Keithley 2450 SMU bug when sourcing current (currently spikes from instrument!)

[Sparky]

Hi friends!

I thought I would share a bug report on Keithley 2450 SMU to inform others and check if anyone here has encountered this issue.

Setup: Keithley 2450 sources 1.1mA through a ~100 ohm resistor.  Keysight 34465A measures the voltage across the resistor in digitize mode (50kHz) and the results are graphed.

Result: See plot attached.

Bug: When sourcing fixed current an artifact is introduced.  The plot of voltage across the resistor shows periodic spikes.  At first I thought this was electromagnetic interference coming from somewhere in my lab, but the frequency is relatively low and the artifact precise and repetitive.  Furthermore turning the SMU output OFF the artifact would continue, but if I turn the SMU off itself then the artifact disappears.  Also if I disconnect the leads from the SMU the artifact disappears.

Artifact:
- Alternating high then low voltage/current at 0.5ms intervals.
- A burst of glitches lasts 16ms.
- 3.5ms delay between each burst of glitches.
- A series of three bursts of glitches repeats precisely every 500ms.

It looks like a 1kHz clock signal is coupled into the current source.  Perhaps something in the front panel or elsewhere that is clocking 16-bits in 3 bursts, and then repeated every 500ms. 

In a follow up test I connected a Keithley DMM7510 in series with the SMU and resistor while the 34465A is measuring the voltage across the resistor.  I attached photos of each instrument.  (In this case the SMU is sourcing 0.95mA, and I've connected the DMM7510 terminals such that its measuring negative current).  See photos attached.  I've also found that manually setting the "range" and changing the source current setting has an impact on the severity of the problem.

I'm on the latest firmware (v1.7.3) now but originally discovered the problem on an earlier firmware.  I decided to update to the latest hoping it was fixed but it is not.

So far I contacted Keithley and they've been able to reproduce the problem.

Has anyone else experienced this?

Sparky

[jjoonathan]

Another spike glitch issue? And here I thought the recent DMM7510 "update" was an anomaly 

[Sparky]

Another spike glitch issue? And here I thought the recent DMM7510 "update" was an anomaly 

omg, what is going on there?  What firmware version was that and do you have any info/link on that issue?  Is it reported to Keithley?

[HighVoltage]

Hello Sparky,
I made the test as per your instructions on my 2450 and 2460 SMU
It seems my 2450 has similar spikes but the 2460 does not.

[jjoonathan]

Yeah, Keithley knows about the DMM7510 issue and they don't care because it usually doesn't matter and can be worked around if it does. More info: https://www.eevblog.com/forum/testgear/flebay-dmm7510-has-issues/msg3399432/#msg3399432

Personally, if I were selling $4500 premium multimeters I would be very concerned about highly visible blemishes in their default settings regardless of whether or not they actually impaired typical measurements, but I guess Keithley sees it differently 

EDIT: Yep, still true as of latest firmware (1.7.5)

[Sparky]

Thanks HighVoltage for performing tests on your 2450 and 2460.  It is fortunate the problem is not exhibited on 2460.  I hope this means the root cause can be resolved in firmware.  Right now the issue is a nuisance because it interferes with my tests.  How can I be sure what I am measuring in other part of the circuit with such artifacts...  well, at least it is easy to spot!

Thanks jjoonathan for the info and links.  Looks like I should go and enable Line Sync immediately on the DMM7510! 

I agree with you and MegaVolt that even though some artifacts may be within spec these "spikes" are clearly not anticipated in default behavior of equipment.  Who is on the lookout for issues such as these when they are using $5000 instruments? 

[HighVoltage]

Thanks HighVoltage for performing tests on your 2450 and 2460.  It is fortunate the problem is not exhibited on 2460.  I hope this means the root cause can be resolved in firmware.  Right now the issue is a nuisance because it interferes with my tests.  How can I be sure what I am measuring in other part of the circuit with such artifacts...  well, at least it is easy to spot!

The 2460 Hardware is very different from the 2450 Hardware.
It is possible that this problem can not be solved through firmware update that that is the reason why Keithley has not moved on this.

[justanothername]

I was just writing that I can confirm this, now its gone all the sudden.
I swear I saw it on the picoscope.

[Sparky]

I was just writing that I can confirm this, now its gone all the sudden.
I swear I saw it on the picoscope.

Thanks for testing!  The spikes appear at 500ms intervals, and repeat continuously.  Depending on the picoscope setup maybe it triggered and other times didn't?  I'm not familiar with picoscope.

I attached two more images showing the settings on 2450 -- notice the output is not turned on yet, and the artifact still shows (see the image on 7510).  When the SMU output is enabled the artifact is still there.  On 2450, the "range" setting and "source" value both impact the artifact.

Thanks HighVoltage for performing tests on your 2450 and 2460.  It is fortunate the problem is not exhibited on 2460.  I hope this means the root cause can be resolved in firmware.  Right now the issue is a nuisance because it interferes with my tests.  How can I be sure what I am measuring in other part of the circuit with such artifacts...  well, at least it is easy to spot!

The 2460 Hardware is very different from the 2450 Hardware.
It is possible that this problem can not be solved through firmware update that that is the reason why Keithley has not moved on this.

Thanks for the note, HighVoltage.  Well...I will cross my fingers and really hope Keithley can fix this -- the artifact is very prevalent.

[Sparky]

Response from Keithley:

The periodic nature of it was curious to me. I checked with the factory/design engineering. The periodic nature of the increased source current is due to some internal sensor monitoring that the instrument performs. We must have a little internal coupling into the signal path. They were not alarmed by the observation.

They reminded me to keep in mind: our performance specs are all defined at 1NPLC settings using an integrating A/D converter.
Using a relatively wide bandwidth tool such as the 1MHz digitize feature of DMM7510 may show signal characteristics that are well above the effective bandwidth of any IV data the 2450 will return.

To illustrate, compare the DMM7510 reported current when the DMM7510 is set for DCI using the integrating A/D at 1 NPLC.
Using integrating A/D at 1NPLC on either the 2450 or the DMM7510, both will show current values that are well within the allowed sourcing error of the 10mA range of the 2450.

Overall Keithley claim the instrument is within spec and from this point it seems there will be no further investigation on the issue.  I understand their spec definition and requirement to have those conditions that define the performance, but it's still hard to find these spikes acceptable in practice.  It seems entirely unexpected and very likely can introduce artifacts into someones circuit, like it did for me.

I'm now considering work arounds to the problem -- a simple option is to place an inductor in series with the SMU output.  Just wondering if anyone has some thoughts or prior experience on resolving this type of problem practically?  That would help pointing me in the right direction.  Thank you!

In any case, I'll update here with what I'm able to come up with.