All about Keithley DMM7510. Bugs and features, recipes, advice, notes.

[MegaVolt]

Can someone describe roughly (overall) how the issues discussed in this thread relate to the 6500?  For example, when 7510 issues surface are they likely to be similarly present in the 6500 and then fixed in the 6500 after being fixed in the 7510?  Or does it work in the reverse sequence?  Or some of both sequences?  Or are the issues lists for the two models dissimilar enough that the features requests and bug lists have relatively little relationship to one another in terms of being found and addressed?  Or something else...?

I think the only way is to learn to repeat the described steps and check if there is a problem in 6500.

I tried to post problems in the topic about 6500 but I was asked to start the topic separately because the DMM7510 "is very different from 6500."

I will be grateful if someone checks the problems found at 6500.

[MegaVolt]

Schedule deviation of time from the expected. Ideally, each sample should be in the time position t = i * T0. The graph shows the difference between the time given by the device and the ideal delta t = t - i * Т0
Graph for frequency 1 MHz

[Sparky]

Schedule deviation of time from the expected. Ideally, each sample should be in the time position t = i * T0. The graph shows the difference between the time given by the device and the ideal delta t = t - i * Т0
Graph for frequency 1 MHz

Thanks MegaVolt.  I really appreciate the testing you are doing on DMM7510 and bringing to light all the peculiarities of the meter, especially related to digitizer mode and aspects of logging/recording data.  I really hope Keithley is paying attention to your findings and follow this thread.

This result is really bizarre.  I wonder if the deviation in sample times is real, or a floating point rounding/truncation error, or something else.  It would be good to have clarification of this issue.  E-Design do you have any thoughts on this?

[Zucca]

A big Thank to MegaVolt also from my side. Well done, I have a 7510 and it is my top DMM on my bench. Unfortunately I do not use it too much. 

I hope to share my adventures here in the future.

[MegaVolt]

Thanks MegaVolt.  I really appreciate the testing you are doing on DMM7510 and bringing to light all the peculiarities of the meter, especially related to digitizer mode and aspects of logging/recording data.  I really hope Keithley is paying attention to your findings and follow this thread.

Thank you for your feedback. I am very pleased to know that someone else needs this. I do not understand Keithley's policy. I could not find contact with people who understand how the device works. Everywhere there are lawyers who say that the device is good and within the specification. Forum forum.tek.com is also silent. Therefore, we can help ourselves.

This result is really bizarre.  I wonder if the deviation in sample times is real, or a floating point rounding/truncation error, or something else.  It would be good to have clarification of this issue.

Bred O. mentioned an error when switching from one clock domain to another. And what I see in the two down pictures is very similar to this. But what was happening in the upper pictures was a complete surprise for me yesterday. Perhaps this is some kind of software frequency adjustment. For example, under frequent networks or something else ....

I have not yet come up with a good experiment on how to understand if there is a problem with digitization or not.

[MegaVolt]

A big Thank to MegaVolt also from my side. Well done, I have a 7510 and it is my top DMM on my bench. Unfortunately I do not use it too much. 

I hope to share my adventures here in the future.

Join !!! Maybe when there will be a lot of us Keythley will pay attention to us )

[MegaVolt]

To me, these kind of effects look like they have their explanation in firmware - So I cant help investigate it directly. What I can do is see if I can get a FW engineer to look at it or an Applications engineer who can spend the time to take a look.

Ask them to look. Thank!!!

My advice is post these questions and results on the Tek forum and if it doesnt get attention, it is easier to get more people involved to help.

Good! I will add them. I just don’t understand what is the problem and what is not. From the point of view of Keithley.

[Sparky]

To me, these kind of effects look like they have their explanation in firmware - So I cant help investigate it directly. What I can do is see if I can get a FW engineer to look at it or an Applications engineer who can spend the time to take a look.

Ask them to look. Thank!!!

My advice is post these questions and results on the Tek forum and if it doesnt get attention, it is easier to get more people involved to help.

Good! I will add them. I just don’t understand what is the problem and what is not. From the point of view of Keithley.

