EEVblog 121GW Multimeter Issues

[npelov]

Hmm I posted this in the wrong forum - so I repost it here:

Since v2.0 the SD card logging stops working properly after a while. I haven't tried that in 2.02. The last time I recorded voltage for 19.5-20 hours with interval 2 seconds, fixed range. I monitored the display - the battery dropped gradually with cut off at about 10.8V. At about 19.5 hours the voltage was 11. At 19.0 hours - 11.8V. However since about 2h 40min (about 4800 recods) the value in csv is  12.529V +/- 3 LSD. The last digit keeps moving around like it's recording real data, but it has nothing to do with what's on the display.

I've reported this before, but I don't see it in firmware change log. Is this fixed? Does anyone else have similar problems?

Edit: Also since v2.0 the multimeter will occasionally freeze when I start logging - switching ranges, pressing buttons have no effect, until I turn it off.

P.S. Yes, I know that bluetooth exists. I don't want to rely on wireless for long term logging.


P.S. 2: I am currently starting a new 20+ hours log with v2.02 - we'll see if it happens again.

[joeqsmith]

Hmm I posted this in the wrong forum - so I repost it here:

Since v2.0 the SD card logging stops working properly after a while. I haven't tried that in 2.02. The last time I recorded voltage for 19.5-20 hours with interval 2 seconds, fixed range. I monitored the display - the battery dropped gradually with cut off at about 10.8V. At about 19.5 hours the voltage was 11. At 19.0 hours - 11.8V. However since about 2h 40min (about 4800 recods) the value in csv is  12.529V +/- 3 LSD. The last digit keeps moving around like it's recording real data, but it has nothing to do with what's on the display.

I've reported this before, but I don't see it in firmware change log. Is this fixed? Does anyone else have similar problems?

Edit: Also since v2.0 the multimeter will occasionally freeze when I start logging - switching ranges, pressing buttons have no effect, until I turn it off.

P.S. Yes, I know that bluetooth exists. I don't want to rely on wireless for long term logging.


P.S. 2: I am currently starting a new 20+ hours log with v2.02 - we'll see if it happens again.

I have done a fair amount of logging with the 121GW using both the SD card and wireless.   I've not had any problems but I wonder if it has something to do with the time or amount of data you are collecting.   I will try to repeat your test using 2.02 and see if I am able to replicate what you are seeing.     

******

It's more than half way and watching the display, everything appears normal.  Taking the number of samples and converting to hours, it checks out.    I've certainly collected more than 4800 samples but normally I collect at the highest speed or 1sps, not 2sps.     I should be be able to provide an end result yet today.   

[npelov]

Same results with 2.02 firmware. 39003 points collected, at some point the voltage just stops dropping. The last value is 11.65V, but I've seen that at the end it was about 11.1V (cutoff is at 10.5 and then the battery voltage recoveded a bit).

Btw I tested the cut off with a bench power supply - the cut off at 10.50V works fine.

The drop near 5000th record is the end is because I physically bumped the probes off the battery. You can see after 26000-27000 records the graph flattens and stays at about 11.65V, and I know that can't be true because I can see different value on the display of the DMM

[exe]

Same results with 2.02 firmware. 39003 points collected, at some point the voltage just stops dropping. The last value is 11.65V, but I've seen that at the end it was about 11.1V (cutoff is at 10.5 and then the battery voltage recoveded a bit).

Btw I tested the cut off with a bench power supply - the cut off at 10.50V works fine.

The drop near 5000th record is the end is because I physically bumped the probes off the battery. You can see after 26000-27000 records the graph flattens and stays at about 11.65V, and I know that can't be true because I can see different value on the display of the DMM

Did you measure the voltage of DMM by using the same DMM?

[joeqsmith]

Same results with 2.02 firmware. 39003 points collected, at some point the voltage just stops dropping. The last value is 11.65V, but I've seen that at the end it was about 11.1V (cutoff is at 10.5 and then the battery voltage recoveded a bit).

Btw I tested the cut off with a bench power supply - the cut off at 10.50V works fine.

The drop near 5000th record is the end is because I physically bumped the probes off the battery. You can see after 26000-27000 records the graph flattens and stays at about 11.65V, and I know that can't be true because I can see different value on the display of the DMM

Did you measure the voltage of DMM by using the same DMM?

I thought they have a completely different battery pack with some sort of load attached to it that has some sort of cutoff voltage.  The 121GW is monitoring the battery packs drain.  I assumed what they were looking at has nothing to do with the 121GW's internal batteries. 

I assume they point of the cutoff of their load because it is much lower than what the 121GW is showing in the log file.   

I was wondering if their load with the battery were instead getting into some case where it limits the average draw and this is why they see it flatten out. 

In my case, I used a constant resistive load so I know there is nothing going on that could fool me.   I also changed out the battery for the last 5 hours of data to make sure the meter is still recording new data.    It still has a ways to go but at least looking at the display, everything appears to be fine.

