BM786 Switch Issue

[joeqsmith]

All of the meters I looked at had these scratches in the switch area.  I didn't ask them about it.  Agree that it looks poor but it didn't seem to have much of an effect on the meter I life cycled.   If you watch that video, the meter was still working fine after that test.   Many meters that started out with smooth boards were ground to dust....

Here are a couple of pictures showing this area before I subject it to some high voltage with the meter set to ohms. 

[bdunham7]

I don't know if it has been mentioned before, but it looks like your picture is missing the same chunk off the end of the trace as the other meter that failed?

[floobydust]

There's nothing physically wrong with that PCB portion - it's just bad PCB CADD layout.
You have rectangular-ended switch polygons, rounded-end traces and soldermask connecting there, so it looks wonky.

Many traces exiting the rotary switch leafs are done wrong - they should exit radially. Otherwise you get overlap and extra scuffs. It's not a great switch PCB layout IMHO.

From the way the DMM cuts out when the switch is wiggled on Ohms, I suspect power being shorted out vs open circuit?

P.S. I wonder if you can even do this switch design in Altium, or you have to draw it in AutoCAD and import the switch portion?

[joeqsmith]

I don't know if it has been mentioned before, but it looks like your picture is missing the same chunk off the end of the trace as the other meter that failed?

Interesting is that it would seem that this area of concern for you may actually be a way to improve the design of meters.  Consider the attached picture showing the BM786 after completing a 50,000 cycle life test on the switch.  Not Dave Jones 50K but full CW/CCW making one cycle.  Or 100,000 DJ cycles.   Notice in cases where the mask is removed in area around the contact's leading edge, we see some rounding off where in this case, we start with a smoother transition and less wear.   

Even with scratches the meter held up very good to this test compared with others I have looked at.  Low quality meters are ground to dust and if we want to talk about poor layouts that actually hurt the performance, lets at least mention the end results of placing a via in the center of the switch contacts.   

[AVGresponding]

All of the meters I looked at had these scratches in the switch area.  I didn't ask them about it.  Agree that it looks poor but it didn't seem to have much of an effect on the meter I life cycled.   If you watch that video, the meter was still working fine after that test.   Many meters that started out with smooth boards were ground to dust....

Granted it probably doesn't affect switch life noticeably, but it does indicate a degree of hand finishing which is as we all know something you prefer to avoid due to cost, variability of results etc.

Can't be good for the solder mask, and there has to be at least a small risk of pinging off a component or damaging a trace.

It begs the questions, why do Brymen consider it necessary? What problem with the automated fabrication process is this an attempt to remediate?

[Algoma]

Looks like about 5 quick scrapes with fairly coarse abrasive.

I would think A simple scrub with a fine steel wool would put texture and shine on the pads without scraping off vast gouges across the solder mask.

The scratch pattern looks like they're using an edge of a heavy grit sanding block that barely fits into the given space of the already populated board.

[joeqsmith]

Like I said, I never asked them about the scratches and without taking that first step,  I would have no way to who, why or when it was done.   

Getting back to my comments about HV and the ohms function.
https://www.eevblog.com/forum/testgear/bm786-switch-issue/msg3516068/#msg3516068
https://www.eevblog.com/forum/testgear/bm786-switch-issue/msg3516184/#msg3516184

First, Brymen supplied four different meters.  One meter was a non-functional, incomplete prototype.  This is the meter I used for this test.  As you can see from the pictures, the layout and scratches are similar to what was released.   The meter was supplied with no wiper contacts.  Sorting through the junk pile, I found a wide contact that came close to fitting.  Of course, this is different than the contacts supplied with the meter.  I used my kill-u-dead power supply which you may have seen blow one of my HV 400mA fuses.     

Looking at the attached pictures, you can see how the wiper contact did not ride down the center of the pad and is off to one side.  We can see that the heat was enough to delaminate the board. What is interesting is the size of the area of damage.  Had the contact been centered, would we even see the delamination?  Maybe not.  I would even suggest that the plating could fail some time later.  By chance, it looks like it made better contact on the other pad and there appears to be no damage.  I tested the two transistors that make up the high speed clamp (Q4&Q5) prior to running the test and after.  The both appear fine.   I would have expected that but it's good to know.   

Will the PTC and surge rated resistor limit the current enough to save the switch contacts under all conditions?  I would say no.   
Could the input signal be within the range of the meter and damage it?   I would say yes.
Could the PCB be damaged without there being any damage to the front end and clamping circuits?   I would say yes.
Could the damage be such that it creates enough heat to delaminate the PCB?   I would say yes.   
Are there sources of HV/HC that could do this level of damage that are common?   Seems like the MOT is a common source..

