Handheld meter robustness testing

[joeqsmith]

I waited two years and you sent me old stock.  That's on you. 

Fine. But like I said in my email if you had asked first I would have told you to hold off for a couple of weeks and I could have sent the new build that was on the way.
I do not usually pack and ship, so I had no idea you ordered these meters. If I had noticed I would have known what you had wanted them for and would have contacted you.
I can't help bad luck timing and lack of communication.

If I was looking for preferential treatment, I would have contacted you.  To be clear, I wasn't.  Buying the meters, dealing with service,.... it's all part of the process.   I want the same hardware that anyone else placing an order is going to get.   If you are concerned about selling old hardware, don't sell old hardware.   If you are concerned that there is a problem with people placing orders not understanding what it is you are selling, then explain it on your ad.  Don't expect customers to reach out to you to try and sort it out. 

Again, if you feel running the latest hardware would help viewers in some way,  I would be willing to run it as part of these tests.   It's up to you.   

[joeqsmith]

Looks like she was drifting down.  Don't include the settling time.    Even more funny was seeing that slow Keysight lock right in and hold steady.   

Err, the Keysight displayed 9.7ohms at one point.
As for "drift" in the 1mOhm LSD, it's a switch, just my finger pressure can change by that amount. I checked, the Keysight "drifts" in a similar way, you just didn't happen to see it in the video.

The only one moving was the 121 from what I saw.  If the others drift the same and you didn't include it in the video, that's easy to fix.  We are all about the data.   Your finger was on that button a long time with the Keysight showing out 3 places and that thing was solid.   I was thinking they may do some sort of filter change, like Brymen.

[rsjsouza]

Looks like she was drifting down.  Don't include the settling time.    Even more funny was seeing that slow Keysight lock right in and hold steady.   

Err, the Keysight displayed 9.7ohms at one point.
As for "drift" in the 1mOhm LSD, it's a switch, just my finger pressure can change by that amount. I checked, the Keysight "drifts" in a similar way, you just didn't happen to see it in the video.

Thanks for sharing the video; I can attest the Keysights drift some as well and I even commited that sin when I did my own video about the subject, although the U1273A was still using Keysight's evil smooth filter mode that crapped out everything on it (it became better after disabling it).

[gnavigator1007]

There was not supposed to be any confusion surrounding the prototype

There's still confusion about this meter. See people regularly thinking it's open source. Remember seeing contradictory instructions on which way to install the shim. Various things like this. From what I've seen, joeqsmith has made plenty of effort to make clear that the original was a prototype meter. Suspect more of the confusion has come from people watching videos out of order, reading parts of old forum discussion, and maybe previous episodes of the amp hour. There's various info scattered across so many places. Now with these board revisions I'd expect even more confusion to come. While I'd love to see joeqsmith run all his tests on the meters he currently has as well as the very latest, I'd also hope that you'll do a video explaining the latest changes and why they were made.

[malagas_on_fire]

Looks like she was drifting down.  Don't include the settling time.    Even more funny was seeing that slow Keysight lock right in and hold steady.   

Err, the Keysight displayed 9.7ohms at one point.
As for "drift" in the 1mOhm LSD, it's a switch, just my finger pressure can change by that amount. I checked, the Keysight "drifts" in a similar way, you just didn't happen to see it in the video.

Thanks for sharing the video; I can attest the Keysights drift some as well and I even commited that sin when I did my own video about the subject, although the U1273A was still using Keysight's evil smooth filter mode that crapped out everything on it (it became better after disabling it).

Thanks for the testing video.
The U1282A looks to be the slowest but gets to the value , minus one digit, maybe can confirm the same behaviour for that meter. Looks like the ut136b is fast and accurate but doesn't have much digits to process. Would go for the slow , but hi resolution and accurate than fast and inacurate.

[joeqsmith]

Going over my notes for the prototype to sort out what tests I plan to run.   One of the problems I came across was with the VA mode.  Roger, who ran the YT channel  KainkaLabs, repeated my tests.   He did a really nice job explaining the problem and how to work around it.   I doubt I will go over this problem again, so if you're interested, I suggest you watch his video.

