Suggestions for a millivolt DMM

[Performa01]

I have repeated the test with the Siglent SDS1104X-E and a 60Hz signal.

This time the output signal at 100mV was calibrated using a Keithley 2001. Apart from the calibration error (which is insignificant for this test), only the accuracy of the internal attenuators in the Siglent SDG6052X arbitrary waveform generator remain an issue. From other tests I have every reason to believe that the attenuator in this device contributes less than 0.5% error to the net result, so I can present my findigs with reasonable confidence:


SDS1104X-E_Linearity_60Hz

This looks certainly much better now, with less than 1% absolute and 1.7% linearity error down to 500µVrms and still good 1.3% accuracy and 2.1% linearity down to 300µVrms. Below that, the error increases a lot.

[joeqsmith]

I thought I was making it clear that I was assuming there is only the 100uV source to work with.  Sounds like I wasn't clear enough.  Obviously if there is access to a higher voltage source that is in sync with the signal of interest that could be used for a trigger, this wouldn't be a problem.

[Performa01]

I thought I was making it clear that I was assuming there is only the 100uV source to work with.  Sounds like I wasn't clear enough.  Obviously if there is access to a higher voltage source that is in sync with the signal of interest that could be used for a trigger, this wouldn't be a problem.

I think most of us have referred to the situation as described by the OP in his reply #106:
https://www.eevblog.com/forum/testgear/suggestions-for-a-dmm/msg2761686/#msg2761686
It sounds like there is a strong 140mV signal from the same source available.

How well does your DSO do if you use the 100uV signal for the trigger and try to average?

As already stated, it does not do it at all, since it cannot trigger a 100µV RMS signal.

It does work with a 200µV RMS signal though:


SDS1104X-E_60Hz_200uVrms_Avg16

The expected reading would be about 205µV, and the error is in line with my previous test.

EDIT: Please notice the low standard deviation of the result in the measurement statistic ...

Btw, just to be safe: "my" DSO happens to be also the DSO of the thread opener. Therefore I thought this demonstration might be helpful in his situation. Otherwise I would not recommend any DSO (not even a PicoScope 4262) without suitable external preamplifier for measuring low frequency microvolt signals.

[Kleinstein]

For the trigger one does not need to amplify the actual signal, just a signal in phase (e.g. same generator / mains if mains signal) would be good for the trigger.

Just using a longer data window can also be an option, though it includes more higher frequency noise. So ideally one would want a lower than 20 MHz BW limit.

Using a high degree of averaging and a rather short data window (e.g. 1 or 2 periods) nearly turn the DSO into a lock-in amplifier. 

I use my scope as a null detector in a bridge, synced to the generator, and using a uCurrent to amplify the current into a ratio transformer "variable tap".  I am getting 0.1ppm resolution by looking at the phase relationship over 1 period - the ability to average the noise out of a noisy AC waveform is definitely a killer feature of a DSO, I can't think of any other instrument that could do this.

The classical instrument to measure small AC signals including phase to a reference is a lock in amplifier, especially the better 2 phase version. Compared to an DSO modern digital Lock-ins tend to use higher resolution ADCs, but lower speed. The other related old time instrument is a box car integrator - the equivalent to the averaging mode one an DSO. This one could also be done in analog for a limited number of channels.

A good working averaging mode can be an important and very useful feature on an DSO.

[joeqsmith]

As already stated, it does not do it at all, since it cannot trigger a 100µV RMS signal.   It does work with a 200µV RMS signal though:

.... since even the excellent trigger system in an SDS1004X-E cannot deal with a noisy 100µVrms signal.

I missed that part.   Even at 200uV, it seems impressive.     

[SilverSolder]

The classical instrument to measure small AC signals including phase to a reference is a lock in amplifier, especially the better 2 phase version. Compared to an DSO modern digital Lock-ins tend to use higher resolution ADCs, but lower speed. The other related old time instrument is a box car integrator - the equivalent to the averaging mode one an DSO. This one could also be done in analog for a limited number of channels.

A good working averaging mode can be an important and very useful feature on an DSO.

Using a DSO, you can act as a "human lock-in amplifier" by varying the capacitance across one of the bridge arms until the waveforms show that you have nulled out both the real and quadrature currents.   You can even act as a human capacitor if the circuit is somewhat high impedance (e.g. 100K), playing it like a Theremin to balance it out!   

Ghetto metrology!   

Just for fun, I tried inputting a nice noisy 100uV signal on an Agilent 54622D,  which (unlike the Siglent) only has a 2mV/division vertical amplifier (the 1mV setting is "faked").


Without averaging:



With averaging:



100uV amplitude is too low to measure properly - but good enough to see the phase!

Further experiments showed that you can get down to about 5uV - 10uV and still just about make out that there is a signal there...

If you put Dave Jones' uCurrent in front of this not particularly sensitive scope, with accurate amplification of x100, you can actually peek into the nanovolt region!

7 digit resolution is possible (has been achieved in the basement) this way, and with another x10 more amplification, that elusive 8:th digit might be within reach using the DSO method...

But, who knows, perhaps there is a lock in amplifier or a box car averager in the future! 

[joeqsmith]

Using the same setup I used with the handheld meters, but using a separate signal for the trigger,  my vintage DSO is about 10% out at 100uV.   The Fluke 101 with a simple amplifier would easily out perform it at a much lower cost.  Not too surprised. 

[Frex]

hello all,

For those who want make low level AC measurements, i designed few months ago
a small wide-band low noise amplifier to do power supply wide-band noise measurements.
It include two stages with x10 and x100 gain output with flat bandwidth  over 20Hz - 20 MHz.
It accept DC offset of + or - 20V max. The total bom count is about 30€,
and the full project is open source and available on Github here.

You can also look here where i describe the project with some measurements results.
Regards.

Frex