HP34401a DC reference max. out resistance

[MrYakimovYA]

Hello!

I'm just interested in how much the output resistance of a DC reference can be if I'm going to calibrate HP 34401a at all DC ranges?

I looked over the Service Manual. There is only Fluke 5700a is recommended as a reference. Or less accurate DC reference (with unknown output resistance) that must be measured with HP 3458a.

No word of output resistance(((( 10 Ohm, 50 Ohm?

[Kleinstein]

The input current on the 34401 is supposedly less than 50 pA in the 10,   1 and 0.1 V range.  It take some 2 kOhm to get 100 nV from this at least resolve this. So for the lower ranges there is not problem.

The ranges with divider are 10 Mohm.  So get 1 ppm chage with would be 10 Ohm of source resistance and the higher ranges are quite a bit less accurate. So 50 Ohm are probably still OK and if really needed one could correct much of this, if the output resistance in known.
The AC ranges are likely 1 Mohm, but also quite a bit lower accuracy. So chances are good a 50 Ohm output would be OK.

[dietert1]

It's fairly easy to insert an extra resistor into the cabling in order to check whether/how much the measurements depend on source resistance. Also this can become a test of the DVM input current, so you check whether it's input stage is working as expected.

Regards, Dieter

[MrYakimovYA]

It's fairly easy to insert an extra resistor into the cabling in order to check whether/how much the measurements depend on source resistance. Also this can become a test of the DVM input current, so you check whether it's input stage is working as expected.

Regards, Dieter

I fully agree with you. I can do it. But it is something like an trying with unknown results. I don't know how HP 34401a is calibrating internally. I mean which circuits and algorithms are executed. So, I would expect an official requirement regarding to max output resistance of DC reference (standard). But it seems to me there is no such requirements

[DavidAzulay5000]

[Kleinstein]

The zero adater shown in the last video is wrong. The way with 4 connections to the next terminals one gets 1/4 of the wire resistance. This is a low value be not really zero. One can however fix it by cutting one of the links from the sense to drive side, e.g. the link from H to Sense H.  The idea with the zero ohm adapter is to have the test current flowing in a way that it does not interfere with the link of the sense terminals.

[DavidAzulay5000]

The zero adater shown in the last video is wrong. The way with 4 connections to the next terminals one gets 1/4 of the wire resistance. This is a low value be not really zero. One can however fix it by cutting one of the links from the sense to drive side, e.g. the link from H to Sense H.  The idea with the zero ohm adapter is to have the test current flowing in a way that it does not interfere with the link of the sense terminals.

i build this zero ohm adapter work perfect

[Kleinstein]

The zero ohm adapter from the photos is OK, if used in the right orientation.

[alm]

I fully agree with you. I can do it. But it is something like an trying with unknown results. I don't know how HP 34401a is calibrating internally. I mean which circuits and algorithms are executed. So, I would expect an official requirement regarding to max output resistance of DC reference (standard). But it seems to me there is no such requirements

I imagine they expect any cal lab to use a calibrator like the Fluke 5700, Fluke 5440, maybe Fluke 5100, Datron 4700, etc. These calibrators will all have a very low output impedance, except at the lowest ranges, where the meter would be >> 10 MOhm anyway. So I guess the output resistance was never a consideration.

Before there were these calibrators, the practice would be like the Fluke 7105 system, where they had a lower accuracy calibrator and adjusted it to a standard cell using a voltage divider. But the calibrator was used to drive the device under test, not the divider.

[DavidAzulay5000]

The zero ohm adapter from the photos is OK, if used in the right orientation.

[MrYakimovYA]

I imagine they expect any cal lab to use a calibrator like the Fluke 5700, Fluke 5440, maybe Fluke 5100, Datron 4700, etc. These calibrators will all have a very low output impedance, except at the lowest ranges, where the meter would be >> 10 MOhm anyway. So I guess the output resistance was never a consideration.

Before there were these calibrators, the practice would be like the Fluke 7105 system, where they had a lower accuracy calibrator and adjusted it to a standard cell using a voltage divider. But the calibrator was used to drive the device under test, not the divider.

In case if using calibrators - yes, no question about output resistance. My question came from situation if I have only 10 VDC standard.
And I'd want to calibrate 1 and 0.1 V ranges with using Hamon Divider or any 1:10, 1:100 divider.

[alm]

If you have a short term stable adjustable voltage source, then you could adjust it to the 10 V standard using the Hamon divider. Otherwise output impedance of the divider would indeed be a concern, but on the other hand you don't want to load the 10 V standard too much.

The Fluke 5700A has an output impedance of 50 Ohm in the 220 mV range, but a low impedance (they don't specify) for higher ranges. The older Fluke 5440 had a ~500 Ohm output impedance for 220 mV and 2.2V ranges, and low impedance for higher ranges.

[Kleinstein]

With the divider one may get to a range (e.g. > 10 K) where the source impedance can matter.  It may be worth testing the effect by increasing (e.g. double) the impedance as suggested by Dietert.
The idea behind this is that the error should be proportional to the source impedance.  This is OK for input bias and resistance, though not necessary for an error due to a settling. So there are limits to what can be corrected.
 
In most cases the 34401 is very good with long term drift. So an adjustment is rarely needed.

A first test for a check or calibration would be to check if the input current is actually as small ( < 50 pA AFAIK) as it should be.

The suitable resistance for the divider depends on the voltage range.  The 100:1 divider may be tricky in the classic way with copying the 1:10 ratio. A 10 in series and 10 in parallel case may be more practical.