Low level AC performance of Fluke 8846A?

[grenert]

Joe Geller of voltage reference and JCan fame (www.gellerlabs.com) has done a nice study looking at the performance of the Agilent 34401A, 34410A and 3458A multimeters when they measure very low level AC signals (<1 mV, below their specifications).  He found that the 34401A becomes very nonlinear below about 0.4 mV (again, noting that the meter is not actually specified at these levels), while the 34410A can go down to around 0.1 mV before this happens.  The 3458A is linear to well below 0.1 mV.

Does anyone with access to a Fluke 8846A have the equipment to see how it performs on this test?

EDIT:  Here's the link to the study:
http://www.gellerlabs.com/34401A%20AC%20zero.htm

[Wartex]

Too late, just got rid of one. I have Agilent 34405 tho. There is another member who definitely has a brand new 8846a - robrenz https://www.eevblog.com/forum/index.php?action=profile;u=6155

[amspire]

It is incredibly hard for RMS converters to be linear over a big dynamic range as they have to average the square of the voltage.

So if you look at the 1V range and you have a RMS converter that can cope with a peak voltage of 2.25V DC and that can internally cope with up to 10V DC.

1mV AC in would mean the average squared volts in the RMS converter would be down around 2uV DC. An input of 0.1mV AC means the RMS converter is trying to average squared volts down around 0.02uV DC which is near to impossible to achieve.

It would be very interesting to find out how the big dynamic range is achieved in the 3458A. They could be running the RMS converter at a higher then usual voltage, or they may have added extra circuitry to somehow handle very low AC signals.

At least it is better to have meters that are non-linear rather then do what Agilent did in one of its recent handheld DMMs. When the AC signal reached the non-linear part of the RMS converter's range, they just set the display to 0.000 V.

Richard

[grenert]

At least it is better to have meters that are non-linear rather then do what Agilent did in one of its recent handheld DMMs. When the AC signal reached the non-linear part of the RMS converter's range, they just set the display to 0.000 V.

Actually, the 34401A is guilty of this!

[alm]

The AC ranges are probably only specified for 10% (maybe 1%) of FS or so, so it's all within specs. I would expect meters that calculate the RMS by digitally sampling the signal, like the 3458A supports, to be superior in this regard. I think the 34410A also does something similar. Older models usually used analog or even thermal RMS converters.

[robrenz]

The specs on the 8846A are worse than the 34410/11 specs  for the 100mV AC scale.
 
8846A
AC Voltage specifications are for ac sinewave signals >5 % of range. For inputs from 1 % to 5 % of range and <50 kHz,
add an additional error of 0.1 % of range, and for 50 kHz to 100 kHz, add 0.13 % of range.

34410/11
Specifications are for sinewave input >0.3% of range and > 1mVrms.
Add 30 uV error to AC voltage specification for frequencies < 1kHz.

When I am done stability testing I will do a rough test with signal generator and divider to see what it actually does below 5mV AC

[grenert]

I would expect meters that calculate the RMS by digitally sampling the signal, like the 3458A supports, to be superior in this regard. I think the 34410A also does something similar.

Ah, thanks for pointing that out.  I now see that mentioned in the 34401A vs 34410A brochure below.  It does indeed use sampling, while the 34401A uses the venerable AD637.  They specifically mention that performance below 10% of full scale is improved, and crest factors can be up to 10.  Both the 34401A and the Fluke 8846A are specified for crest factors up to 5, so I suspect they use similar methods for ACV.

When I am done stability testing I will do a rough test with signal generator and divider to see what it actually does below 5mV AC

Thanks!   

[robrenz]

8846A AC volts 10mV and below
This is a very crude test but I think it is still valuable.
Rigol 1022 signal generator  with sine wave output at 1KHz feeding a 100 Ohm 10 turn pot as a voltage divider.
Adjusted the Rigol impedance setting so the displayed 2V output equals an actual 2V across the 100 Ohm pot.
Adjusted the voltage divider to get 100:1 attenuation, 2V on Rigol = 0.02V out of the voltage divider. (Rigol only goes to 2mVrms)
Shorted inputs give total min max span of 2.73uV with a SD of .7uV  (noise floor)
These readings are non zeroed but remember I set my voltage divider ratio with the meter.
meter is in 6.5 digit mode with 100NPLC and analog filter on.
Stepped the Rigol down thru the voltages and and gave each reading about 5 seconds of settling time.

The data has readings from 10 mV down to 2mV that are not plotted.  The plotted values are from 0.02mV to 1.0mV
The only thing this shows is the 8846A does not show the nonlinearity that the gellerlabs plots show for the 34401 or 34410 and looks like it would be linear right into the noise floor (Rigol stops at 2mV giving 20µV out of the voltage divider) No accuracy determination can be made from this crude setup but I think the Rigol and voltage divider are linear enough to see if that graph bulge existed or if the meter had a bogus software zero.

I dont know about you but I am impressed