Handheld meter robustness testing

[joeqsmith]

My old Fluke 73/III had a real switch, and a separate spring assembly, that spring assembly had separate part number...
I guess they really don't make them as they used to... 

I have not seen any imaginary switches used in any of the meters I have looked at.  Or at least they appear real to the touch to me anyway.      Sorry...  Anyway it looks like someone had posted some decent pictures of what yours looks like.   It appears to be a fiber wafer switch similar to what was in the TV sets from my youth.   It's too bad we can't get that time machine working or I would put together enough old cash, go back in time and buy some of these brand new and run them.  You would think the life would be much longer with the two contact but I have seen a lot of these go intermittent as well.     That Tachikawa I recently restored has a separate wafer switch as well.  This is not a high end meter by any means. 

https://www.eevblog.com/forum/testgear/t21362/

[joeqsmith]

While the H-PAK meter is apart, I tried to take some better pictures of the damaged areas of the spring. 

[2N3055]

My old Fluke 73/III had a real switch, and a separate spring assembly, that spring assembly had separate part number...
I guess they really don't make them as they used to... 

I have not seen any imaginary switches used in any of the meters I have looked at.  Or at least they appear real to the touch to me anyway.      Sorry...  Anyway it looks like someone had posted some decent pictures of what yours looks like.   It appears to be a fiber wafer switch similar to what was in the TV sets from my youth.   It's too bad we can't get that time machine working or I would put together enough old cash, go back in time and buy some of these brand new and run them.  You would think the life would be much longer with the two contact but I have seen a lot of these go intermittent as well.     That Tachikawa I recently restored has a separate wafer switch as well.  This is not a high end meter by any means. 

https://www.eevblog.com/forum/testgear/t21362/

Take a look here.. https://youtu.be/9ptPe_AeZiQ?t=509

By all means, switches can be made right on a PCB. It's just that its not that easy as just thinly gold plating some pads and slapping some brass to slide on it...
Switch on 73 looks low tech, but seems robust enough.. And it is, that damn thing still works no problem.

Regards,

Sinisa

[IanB]

One of the guidelines in studies of friction in bearings and sliding contacts is that the two mating surfaces should be made of dissimilar materials. In the case of the multimeter switch spring, I would think the plastic springy bit should be made of something hard and durable (maybe like nylon), and the outer ring with the detent slots should be made of something soft and slippery (like polypropylene?).

In the different cases of the meter switch that was durable (Fluke) and the one that wasn't (Keysight) it would be interesting to do a materials analysis of the different components to find similarities and differences. YouTube blogger Ave has often given hints on how to identify different plastics. One simple technique is to find the melting point with a soldering iron set to different temperatures. Sometimes there is even a meterial code embossed on the part.

[joeqsmith]

One of the guidelines in studies of friction in bearings and sliding contacts is that the two mating surfaces should be made of dissimilar materials. In the case of the multimeter switch spring, I would think the plastic springy bit should be made of something hard and durable (maybe like nylon), and the outer ring with the detent slots should be made of something soft and slippery (like polypropylene?).

In the different cases of the meter switch that was durable (Fluke) and the one that wasn't (Keysight) it would be interesting to do a materials analysis of the different components to find similarities and differences. YouTube blogger Ave has often given hints on how to identify different plastics. One simple technique is to find the melting point with a soldering iron set to different temperatures. Sometimes there is even a meterial code embossed on the part.

The bearing surfaces show little signs of wear on the H-PAK.  However, once the plastic cracked we lost all the tension which would greatly reduce the wear. 

My plastics knowledge is VERY little.  I may be able to tell you if they are a thermoset or not by seeing if they melt at all.   

[IanB]

The bearing surfaces show little signs of wear on the H-PAK.  However, once the plastic cracked we lost all the tension which would greatly reduce the wear.

Yes, I saw the pictures of the crack after I posted. That basically looks like a catastrophic failure. The fact that it failed like that suggests that the mechanical engineers were either absent or failed to do a good enough job on the design.

