Keithley 220 current output stability?

[acts238willy]

Is the 220 stable enough in its old age to be ppm?
I'd like to force current with the 220 and read voltage
with a 3458 - to check resistors.
Or is the 3458 ohms circuit better?

[ap]

So you want to force current with the K 220, measure its value with the 3458A and then the voltage accross the resistor? Or how do you determine the exact current (the K 220 allone will be way too unprecise)? You would need two 3458A meters at a time, on for current measurement and one for voltage (as you cannot assume that the K 220 current output is fully independent of the load when you changeover from current to voltage measurement with one meter).
But what sense would that make? The current uncertainty of the 3458A is worse than the ohms uncertainty (simple spec comparison, pretty obvious). Plus you add the drift (temp, noise...) of the K 220. I am not getting the rational of why you even consider this, maybe you can explain.

[VintageNut]

From my reading of the datasheet, the answer is no. p-p noise is specified at best 100 ppm and at worst 400 ppm.

At 1nA, the total uncertainty is 0.6%if I did the math correctly. If you are trying to force 1pA, the uncertainty is 200%.

That said, specs are for mass production and covers all units. You can do better but you have to own better calibration gear and calibrate it enough times to be able to prove that you have a better cal results than the datasheet.

[acts238willy]

The noise numbers shoot the idea down.
I was going to hook 2 resistors in series and force the current through them.
Then, measure the voltage across each resistor, one being my standard and the other the
unknown. Then the ratio would tell me what the unknown resistance was.
With the current noise being tens of ppm or even more, the value would be useless.
Is there a 'cheap' old source that would be good to a ppm?
Thanks!

[CalMachine]

The noise numbers shoot the idea down.
I was going to hook 2 resistors in series and force the current through them.
Then, measure the voltage across each resistor, one being my standard and the other the
unknown. Then the ratio would tell me what the unknown resistance was.
With the current noise being tens of ppm or even more, the value would be useless.
Is there a 'cheap' old source that would be good to a ppm?
Thanks!

A Datron 4708 should suffice 

[SilverSolder]

Any reason not to simply use a stable voltage source/reference to power your two resistors?  After all, if one of the resistors is "well known", you will know what the current is from the voltage across it?

[e61_phil]

How much current you would like to force and how high are your resistors in value?
Keep input current into your meter in mind.

[TiN]

This arrangement with external current source and voltage detector usually makes sense for low-value resistors (<10 ohm) when you can pump more current than 3458A to get good data (beware self-heating and PCR!) or when you trying to measure high-value resistors (where you need high-voltage source + second divider to provide guard potential to avoid leakage). In 1k-1Meg range typical 3458A with simple substitution very likely to render much better result than random old current source + DVM in divider setup.

Also watch out for thermal EMFs and connection resistances with just simple divider. To automate setup and get decent results you need measure those parasitics and use current source and DVM that support reversal. Kinda like "TrueOhm" feature in F8508A DMM. To mitigate some of these issues often 4-resistor bridge is used instead.

[acts238willy]

Wanted to start with my SR104 10K and go up and down a decade.... after a bunch of readings
over a few weeks, I wanted a good value for 1K and 100K. Then 100 and 1M.
Looks like I'll just use the 3458 ohms function, even though it's not the greatest.
Thanks for the help, all...

[TiN]

Make sure to use correct settings for 3458A OHMF. Wiring, Delay and OCOMP can cause big errors. Search for Dr.Frank's explanation thread.