[joeqsmith]

In part 3 we compare the old production hardware with some of the other surviving meters.   We attempt to measure a small capacitor, low voltages and currents.   Because I know Dr Frank like those really small signals, I drove it into the muck.   The tape eraser is brought out to see if the production meters are less sensitive to it.  Finally we finish up with some higher voltage tests and install the latest firmware, because I heard that it fixes everything.     



There's a lot of testing to go so stay tuned for part 4.

[dcac]

Thanks for the video Joe, interesting as always, and thank you for continuing testing with your "new" 121gw's.

And just a note, don' know if you mentioned this but your 1nF reference cap which you measured to 992pF with the BK Precision, the BNC adapter you then used probably added 8-10pF in parallel, so reference cap the other meters saw was likely very close to exactly 1nF. Just to show how accurate some of them really were.

[dcac]

About the 121gw capacitance drifting, my (uneducated) guess is that it might have something to do with the SM6T22CA (D13 on the original 121 schematics), these devices seems to have a lot of nonlinear, voltage and probably temperature dependent capacitance behavior.

Perhaps a reason D13 now seems to been replaced with a traditional transistor clamp circuit in the latest hardware revision.

[joeqsmith]

Thanks for the video Joe, interesting as always, and thank you for continuing testing with your "new" 121gw's.

And just a note, don' know if you mentioned this but your 1nF reference cap which you measured to 992pF with the BK Precision, the BNC adapter you then used probably added 8-10pF in parallel, so reference cap the other meters saw was likely very close to exactly 1nF. Just to show how accurate some of them really were.

You are correct.   FYI, all three Flukes measure 9pf for the jig by itself.   Added to the 992 measured with the BK,  1001pf read by the Flukes.   Funny how that works out.       

Normally, for smaller values I will attach them to a test board and use the VNA which provides more detail.  The test fixture is the same PCBs used for the standards used to calibrate the VNA.   

About the 121gw capacitance drifting, my (uneducated) guess is that it might have something to do with the SM6T22CA (D13 on the original 121 schematics), these devices seems to have a lot of nonlinear, voltage and probably temperature dependent capacitance behavior.

Perhaps a reason D13 now seems to been replaced with a traditional transistor clamp circuit in the latest hardware revision.

It's very possible they have corrected it with the latest revision.   Someone would need to check it. 

I was surprised how they coerce the reading to zero if the meter is powered on with no rel but if you are fast enough to rel it out, it will start to read negative values.   Rel it after it have gone negative and it will again coerce to zero, no mater how far the meter continues to drift.  I call this a bug.

[CDaniel]

I don't think it's D13 , that input is used for some functions like mV where the drift should be very noticeable .

[dcac]

I was surprised how they coerce the reading to zero if the meter is powered on with no rel but if you are fast enough to rel it out, it will start to read negative values.   Rel it after it have gone negative and it will again coerce to zero, no mater how far the meter continues to drift.  I call this a bug.

Yeah as I earlier wrote in the other thread if they only allowed for negative capacitance values this drift issue could be handled with the rel button. At least if you are quick enough to do the measurement and remember to recheck if it drifted further before the next measurement.

So what! if it then perhaps initially will read something like negative 50pF or whatever with no probes attached, if that is what they are concerned about...
 

[joeqsmith]

So what! if it then perhaps initially will read something like negative 50pF or whatever with no probes attached, if that is what they are concerned about...

Guessing they felt the negative capacitance would cause a bit of confusion and decided to coerce it to zero.  Lots of ways to skin this problem but what they have now isn't at all what I would have expected from a company who designs and produces DMMs. 

[IanB]

Guessing they felt the negative capacitance would cause a bit of confusion and decided to coerce it to zero.  Lots of ways to skin this problem but what they have now isn't at all what I would have expected from a company who designs and produces DMMs.

Do we know for sure that UEI "designs and produces DMMs"? For instance, do we know of other meters they make apart from the 121GW? I have the feeling their concerns lie elsewhere and the 121GW has been a bit of a departure from their normal product lines.

