C Andrea

My hunch is that the SMU control loop is destabilized by your 100nF.

And evidently the situation is worse for compliance setting of 10mA vs 5mA.

What is the measure range in each case?  Is it auto ranging or set to fixed current measure range?

Turning on high C mode should give you the consistent behavior of appx 1mA measured when forcing 5V regardless of current limit setting at 5mA or 10mA.

From data sheet of 2602B:   Maximum Load Impedance Normal Mode: 10 nF (typical). High Capacitance Mode: 50 µF (typical).

Since you have 100nF, try enabling the high capacitance mode.

Glad to hear you solved it.

FYI for others, if you turn off auto range for current measure, and set a fixed range, the allowed values of current compliance limit must be within that selected measure range.  This is why too small for sense range error is raised.

If auto range is on, then it is possible to have a much higher value for current compliance than the presently in use measure range by the auto range algorithm.

Since your *IDN? query is working well, I'd say your communication is proved.

You are getting a timeout on the other query because it is not a valid command for the 2611B.
Therefore, that command will not cause any response message and your gpib read times out.

I suggest you consider updating the firmware level.  You are at 3.0.3.  Version 3.4.2 is available for your unit.
You need a login to the tek.com site, but otherwise is a free download.

Also while there, download the reference manual which details the commands that the 2611B does understand.
The user manual offers a subset of that info and some typical use case scenarios.

If you want to know the line frequency you can write a command:  print(localnode.linefreq).
Then read the response that is "printed" to the output queue.

Downloads for the 2611B

 

For your incorrect current measurement, try some manual operation and progressively add cable and probes.
So how low are measurements of current for just instrument?
If nice and small, restore zero check and modify connection to just one cable or connector.
Then measure again.

Some BNC cables can give quite a lot of “popcorn” spurious current noise.  Make sure you don’t have that before measuring the sample.

Is your piezo material in a shielded enclosure?  If not, then I image you will get a lot of measurable change just due to environment.  Wave your hands or walk by while measuring.  You can probably detect the motion as charged dust from air impacts the sample surface.

Hi Michael.  Great feedback!  I’ll make sure the factory/firmware engineers are aware of your report.

So, you will generate V and I data as function of some changing amount of mechanical stress?

For voltage, how high is high?  I suspect the sample also has high source impedance.  Therefore your volt meter needs input R spec larger than the source R to avoid loading errors.  Suggest the electrometer 6514 or 6517B.
Alternatively, a SourceMeter with low current ranges can be put into force 0 amps and limit voltage to some value higher than what you expect to measure.  Something like 2636B or 2635B has 1nA source range and offers quite good input R when behaving like volt meter good up to 200V.  Use 2657A if you need higher voltages.

By definition, a good v-meter has zero current flowing into or out of the meter.  So makes it not feasible to also measure the current on those same connection points with same instrument.  Unless the instrument is a SourceMeter, then it can change its configuration and force 0 V with current limit higher than you need to measure.
But understand I am describing two sequential measurements:  measuring V when force 0 Amps, then measuring I when forcing 0 V.  Can this work for you?

If envisioning two meters, how will you connect them?  Seems the amp meter will compete with the V meter for what the charge stored on your sample should do.

 

Not sure about pyMeasure and what actual SCPI commands it is sending to the 2400.
Looks like a loss of synchronization between the code and the instrument state since the error handler is itself throwing an error!

But since it is working mostly, I am wondering if baud rate of the RS-232 is playing a role.
On our old forum, there are some posts about dodgy RS-232 behavior at the fastest baud rate.  
Try slowing down to 19K baud rate.

If trouble persists, can you reduce the code into smaller bit that shows the issue?
Alternately, try using pySerial or pyVISA and directly send the SCPI commands understood by the 2400.

There is a Verification procedure and a separate Adjustment procedure.
When I used "tweak" I was thinking to carry out the Adjustment procedure.

For just the DC Amps, you've indicated the zero is out of tolerance.
Is it only 10uA range, or is the 100uA range also affected?
What about the full scale 10uA level?  Is that in spec?  Have you adjusted it whether it seems to need it or not?

And I was just guessing that perhaps the adjusted DCV zero might have some impact on the DC Amps A/D converter readings too.

If it continues to be out of tolerance, it seems an RMA for service to understand root cause is needed.

 

Hi Sergey,

The tspnet.write is not limited to just a few characters.  Not sure what the upper limit is, but is not 5.
Here is some sample code where the feature is used to interact with a Watlow temperature controller and many of the commands used there with tspnet.write are longer than 5 characters.

https://github.com/tektronix/keithley/tree/main/Instrument_Examples/DAQ6510/Use_the_DAQ6510_to_Control_a_Temperature_Chamber

Sometimes we have to deal with string is really a null terminated array of characters.
Google up ASCII (decimal) code for certain character.  the Zero is 48.  Null is 0.
Line feed is 10;  Carriage Return is 13.