write to the display 

So I had to switch it to a soft SPI to actually free up the pin. Not a big deal as it was running just 1 MHz SPI and waiting around for each byte to complete anyway.

It appears to be working correctly, but still needs a bit of software fine tuning.

[jaxbird]

Just a short update on the thermal EMF compensation modification experiment.

The compensation works very well now, display updates every ~1.5 seconds in slow mode with some averaging.

It's easy to test the effect when e.g. measuring a resistor. Without compensation, when gently heating one side of the resistor leads, breathing on it is enough, the measurement will change by something like 30 micro Ohm, + or - depending on side.

With the compensation enabled there is little or no change.

It was a bit tricky to implement in software as the code structure is build around running the ADC in continuous mode. It's doing 19.2 ksps internally with an output rate of 10Hz from the SINC3 filter for best performance. So when switching between offset and ohm measurements the first few samples are inaccurate and must be discarded.

A nice side effect is that it auto zeroes the measurement. I haven't added relative measurements with thermal EMF compensation, maybe some other time.

[jaxbird]

In case anyone want to try it out, here are a couple of detail pictures of the modification:








I used AO3400 for the MOSFETs and BC847 for the NPN, as that is what I had. Shouldn't cost more than $1 or so in components.

PM me for a copy of the modified firmware hex file.

[jaxbird]

Decided to print a small foot for it so it will sit on the bench at a nice angle:



Just a quick design and print, nothing too fancy:



Not too tall, good angle for easy readability:




OK, now I think I'm done with pimping this little meter kit

Overall pretty happy with it and it also served my secondary goal of getting to know the Silabs F350 better. It's actually a good kit to experiment and get familiar with the features and performance of the mcu and some of it's main peripherals.

[jaxbird]

I'm done with pimping this little meter kit

I lied  , just one more little modification, my battery is a bit too thick, about 0.5 - 0.75mm, but it sits partly on top of the reference, adding a bit of pressure.

So I routed a small 1.5mm deep pocket in the back plate for a much better fit.



Old phone battery, wrapped in kapton tape, as it's in an aluminium package.