New Keithley DMM6500 and now DAQ6510

[maxwell3e10]

If this is a problem for all DMM6500 meters and even exists to a significant extend on the 10V (and perhaps other) ranges, its a fairly major flaw. Its surprising it took over a year since the instrument was introduced to discover it. These boards really help to ferret out the flaws.

[cozdas]

Here is the simplified bug repro steps for people to test it (Edit: even simpler steps)
- switch to DC volts, 100V range
- short the Input High and Input Low: you should read ~0V on the display
- While still in DCV mode pass DC 1A through the Amps-Input Low terminals: Now you will read a large value (-2.6V in my case)

I also tested with the rear terminals: If you use the rear 10Amp therminal, the error in voltage reading is around only -5mV in all ranges. Rear 3Amp terminal behaves the same as the front terminals: -2.6V error for 1A in 100V range.

therefore the workaround is:
If you need to pass current through the DMM while measuring voltage use the rear 10Amp terminal. This will reduce the voltage reading error significantly.

[Mike G]

Hi, thanks to everyone for highlighting this serious problem. I have just checked and my 6500 exhibits the same problem. There is indeed a small (but totally unnacceptable) error introduced on the 10v range and an enormous error on the 100v and 1000v ranges, the meter certainly is not within spec under these conditions. I will try to upload a couple of pictures later today.
Regards all, Mike

[Hugoneus]

I couldn't fall asleep, so I went and tried it again.

I can confirm that I see the problem with the DAQ6510 as well. At 1A there is an offset of -2.6V present in the 100V range and about 80mV in ranges < 100V.

I will discuss this with my Keithley support contact.

[JxR]

I couldn't fall asleep, so I went and tried it again.

I can confirm that I see the problem with the DAQ6510 as well. At 1A there is an offset of -2.6V present in the 100V range and about 80mV in ranges < 100V.

I will discuss this with my Keithley support contact.

Thanks for checking again and confirming.  I have to say I was kind of hoping the DAQ wouldn't have the problem as well.

[Hugoneus]

I couldn't fall asleep, so I went and tried it again.

I can confirm that I see the problem with the DAQ6510 as well. At 1A there is an offset of -2.6V present in the 100V range and about 80mV in ranges < 100V.

I will discuss this with my Keithley support contact.

Thanks for checking again and confirming.  I have to say I was kind of hoping the DAQ wouldn't have the problem as well.

Both the DMM and DAQ have the same architecture, so it makes sense that they share the issue. I am curious as to why the DMM7510 doesn't have this problem.

I wonder if this can be fixed in firmware; I am skeptical, because the instrument normally can't tell if there is current going through the other terminal without switching some relays at low current ranges.

[maxwell3e10]

I also tested with the rear terminals: If you use the rear 10Amp therminal, the error in voltage reading is around only -5mV in all ranges. Rear 3Amp terminal behaves the same as the front terminals: -2.6V error for 1A in 100V range.

If the error is present at some level on all current and voltage ranges it suggests a problem with the ground reference point. This might not be so easy to fix. Even a 5 mV error is well outside the spec on most ranges. I wonder if it says in the manual somewhere that simultaneous current flow and voltage measurements are not allowed. Or perhaps in the process of simplifying the design of the DMM7510 they went a bit too far.

[JxR]

If this is a problem for all DMM6500 meters and even exists to a significant extend on the 10V (and perhaps other) ranges, its a fairly major flaw. Its surprising it took over a year since the instrument was introduced to discover it. These boards really help to ferret out the flaws.

I reported the problem back on page 21 about 1.5 months ago, and started engaging with Keithley about the problem at the same time.  I think the fact that they went completely silent and refused to continue to communicate with me after I sent them a video I recorded fully demonstrating my setup and the problem speaks volumes.  They must have confirmed the problem but don't currently have a solution to it.

I didn't want to make a fuss about it since I never knew until now if it was just my unit or not.

This might not be so easy to fix. Even a 5 mV error is well outside the spec on most ranges. I wonder if it says in the manual somewhere that simultaneous current flow and voltage measurements are not allowed.

Secondary measurements are feature of the device and I never remember reading anything that would imply such a problem.

[maxwell3e10]

When measuring current, one should sense voltage directly across the current shunt. When measuring voltage, one should sense voltage directly across input terminals. One possibility is that they "saved" on one switch that moves the ground reference of the ADC from one point to the other, so it now includes a voltage drop across some internal traces.

[The Soulman]

When measuring current, one should sense voltage directly across the current shunt. When measuring voltage, one should sense voltage directly across input terminals. One possibility is that they "saved" on one switch that moves the ground reference of the ADC from one point to the other, so it now includes a voltage drop across some internal traces.

That sounds like it.

[JxR]

When measuring current, one should sense voltage directly across the current shunt. When measuring voltage, one should sense voltage directly across input terminals. One possibility is that they "saved" on one switch that moves the ground reference of the ADC from one point to the other, so it now includes a voltage drop across some internal traces.

I thought I remember someone where in this thread the shunt resistor values were confirmed.  Although not sure where exactly that was.  Maybe that would help you with your theory.

[iMo]

A year back this poster showed similar results - aprox 0.3V/1Amp.
With 100V range aprox 3V/1A then?

https://www.eevblog.com/forum/testgear/new-keithley-dmm6500/msg1781402/#msg1781402

[JxR]

A year back this poster showed similar results - aprox 0.3V/1Amp.

https://www.eevblog.com/forum/testgear/new-keithley-dmm6500/msg1781402/#msg1781402

I don't any problem at all on the 1V range, which all those pictures indicate.  Doesn't appear related unless I'm missing something.

[The Soulman]

Does anyone have link to the manual that shows this setup?
Couldn't find it myself.
I'm curious to know how this is supposed to work with a shared common terminal without introducing errors.

