Handheld meter robustness testing

[joeqsmith]

One option would be to DL the picture, then reupload under your account.  Or the simple option, don't link it at all.  After all it was only one previous post. 

I fixed it.

PS: The joke was that I know perfectly well what calculator you use, I've seen it in your videos. It's easy to remember because it's the same one I use. Thanks for derailing that.

Thanks.  Want to tell a joke, finish this one:

Two 121GWs walked into a bar and sat next to a Brymen and a Fluke. 

[rsjsouza]

Two 121GWs walked into a bar and sat next to a Brymen and a Fluke.

...one of them says: can you believe that Keysight dude? It can't take a few rounds without falling apart!

[Zbig]

So could they really be doing no cold junction compensation at all? The Temp mode really appears to just read the DCmV and translate this to temperature using a LUT, though a LUT that gives much more resolution than the one for the NTC sensor.

The cold junction compensation done on multimeters is to add the ambient temperature to the thermocoupler temperature and the ambient is often measured inside the multimeter chip, i.e. a long way from the correct location. This also means a multimeter may need one to two hours at a specific temperature before the internal sensor matches the actual cold junction.

Just to rephrase HKJ's response, there is little point speculating whether a thermocouple-based temperature measurement function in a multimeter uses cold junction compensation or not. As long as it doesn't show "0" every time the probe is in the same environment (same temperature) as the meter itself, then of course it does. In this scenario, both ends of the thermocouple (that is the "tip" and the whole length of wire) being the same temperature, i.e. with zero temperature gradient, the probe produces a voltage of zero. So there really is no lazy, cut-corners way of implementing a thermocouple-based temperature measurement with the cold-junction compensation omitted altogether. Of course it doesn't stop you from implementing it improperly.

[joeqsmith]

Destructive testing and final thoughts.  There's a joke in there somewhere.....

[CDaniel]

   Good job , so those 2 small SMD transistors for resistance protection are far worse than 2 x 1N4007 ... without further modifications . When you look at the schematic in series is just the thermistor ( 1K2) + 1K resistor + 100ohm SMD resistor the one that probably burned . The thermistor has a time lag to heat up and increase its resistance .

So the switch for how many cycles do you think will be good ?

[Fungus]

There's a joke in there somewhere.....

I watched it all but I didn't hear any jokes. Anybody...?

[Dr. Frank]

Provided that Dave really has replaced these two 1N4007 by the reversed transistor diodes pair, (aka  FLUKE protection), which he had promoted in two of his former videos, my prediction that this was useless, and the wrong approach, obviously has come true, as the Ohm range was far off after the overload experiments by Joe.

I'm no real friend of Joes further brutal destructive tests, as they seem to me being exaggerated and not conforming to the IEC et.al. test standards.

But anyhow, the 121GW really has big problems not to detect AC+DC signal overloads.
This has been discovered already when AC+DC current overloads were investigated in another thread.

So this is really a safety issue at high voltages, which should be solved by UEI.
Hopefully this can be done in SW, but I fear that this might be an intrinsic problem of the circuit, or of the chip set, precisely, of the way how AC signals were measured.

Frank

[joeqsmith]

There's a joke in there somewhere.....

I watched it all but I didn't hear any jokes. Anybody...?

I doubt anyone can top the 121 for a joke. 

  Good job , so those 2 small SMD transistors for resistance protection are far worse than 2 x 1N4007 ... without further modifications . When you look at the schematic in series is just the thermistor ( 1K2) + 1K resistor + 100ohm SMD resistor the one that probably burned . The thermistor has a time lag to heat up and increase its resistance .

So the switch for how many cycles do you think will be good ?

 

I am looking at the meter today.  The contacts that were damaged were #1 & #7 which is for the two PTCs used for LZ mode.  That bank was not selected and the wimpy little transient generator jumped it.   I'm fine with suggesting that I didn't get all the copper debris cleaned up as an excuse.   Had that not jumped, all the energy would have gone into that clamp..     R82 (100 ohm) is open.    U9 (mux) is gone.

It appears the cobbled on mezzanine clamp circuit is still fine.  Both transistors check.  Rather than the clamp saving U9, it appears U9 was what saved that clamp.   

