EEVblog 121GW Multimeter Issues

[joeqsmith]

Comparing the 121 prototype with some of my lower end meters.  I thought I would use the Kasuntest ZT-102 and UNI-T UT61E as both of these meters were specifically modified for making this measurement.    It's becoming apparent that the 121GW prototype is unique in how sensitive it is.  I would be interesting in knowing how the hardware from the kickstart behaves.     

https://youtu.be/_u_34E5ZUOI

[Candid]

I only did a quick test. But let me see... shit no tape eraser around  .

What I did was DMMCheck Plus with 1mAdc and the 121GW with my Fluke 28II in series. Then I used a strong neodymium magnet and the 121GW reacts on it (not as much as to the electrical field of your tape eraser but noticeable) and the Fluke 28II does not. I did not do any further or deep testing and did not check other DMMs.

[chronos42]

Hi,
today I also made a test with a DC current calibrator with 1mA DC and a tool / head degausser. There is a clear fluctuation in the display, the same like in Joe`s video. So the 121GW is indeed still very sensitive for alternate magnetic fields, even with the latest hardware revision. (I bought my 121GW today directly by Welectron, the company is very close to my home.)

[Candid]

So with better shielding like you made for the Gossen this should be addressable?

[IanB]

I repeated the same tests with a 1 mA DC source, and got essentially the same results, here are photos showing the poor measurement on the 5 mA DC range, but the perfect measurement on the 50 mA range, on my Amprobe AM270 meter, and also using the uCurrent Gold.

Summary: On the 5mA DC current range, the 121GW reads off, and is quite significantly affected by touching the leads. None of these effects are seen on the 50mA range, on the Amprobe meter, nor when using the uCurrent.

Can anybody replicate these findings?

I just did a quick test on my meter with firmware 1.26, using the 1 mA current source on a DMM Check device.

The BM869s indicated a stable 1000.2 µA in the 5000 µA range.

The 121GW indicated a stable 1.0012 mA in the 5 mA range, with some wavering of ±1 in the last digit.

I did not detect any influence of touching the test leads or waving my hands near the meter.

My guess would be that maybe you have a problem with your test setup or measurement procedure?

[Towger]

My guess would be that maybe you have a problem with your test setup or measurement procedure?

Or some meters/batches are more susceptible to the problem than others. 

[Candid]

Remember that Joe has a prototype that he highly modified. So everything and nothing is possible.

[IanB]

Hey IanB, please try downgrading to 1.0 and see if your low resistance range becomes stable. 

I ran the same test using the UNI-T UT181A.  This meter has a 10mohm res, so you loose a digit.  Even with that handicap, the 181A still appears slightly tighter than the 121 prototype with the 1.26 firmware installed.     

Just to verify, you are measuring a 0.5 ohm resistor, not a 0.5 meg resistor as suggested by the axis label?

[IanB]

0.5 ohms.  You are the one who brought it up, so what ever values you were trying to measure will be fine.  Just do a before and after comparison and report the results.  If the noise you were seeing is reduced, I think we have our answer.

Maybe I'll give it a try after work tonight.

However, the problems I was seeing seemed to be related to static fields, caused by proximity of my hands to the meter or test leads.

The only way I think firmware can fix that is by having lots of software filtering. So it might be a choice between "slow and steady" or "fast and erratic".

[M4x]

Joe should get a "regular" 121GW ;-)

[HKJ]

If/when they release it and we start seeing positive reviews,  I will be more than happy to put it through it's paces.   Seeing its a bit of a novelty, I may give it some special treatment like the Gossen and take it on a road trip.  That's big investment on my part so I want to make sure its stable first.

What do you mean release it? Dave sells it in his shop (It is out of stock) and and welectron sell it in his (And it is in stock).

[IanB]

Hey IanB, please try downgrading to 1.0 and see if your low resistance range becomes stable.

Here are the results of a test on a 0.5 ohm resistor sampled on a human time scale. On looking at this data I think sampling over a much longer period might also be interesting, but I don't think this is a common use case for a handheld meter. Normally you would want to apply the probes, take a reading in a few seconds, and move on to the next measurement. For such a use case I don't know if the v1.00 firmware has significant advantages over the v1.26 firmware. Even though v1.00 appears to have stronger high frequency filtering, the overall measurement uncertainty doesn't seem to be any different.