MegaVolt: can you post a link to your thread on the Tek forum?  I want to track it and post there also.

It appears Keithley are ignoring these issues.      Why can't they have one firmware guy reading these forums?   

[MegaVolt]

MegaVolt: can you post a link to your thread on the Tek forum?  I want to track it and post there also.

https://forum.tek.com/viewtopic.php?f=363&t=142217
https://forum.tek.com/viewtopic.php?f=363&t=142031

It appears Keithley are ignoring these issues.      Why can't they have one firmware guy reading these forums?   

Bred O. was such a man. But for some reason he stopped writing on the forum.

[MegaVolt]

DigiV 100V behaves very strange. Maybe of course these are problems of my device. If someone can check please what your devices show?

At 1000V all this is present but not so bright. And at least fit into the tolerances )

[MegaVolt]

This effect has improvement in next hardware revision.

I am happy to be able to participate in the free beta test program for future true Keithley customers.     

I understand the reason for the glitch in the digital track lying next to the divider? Or is it more complicated? Can we get more information to fix the problem ourselves? 

I understand that the glitches that I saw in the 100V range in the normal measurement modes are of the same nature?

[MegaVolt]

Different numbers of decimal places depending on the selected screen or the value of a number.

[Kleinstein]

It is probably the values to show. They seem to use the same number of total digits. So one more digit resolution once below 10 V. For the average the higher resolution may actually make some limited sense.

The peak to peak and StdDev values still have some digits well in the noise / fantasy. It is not a big deal, but ideally there should not be much more digits than actually real. If at the edge one more digit can be good - so the average could get one more digit, while the peak to peak and StdDev case should get less.

[JxR]

Temperature

* Does not support measurements for Pt1000. No solution.

Both the DMM7510 and DMM6500 should be able to use a PT1000 RTD.  Change 3/4-Wire RTD to "User" and set RTD Zero to 1000.

[essele]

I've just posted a couple of bugs around the temp measurement stuff in the Tek forum, so I though I'd add them here for completeness.

(a) Using a thermistor, with 1.7.0e the display reads in kilo degrees ... not very helpful! It was fine in the older firmware.
(b) Using a PT1000 (not sure that's relevant), with 1.7.0e the display reads "-DispNan-" ... graphing is fine, so the value is actually read ok. Haven't tried this in the older firmware.

I'm assuming these are generic temperature display problems with 1.7.0e firmware. Quite how you can release a major firmware release without testing this stuff is beyond me!

I've also found quite a few display issues, but in the scheme of things they seem less important. I must say my first impressions (after only 2 days) is that the hardware seems great, the capabilities (esp the triggering stuff) are superb, but the firmware quality is awful!

[MegaVolt]

Both the DMM7510 and DMM6500 should be able to use a PT1000 RTD.  Change 3/4-Wire RTD to "User" and set RTD Zero to 1000.

Thank you! I missed that moment.

I've just posted a couple of bugs around the temp measurement stuff in the Tek forum, so I though I'd add them here for completeness.

Thank you!

(a) Using a thermistor, with 1.7.0e the display reads in kilo degrees ... not very helpful! It was fine in the older firmware.

Yes it is not convenient

(b) Using a PT1000 (not sure that's relevant), with 1.7.0e the display reads "-DispNan-" ... graphing is fine, so the value is actually read ok. Haven't tried this in the older firmware.

This is strange. I connected a 1 kΩ resistor and got the measurements.

I've also found quite a few display issues, but in the scheme of things they seem less important. I must say my first impressions (after only 2 days) is that the hardware seems great, the capabilities (esp the triggering stuff) are superb, but the firmware quality is awful!

Unfortunately, a number of modes work either with an error or with noise that can hardly be called Gaussian. It seems that tested only some ranges that are really great.

[MegaVolt]

It is probably the values to show. They seem to use the same number of total digits. So one more digit resolution once below 10 V. For the average the higher resolution may actually make some limited sense.

