What would be the best accurate low ohm meter possible available

What would be the most accurate low, ~ 50mili - 15, ohm meter possible either as finished product or as components to be assembled on PCB (shown with its circuit diagram)?
The actual need is about 0.3% tolerance is adequate

**broken link removed**


Regards, Dana.

Relatively easy, either home made, kit, or commercially available (at low cost from China).

You simply pass a constant current through the resistance, and measure the voltage drop across it - ohms law tells you the resistance (directly in milliohms if you select the constant current carefully.

However - your requirement for 0.3% tolerance is the main issue - how are you going to calibrate it to that degree of accuracy?, and can you afford resistors of that tolerance?. Why do you imagine you need 0.3%?.

Generally speaking you want to do this as a Kelvin measurement.

Here is a handbook that discusses many issues when doing precision measurements -



Regards, Dana.

0.3% of your upper limit, 15Ω, is 0.045Ω.
0.3% of your lower limit, 0.050, is 0.00015Ω.

To send 1A through 0.1Ω. you'll need 1 * 0.1 = 0.1V.
To send 1A through 0.15Ω, you'll need 1.5 * 0.1 = 0.15V.
To detect the difference, you'll need to detect the difference between 0.1V & 0.15V.

To send 1A through 15Ω. you'll need 1 * 15 = 15V.
To send 1A through 15.05Ω, you'll need 1 * 15.05 =15.05V.
To detect the difference, you'll need to detect the difference between 15V & 15.05V.

15/0.05 = 300. An 8-bit ADC gives you 256 levels. so you need at least a 9-bit ADC. 10-bit is more common and gives some room for better discrimination.

Many microcontrollers -- eg. PIC16(L)F18313/18323 -- as well as many standalone ADCs can produce 10-bit output.

Then there are two problems:

1) The accuracy of the reference voltage.

2) The accuracy and stability of the constant current source.

Of course, you may not want to send a 1A current through whatever it is you are measuring. If you need to reduce that by an order of magnitude, then the accuracy of your reference voltage and constant current source needs to improve by an order of magnitude.

Buk said:

0.3% of your upper limit, 15Ω, is 0.045Ω.
0.3% of your lower limit, 0.050, is 0.00015Ω.

To send 1A through 0.1Ω. you'll need 1 * 0.1 = 0.1V.
To send 1A through 0.15Ω, you'll need 1.5 * 0.1 = 0.15V.
To detect the difference, you'll need to detect the difference between 0.1V & 0.15V.

To send 1A through 15Ω. you'll need 1 * 15 = 15V.
To send 1A through 15.05Ω, you'll need 1 * 15.05 =15.05V.
To detect the difference, you'll need to detect the difference between 15V & 15.05V.

15/0.05 = 300. An 8-bit ADC gives you 256 levels. so you need at least a 9-bit ADC. 10-bit is more common and gives some room for better discrimination.

Many microcontrollers -- eg. PIC16(L)F18313/18323 -- as well as many standalone ADCs can produce 10-bit output.

Then there are two problems:

1) The accuracy of the reference voltage.

2) The accuracy and stability of the constant current source.

Of course, you may not want to send a 1A current through whatever it is you are measuring. If you need to reduce that by an order of magnitude, then the accuracy of your reference voltage and constant current source needs to improve by an order of magnitude.

Click to expand...

Most PIC's have 10 bit A2D (assuming he's wanting to use a microcontroller), you'd struggle finding one that's only 8 bit - and quite a few are 12 bit, which comes in handy. The 18F27K42 is 12 bit, and comes with lot's of extra features and oodles of memory.

Generally you tend to keep the voltage to be measured fairly low (1.99V maximum for example) so as to keep the current low.

Nigel Goodwin said:

Most PIC's have 10 bit A2D

Click to expand...

Why do you feel the need to repeat exactly what I said:

Many microcontrollers -- eg. PIC16(L)F18313/18323 -- as well as many standalone ADCs -- can produce 10-bit output.

Click to expand...

whilst phrasing it in a fashion that makes it seem -- on the surface -- like you are actually contributing something useful to the thread?

Nigel Goodwin said:

and quite a few are 12 bit,

Click to expand...

Useless information!

Nigel Goodwin said:

Generally you tend to keep the voltage to be measured fairly low (1.99V maximum for example) so as to keep the current low.

Click to expand...

If you need to measure a range 0.05Ω through 15Ω; then how would you suggest doing that, whilst confining the voltage to <2v?

Don't tell me -- you invisage an auto-ranging bias voltage applied to the reference voltage of the ADC.

And your sample rate?

abdulbadii said:

What would be the most accurate low, ~ 50mili - 15, ohm meter

Click to expand...