[joeqsmith]

A recent post from a viewer.  To be clear, my goal in posting this here in the forums is to open a discussion about running a test like this.  Do you feel all the meters would have to be brand new to gain any understanding about the switched life?   Do you feel we need to have different tests for different meters?   If so, what do you feel should be the criteria?    Do you feel there is no value in running the test because we are only looking at one meter?

Again, to be clear, I am leaning towards running the functional tests, followed by the transient tests just like I always have since I started doing more of a review.    After this, I would then life cycle them.  If there is anything left, I would then run the half cycle if I thought the meter would give us some sort of light show...     It's obviously a sort of crap shoot with my videos as there are times when I make some pretty big changes to these meters just to try and get a better understanding about their designs.     

Feel free to chime in.  As far as I am concerned at this stage, there really are no wrong answers. 

 
Joe, very informative test as always! However, I think the fact that one of the pads
on the rotary switch was damaged initially fom prior tests you ran, should have negated
this particular meter  from your selector switch stress test no matter what you did to
repair/negate the problem. In a real word stress/performance scenario you just can't
bench-test a product and achieve a fair /accurate result if that product was altered
in any way, no matter how good the intentions.

(me)
I was very clear about the damage that had been done to the meter prior to the testing
along with what I had done to  mitigate it.   If you feel running a brand new unit with
near SNs would change the results,  I am perfectly fine with that.  I have no data that
shows one way or the other.    You are welcome to repeat the test and show your own
results
       
Indeed Joe. 3rd party Refurbished isnt a factory standard and as such I would take
any testing results with a grain of salt no matter how pretty the graphs might appear.
It's one thing to take a bevy of refurbished meters that had similar issues and produce
a final test result but it's quite another to compare dissimilar meters and try to accrue
some sort "precision" results based on the aftermath of varied component failures.
       
(me)       
I have disclosed if the meters were new, used or had damage to the switch area.  I assume
you are considering the repaired meters, like the Fluke a refurbished meter as it was
damaged during the transient tests and repaired by myself (3rd party) .   That's fine. 
It's only one meter, not really a bevy.    If you feel the Fluke would have performed
better (or worse) had Fluke performed the repairs, I am fine with that.  It could also
be that a brand new Fluke 17B+ would also perform better than the one I looked at.   
Again, I don't have any data one way or the other to suggest different.   If you feel
the same model Keysight meter from the same lot would yield different results, you are
certainly free to show your own data.  I would welcome it.

[joeqsmith]

The bearing surfaces show little signs of wear on the H-PAK.  However, once the plastic cracked we lost all the tension which would greatly reduce the wear.

Yes, I saw the pictures of the crack after I posted. That basically looks like a catastrophic failure. The fact that it failed like that suggests that the mechanical engineers were either absent or failed to do a good enough job on the design.

I agree with your last comment.  Running an FEA on the parts my show something.  Then again, it may be more effort than it's worth.   

My iron will go down to 400F on the display.  Both the case and the spring will melt at whatever temperature this is at the tip.   The case has a much more even flow to it.  The spring appears to burn somewhat.  I agree with SeanB and suspect a filler.  The case appears to be a lot more flexible than the spring.     

[saturation]

For plastics, its straightforward for pure polymers, however, it will give confusing results for composites.

http://www.boedeker.com/burntest.htm

http://www.chymist.com/Polymer%20Identification.pdf

Except in the case of fiberglass, where after melting the remnants will be hard and dense other composites: wood, carbon fiber, etc., will burn with the plastic polymer, alter the color profile but will give mixed results for density tests.  Burn tests are toxic and potentially carcinogenic.

[joeqsmith]

A recent post from a viewer.  To be clear, my goal in posting this here in the forums is to open a discussion about running a test like this.  Do you feel all the meters would have to be brand new to gain any understanding about the switched life?   Do you feel we need to have different tests for different meters?   If so, what do you feel should be the criteria?    Do you feel there is no value in running the test because we are only looking at one meter?

Again, to be clear, I am leaning towards running the functional tests, followed by the transient tests just like I always have since I started doing more of a review.    After this, I would then life cycle them.  If there is anything left, I would then run the half cycle if I thought the meter would give us some sort of light show...     It's obviously a sort of crap shoot with my videos as there are times when I make some pretty big changes to these meters just to try and get a better understanding about their designs.     