Edit: Fixed incorrect firmware versions

[IanB]

Where did you find firmware version 1.7?  What is the sample rate you are using?   Are you collecting to the SD card or using BT?   

I would like to try and repeat your 90 second test.

I'm sorry, that was a brain fart resulting from working at 3am when I couldn't sleep. It was v1.00 vs v1.26 of course. I'll fix my post.

I am collecting using the SD card at the maximum sample rate (0 seconds in the setup menu). There is no time stamp included with the data values, so I am assuming 0.2 sec per sample as per the manual. It seems about right.

I tried to use Bluetooth, but the v1.00 firmware doesn't appear to work with the app. The maximum sample rate with BT is also limited to 0.5 sec.

[IanB]

Can you do related test for me on your hardware?

While measuring the 0.5 ohm resistance, can you repeatedly reach out and lightly touch the Hold button with your finger and then pull it away again? Don't press the button, just touch it. Repeat the touch/withdraw cycle a few times. What happens to the reading when you do that? On my meter I can make the reading vary between about 0.2 ohms and 0.9 ohms.

[IanB]

So the question now is if that 0.01 you are seeing is good enough.  It sounds like you are expecting much better results.  For some reason, I was thinking your meter had a lot more variance when you first posted about it.  IMO, on a 2 wire handheld, I wouldn't have expected it to be much better than this.   In the second attachment, I compare it with the data I collected at 1 second intervals with the other meters.  The yellow is the UNI-T UT181A.   0.01ohm res and 4 counts, its a pretty big swing.

Yes, I'm expecting much better results than I'm seeing. In my next post I'll show some other results to support my contention.

[IanB]

Here's an interesting data set. I used the 121GW in two ways. On the green plot I used the 121GW to measure mV drop across the resistor when subjected to a 1 mA constant current source. On the blue plot I used the 121GW to measure the resistance directly after using the Rel button to null out the leads. Since I can't physically touch the Rel button without disturbing the reading I activated it remotely using the Bluetooth app. I also collected the data using Bluetooth since (a) it is too much trouble to reach in and retrieve the SD card and (b) when I tried to press the Mem button to start logging it appeared to disable the Rel mode.

Observations:

1. The mV reading is mostly stable to ±1 count. This for me is the ideal situation.

2. The resistance reading climbed dramatically over the test interval and was not stable. I was surprised by this so I repeated the test again, with the same outcome. The reading started around 0.5 and climbed to over 0.53.

3. At about 100 s on the green trace there are some glitches with sharp spikes in the reading followed by a decay. I did not move or touch any part of the experiment at this time, so I attribute the glitches to some environmental influence beyond my control.

[joeqsmith]

Look at the drift in my post #583.   Once the meter has had a chance to warmup it does alright. 

I would expect the voltage would be better with that new reference in there.  That was one of the things they changed in the kickstart hardware from the prototype I have.   Dr Frank had made a few comments about the poor choice in components used in the divider network.  The short term drift as I have mentioned a few times now, I believe is the Hycon device.  Maybe UEI could chime in on this.   

I'm not expecting any equipment to come up from a cold start and instantly provide optimum results and I'm not too concerned with the drift yet.  I will leave that to Dr Frank. 

Can you do related test for me on your hardware?

While measuring the 0.5 ohm resistance, can you repeatedly reach out and lightly touch the Hold button with your finger and then pull it away again? Don't press the button, just touch it. Repeat the touch/withdraw cycle a few times. What happens to the reading when you do that? On my meter I can make the reading vary between about 0.2 ohms and 0.9 ohms.

When I tested the Gossen, I had it in a screen room, me grounded with a shorting bar or a 50 ohm terminator installed in the meter.  That meter was really bad.  The worst I have ever seen.    The 121 proto is no where near that bad but it's sort of a subjective test.   Are you rolling around on carpeted floor with your chair building up a charge?  I really have no idea what you are doing other than touching the button. 

The best thing I can do is try and show you the prototype 121 with some sort of controlled test.   I will use the 1.0 firmware as with the prototype, it seems to at least give the same noise levels as your hardware.  You do not specifically mention logging but only measuring so I will not enable logging.  The BT radio on this meter does not appear to be enabled so consider that off.  The meter is located on a grounded static mat.  The 0.5 ohm resistor is directly inserted into the meter without test leads.  I will have a wrist strap on.  For me, this would be the normal way I would be using a meter anyway.   I have no other tests running.  There's no high voltage supplies turned on.  Should be a fairly stable environment right now.   