The peak to peak and StdDev values still have some digits well in the noise / fantasy. It is not a big deal, but ideally there should not be much more digits than actually real. If at the edge one more digit can be good - so the average could get one more digit, while the peak to peak and StdDev case should get less.

Yes, I understand. I was uncomfortable getting an extra digit by switching the display. I'd love to have all the numbers on any display.

[MegaVolt]

Ohhhhhhhh !!! I found curious documents. True, they are in Russian.

1. This is a test procedure for the device.
2. Description of the type of measuring instrument.

These documents describe the conditions under which the device is accepted in the state registry of Russia. And the presence of these documents means acceptance by the state as a metrological device. Those. these are the parameters that an independent metrology laboratory checks and guarantees.

[MegaVolt]

Nameplate describing noise quality with shorted inputs in DigiV mode:

[essele]

(b) Using a PT1000 (not sure that's relevant), with 1.7.0e the display reads "-DispNan-" ... graphing is fine, so the value is actually read ok. Haven't tried this in the older firmware.

This is strange. I connected a 1 kΩ resistor and got the measurements.

That's bizarre ... I've just attached a 1k resistor, using 4W resistance mode it's fine .. I switched to temperature and 4RTD and it said '-DispNAN-', then I went through some of the other transducer devices and it's started working ... very weird.

Just tried a reset and still it appears to be working, although it said "PT100" but was reading a correct value ... so I don't think "reset" actually reset the zero value. That seems like another bug. (This is reproducible -- it's nice to know reset isn't actually a reset!)

Power cycled, and it then seems to be working fine ... PT100 shows "overflow", and then setting the zero value to 1000 works ok.

So I think there's some sequence of events that I went through (I had been using a thermistor before) that got it into this confused state. I've tried going back over some similar steps but can't reproduce it.

Further -- the "zero" setting actually persists if you go from 4RTD to Thermistor and then back to 4RTD even though it says "PT100", so they are not properly resetting values. If you manually select "PT100" then it's ok.

Horrible.

[MegaVolt]

Unfortunately, this is a frequent occurrence for this device. Any deviation from the correct path can lead to glitches which can be eliminated only by rebooting

[MegaVolt]

The device allows you to set up a fractional aperture; maybe it works; maybe not. I haven’t figured it out yet.

I checked the ability to install a fractional aperture. Unfortunately this does not make sense. In manual mode, the aperture can be a multiple of 1 μs and nothing else.
Fractional aperture only becomes in AUTO mode as I described earlier.

If you want to get a signal suitable for FFT processing, I recommend setting it to either 1µs mode or AUTO mode. In this case, the AUTO mode averages several samples, which improves the signal-to-noise ratio.

Of course, remember that the device has a wide band and if you have not limited the frequency of the signal, then you will get frequency aliasing.

[MegaVolt]

New Test Script Builder released.

https://www.tek.com/software/testscriptbuilder/kts-850j05

[essele]

I've just updated another item I've logged on the Tek support forum...

https://forum.tek.com/viewtopic.php?f=363&t=142266

Its basically saying that, as the graphs start to cover longer time periods, they seem to round the data values and have quite a signficant loss of precision (although there are strange anomalies where this doesn't appear to happen.)

I seem to be talking into a void on the other forum (although did get a response on one of the temperature issues) but thought I'd include it here for completeness.

More detail is in the other forum, but basically I've taken the same (almost) data and bucketed it into 300 buckets (storing min and max) and then plotted using excel, so this is assuming they have 300 pixels to display the data on the screen, and this is roughly what it should look like...


But on the 7510 it actually looks like this...



.. clearly awful.

[Kleinstein]

The reduced resolution is really odd. There may be  reason to use reduced resolution for intermediate data to save on the memory - though still odd, as 1 µV resolution would be more than 24 bits and 32 bits would give enough resolution.

So it looks like another small points of sloppy implemented graphics software.

There is another not so nice feature that was noted before: the grid lines are 1.1 µV apart, which is an odd choice. At least it looks like exactly 1.1 with no extra rounding error at the labels.