I bought a Keithley 6514 electrometer!

[Anders Petersson]

Maybe this post can be educational or invite some helpful replies.



Keithley 6514 is a 5 1/2 digit electrometer with vacuum fluorescent display (VFD). It can measure 0.1 fA - 20 mA, 10 uV - 200 V, 10 mOhm - 200 GOhm and 10 fC - 20 uC, all with low burden voltage and bias current.
Datasheet: https://www.tek.com/datasheet/high-resistance-low-current-electrometers-series-6500-6430/model-6514-programmable-electro
Manual: https://download.tek.com/manual/6514-901-01(D-May2003)(Instruction).pdf
Teardown photos and more info: https://xdevs.com/fix/kei6514/

6517 is more popular, with wider measurement range, better accuracy, voltage sourcing and dot-matrix display but naturally also higher price. I guess 6517 is newer than 6514, but I'm not sure. I opened a separate thread at https://www.eevblog.com/forum/testgear/history-of-keithley-electrometers/ for the history of Keithley electrometers.
Surprisingly, 6514 can still be bought new for an eye-watering list price of 7,240 €, which makes for a good feeling when picking up a used one! On Ebay the general price range seems to be $2000 - $3000.

https://www.tek.com/keithley-low-level-sensitive-and-specialty-instruments/keithley-high-resistance-low-current-electrom

The model for sale is called 6514/E, but I don't see the difference from the regular 6514 I got.

A Keithley leaflet cheekily pronounces 6514 as "Easy to use, even easier to afford": https://www.imperial.ac.uk/media/imperial-college/research-centres-and-groups/centre-for-bio-inspired-technology/7293196.PDF

I guess there's no great pressure to develop new models in this niche. For a more modern appearance (at least from Keithley), one would need to turn to SMUs like the Keithley 2400 series, for prices not adapted for occasional users. (Keysight B2987A is more modern, though.) The limited user interface compared to modern touchscreen models isn't too much of a problem as experiments are best done by remote-control from a computer over the RS232 or GPIB ports anyway. This will take some programming as I've only found support for 6514 in the Python-IVI library.

Anyway, the instrument I got is well-kept and looks pristine after cleaning the top lid with alcohol. It runs with an audible but acceptable hum. There's a faint chirp when switching ranges which I can't tell if it is a buzzer or creeking of a relay.
Basic operation was intuitive without the manual, except the "ZCHK" and "ZCOR" buttons. Apparently some modes start off with ZCHK (zero check) enabled which must be manually turned off to enable the input. Here the limitations of the display is most obvious as the designations "ZC", "CZ" and "ZZ" are not particularly explanatory.
When testing to read pH probes in voltage mode, the zero check mode seemingly short-circuits the probe which needs a minute or two to recover the voltage level after disabling zero check.

Although the instrument works across a "normal" voltage range, current measurements are reserved for specialty use cases as the range tops out at 21 mA and the manual states that "Exceeding either of these values may cause damage to the instrument that is not covered by the warranty."

Confusingly, the instrument is advertised as 5 1/2 digits, but can be set to display 6 1/2 digits in all modes. The datasheet specifies the coulumbs range in 6 1/2 digit resolution, but ohms, volts and amps in 5 1/2 digits resolution. What gives? Maybe the sixth digit is only useful when averaging is activated?

Now what?
I've ordered a triax cable from Ebay and will eventually build a guarded box for sensitive measurements.

I would also like to calibrate it. Checking against DMM6500, the voltage reading is off by 0.1%. The datasheet gives 1-year accuracy as 0.025% so maybe the error is reasonable after years without calibration.

As for amps calibration, it doesn't look like generating an accurate nA or pA current is that easy. A 2 V reference would need a 1 GOhm resistor to reach 2 nA and 1 TOhm to reach 1 pA. And high-value resistors are neither cheap nor accurate. A PDVS2mini could generate 10 uV, which with a 10 kOhm 0.01% resistor would give 1 nA. But I fear that the burden voltage (20 uV) or any leakage quickly invalidates current calibration with a voltage reference. I guess this is not the way to go.

The manual recommends to calibrate with some exotic equipment:
Fluke 5700A calibrator
Fluke 5450A Resistance calibrator
Keithley 5156 Electrometer Calibration Standard

Maybe the cheapest option is to find a commercial calibration lab. Have others here calibrated their electrometers?

[bsw_m]

Congratulations on your purchase!
You can try to take a source of linearly increasing voltage(ramp generator) with a speed of 10mV / s and an air capacitor with a capacity of 100pF, you can form a current of 1pA

[Anders Petersson]

Thanks! 
I don't have a ramp generator and for capacity, I can only verify the value to 0.8% accuracy (using DMM6500).

[antintedo]

I can only verify the value to 0.8% accuracy (using DMM6500).

0.8% of reading and 0.5% of range, so 5.8% for 100pF, not including effects from your setup and type A uncertainties.

[HighVoltage]

I had a 6514 a while back, sold it and got an almost new 6517B

You will need to invest heavily in to triax cables and connectors and build yourself a really good shielded box.
For some measurements you will need a shielded box within a shielded box to get stable readings.

[Anders Petersson]

You will need to invest heavily in to triax cables and connectors and build yourself a really good shielded box.
For some measurements you will need a shielded box within a shielded box to get stable readings.

Yes, I have ordered a triax cable and hope to get creative in assembling a double-shielded box.

[bsw_m]

Double shielding really not needed.

[HighVoltage]

Double shielding really not needed.

For to 1 TeraOhm you are right, simple shielding is enough.
But if you want to measure reliable 10 or 100 TOhm, you need double shielding to get stable readings.
Of course, there are also other ways to achieve that same level of stability.
Double shielding in a box is very simple.

Here is my TOhm resistor in smaller box inside the shielded box with 4 Triax connectors.

[bsw_m]

Double shielding really not needed.

But if you want to measure reliable 10 or 100 TOhm, you need double shielding to get stable readings.

Where I'm measure small current, such 1E-17, using measuring box without double shielding, I not have any problems with stability.

[Anders Petersson]

For to 1 TeraOhm you are right, simple shielding is enough.
But if you want to measure reliable 10 or 100 TOhm, you need double shielding to get stable readings.

In my case, 6514 only measures up to 200 GOhm, so no problem then. 

Anyway, I've read posts by both HighVoltage and shodan on this topic and I appreciate you're taking the time to write these educational posts.

[HighVoltage]

This Keithley engineer has a pretty interesting way of testing Peta Ohm (10^15 Ohm) with a water cup.

[HighVoltage]

And.... Dear HighVoltage, please remember:
Guard potential pin always have small(mV...uV) offset to ammeter input. Therefore i see at least 2 leakage ways into you're double box.

Indeed, that was not my best example to T-Ohm measurements, just a picture I had at hand.
Usually my inner box has also PTFE insulation and I am using PTFE cables.

 

[Anders Petersson]

Sounds good! 1V or even 10V wouldn't be that hard to generate with precision.