EEVBlog 121GW Discussion thread

[joeqsmith]

IMO, the weak link of the meter's front end is that 15V diode check mode.   I marked the switch to make it easy to follow on the schematic.   

Also, note that a while back another member had posted some good pictures of their switch showing a fair amount of contamination in this same area.  I am not sure why it would accumulate here. 

A bit of damage for a low energy source.  Is there more to this story??

[Kean]

Do you remember what jacks were in use when measuring the DC/DC?
Were you using the meter for measurement of current, power or voltage?

The probes were definitely in the V and COM jacks for voltage measurement, and meter in auto ranging DC V.

I wasn't expecting to find this kind of damage, or any damage at all.  When I was getting stange measurements (like 166V at first and then OFL) I removed the probes, and then double checked the range switch position and jacks I was using, and tried the measurement again.  Then I switched to another meter.  It was only because an astrophysicist friend/client was visiting and he saw the PCB (he noted the 3 different crystals) that he asked about the rotary switch mechanism.  So I removed the contact housing to show him it in all its glory, and thus found the damage.

The energy to do that couldn't have come via the V jack, there is at least a PTC + 1K in series with every path. Even if the contacts shorted the VA range contacts and the ohms contacts, worst case path is V jack to GND via a PTC + 1K. So, it's weird 
So I can only presume the path is via the low impedance A jack somehow...
Maybe one of the range switch contacts wasn't installed properly and was floating around? I presume you hadn't taken off the range switch before?, as that's required for the shim replacement.

Agreed, that's basically what I thought - the protections circuits (even if not quite up to Joes standards...) shouldn't have allowed this.

Yes, I had swapped out the range switch and installed the shim, but I had not at this point taken off the part of the switch that houses the contacts off the PCB.  I had been pretty careful to ensure (per the instructions) that the orientation of that didn't change.  It actually takes a fair amount of force to remove the contact housing, and even then the contacts were well seated.  The contacts coming astray and connecting the wrong circuits was one of my thoughts as well - but I can't imagine the low impedance current path would go through the switch, just the kelvin connections off the shunt.

Kean, can you post a photo of the range switch contacts please.

Sure, see attached photo.  They look squeaky clean to me.  Sorry for the potato cam photos - best I can do at 4AM after a couple of rather awful days...  but I'm happy to arrange a meet up with you for a closer look and some forensics.

I also cleaned up the PCB with some IPA, and the PCB contacts actually look basically undamaged.  It almost looks like a track in the middle layers has burnt and created a crater.  It is a 4 layer PCB isn't it?  Maybe a PCB fault, a slightly under etched track clearance.

Edit: you can probably work out the orientation of the range switch based on the contamination shown on the plastic underside.  You can see the carbon deposit is not near any of the contacts points.

Joe, I'm not sure what other damage you are referring to.  A bit of carbon contamination has spread, but once cleaned nothing else but the crater - and other side of PCB looks OK.

[ChunkyPastaSauce]

Edit: you can probably work out the orientation of the range switch based on the contamination shown on the plastic underside.  You can see the carbon deposit is not near any of the contacts points.

To me, it looks like switch would have been in V, or possibly completing a switch between mv and V (closer to V)

Also, note that a while back another member had posted some good pictures of their switch showing a fair amount of contamination in this same area.  I am not sure why it would accumulate here. 

If it was this one https://www.eevblog.com/forum/testgear/eevblog-121gw-multimeter-issues/msg1561861/#msg1561861 it don't think it's related (the contamination plastic and possibly some mask as the unit was shipped with switch shim washer installed upside down, causing switch to rub against pcb).

[EEVblog]

Do you remember what jacks were in use when measuring the DC/DC?
Were you using the meter for measurement of current, power or voltage?

The probes were definitely in the V and COM jacks for voltage measurement, and meter in auto ranging DC V.

So you weren't measuring Power (VA) at all?
Hmm, there goes one theory.

Sure, see attached photo.  They look squeaky clean to me.

And there goes theory #2

I'm now at a loss as to how this happened. Need to think some more...

[Kean]

So you weren't measuring Power (VA) at all?

Nope, don't think I've even tried the VA mode as yet.

I'm now at a loss as to how this happened. Need to think some more...

Indeed.  Here is a couple of close ups of the PCB - one with backlighting.  You can actually see some minor damage to the gold plating in the corners, so maybe there was an arc across those (36&37) but via what path?
I'm not so sure about the internal trace now, but you can see there is one that goes directly under that crater.