Again, I have no idea what happened in this particular case.   I do find it odd that the failure just happened to be on one of the modes using the high speed clamp. That the delamination is only on one spot of the entire board.  That spot just happens to be where the wiper contact would sit when the ohms mode is engaged. 

[Algoma]

I can assure you, this meter only encountered light usage in a computer repair shop before it was set aside for exchange from Dave due to the noted issues as before. The replacement meter has worked flawlessly for weeks now.

the original meter in question was only subjected to the following list over the few days it was out of the box:

1) DHL's shipping process, they certainly did their best effort to indent the lower right corner of the the packing box.
2) Dell Laptop 19.5VDC 5A power supply, all tested good.
3) RH05020R00FE02 20Ohm 1% Vishay Dale Chassis Mount Load Resistor
4) CMF601K0000BHEB 1K 1W 0.1% Vishay Dale Axial Resistor
5) CMF60100R00BHEB 100Ohm 1W 0.1% Vishay Dale Axial Resistor
6) HEATHKIT IN-3117 DECADE RESISTANCE KIT

Highest voltage nearby would be 120v Mains power bar +- Static electricity .. at best.

I'm neither qualified or experienced enough go anywhere near the limits of a brand new meter I paid good money for, especially poking conductive things towards dangerous energy sources, on any range. Your original testing videos assisted in the decision to select this meter as a safe and well qualified meter for my usages. I'm simply a Network Engineer with curiosity about the foundations of the technology I work with. Not an electrician

The minimal usage it encountered is very unlikely to have caused any such damage to the trace, unless it was some form of high voltage qualification testing done at the factory.

[Fungus]

I don't know if it has been mentioned before, but it looks like your picture is missing the same chunk off the end of the trace as the other meter that failed?

Who says it's missing a chunk? In Dave's video all the traces have square ends, even though there seems to be a round end on the PCB.

The trace at the top looks like a boolean union between the exposed gold trace (with square end) and the non-exposed trace under the solder mask:

[joeqsmith]

I can assure you, this meter only encountered light usage in a computer repair shop before it was set aside for exchange from Dave due to the noted issues as before. The replacement meter has worked flawlessly for weeks now.

the original meter in question was only subjected to the following list over the few days it was out of the box:

...

Highest voltage nearby would be 120v Mains power bar +- Static electricity .. at best.

I'm neither qualified or experienced enough go anywhere near the limits of a brand new meter I paid good money for, especially poking conductive things towards dangerous energy sources, on any range. Your original testing videos assisted in the decision to select this meter as a safe and well qualified meter for my usages. I'm simply a Network Engineer with curiosity about the foundations of the technology I work with. Not an electrician

The minimal usage it encountered is very unlikely to have caused any such damage to the trace, unless it was some form of high voltage qualification testing done at the factory.

As I have said, I would have no way of knowing what happened.  For all I know, the meter was left sitting out and someone borrowed it for a test run.   I wasn't there, I didn't receive your meter to evaluate and looking at it now would be rather pointless as any evidence left would have been taken care of by the end of a metal blade.   All I can do is present data that shows that it is possible to delaminate the PCB while in the ohms mode while injecting a signal to the inputs of the meter  as well as point out other details we can see from the video and pictures.       

[Algoma]

Yes, and I'm defending the point that this was NOT caused by suicidal abuse of the multimeter. the issue was present within the first turn of the dial with no leads attached. This case was fully sourced from the manufacturer.

I state clearly, that it is false, slanderous hearsay to claim that in this case the meter was abused, and the evidence hidden in this case. Dave's blade left far more scratches in his video than any other markings on that trace,.

[bdunham7]

Interesting is that it would seem that this area of concern for you may actually be a way to improve the design of meters.  Consider the attached picture showing the BM786 after completing a 50,000 cycle life test on the switch.  Not Dave Jones 50K but full CW/CCW making one cycle.  Or 100,000 DJ cycles.   Notice in cases where the mask is removed in area around the contact's leading edge, we see some rounding off where in this case, we start with a smoother transition and less wear.

OK, from that angle I can see how the traces intersected and I see the worn triangle of soldermask on the life-cycle tested meter. 

[joeqsmith]

Yes, and I'm defending the point that this was NOT caused by suicidal abuse of the multimeter. the issue was present within the first turn of the dial with no leads attached. This case was fully sourced from the manufacturer.

I state clearly, that it is false, slanderous hearsay to claim that in this case the meter was abused, and the evidence hidden in this case. Dave's blade left far more scratches in his video than any other markings on that trace,.