[CDaniel]

The drift is not acceptable ... even for a chinese multimeter .
Positive offset is "normal" if you look at a Fluke ~60pF , but it would be possible for 121GW's microcontroller to zero this at startup - of course when the drift issue is resolved .

[joeqsmith]

Guessing they felt the negative capacitance would cause a bit of confusion and decided to coerce it to zero.  Lots of ways to skin this problem but what they have now isn't at all what I would have expected from a company who designs and produces DMMs.

Do we know for sure that UEI "designs and produces DMMs"? For instance, do we know of other meters they make apart from the 121GW? I have the feeling their concerns lie elsewhere and the 121GW has been a bit of a departure from their normal product lines.

These would be good questions for Dave.   He did post that video on the history of the 121 but I haven't watched it.   You may find some of your answers there. 

There was an early discussion about UEI getting out of the DMM business.    With the data spread out, you may want to just try the advanced search and see what you can find.   

The drift is not acceptable ... even for a chinese multimeter .
Positive offset is "normal" if you look at a Fluke ~60pF , but it would be possible for 121GW's microcontroller to zero this at startup - of course when the drift issue is resolved .

First, to avoid any confusion, the attached data is taken from the prototype.  This is AFTER I had installed more recent firmware and had the BLE interface working.   This meter had been aligned by myself and is in no way in current calibration.   The meter had also been damaged a couple of times and I had implemented the changes I mentioned earlier with the clamps, different Mux, etc.   The prototype uses a different reference.  There may be other differences as well.

These graphs show the 121's internal temperature (obviously no longer correct) and measuring a 2nF capacitor over an 8 hour period.   What is interesting about this data is that we saw in the last segment how the old production meter drifted about 10pF in maybe 10 minutes.     At least with the prototype the drift appears to correlate with the reported temperature.  I would have expected old production hardware to also show an increase in capacitance, rather than the decrease.  But it does appear the two have similar slopes, just opposite in magnitude.   

At some point, I will toss the 121 into that meat packing box and we can get some idea how it behaves over temperature.   

[CDaniel]

I didn't said the drift is temperature related , in fact is highly unlikely because when you cycle through functions and back to cap , usually is reseted .

[joeqsmith]

I didn't said the drift is temperature related , in fact is highly unlikely because when you cycle through functions and back to cap , usually is reseted .

No one suggested you did.  I am only presenting you with the data I collected off the prototype.  The prototype was one of the worst meters I had looked at for temperature drift.  It will be interesting to see this old production hardware compares.   

[rsjsouza]

Do we know for sure that UEI "designs and produces DMMs"? For instance, do we know of other meters they make apart from the 121GW? I have the feeling their concerns lie elsewhere and the 121GW has been a bit of a departure from their normal product lines.

These would be good questions for Dave.   He did post that video on the history of the 121 but I haven't watched it.   You may find some of your answers there. 

There was an early discussion about UEI getting out of the DMM business.    With the data spread out, you may want to just try the advanced search and see what you can find.   

I've known UEI as a DMM manufacturer for quite some time. By the time the 121GW came about, they really had no meters on their lineup (all discontinued). They have now four meters, alongside a number of clamps and other measuring devices.

(...) measuring a 2nF capacitor over an 8 hour period.  

Unless the tests are done by measuring a capacitor for 10s at every minute, which reflects a more practical scenario, what is the point of this? Capacitance testing on a DMM is useful only for a quick check, especially around its lower edge.

At some point, I will toss the 121 into that meat packing box and we can get some idea how it behaves over temperature.

That is a much nicer test, especially since autumn is upon us!

[joeqsmith]

(...) measuring a 2nF capacitor over an 8 hour period.  

Unless the tests are done by measuring a capacitor for 10s at every minute, which reflects a more practical scenario, what is the point of this? Capacitance testing on a DMM is useful only for a quick check, especially around its lower edge.

At some point, I will toss the 121 into that meat packing box and we can get some idea how it behaves over temperature.

That is a much nicer test, especially since autumn is upon us!