[Seppy]

IMO, the weak link of the meter's front end is that 15V diode check mode.   I marked the switch to make it easy to follow on the schematic.   

Also, note that a while back another member had posted some good pictures of their switch showing a fair amount of contamination in this same area.  I am not sure why it would accumulate here. 

A bit of damage for a low energy source.  Is there more to this story??

Hello, thanks for the post yeah still a little at a loss. I think there might be an issue with your labeling here is why I think so:

The inside row on the PCB has pad pattern when read from right to left (ignoring routed connections):

ROW 1 _ _ _ __ _ __


One row further in you see:

ROW 2 __ _ __ _ __


On the schematic:
Reading pins 34, 33, 13, 36, 12, 11 follows the following pattern:

_ _ _ __ _ __ <- Pattern matches ROW 1

Reading pins 35, 16, 37, 15, 14 has the following pattern:
__ _ __ _ __ <- Pattern matches ROW 2


However, a discrepancy exists where:

1. 13 and 11 are ground, buzzed out to the current shunt
2. 14 and 16 are connected to FB4, buzzed out to ferrite

Lets refer to these as Pair 1 and Pair 2 respectively.
It appears these pairs have been swapped on the PCB.

From Joe's annotation if you measure what is labeled 37 and 15 you get an open circuit, on the schematic these should be a short.
Whereas if you measure between 12 and 36 it is a closed circuit indicating these rows are swapped.

Attached are what I believe the connections to be.

[Kean]

Thanks Seppy.  So, the switch contacts concerned are still 36 & 37?
I'm tied up with clients till the weekend, so I haven't even had a chance to reassemble it and see what still works.

[joeqsmith]

From Joe's annotation if you measure what is labeled 37 and 15 you get an open circuit, on the schematic these should be a short.
Whereas if you measure between 12 and 36 it is a closed circuit indicating these rows are swapped.

Attached are what I believe the connections to be.

I had not checked my work.  The short between 15 & 37 should have been a dead give away.       Good catch.   

The problem is still the same.  I wonder what R82, U9 area looks like.   I've said it a few times that the HEF parts have an absolute maximum VDD of 18V.  When I modified the pre-production meter,  I was having a problem finding a part with a sharp enough knee to stay under this.  I ended up changing to the part for a CD which added two more volts of margin.  Enough to get it to survive some basic transients after this.  But again, these are very low energy transients.  Nothing that would do this level of damage. 

[Kean]

I was thinking during dinner tonight that I should probably check with my colleague.  Maybe he visited the office while I wasn't around and used this meter in a "problematic way".  I doubt it, and I can't imagine what he could have done to cause this anyway 

* I did allocate him an EEVblog BM235 meter when he started working for me to try keep him away from my "toys" 

[PA0PBZ]

I'm not so sure about the internal trace now, but you can see there is one that goes directly under that crater.

That track has nowhere to go but 36.

[IanB]

Can someone explain how to display the burden voltage when measuring current?

I can follow the instructions by pressing the SETUP button and changing the setting to "bd.on", but after that I cannot see how to get a voltage shown on the display? (I assume the voltage should appear in the secondary display above the current reading?)

[ChunkyPastaSauce]

Can someone explain how to display the burden voltage when measuring current?

I can follow the instructions by pressing the SETUP button and changing the setting to "bd.on", but after that I cannot see how to get a voltage shown on the display? (I assume the voltage should appear in the secondary display above the current reading?)

Noticed the same and reported in https://www.eevblog.com/forum/testgear/eevblog-121gw-multimeter-issues/msg1566907/#msg1566907

[firewalker]

Is there any news from the Europe distributor? Did they received the meters?

Alexander.

[joeqsmith]

Any thoughts?

The energy to do that couldn't have come via the V jack, there is at least a PTC + 1K in series with every path. Even if the contacts shorted the VA range contacts and the ohms contacts, worst case path is V jack to GND via a PTC + 1K. So, it's weird 
So I can only presume the path is via the low impedance A jack somehow...
Maybe one of the range switch contacts wasn't installed properly and was floating around? I presume you hadn't taken off the range switch before?, as that's required for the shim replacement.

At 600V 2.5Kohm would still be 250mA.  Maybe enough to cook the PCB.  Seems possible.   

https://youtu.be/6j8i3LfKm5A

[EEVblog]

Any thoughts?