Just looking over the board for other signs of damage, I came across this gem.   It appears C67 is a nopop, so having pin 3 of U13 lifted doesn't hurt anything.   First, I am surprised to see such skills.  Then, I'm surprised that no one caught it before it left the factory.   

A UEi and a UNI-T went on a date.  9 months later, they had a baby CEM.   

[dcac]

So should we make a guess that next hardware revision will have R82 (100 ohms) with a higher voltage rating, or perhaps higher resistance too?

And I seem to remember long long time ago Joe already predicted R82 will probably fail and it seems UEi/Dave did not get that message then, but perhaps now? - we can at least hope.

[joeqsmith]

Provided that Dave really has replaced these two 1N4007 by the reversed transistor diodes pair, (aka  FLUKE protection), which he had promoted in two of his former videos, my prediction that this was useless, and the wrong approach, obviously has come true, as the Ohm range was far off after the overload experiments by Joe.

I'm no real friend of Joes further brutal destructive tests, as they seem to me being exaggerated and not conforming to the IEC et.al. test standards.

But anyhow, the 121GW really has big problems not to detect AC+DC signal overloads.
This has been discovered already when AC+DC current overloads were investigated in another thread.

So this is really a safety issue at high voltages, which should be solved by UEI.
Hopefully this can be done in SW, but I fear that this might be an intrinsic problem of the circuit, or of the chip set, precisely, of the way how AC signals were measured.

Frank

Hey Frank.  I shutter to think what would happen if I ran the actual surge test and that switch broke over like that.   

I was going to do some experiments with the new changes but I thought, I've already gone down that path with the prototype and here we are 2 years later....    Another problem is the video was a couple of weeks packed into an hour.    Then there is the fact that it was painfully obvious that it would not save the mux.  I had commented to a few people that I was concerned that the simple grill ignitor would take it out.  Like the shim, it just wasn't thought out.   

So should we make a guess that next hardware revision will have R82 (100 ohms) with a higher voltage rating, or perhaps higher resistance too?

And I seem to remember long long time ago Joe already predicted R82 will probably fail and it seems UEi/Dave did not get that message then, but perhaps now? - we can at least hope.

Not suggesting this is the solution but I clamped on the other side of R82.  I suspect I could have removed my TVS and both 400x and it would be fine having R82 limit the current to the mux but I never tried it.  It's not my job to design it.  I just wanted to find the next weak link in the chain.   


That 121GW has some mighty big balls,   solder balls that is....   

[CDaniel]

Something was dammaged ( leaky ) from the grill starter ... those 2 transistors were the perfect candidats  Maybe should be investigated more .

[joeqsmith]

Something was dammaged ( leaky ) from the grill starter ... those 2 transistors were the perfect candidats  Maybe should be investigated more .

You and I are thinking along the same path.  The two switch pads had  become hot enough to breakdown the adhesive.  I ground out the bad area and filled with epoxy.  I had an extra mux from my experiments with the prototype so I have replaced the damaged one.   I also replaced R82.  I then cleaned up some of the poor soldering and removed all of the clamp circuits D13, mez, Q5, MOV.  Checked all the clamps and they don't appear to have been the cause but I left them out.   After cleaning it,  I powered up the meter.    The frequency now works.  Looking at the resistance and capacitance, they are both still low.  It seems back to where it was when the grill starter had damaged it.    I doubt U16 and am suspecting the Hycon took the hit.  Not a big surprise.....   

The meter is in pretty rough shape after running the life test.  I've already spent more time with my failure analysis than I had planned.  The meter is good enough now that if we come up with some other destructive test, it should be usable. 

**************
Removing U16, the 10Meg range jumped.  A bit unexpected.  I now suspect this was from the heat.  Replaced the part, cleaned and dried.   Right back to where we started.   There is not much left and getting these controller ICs seems to be a bit of a problem.   

[CDaniel]

I think you did the thermal test  before with the prototipe , was the same ( bad ) ?

[joeqsmith]

I think you did the thermal test  before with the prototipe , was the same ( bad ) ?