I have made no false claims about what happened to your specific meter.  I have stated multiple times that I would have no way of knowing.   

"... and it's in ohms range so you would never get high enough current passing through the uh input protection resistors here and the PTCs to cause some sort of ridiculous you know arching of like a across that contact. Its just its just not possible"

At the rated voltage (1KV) it certainly is possible to draw an arc large enough to damage the contacts.  I had made a video to demonstrate this after anther member had damaged a 121GW.    While you should never rotate the function switch with the meter connected to an active circuit,  a few of us have bad habits.  Mix that with a little HV..... 

What I have done is conducted a simple test to see if the original statement from the video was true or not and presented my findings.   I'm sorry that makes you become defensive.  It shouldn't.   

[Algoma]

@joeqsmith I do apologize.

Your work and testing is valuable to the community, and this in depth investigations indeed this shows what types of damage can occur to a meter if mishandled. Thank you for all the work you put into testing these meters.

[joeqsmith]

Interesting is that it would seem that this area of concern for you may actually be a way to improve the design of meters.  Consider the attached picture showing the BM786 after completing a 50,000 cycle life test on the switch.  Not Dave Jones 50K but full CW/CCW making one cycle.  Or 100,000 DJ cycles.   Notice in cases where the mask is removed in area around the contact's leading edge, we see some rounding off where in this case, we start with a smoother transition and less wear.

OK, from that angle I can see how the traces intersected and I see the worn triangle of soldermask on the life-cycle tested meter.

You can see how this leading edge can cause problems.   I think you may be onto something.  Then again, when we look at the Fluke 87V (damage 2), we can see how the design of the wiper contact can also have an effect on wear.   There is an art to getting the switch right.   

[EEVblog]

I can assure you, this meter only encountered light usage in a computer repair shop before it was set aside for exchange from Dave due to the noted issues as before. The replacement meter has worked flawlessly for weeks now.

the original meter in question was only subjected to the following list over the few days it was out of the box:

...

Highest voltage nearby would be 120v Mains power bar +- Static electricity .. at best.

I'm neither qualified or experienced enough go anywhere near the limits of a brand new meter I paid good money for, especially poking conductive things towards dangerous energy sources, on any range. Your original testing videos assisted in the decision to select this meter as a safe and well qualified meter for my usages. I'm simply a Network Engineer with curiosity about the foundations of the technology I work with. Not an electrician

The minimal usage it encountered is very unlikely to have caused any such damage to the trace, unless it was some form of high voltage qualification testing done at the factory.

As I have said, I would have no way of knowing what happened.  For all I know, the meter was left sitting out and someone borrowed it for a test run.   I wasn't there, I didn't receive your meter to evaluate and looking at it now would be rather pointless as any evidence left would have been taken care of by the end of a metal blade.   All I can do is present data that shows that it is possible to delaminate the PCB while in the ohms mode while injecting a signal to the inputs of the meter  as well as point out other details we can see from the video and pictures.     

Joe, why are you getting involved in this way? Algoma is the owner of the meter in question, there is no need for speculation like this, he has categorically stated that this happened out of the box, he contacted me the day he got it and tried it. There is no need for speculation here on your part, it is not helpful. This is a factory fault.

[joeqsmith]

Joe, why are you getting involved in this way? Algoma is the owner of the meter in question, there is no need for speculation like this, he has categorically stated that this happened out of the box, he contacted me the day he got it and tried it. There is no need for speculation here on your part, it is not helpful. This is a factory fault.

Because you chose to make it public and make a claim which I questioned the accuracy of.   I'm sorry you feel that my attempt to substantiate your claim wasn't helpful.  If you like, you may remove the posts if you feel that strongly about them.   Personally, I like being data driven but its your site. 

https://www.eevblog.com/forum/testgear/bm786-switch-issue/msg3516068/#msg3516068
https://www.eevblog.com/forum/testgear/bm786-switch-issue/msg3532326/#msg3532326

[bdunham7]

This is a factory fault.

That still leaves the question of what happened at the factory--and how likely it is to be a common issue.  I suspect that it is actually not going to be a common issue, as the most likely thing I can think of that would cause this would be technician error or equipment malfunction during the calibration process. 

I have had three incidents in the past year while repairing multifunction calibrators.  One was my error, I fed my best DMM 900VDC while in ohms range (no harm done), one was a calibrator that would surge a bit while settling on AC and it went above 1000VAC enough to cause the DMM clamping to activate, which then tripped the calibrator overload (again, no apparent harm--I think).  I started using 'disposable' meters for initial checks and just last week I was testing one at 1000VAC/1kHz when I looked over to see the disposable meter of the week, an old Fluke 8800A, reading 1435 volts!  Again, amazingly, nobody died.

