121GW RF Issue

[joeqsmith]

From my testing, Brymen has proven to be right there with the name brands.  Not only some of the most electrically and mechanically robust products on the market but also are withstanding the test of time.

As someone who sells a lot of Brymen's, I can attest that they have a noticeable non-zero long term failure rate. It's not high, but it's certainly not zero.

No brand has a zero long term failure rate.

True.  When I worked in automotive, we would burn-in our products to address the infant mortality.   If anyone is interested in a book on reliability, see attached.   It doesn't really cover the basics like burn-in but is more advanced.  Heavy math.

Thanks everyone who ran their 121GW in the mVDC mode.   I spent some time today to see if I could sort out why mine is so unstable.   

[EEVblog]

Thanks everyone who ran their 121GW in the mVDC mode.   I spent some time today to see if I could sort out why mine is so unstable.   

Your meter may not have the extra shielding that was added quite a few years ago now.

[joeqsmith]

Thanks everyone who ran their 121GW in the mVDC mode.   I spent some time today to see if I could sort out why mine is so unstable.   

Your meter may not have the extra shielding that was added quite a few years ago now.

That's certainly possible. 

Looks like I purchased the two 121GWs from Dave around 11/3/2019.

Post a few photos of the current shielding and I will pull mine apart and post a few of what I received. 

***
Attached showing one of the production meters.  These were purchased at the same time and from what I remember had the same construction.

[EEVblog]

Attached showing one of the production meters.  These were purchased at the same time and from what I remember had the same construction.

That does have the shielding.
What firmware are you running? Try the latest version.

[BillyO]

So don't necessarily expect your Brymen to last 20 years. One of the failure modes seem to be related to some sort of processor die rot.

That's not great to hear.  I have a 45 year old RadioShack DMM that's still going strong.  20 years should be nothing, especially for what my  Brymen BM786 cost.

[EEVblog]

So don't necessarily expect your Brymen to last 20 years. One of the failure modes seem to be related to some sort of processor die rot.

That's not great to hear.  I have a 45 year old RadioShack DMM that's still going strong.  20 years should be nothing, especially for what my  Brymen BM786 cost.

I'm not saying they are going to die, I'm just pointing out that the failure rate is non-zero. Some people think Brymens are the ducks guts and they'll last forever, that's not the case for anything. And always remember survivorship bias.
Remember, I have a vested financial interest in not telling you this about the Brymens, but I'm just being honest, failures do happen.

The worst example of meter failure rate I know of is the Fluke 19. Getting a working one today is as rare as hens teeth. They were infamous in the industry for having the main chipset fail.

[BillyO]

Remember, I have a vested financial interest in not telling you this about the Brymens, but I'm just being honest, failures do happen.

I appreciate the honesty.  I'll be 86 in 20 years and will probably need a meter with a BIG display.

BTW, when will the differential probe be back in stock?

[wolfy007]

I watched Joe's videos, and I did a quick test and found I had a similar problem with my GW121, Ill put my replies to his video here also;

"WOW this is fascinating (this vid, the last one and blog comments), and as I have the BM869s (164490215), BM786 (204520906) and a 121GW (180202062, from the crowd funder), I had to try this against a couple references I have (incl, Fluke 731b, 732A, 343A, 752A divider). My 121GW (U-2.00 updated) is reading high on mV range so I ended up using 100mV as a test voltage (I used shielded PTFE copper cables). Sure enough, noisy till I put foil around it and grounded, then stable as the other two. Unexpected, but nice work, definitely need to remember that."

"PS: couldnt help it, also tried an Agilent U1253A (MY51400064), Fluke 189 (75240037), 289 (35740002), 8060a (7720020), Uni-T UT181a (C171835804), Tektronix DMM916 (144966) and even an Owon B41T+ (21340133), thankfully no other noise issues (even tried with a MHz noise generator transmitting in the background)."

[EEVblog]

BTW, when will the differential probe be back in stock?

ETA for shipping by 10th Sept.
Holster will be yellow instead of blue due to a supply issue.

[joeqsmith]

Attached showing one of the production meters.  These were purchased at the same time and from what I remember had the same construction.

That does have the shielding.
What firmware are you running? Try the latest version.

Thanks for checking. 

The production meter has 2.02 dated 09/27/2019.  I believe this was the latest available when I purchased the two meters and made the review.   

Any idea what the changes were from 2.02 to 2.04 and to 2.05?  I don't see any sort of change log and there is only the binary in the ZIP file.   

I did install 2.05 and injected RF into the shorted leads as before.  The new firmware did not have an effect on the sensitivity.  I guess I wasn't expecting it to as there is not a lot the firmware can do at 100MHz.  It was worth a try.    Foil seems to do a fair job. 