Dave was quick to point out it had a different reference.  Yes, it was a high drifter compared with the other meters I had looked at.  That UT181A was the most stable out of the group (which is why I show it).   The UT61E was also really bad.  Not sure which was worse, it or the CEM.  After I took a crack at compensating the UT61E for a few pennies and it's now one of the more stable meters I have.   

I was surprised to see the 121GW moving as much as it did.  Someone asked about the humidity which in the video, the other two meters were ran during the humid summer months.   I really dropped the temperature of the Brymen BM869s.   Then I ramped the temperature as fast as I could.   You may have seen me open the box on the UT181A to help speed things up.  Those two meters, even with that level of abuse were fairly stable at both ends.   We are nearing winter and the house is now very dry.  Static is starting to be a problem again.   I only took the 121 to -10 and ramped it in about an hour.   Fairly slow.       

https://youtu.be/mzBDgxkO3HE?t=239

*** 
Correct time stamp.

[CDaniel]

OK , I asked if the prototipe was the same in the tests you did back then , if you did . For voltage .

[joeqsmith]

I am not sure what you are asking then. 

I think you did the thermal test  before with the prototipe , was the same ( bad ) ?

"was the same (bad)"  Are you asking how much drift the prototype had? 

OK , I asked if the prototipe was the same in the tests you did back then , if you did . For voltage .

I did run the prototype back then.  Yes, I ran it with a constant voltage.   

It almost comes across that you are asking if the meter shown being temperature cycled in this video is the prototype meter.  If so, no it was not.  It was one of the two meters that Dave had sold me.   (actually, it was the one marked Production 1)    I ran all of the non-destructive tests on Production 1 in order to preserve it for future tests.   

[dcac]

About the grill starter test, when you tested the prototype not only did it survive but the test never seemed to reset the meter either, so that clamping design was better or more robust if you like. But here with the production meter it went into reset several times as if the voltage spike found a way in even to affect the MCU voltage rails.

[CDaniel]

I am not sure what you are asking then. 

I ask if the prototipe is worse , better , or the same for voltage temperature stability .

[joeqsmith]

About the grill starter test, when you tested the prototype not only did it survive but the test never seemed to reset the meter either, so that clamping design was better or more robust if you like. But here with the production meter it went into reset several times as if the voltage spike found a way in even to affect the MCU voltage rails.

Keep in mind that the prototype also had a different PCB and other changes.  I can't say for certain on what change/s caused the meter to less robust than the prototype.   And mind you, I am certainly NOT suggesting that the prototype was the holy grail.  That meter wasn't even close to what I would consider being electrically robust. 

For the path of the ESD, it does appear that it may have involved the Hycon IC.  Too bad really.   It appears the mVDC and temperature functions are also dead.  mVAC is working.  I can trace the signals to the Hycon part and it seems fine.   

Beating on the prototype with the stupid grill starter.  That test started out as a bit of a joke.   
https://youtu.be/X28bwdTBW8g?list=PLZSS2ajxhiQDBDdtQNjVnGxShaVQ3nUMY&t=1013

I am not sure what you are asking then. 

I ask if the prototipe is worse , better , or the same for voltage temperature stability .

   Now I understand.  Sometimes a hammer works best.    The ones I purchased appear to be much better.   

[Fungus]

There's a joke in there somewhere.....

I watched it all but I didn't hear any jokes. Anybody...?

I doubt anyone can top the 121 for a joke. 

I'm more glad then ever I didn't get one. Between the firmware problems, the shims to get the selector switch to make contact with the PCB, the SD card madness, and now this... it's a disaster.

It even seems like the CAT rating could be bogus.

[rsjsouza]

There's a joke in there somewhere.....

I watched it all but I didn't hear any jokes. Anybody...?

I doubt anyone can top the 121 for a joke. 

I'm more glad then ever I didn't get one. Between the firmware problems, the shims to get the selector switch to make contact with the PCB, the SD card madness, and now this... it's a disaster.

It even seems like the CAT rating could be bogus.

Fungus, just keep in mind the CAT requires the operator survivability, not the meter. Joeqsmith's tests validate the electrical robustness. Although I suspect there is a correlation between the two.