That's been a long time ago but I suspect what drove it was me seeing that 121 proto moving all over and wanting to know if it would settle.   

After watching Dave's video,  I plan to have a look at the resistance stability in the 50 ohm range.   I put together a poor mans reference using a Vishay 0.005% 1ppm S-Series foil resistor.   I'm surprised the numbers the old 34401A throws up after an hour warm up are this close.  It's been a good meter. 

I plan to repeat the stopwatch test as well. 

[bitseeker]

Going over my notes for the prototype to sort out what tests I plan to run.   One of the problems I came across was with the VA mode.  Roger, who ran the YT channel  KainkaLabs, repeated my tests.   He did a really nice job explaining the problem and how to work around it.   I doubt I will go over this problem again, so if you're interested, I suggest you watch his video.

That was a slick fix, Joe.

[joeqsmith]

Going over my notes for the prototype to sort out what tests I plan to run.   One of the problems I came across was with the VA mode.  Roger, who ran the YT channel  KainkaLabs, repeated my tests.   He did a really nice job explaining the problem and how to work around it.   I doubt I will go over this problem again, so if you're interested, I suggest you watch his video.

That was a slick fix, Joe.

When I demo'ed that back in 2017, I caught some flack over it.   I was glad to see another person finally post a video on the VA mode and was even more pleased when he repeated my test.   I haven't downloaded the latest manual yet to see if they ever did anything with it.  You know the old saying  "you can't push a rope"? 

I made an attempt to repeat Dave's settling time test using roughly the same resistance value.   I've ran this same test before with the prototype and even older firmware.    The left 121GW old production hardware is using firmware 1.57, the right 2.02.   The counter is clocked at 1KHz.  MOSFET switches are used for controlling all three meters (can't have that switch resistance changing with my thumb pressure).   The FETs trigger is derived from the 1KHz clock source (my Arb).     

The video was first shot in real time.  It was then stretched to slow it down by 4X.   The last segment was shot at 1000fps.   The camera can only capture a few seconds of data in this mode.   

I used the Gossen the last time I ran this test as well as it's the only other handheld meter I have that will display three places beyond the decimal point.  Obviously we would expect my other lower resolution meters to be much faster.   I'm fine with the settling time of the 121Gw, if the meter would actually settle.

This Gossen meter is not stock.  I had added some magnetic shielding (Netic) to the rear and some copper shielding to the front.  Otherwise  it was a great sensor for detecting all sorts of environmental conditions.        I would actually say the meter could be dangerous to use in CAT III.   Like the 121, there were modes where the meter would show very low voltages in the presence of high voltage levels.  Really bad IMO.   
 
See Part 4 @ 7:40

[bitseeker]

Yeah, I remember the shielding you added to the Gossen. I quite enjoy all the fixes you come up with.

[CDaniel]

About the 121gw capacitance drifting, my (uneducated) guess is that it might have something to do with the SM6T22CA (D13 on the original 121 schematics), these devices seems to have a lot of nonlinear, voltage and probably temperature dependent capacitance behavior.

Perhaps a reason D13 now seems to been replaced with a traditional transistor clamp circuit in the latest hardware revision.

Yep , appology because I said it's not D13 , but it is .
Today I made some experiments , removed D13 and the drift is gone as far I had time to check .
Definitively the transil diode has capacitance very dependent on temperature and voltage if you check it with a LCR meter . Probably leakage resistance too.
Without it in circuit the reference cap 510pF now measure 200pF but as I said I couldn't see drift .
I didn't checked yet if it's possible to calibrate the meter and how much other functions are affected ...
If it's confirmed on the PCB is the place where the 2 transistors unpopulated in the schematic could be soldered .

[dcac]

Great to have that confirmed, these TVS diodes really does not seem like devices you want in a capacitance measuring circuit. Interesting both the schematics and PCB have allowance for alternative components with Q1-Q2, question is what type transistors to use though. I wonder why UEi opted for the TVS, seems like a bad choice anyway. 

I think you should be able to re-calibrate the caps mode.