[Scottjd]

In my case, I used a constant resistive load so I know there is nothing going on that could fool me.   I also changed out the battery for the last 5 hours of data to make sure the meter is still recording new data.    It still has a ways to go but at least looking at the display, everything appears to be fine.

I meant to set mine up last night, but forgot as I fell asleep waiting for my meters and voltage ref to warm up. I needed to check the calibration since I moved and the equipment was not in use for I months. The good news, bench meter is still looking good. But I already created the profile for this log test yesterday, so I figured I would still run it.

In my case I’m using the Rigol PSU to run a profile that starts at 20VDC, and stops at 10.8VDC
It will drop the voltage every 35 seconds over a 2048 points exp fall curve down @ a fall index of 3. This will give it a log and runtime of 19.91 hours.

If I knew how big the battery was and what the load is I could set something up the same, but for now I think using the PSU profile should result in the same logging characteristics.
I’ll be running firmware version1.22 and 2.02 logging at 2 sec interval.
Scott

[npelov]

Did you measure the voltage of DMM by using the same DMM?

I did try second DMM - braymen BM869 to confirm voltage

[npelov]

The battery is 9Ah lead acid (rated at 0.45A for 20h). The load is an embedded pc with DC-DC buck converter. it started at 0.33A at 12.8V . First time it took 20 hours, second time less - I guess battery wasn't fully charged. Sample time is once a 2 secodns

[uer166]

Wow so first the A/mA jacks cracked at top, now entire Volt jack sleeve came off from next to no use of meter 

See attached..

[tautech]

Wow so first the A/mA jacks cracked at top, now entire Volt jack sleeve came off from next to no use of meter 

See attached..

Bad batch, see here:
https://www.eevblog.com/forum/testgear/eevblog-121gw-multimeter-issues/msg2792850/#msg2792850

[uer166]

Whoops apparently I suck at searching. It's surprising how consistent it is, only the GND jack is left intact! I requested jacks at the "submit issues" page, hopefully that would be enough to get Dave's attention.

[joeqsmith]

The battery is 9Ah lead acid (rated at 0.45A for 20h). The load is an embedded pc with DC-DC buck converter. it started at 0.33A at 12.8V . First time it took 20 hours, second time less - I guess battery wasn't fully charged. Sample time is once a 2 secodns

I was going to abort the collection as soon as it rolls past 50,000 samples.  It's getting very close.     I've cycled out the battery a few more times just to make sure that I have some known changes in the data.  Still there's nothing apparent with the meter's display but it sounds like yours looked normal as well.  Just the data was a problem.   

I suspect that something is happening with your setup where the load on the battery is changing that causes it to go flat, but hey, that's why I am running a constant resistive load.   If its a problem with the logging,  I should see something similar.   

[EEVblog]

Wow so first the A/mA jacks cracked at top, now entire Volt jack sleeve came off from next to no use of meter 

Email me dave AT eevblog DOT com with 121GW Connector in the title and your address and I'll ship you a new one.
Someone already posted the link to the issue.

[joeqsmith]

I collected 50,285 samples to the SD card at 2sps, using version 2.02 firmware.  The SD card is marked SanDisk Edge 8GB  8195ZTAZE030.   The meter was set to it's DCV mode with the BLE radio turned off. 

The test signal was a battery that was discharged with a resistor.   During the test I had swapped batteries a few times to create a unique signature in order to make sure that the logged logged data tracked the actual signal.   

I am pleased to say it worked perfectly.  It appears the meter's LCD matches with what was captures.     

Again, I have logged many hours of data to the SD card with the prototype and later with what ever version of production meters I now have.   This included firmware ranging from 1.0 to 2.02.    I have yet to see a problem with logging.   

If you really feel that it's you meter, I suggest you change your setup to remove that PC from the equation and repeat the test with something that is stable, like a resistor.  Then report back. 

[EEVblog]

I just logged 30,000 samples at 1sps of a 50uHz triangle wave on DCV autoranging, v2.02, supplied 8GB Sandisk card. No issues at all.

FYI UEi are adding date/time sample stamp but it will take until end December

[joeqsmith]

Thanks for repeating this test as well.  Hopefully we will see Scott's today.  His setup should be much closer to the OP's. 

[Scottjd]

Well I stopped this around 36,000 logs.

On firmware 1.22 logging interval 2, Eneloop rechargeable batteries at 4.9VDC and 8gb stock Micro SD card I was able to replicate the issue. It stated at log 660 and continued until the end at log 44,050. Over about 1.5 minutes with a span of 45 logs it would repeat logging between 19.558VDC drown to 19.519VDC. I know I went into the lab at least a dozen times to look at the meter and it was displaying the same thing as the PSU.
I would expect the older firmware to have bugs, I just wanted to point out that it was repeatable when I used version 1.22

On firmware 2.02 logging interval 2, alkaline batteries at 5.6VDC and 8gb stock Micro SD card the whole thing ran as expect from start to finish.

I think I want to run it again for a shorter period this time using firmware 2.02 and the envelop rechargeable cells just to rule this out.