I do the same in my videos but with much less voltage (~60Vac but on several seconds) just to show a bit of what to expect if the ranges are incorrectly selected.

[joeqsmith]

There's a joke in there somewhere.....

I watched it all but I didn't hear any jokes. Anybody...?

I doubt anyone can top the 121 for a joke. 

I'm more glad then ever I didn't get one. Between the firmware problems, the shims to get the selector switch to make contact with the PCB, the SD card madness, and now this... it's a disaster.

It even seems like the CAT rating could be bogus.

Fungus, just keep in mind the CAT requires the operator survivability, not the meter. Joeqsmith's tests validate the electrical robustness. Although I suspect there is a correlation between the two.

I do the same in my videos but with much less voltage (~60Vac but on several seconds) just to show a bit of what to expect if the ranges are incorrectly selected.

To this day, I still can't tell you the correct interpretation of the safety standard.   Maybe this is why Fluke had told me outright that they designed their meters to survive.  Maybe they error to the side of making damn sure they don't injure a worker by making sure the meter survives?? Too bad the ex Fluke employee doesn't jump in.   

I don't hold a lot of stock now in these standards and seeing the 121 unable to show voltages didn't boost my confidence.   I was really surprised to see that switch contact arc over as well.  Then again, seeing that Gossen with it's latching relays and no magnetic shield...  why should I be surprised. 

You and Frank are both right in that I don't test to the IEC standards.  Looking at the meters from a safety point of view has never been a goal. 

[dcac]

That temperature drift from -10 to 60C is very concerning, and what the hell is causing it.

As I've mentioned earlier 121gw specified operation temp is 0 to 50C according to the manual. I interpret this to mean within this range it should be reasonable within its specified accuracy, in all modes/ranges(?).

Dave did his own thermal test but at a 4.5000V reference, And this was with the new reference chip for the 121gw, I believe. And here drift did not really seem to be a problem:



But in Joe’s test with a 1.000mV reference I saw a reading of 0.870mV at 0C (according to 121gw internal temp sensor) and again 0.870mV at 20.2C and then 0.909mV at 50C. Way of the specified accuracy for DCmV ±0.1% + 10 counts.

But then the meter, after being allowed to stabilize 2h in room temperature 20.2C is now showing 0.997mV.

Strangely the only time I saw 0.997mV in the thermal chamber was when internal temp was showing -18C.

I realize just because the 121gw internal sensor is showing a certain temperature doesn’t mean the rest of the meters internals is at that same temp (yet). But 0.870mV at 20.2C in the thermal chamber seems very far from 0.997mV at “room temperature” 20.2C.

I just don't get it.

[malagas_on_fire]

The uni-t 139C is the only uni-t that have survived the minimal energy ESD  test ? Seems the ut139 series ( A/B/C/D/E/S) is more robust, but this is a line of meters behind the 121GW and this got into ESD low energy failed.. A nearby taser such as the electrobooomm video would put that meter down if you remember ...

 Not to curse the 121GW, but there is a lot of room for improvement....

The uni-t has another new series of meters called pro.... UT195 :

https://www.uni-trend.com/html/product/General_Meters/Digital_Multimeters/UT195/

They have dual dispay but 6000 counts only... wonder iff they are built to last has they anounced ....   the ringing generator , the ESD, and transients




PS.: ohh was forgeting the clamp meter ut210E, the nice and compact one.

[joeqsmith]

That temperature drift from -10 to 60C is very concerning, and what the hell is causing it.
...
I just don't get it.

Different parts inside the meter are changing at different rates when I ramp the box, even at this slow rate.  With this meter, it causes a fair bit of error.      I allow it to sit at the two extremes for a half hour to stabilize.  There's not a lot of room in that box and there is a decent sized fan.  At -10 we had 1.003mV, 20ish 0.997mV and 60C 0.928mV.   

But if you leave it in your car overnight in the dead of winter (much colder here than -10C) and bring it inside to use it, expect there to be a fair bit of error until it settles.  Normally we would test something like this with a shock chamber but in my home lab, you have to settle for my cardboard box.