[joeqsmith]

I watched Joe's videos, and I did a quick test and found I had a similar problem with my GW121, Ill put my replies to his video here also;
...

"PS: couldnt help it, also tried an Agilent U1253A (MY51400064), Fluke 189 (75240037), 289 (35740002), 8060a (7720020), Uni-T UT181a (C171835804), Tektronix DMM916 (144966) and even an Owon B41T+ (21340133), thankfully no other noise issues (even tried with a MHz noise generator transmitting in the background)."

Thanks for trying to replicate what I saw with the 121GW.   The only others I have that may be worth looking at are my TPI (Summit) 194II, Yokagowa TY720 and maybe the CEM DT-9938.  I would need to check their specs.   

Still,  this a poor way to test meters as it's not controlled.   I expect IEC 61326 requires the meters pass the industrial electromagnetic environment at 10V/m, performance criterion A.  This means no degradation of performance during the test.    ....   OK, I think I see the problem.  Looking at the backside of my 121GW, they are not tested to 61326.    Even if they were, then do I trust the people testing the products.....   

[EEVblog]

Confirmed. The reading peaks (about 0.25mV) at 110MHz (500mV level).
Above and below 110MHz-ish it's basically zero.
I was also able to reproduce this on the Keysight U1282A, but that peaks at 66MHz and gives a higher reading than the 121GW for the same RF signal level.
The Keysight also gives a reading about half at 36MHz. Nothing on the 121GW at 36MHz.
For reference, both meters use the HY3131 chipset.

I will branch out this to a new thread, it does not belong here.

UPDATE:
The Fluke 17B also does this at about 120MHz, at 0.4mV reading. Doesn't look as bad though of course it's only 100uV resolution, not 1uV resolution, but the reading is actually higher than the 121GW.
The BM235 also does this, up to 0.4mV depending upon the location of the leads physically.
The Keysight U1272A also does this (0.4mV) at 36MHz.

I have a recollection this is nothing new and has been discussed a few years back?

[joeqsmith]

I have a recollection this is nothing new and has been discussed a few years back?

It's very possible. 

My goal was to provide some idea how accurate my 7 YO Brymen is in the DC modes after it was suggested they need yearly alignment.  To gain some confidence in my measurements, I wanted to run a different brand meter.   I don't own a lot of high res handhelds and the first meter I grabbed was the 121GW.   I wasn't thinking RF and never thought I would run into problems making such a simple measurement.  I tossed it aside and grabbed the GMW for a comparison.     

I suspect every meter is susceptible to RF.   One time during my career I had to use an outside lab for testing and when I arrived, there was a tomahawk cruise missile in the lobby.   They had capability for testing several GHz at 1000V/m.   At that time most of my involvement was at 100V/m.  I've damaged electronics at that level.    For the last few decades anything I have been involved with is only tested to 10V/m.  As I mentioned during the video, even at these low levels, I have seen handheld meters (even Fluke products) that were effected.       

61326 radiated immunity starts at 80MHz.  Below this is conducted. 

[EEVblog]

As I mentioned during the video, even at these low levels, I have seen handheld meters (even Fluke products) that were effected.

Fluke famously had to redesign the 87V after my video showing that a GSM mobile phone nearby could brick it. Took them a year to get the new model rolled into production.

[EEVblog]

Still,  this a poor way to test meters as it's not controlled.   I expect IEC 61326 requires the meters pass the industrial electromagnetic environment at 10V/m, performance criterion A.  This means no degradation of performance during the test.    ....   OK, I think I see the problem. 

The only meter I know of that's specifically EMI marketed 61326-1 is the "double shielded" Brymen BM2807CSE. That's it's big marketing claim, for high EMI enivironments.
But even that is only specified for 3V/m, and even 1V/m for some functions (all +100 digits)
The BM869 is only tested to 3V/m under 61326, but it's +1000 digits.
And AFIK it doesn't mean "no degredation", which is why Brymen for example specify nominal accuracy +1000 digits for 3V/m

[joeqsmith]

I remember the GMW being marketed as an industrial meter which to me suggests 10V/m, criterion A.    Looking through the standard, it appears that all three level (1,3 & 10) require performance criterion A.   But looking closer, at the definition of A, "No degradation of performance or loss of function is allowed below a performance level specified by the
manufacturer, when the equipment is used as intended."  which backs up your comment about Brymen.   

When I swept those four meters, the GMW was by far the most effected, followed by the UNI-T UT181A.

***
Should add, that was with the meters terminated to 50ohms, no test leads and set to their lowest DCV scale.