Since the ranges have to be selected manually during calibration, how easy would it be for that tech to select 1kV and then realize that the selector is still in ohms?  Or, to turn the knob to ohms and then realize that the calibrator is still putting out 1kV?  Even if the process is more automated, there's still ways that something like this could go wrong.

[floobydust]

The Ohms switch ring has what appears to be a runt which may be aggravating things, looks too short compared all others, not much overlap? That might be a design issue showing up with manufacturing tolerances in production.
You'd have to remove the detent mechanism and turn the switch side to side, to see how much slop there is, where it still works.

It's extra hard doing PCB layout with mechanical constraints- tolerances, moulding needs, etc.  Any disconnect between the MECE's and EE's can cause grief.
I suspect the switch cannot be drawn in the PCB CADD software, so it is drawn in something mechanical (AutoCAD) or wherever the switch mech is being done for the plastics.
Then it's imported into the PCB layer or Gerbers. You can tell because many traces seem to make a wonky exit from the switch leafs.
I have seen this (import as graphic) to achieve oddball geometry/shape copper, but it's a total nightmare for netlists and error checking.

Regardless, I'm sure Brymen is on it.

[EEVblog]

This is a factory fault.

That still leaves the question of what happened at the factory--and how likely it is to be a common issue.  I suspect that it is actually not going to be a common issue, as the most likely thing I can think of that would cause this would be technician error or equipment malfunction during the calibration process. 

As several people have pointed out, I think it's most likely a one-off issue involving the PCB manufacture. There is no evidence it's been burned or pitted away by a high voltage arc in testing. And such testing is usually automated.
It looks like there was a copper lamination issue that ultimately resulted in the copper tearing due the switch contact movement.

[EEVblog]

Joe, why are you getting involved in this way? Algoma is the owner of the meter in question, there is no need for speculation like this, he has categorically stated that this happened out of the box, he contacted me the day he got it and tried it. There is no need for speculation here on your part, it is not helpful. This is a factory fault.

Because you chose to make it public and make a claim which I questioned the accuracy of.

No, it's you questining the owners account that I had a problem with.

[joeqsmith]

Joe, why are you getting involved in this way? Algoma is the owner of the meter in question, there is no need for speculation like this, he has categorically stated that this happened out of the box, he contacted me the day he got it and tried it. There is no need for speculation here on your part, it is not helpful. This is a factory fault.

Because you chose to make it public and make a claim which I questioned the accuracy of.

No, it's you questining the owners account that I had a problem with.

As I had stated in most of my posts, I would have no idea what happened in this particular case.  I can only post on what I see and heard in your video.  Like I said, feel free to pull it all if it bothers you too much or just don't respond to them. 

[bdunham7]

As several people have pointed out, I think it's most likely a one-off issue involving the PCB manufacture. There is no evidence it's been burned or pitted away by a high voltage arc in testing. And such testing is usually automated.
It looks like there was a copper lamination issue that ultimately resulted in the copper tearing due the switch contact movement.

Maybe so, and if it is (hopefully) truly a one-off, we'll likely never know.  I've never seen a defect like that on an unmolested PCB trace and that looks like a decent board.  It typically would take bit of work to tear off a chunk from the middle like that unless there was something like a buried via underneath causing some stress--and since it looks like a 2-layer board, there shouldn't be one. 

An electrical event wouldn't need to cause much arcing to cause a bit of localized heating.  The next time I have a calibrator fired up I'll see how it welds.

[joeqsmith]

"... unless there was something like a buried via underneath causing some stress--and since it looks like a 2-layer board, there shouldn't be one. "
No buried vias.  There's nothing in that area.   

Maybe a SEM could have detected something.   Even with the PCB in the current state, it could be sectioned.   Both may tell you if there was any sort of arc or heat damage. 

I don't know anything about modern calibration equipment but wouldn't expect them to have enough output capacitance or current in the HV setting to do much.  Then again, it may not take a whole lot to damage the board.  It will be interesting what you find.   

Maybe post the model, open circuit voltage, short circuit current, output capacitance or use a 2.3ish kohm load and watch it with a scope.  The more details you can provide about your setup, the better. 

[Fungus]

"... unless there was something like a buried via underneath causing some stress--and since it looks like a 2-layer board, there shouldn't be one. "
No buried vias.  There's nothing in that area. 

Maybe a bug crawled into the PCB plant and got stuck under the copper. I'm sure it happens.

The good thing is that this definitely isn't a systemic production fault. It's a one-off that got through quality control by needing a user to press the dial just right.