How to measure dcV with full resolution in the middle of a DVM range?

[cape zoloh]

I just upgraded from a 5.5 digit handheld to a 6.5 benchtop meter, thinking I would have greater resolution while doing dcV read outs.

Now having a 34401A, I want to read 5.000,000vdc but for some reason i get 05.000.00vdc instead. Is this common amongst many benchmeters, 6.5 digits and higher? I get 5.5 digits just like my Brymen hand held.
Lets say that this is not good enough. Are there any workarounds for this? (I suppose you need another known voltage which I don't have)

[Kleinstein]

The half digit is usually a digit that can only have the value of 0 or 1. With the lower resolution meters this implies something like +- 2000 / 20000 counts. With the 5.5 and 6.5 digit meters there are quite a few that have a maximum like 12 V. The 34401 is such a example with some +-1200000 counts - so just a little better than 6 digits, but not really full 6.5 half.
5.000,000 V would be already close to 7 digits.

With the 34401 like some other meters one may get more resolution when reading the result to the computer or maybe using math functions. However the actual limit is not so much the displayed number, but more the noise level and the linearity.

[bob91343]

There are some options.  First, purchase a voltage reference IC, say 5 V.  Energize it and measure its voltage with your good meter.

Then put that in series with the test leads (bucking) and read the difference, then add that to the reading gotten previously.

Some meters have options to add digits; my HP 3456A does.

[cape zoloh]

I would think that the loading of it being in an unknown circuit might change a number or two at the far ends...but if its buffered, than ok, it might work pretty well.. Seems like such a mechanical solution to something I did not plan for at all (I recieved the Multimeter yesterday, and the 5vdc resolution is the reason I bought it. (and the ohm meter). Been really turning the pennies so to speak, and finaly I found one, pretty afordable too. Now I thought I was set. Nope. The meter is sensitive to 100s of nanovolts but no, no 6.5 digits at your primary working space so to say. Its sad. I don't know how fast it is, but it seems like I can get higher resolution if hooking up the meter to the computer and do some filtering or averaging, correct?

[edpalmer42]

I have a Philips PM2534 6.5 digit meter.  To get the 6.5 digits you have to set the Speed control.  i.e. how many readings per second.  More digits = fewer readings per second.

Does the 34401A have something similar?

[rodpp]

The 34401A DC Voltage ranges are:

100.0000 mV
1.000000 V
10.00000 V
100.0000 V
1000.000 V

20% overrange on all ranges, except 1000.000V.

So, to measure 5V the best fit is the 10.00000V range, that will measure 05.00000V as you said.

The range down below is 1.000000V, with 1.200000V maximum.

The Brymen BM869s ranges in the 500,000 counts mode are:

500.000mV
5.00000V
50.0000V
500.000V
1000.00V

Supposing 10% overrange (I do not found that specification), only while measuring 0-120mV, 550mV-1,2V, 5.5V-12V, 55V-120V and 550V-1000V the 34401A will show one more digit.

[rodpp]

It is important to note that even having similar resolution, the accuracy of the 34401A is way better:

34401A (% of reading + % of range)
100.0000 mV - 0.0050 + 0.0035
1.000000 V - 0.0040 + 0.0007
10.00000 V - 0.0035 + 0.0005
100.0000 V - 0.0045 + 0.0006
1000.000 V - 0.0045 + 0.0010

Brymen 869s (% reading digits + number of digits)
500.00mV - 0.02% + 2d
5.0000V - 0.02% + 2d
50.000V - 0.03% + 2d
500.00V - 0.04% + 2d
1000.0V - 0.15% + 2d

So, when reading 5V on the display of both meters, you know that the real voltage value can be between:

Brymen 869s: 4.9988V and 5.0012V (error +- 1200uV)
34401A: 4.999775V and 5.000225V (error +- 225uV)

In reality that 5.00000V on the Brymen display can be trusted only as something around 5.000V, and the same measurement on the 34401A display can be trusted as something around 5.0000V.

[bdunham7]

Instead of digits, look at counts.  The 34401A is 1,200,000 counts, the Brymen is 500,000 counts.  So the 34401A only has 2.4X the counts.

The Brymen just happens to have its maximum resolution (in counts) right at 5 volts, but unless it overranges, as soon as  you go to 5.00001 volts you lose a digit.

I have no idea what you are doing, but unless you want to look at noise, this whole thing doesn't make sense. At exactly 5 volts the Brymen as a resolution of 2PPM, but a rated accuracy of more than 100X worse--200ppm + some counts.  Even on the 34401A, the last digit is very questionable. 

[Andreas]

Hello,

That is what you get on the serial interface in your terminal program
 when you set up 6.5 digits with 100 NPLC and GPIB-Address 31 (talk only)

with best regards

Andreas