The energy to do that couldn't have come via the V jack, there is at least a PTC + 1K in series with every path. Even if the contacts shorted the VA range contacts and the ohms contacts, worst case path is V jack to GND via a PTC + 1K. So, it's weird 
So I can only presume the path is via the low impedance A jack somehow...
Maybe one of the range switch contacts wasn't installed properly and was floating around? I presume you hadn't taken off the range switch before?, as that's required for the shim replacement.

At 600V 2.5Kohm would still be 250mA.  Maybe enough to cook the PCB.  Seems possible.   

Yes, you are right, seems possible if you can get the contacts to arc over.
Here is quick video showing 1200VDC sustaining, twice the rated voltage of the 121GW.
A test which i have done many time before.

[Rutherfordium]

Is it reasonable to feel nervous when your hand goes near where the banana jacks are plugged in Dave?  Looks like exposed contacts.

[EEVblog]

Another test at 1100V AC

[EEVblog]

It looks like your DC supply is good for about 10mA.  You could try to turn it up to 600V with the 121GW set to the DCV mode, then turn it to ohms.  You may not have enough current to damage it, if your goal was to replicate the damage. 

My goal was only to show that the contacts do not arc over at these voltages.

[IanB]

My goal was only to show that the contacts do not arc over at these voltages.

Yes, but the important factor is turning the selection knob while the voltage is connected. Under these conditions the contact springs can partially bridge the gap between traces on the circuit board and increase the chance of an arc occurring. That's exactly the test Joe was doing in his videos.

[Kean]

Please take a few pictures of R82 and U9 under the microscope.  Maybe even ohm out R82.  What ever went though these two traces had to go through more parts.   There should be some pretty good damage to the top side as well.

R82 measures fine (100.3ohm), and no visible damage in the U9/R82 area.  In fact no visible damage anywhere else on the PCB, although I haven't looked under the LCD assembly.  The fuses are intact.

I also checked with my colleague, and he hasn't use this meter at all.  I will reassemble it and run some functional tests shortly, but don't want to do too much before I meet up with Dave.

[Kean]

Please take a few pictures of R82 and U9 under the microscope.  Maybe even ohm out R82.  What ever went though these two traces had to go through more parts.   There should be some pretty good damage to the top side as well.

R82 measures fine (100.3ohm), and no visible damage in the U9/R82 area.  In fact no visible damage anywhere else on the PCB, although I haven't looked under the LCD assembly.  The fuses are intact.

I also checked with my colleague, and he hasn't use this meter at all.  I will reassemble it and run some functional tests shortly, but don't want to do too much before I meet up with Dave.

And here is the photo I meant to attach to that post.

[FrankBuss]

My goal was only to show that the contacts do not arc over at these voltages.

Yes, but the important factor is turning the selection knob while the voltage is connected. Under these conditions the contact springs can partially bridge the gap between traces on the circuit board and increase the chance of an arc occurring. That's exactly the test Joe was doing in his videos.

Is it allowed that you turn the knob when voltage is connected? But this could be a problem in combination with the lose knob without the shim.

[Kean]

OK, I just ran through a reasonably comprehensive functional test of my "damaged" 121GW, and it all tested out perfectly, even BT.  The only mode that I didn't test (and haven't used) is the VA ranges.

I used an Advantest R6142 programmable voltage & current generator, an ITECH IT7321 programmable AC source, some resistance & capacitance decade boxes, a function generator, and some random diodes to hand.

Although I've returned to the customer the original equiment I was measuring when I think the meter failed, I will set up a similar scenario and see what measurements I get.  I suspect it will be fine now that I've cleaned the carbon residue off the switch contacts.  It was definitely reading crazy voltages before cleaning it.

[EEVblog]

Is it allowed that you turn the knob when voltage is connected? But this could be a problem in combination with the lose knob without the shim.

Not really, no, and I don't know any meter where they would specify that was a thing.
I've tested it with 240V mains whilst switching, but not with a low impedance HV DC source like has been discussed here.

[Kean]

Dave & other interested parties...

I have now remembered a time in the last few weeks I was measuring the 1200V AC output from a 100W Ultrasonic Transducer driver.  At the time I was pretty sure I only used my 15kV and differential oscilloscope probes - but based on the damage and Joe Smiths excellent test video, it seems likely that I must have tried to measure the voltage with the 121GW.

Very embarrasing, as I should gave known better!

Amazingly the meter still seems to function perfectly after cleaning the burnt area, and appears to still be in spec against my Keysight 34465A.