[JxR]

Does anyone have link to the manual that shows this setup?
Couldn't find it myself.
I'm curious to know how this is supposed to work with a shared common terminal without introducing errors.

I don't remember there being any diagram for a measurement setup demonstrating the dual V/I measurement.  Regardless, you pretty much have to do your current measurement on the low side after your load, and your HI input closer to the source.  I made a diagram of the measurement I made using a battery which should be a page back at the most.

https://www.eevblog.com/forum/testgear/new-keithley-dmm6500/msg2578767/#msg2578767

[The Soulman]

Found yours, There you show two separate gnd lines from the voltage and the current meter, in reality
it is one and any voltage drop on it (caused by the current flow) wil add (or subtract rather) to your
voltage reading.
Inside the meter there is also voltage drop on the current shunt and fuse also introducing error unless
 they switch around the gnd-ref point for the adc.

Dave's 121gw also has a similar "feature".

[iMo]

What you get if the Voltage is secondary?

[KedasProbe]

I guess the problem will be similar when I tried to measure the voltage drop
https://www.eevblog.com/forum/testgear/new-keithley-dmm6500/msg2427312/#msg2427312
For the 100mA ranges there was a relays switching that made the readings wrong:

This was the given explanation:

Ah, my bad on not explaining this, the relay switching will cause brief discontinuities in the AMPS terminal so the Rigol that's not synced to the switching would measure a higher resistance.  If you hook a scope up to the terminals you should be able to see this.  The discontinuity happens because voltage measurements happen while the current range relay is in the 1A/3A position.  So if the relay isn't already in that position it will be forced there.  I'm not totally sure all the reasons that position is required, but I believe part of it is to prevent noise from reaching the voltage measurement circuitry. 

So anyway, it's best to use the 1A/3A ranges if you want to continuously measure burden voltage with this method.

Maybe an old firmware should be tested to see if it was introduced later or if it was always like that.

[JxR]

Found yours, There you show two separate gnd lines from the voltage and the current meter, in reality
it is one and any voltage drop on it (caused by the current flow) wil add (or subtract rather) to your
voltage reading.
Inside the meter there is also voltage drop on the current shunt and fuse also introducing error unless
 they switch around the gnd-ref point for the adc.

Dave's 121gw also has a similar "feature".

Are you referring to my little circuit drawing for the test from the battery?  It was just a drawing and there is only a single reference point to ground regardless of how many wires I showed.  Either way, I showed actual pictures of the wires from the DMM and connected to the battery, there really shouldn't be any confusion.

As to the voltage drop across the test leads, we are talking about 40mOhm leads here.  The drop across them is completely insignificant compared to the 2.6-2.8V drop we are seeing at 1A.  It is arguing that the number should be 2.56V instead of 2.60V.

Sorry if I'm misunderstanding what you are trying to say, pretty tired atm.

[KedasProbe]

Hi,

I tested it with my DMM6500 but I simplified the schematic a little (yes that's possible)
I just connected a current source to the DMM6500 and shorted the voltage terminals with 50 Ohm.
So I should get 0 Volt all the time:

I got this: no sound, sorry, when the voltage changes to -2.47V instead of -36mV, it is still -36mV on the terminals.



edit: one more thing I didn't have a relays sound on 100V range (the same like lower ranges)
edit2:
When no resistor is connected then for 1.00A I get -21mV for voltage (and much more in the 100v range and up)
When I short the voltage terminals I get -38mV (and again much more for 100V range and up)

[JxR]

Hi,

I tested it with my DMM6500 but I simplified the schematic a little (yes that's possible)
I just connected a current source to the DMM6500 and shorted the voltage terminals with 50 Ohm.
So I should get 0 Volt all the time

Nice demonstration.  Thanks. 

Edit: I did see about the same sourcing 1A from one of the SMU when I tried.  I like this method, but I wanted to post what I thought was simpler: a 9V battery and a 10Ohm resistor.  I would guess that a lot of people don't actually have a precision current source (but have batteries and voltage sources) and my goal was to have as many people as possible test their DMM6500. 

[maxwell3e10]

This is a good general test setup: flow 1 A into current input, short out the voltage input and measure the voltage. It would be interesting to test DMM7510 and other meters in this way to measure the internal common terminal voltage drop. It depends  on details of construction, can be different for front vs. rear terminals.

In DMM6500 it seems they make a mistake of connecting the 100:1 voltage divider to the common before the current shunts.  This has the effect of amplifying the voltage drop by a factor of 100 on 100V and 1000V ranges.

[KedasProbe]

Edit: I did see about the same sourcing 1A from one of the SMU when I tried.  I like this method, but I wanted to post what I thought was simpler: a 9V battery and a 10Ohm resistor.  I would guess that a lot of people don't actually have a precision current source (but have batteries and voltage sources) and my goal was to have as many people as possible test their DMM6500.

I'm pretty sure anyone with a DMM6500/DAQ6510 has a power supply with max. current protection.
That's what you see in my video, the max current protection doing its job.

[maxwell3e10]

OK, let's not argue. JxR gets the credit of finding this flaw after a lot of people used the meter for a year.

[JxR]

I'm pretty sure anyone with a DMM6500/DAQ6510 has a power supply with max. current protection.
That's what you see in my video, the max current protection doing its job.

Cool.  My mistake, and I didn't actually try it this way myself.  But yeah its essentially the same as using a dedicated current source.

Overall, I guess I'm glad I am no longer alone with the flaw since it tells me I'm not crazy ...but then again it would have been better if it was only my unit since I could have just sent it in for warranty service and no one would have to suffer with me.

I guess the next question is what should we do about it.  All start requesting warranty service from Keithley for our meters?