### Author Topic: Calibrating DVMs with a Kelvin Varley Divider  (Read 1341 times)

0 Members and 1 Guest are viewing this topic.

#### retroware

• Contributor
• Posts: 17
• Country:
##### Calibrating DVMs with a Kelvin Varley Divider
« on: November 30, 2016, 01:13:27 pm »
I'm a bit puzzled with the calibration procedure for the Fluke 8505.  On page 4.8 of the service manual they describe a procedure in which a stable voltage is first calibrated to a standard cell via the 720A KVD. Next, they run that voltage directly thru the 720A KVD to the input of the 8505. The reason I'm puzzled is that the 720A output resistance is ~50K (depending on the exact setting) and they run that into the 8505 with an input impedance of >10G.  But if the input impedance is 10G, this would result in a 5ppm voltage drop across the output of the KVD. The 8505 is speced at 6ppm thus I'm at a loss as to how this procedure can be accurate. And that is before one factors in the linearity of the KVD and the input bias current of the DVM.

Am I missing something? When they spec the input impedance as >10G do they really mean a lot more the 10G?

#### VintageNut

• Frequent Contributor
• Posts: 504
• Country:
##### Re: Calibrating DVMs with a Kelvin Varley Divider
« Reply #1 on: November 30, 2016, 05:14:25 pm »
The input impedance is the voltage divided by the input bias current. The input bias current is most likely fixed. You can characterize the input bias current by plugging a 10M resistor into the input and allowing the instrument to tell you what the bias current is.

This is discussed on other threads in this forum.
working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)

#### Alex Nikitin

• Frequent Contributor
• Posts: 974
• Country:
##### Re: Calibrating DVMs with a Kelvin Varley Divider
« Reply #2 on: November 30, 2016, 07:58:17 pm »
The input impedance is the voltage divided by the input bias current. The input bias current is most likely fixed. You can characterize the input bias current by plugging a 10M resistor into the input and allowing the instrument to tell you what the bias current is.

This is discussed on other threads in this forum.

In most cases the input current will vary with the input voltage and with temperature. The only way to characterize it properly is to measure it directly with an electrometer over the whole input voltage range. I did it for my two Keithley 2015 meters and the input current was -15pA at 0V, -95pA at +10V and -7pA at -10V for one meter and +12pA at 0V, -120pA at +10V and +100pA at -10V for the second unit (the specification is <200pA @ 23C).

Cheers

Alex
« Last Edit: November 30, 2016, 08:03:22 pm by Alex Nikitin »

#### VintageNut

• Frequent Contributor
• Posts: 504
• Country:
##### Re: Calibrating DVMs with a Kelvin Varley Divider
« Reply #3 on: November 30, 2016, 08:04:38 pm »
Good points. I will have to try this with my KE236. It can resolve pA and can source +/- 110V. I do not have an electrometer.

Electrometers are nice when you want the current measure on the LO side. SMUs measure current on the HI side and can create hazardous voltages on the guard.

The input impedance is the voltage divided by the input bias current. The input bias current is most likely fixed. You can characterize the input bias current by plugging a 10M resistor into the input and allowing the instrument to tell you what the bias current is.

This is discussed on other threads in this forum.

In most cases the input current will vary with the input voltage and with temperature. The only way to characterize it properly is to measure it directly with an electrometer over the whole input voltage range. I did it for my two Keithley 2015 meters and the input current was -15pA at 0V, -95pA at +10V and -7pA at -10V for one meter and +12pA at 0V, -120pA at +10V and +100pA at -10V for the second unit (the specification is <200pA).

Cheers

Alex

working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)

#### Alex Nikitin

• Frequent Contributor
• Posts: 974
• Country:
##### Re: Calibrating DVMs with a Kelvin Varley Divider
« Reply #4 on: November 30, 2016, 08:21:58 pm »
Good points. I will have to try this with my KE236. It can resolve pA and can source +/- 110V. I do not have an electrometer.

Electrometers are nice when you want the current measure on the LO side. SMUs measure current on the HI side and can create hazardous voltages on the guard.

The Keithley 617 electrometer I use, can measure the current with up to +/-200V floating input and can source up to +/-100V so it is easy to measure a DMM input current in that range on the HI side using it (though a high impedance input on DMMs usually valid only up tor 10/20/30V range max).

Cheers

Alex

#### e61_phil

• Frequent Contributor
• Posts: 663
• Country:
##### Re: Calibrating DVMs with a Kelvin Varley Divider
« Reply #5 on: November 30, 2016, 08:34:13 pm »
The only way to characterize it properly is to measure it directly with an electrometer over the whole input voltage range.

Another method could be a series resistor between your voltage source and the DMM. Measure the voltage source without the resistor and after that with the resistor in series. The difference will tell you the input current (together with the resistance).

And beware of "Auto-Zero" this will add some current spikes for recharging the input.
« Last Edit: November 30, 2016, 08:36:13 pm by e61_phil »

#### retroware

• Contributor
• Posts: 17
• Country:
##### Re: Calibrating DVMs with a Kelvin Varley Divider
« Reply #6 on: November 30, 2016, 11:34:34 pm »
Thanks for clearing that up for me.  The Fluke 8505 is speced with an input bias current of <5pa at adjustment time. Combined with the ~50K output resistance of the 720A, this gives 250nV which is .25 ppm on the 1V range. Good enough for me. The 720A is only speced at 1ppm at a division ratio of 10 so it together with the voltage reference seem to be the limiting factors.

As Alex pointed out one needs a electrometer to verify for sure. It sounds like an excuse to pick up another pieced of gear I also like the idea of simply using a resistor but there goes my excuse to engage in further volt nuttery.

Smf