Look at a Peak ESR70
They have 10 milliohm resolution, read up to a maximum of 40 Ohms and are available with a calibration certificate if required.


Although primarily for capacitor testing, they read resistance perfectly well.

Buk said:

Why do you feel the need to repeat exactly what I said:



whilst phrasing it in a fashion that makes it seem -- on the surface -- like you are actually contributing something useful to the thread?

Click to expand...

Because you implied that only a tiny few PIC's had 10 bit A2D, when almost all do - and I didn't repeat anything you said.

Buk said:

Useless information!

Click to expand...

I suggest you try a little politeness - it's VERY useful information, if using a processor with a 12 bit A2D it inceases your range 4 fold.

Buk said:

If you need to measure a range 0.05Ω through 15Ω; then how would you suggest doing that, whilst confining the voltage to <2v?

Don't tell me -- you invisage an auto-ranging bias voltage applied to the reference voltage of the ADC.

Click to expand...

No, I envisage an opamp (perhaps you should try looking them up?), with switched ranges if required, but historically low range meters didn't need to use micro-controllers, as there's little need.

Buk said:

And your sample rate?

Click to expand...

Has nothing whatsoever to do with anything - are you just throwing random words out?.

rjenkinsgb said:

Look at a Peak ESR70
They have 10 milliohm resolution, read up to a maximum of 40 Ohms and are available with a calibration certificate if required.

Peak Atlas ESR Gold ESR Plus model ESR70 | Peak Electronic Design Limited

Although primarily for capacitor testing, they read resistance perfectly well.

Click to expand...

Designed and built not too far from me, in fact a friend of mine (who we sub-contract to) lives very close and they often give him samples to evaluate - they are also PIC based.

Probably more than you need but a single chip solution, I estimate ~ 25 lines of code,
could be done with this single chip.

Basically you force a current with onboard IDAC into a precision external cal R. Read that
with 20 bit A/D to create correction reading when doing the Runknown. Then switch
muxes to force current into Runknown and switch DelSig inputs to read Runknown V.
IDAC has a > 2 decade range, uA to mA, and with DelSig used to cal its accuracy allows
a lot of flexibility.

I included USBUART component in case you want to pass data to an OS or something.
Could even add SDCARD to record data. And I added graphic component to handle display.
Again this would all be single chip.

Note reference in this part is good to +/- .2% App note on precision for this part -


If you need more accuracy easy to use an external Vref to better the +/- .2% internal one easy to add.

Note there are onboard OpAmps and PGA's, I do not recommend you use them in this application
as they will contribute more error, and DelSig front end has an excellent differential capability, eg. CMR,
and has G options as well, but hi precision.






Note right hand window shows lots of resources left to do other things. 3'rd window over
shows catalog of onchip resources.

You could even add correlated double sampling to the code to get rid of significant
potential amount of noise in measurements.

Compiler and IDE (PSOC Creator) free. Board to use is $ 10. Debugs and protos.



Regards, Dana.

Nigel Goodwin said:

Because you implied that only a tiny few PIC's had 10 bit A2D,

Click to expand...

How does "Many microcontrollers" imply "tiny few"?

Nigel Goodwin said:

it's VERY useful information, if using a processor with a 12 bit A2D it inceases your range 4 fold.

Click to expand...

As demonstrated, the OP's stated accuracy requirement

abdulbadii said:

The actual need is about 0.3% tolerance

Click to expand...

(that you earlier pooh-poohed the need for "Why do you imagine you need 0.3%?."; as if you know his requirements better than he does.)

is more than adequately covered by 10-bits; making arbitrary mention of 12-bit ADC extraneous in this context.

Ie. 'useless', is not "impolite"; just an accurate characterisation.

Let’s take a look at the TS exact words:
“What would be the BEST accurate low ohm meter POSSIBLE available”.
(Capitals mine).

In that is indeed the case, he should be looking at Keithley or Rhode & Schwarz catalogs. Be prepared to spend at least a couple of thousand Euros.

You could get an used instrument, as long as it has NIST traceable or similar calibration.

schmitt trigger said:

Let’s take a look at the TS exact words:
“What would be the BEST accurate low ohm meter POSSIBLE available”.
(Capitals mine).

In that is indeed the case, he should be looking at Keithley or Rhode & Schwarz catalogs. Be prepared to spend at least a couple of thousand Euros.

You could get an used instrument, as long as it has NIST traceable or similar calibration.

Click to expand...

When someone says things like best my first thought with commercial test measurement and diagnostic equipment is budget? Best is just one of those words. Same here you could spend several thousand USD.

Ron