Watch for yourself and let me know if this differs from what you are seeing and how you are conducting your test.

********
Also, I should add that while this is a prototype and I have damaged the meter and rebuilt it three times now and I have changed the design to harden it, I have NOT added any shielding to this meter.  I've changed out one IC and added a couple of TVSs.  I have a few jumpers on the board to route the return paths for the TVSs.  More people would need to run the test with the kickstart hardware but it seems like many may have left them in their original boxes hoping to retire when they sell them in 50 years on eBay.   

https://youtu.be/ikROglTyclM 

 

[IanB]

OK, for you it appears that touching the Hold button has no effect. But it always does for me, and at least one other person has reported the same phenomenon. I will have to look more closely and try to eliminate any environmental factors that may be at play.

Regarding the drift, I didn't see such drift when I did the previous 90 second test with the SD card. So I was surprised to see it today. I didn't like it, so I discarded the data. Then I ran the mV test that gave the flat line, and after that I came back to the resistance test with the same outcome. I think the meter was warmed up by the time I did the last test. One difference is that BT was on when I saw the drift and was off before.

All that is left is to try to eliminate causes and see if I can narrow anything down.

[Dr. Frank]

Why do you (Ian, Joe and Dave) perform these tests on the low end of the range?

If you do it @ around 50 Ohm, this drift might behave much different.

Frank

[IanB]

Why do you (Ian, Joe and Dave) perform these tests on the low end of the range?

I think that if the meter has a feature, then that feature should have a use and should be usable.

For example, milliohm resolution might be useful when troubleshooting possible bad or high resistance connections in wiring, or bad traces on a PCB, or similar things.

If the stability is so poor that the information you are looking for is concealed by noise and drift, then what is the point of designing it into the meter?

I think if a feature is there it is fair game to test it.

[Dr. Frank]

I did not understand Daves video @ 0.5 Ohm, neither. He also did not respond to my question about his experiment.
You can see an increased jitter at 50 Ohm, compared to the upper ranges, but for sure you're not in the residual mOhm specification, but instead in the % specification..so everything is fine..

What I have seen, and not understood, why it's quite tricky to make a REL (zero) setting.. even if the residual reading is quite stable, that will jitter around, and give some +/- 10 mOhm offsets after pressing REL by your hand.. Ian used the mobile app, which would be better.. can't understand that behavior at all.

And that really sucks, because a 1 mOhm resolution is really great for some applications, compared to other DMMs, but it has to be stable, and the zero has to work smoothly.

Frank

[IanB]

Here is a quick illustration of the problem I observe with the Hold button. The resistor under test is connected to a grounded lead and I am also holding a grounded lead during the test, so both meter and I should be at equal potential. I don't know the cause of the effect, but it seems very specifically tied to the Hold button. Touching the Rel button has much less effect, but actually pressing the Rel button does disturb the reading (not shown in this video).

https://youtu.be/p2zuhvFwmIs

[Brumby]

The resistor under test is connected to a grounded lead and I am also holding a grounded lead during the test, so both meter and I should be at equal potential.

My immediate reaction was one of a possible ground loop.  I would be interested to see what effects are observed with a different lay of the test leads.

My second thoughts were of the exact same movement of your hand in every test.  I would have liked to have seen what happens when approached from different directions - and also what happens if you just 'hover' above the button (3" or so) between attempts rather than move your whole hand out of shot and back in.  While simply repeating the same movement does demonstrate the issue, testing other movements may help characterise the forces at play.


It's an interesting phenomenon, though - and if it is a susceptibility in special circumstances that don't exist in most situations, then it's worth knowing about for that odd occasion.

[joeqsmith]

You have the radio on, which I did not and I'm on a grounded mat.  I would start by turning off the radio.  Would be funny if the BT was the cause.   

[IanB]

I just wanted to show what the effect looks like so you can see what I have been describing. I don't have an anti-static mat, however in this test the whole meter was tied to ground at the COM terminal. BT was on simply to show the response more clearly.

To be clear, this effect happens with BT on or off, with the meter in different rooms and orientations, when I touch the Hold button from any direction, with different lengths and orientations of test leads, and with various different ranges (for example temperature). The test setup in the video was in no way unique or special.