On a different note:
The only thing I noticed is I had to set the range manual before I started logging. Since I start logging first then turn on the PSU (or DUT) it starts in the 5V range. So when I turn on the PSU to start at 20VDC it won't auto range to the 50V range and just log/display OFL. Maybe this has been that way from day one and I just never noticed it. I guess I expected more like the Fluke 189/289 for it to still auto range when logging.

Scott

[Scottjd]

Did you measure the voltage of DMM by using the same DMM?

I did try second DMM - braymen BM869 to confirm voltage

UPDATE:
I logged another 4K after updating the meter to 2.02 and still using the Eneloop rechargeable batteries at 4.9VDC. It ran and logged as expected without any issues. I figured if it was to fail again it would be in the first 1,000 logs like the first time.
Maybe it was a firmware issue that was fixed in the past.

npelov, has your failed since you updated to firmware 2.02?

[joeqsmith]

Hey Scott.   They apparently did try 2.02.
https://www.eevblog.com/forum/testgear/eevblog-121gw-multimeter-issues/msg2811382/#msg2811382

I'm looking over the data I had collected with 1.22 and the longest contains 2603 records, no faults.   When you ran the two tests, were you using both meters, or reprogrammed the one?

I too would have liked to have seen autorange work in logging as well as the min/max.    Autorange does work when using the wireless. 

[Scottjd]

I had both meters logging through same source. For the second shorter round I updated the one meter from 1.22 to 2.02. The shorter test again also had both meters running. Swapped the batteries, but for the next test I think I’ll make them all rechargeable batteries since that’s what I use once the alkalines die.
I haven’t found a pattern but it was weird to 37K of logs with the same range stuck higher. Almost if a process was in a loop. I didn’t use BT since he said he didn’t want to depend on wireless.

I may run another long one after I’m done rewiring the bench, the test equipment bench is almost done.

[joeqsmith]

I understand OP did not want to use BT.    It is also slower and not as deterministic.  When looking at the 121's software filters, I have to log using the SD card. 

With your repeating 45 sample pattern using 1.22, with it starting at 660 samples in, I would have thought it would show up somewhere in all the data I collected.   I wonder now if it is tied to the higher voltage.   When I ran this test, I was using 9V transistor batteries and the meter was always logging less than 10V.    So I setup another test using again, 2.02 starting at 20V. 

With OP running much lower voltages, and seeing the problem with 2.02 I still suspect their PC is fooling them.   

[npelov]

npelov, has your failed since you updated to firmware 2.02?

Yes, I did one 10+ hours test  and  the cuttoff voltage is at 10.5V, but I don't see anything below 11.6V. I did not witness the cut off this time, but when I stopped the logging the battery has recovered to 11.1-11.2V (noticed on the DMM display) and I don't see that voltage in the table. Nothing below 11.6.

I will make more tests, but right now I'm getting ready for a surgery, so I might do that after a week or two ... if I survive.

[Andrew McNamara]

Maybe it was a firmware issue that was fixed in the past.

Another possibility is their code doesn't cope with all the vagaries of SD cards - either a write error, or just a write that took longer than normal (which can happen when the SD card is soft-retrying an error, or doing housekeeping). This would be hard to replicate.

[joeqsmith]

npelov, has your failed since you updated to firmware 2.02?

Yes, I did one 10+ hours test  and  the cuttoff voltage is at 10.5V, but I don't see anything below 11.6V. I did not witness the cut off this time, but when I stopped the logging the battery has recovered to 11.1-11.2V (noticed on the DMM display) and I don't see that voltage in the table. Nothing below 11.6.

I will make more tests, but right now I'm getting ready for a surgery, so I might do that after a week or two ... if I survive.

No need.  After running the meter at a higher voltage, 2.02 firmware 50,000 cycles, after 36000 samples it did just as you and Scott claim.  Once it gets into this mode, its a repeating cycle as Scott mentions.   The only real difference between this test an my prior test is the voltage level.    Also, at 50,000 cycles, I had connected the meter to a power supply and swept it around by hand to give me some sort of marker.  Notice that the logged data remained flat as you suggested.   Also, while adjusting the supply, I was watching the LCD and there were no signs that anything had gone wrong. 

It's now clear you have discovered a problem.   I would have bet the farm it was on your side, but that's why we test....

Get well soon.

****
Added a histogram of the data after it hung (roughly 14,000 samples).   In this case, only the four levels were ever recorded.   

[Scottjd]

Glad to see Joe also was able to capture this  , now with multiple confirms of the issue my guess is the manufacture will look into the issue and hopefully resolve it. Like Joe said, that’s why we test.

The weird thing is trying to make it repeatable. I did capture another 44K logs and it didn’t show up this time. So I ran a faster more aggressive fall curve with only 1000 points and 5 seconds between each point change from 20VDC to 5VDC over a few hours capturing a little over 2000 logs. And again I didn’t see it.
I could run another 19 hour test with 50K logs captured, but this is such a long test to do and not fun to watch  It must be a very specific chain of events that happens to cause this loop.
I think having three different meters see the issue is enough for me. My power supply is needed for some other projects.

I hope all goes well and you have a fast recovery.
Scott