Feel free to chime in.  As far as I am concerned at this stage, there really are no wrong answers. 

 
Joe, very informative test as always! However, I think the fact that one of the pads
on the rotary switch was damaged initially fom prior tests you ran, should have negated
this particular meter  from your selector switch stress test no matter what you did to
repair/negate the problem. In a real word stress/performance scenario you just can't
bench-test a product and achieve a fair /accurate result if that product was altered
in any way, no matter how good the intentions.

(me)
I was very clear about the damage that had been done to the meter prior to the testing
along with what I had done to  mitigate it.   If you feel running a brand new unit with
near SNs would change the results,  I am perfectly fine with that.  I have no data that
shows one way or the other.    You are welcome to repeat the test and show your own
results
       
Indeed Joe. 3rd party Refurbished isnt a factory standard and as such I would take
any testing results with a grain of salt no matter how pretty the graphs might appear.
It's one thing to take a bevy of refurbished meters that had similar issues and produce
a final test result but it's quite another to compare dissimilar meters and try to accrue
some sort "precision" results based on the aftermath of varied component failures.
       
(me)       
I have disclosed if the meters were new, used or had damage to the switch area.  I assume
you are considering the repaired meters, like the Fluke a refurbished meter as it was
damaged during the transient tests and repaired by myself (3rd party) .   That's fine. 
It's only one meter, not really a bevy.    If you feel the Fluke would have performed
better (or worse) had Fluke performed the repairs, I am fine with that.  It could also
be that a brand new Fluke 17B+ would also perform better than the one I looked at.   
Again, I don't have any data one way or the other to suggest different.   If you feel
the same model Keysight meter from the same lot would yield different results, you are
certainly free to show your own data.  I would welcome it.

The OP puts out their own handheld meter reviews.   I knew this but had never watched one until tonight.   My wife and I watched the one for a Sanwa meter that they had asked me to look at.  I was hoping to get some context on where they were coming from with their comments.  It's a little more clear now.   If you would like me to test meters more like this, feel free to let me know. 

https://youtu.be/CDsgVkHgE4g

[bitseeker]

The switch spring failure in the Keysight is unfortunate. Hopefully, their higher-end meters have a better/more durable mechanism.

"Unfortunate" isn't the word I'd use - it's probably not bad luck.

"Unfortunate" also means regrettable.

As for "Hope"? I'd want proof. 

Certainly. Failures due to poor materials have resulted in the publishing of service notes and usually free repair/replacement of the bad parts. Alas, there isn't one about the function switch.

https://servicenotes.literature.keysight.com/litapp/SearchSN.do?method=openExternalSNSearch&prodNum=U1232A

PS: I wonder if user 'Keysight DanielBogdanoff' reads this thread. Keysight's reputation for multimeters needs some salvage work here. Maybe he could send Joe some meters to test.

He usually pops in on scope threads, but sometimes on other ones that mention HPAK equipment. I haven't seen him in this one, yet.

[Metermeister]

I like both your style. His video is informative in more different way than yourself. It's good that we have many reviewers our field needs them. Electronics in Europe is on the decline.

[joeqsmith]

I did ask him to run his own test on the Keysight meter.   This one is 2 years old and gets 10s of thousands of cycles yearly. 

https://youtu.be/_EQdxZK0yHw

[Fungus]

When he grabs it and twists the knob it doesn't sound very clicky, not like in your "intro" video.

Would it kill him to open it and take a look instead of just opining?

[HalFET]

Finally managed to get the profilometer long enough to do a scan of a 8x8mm patch near the centre of the switch, but the nickel coating is a bit too reflective it would seem. (It registers as lower than the soldermask )  Didn't have time to redo it as a dual scan with two modulation thresholds so here's one to begin with:

[joeqsmith]

When he grabs it and twists the knob it doesn't sound very clicky, not like in your "intro" video.

Would it kill him to open it and take a look instead of just opining?

He did say tens of thousands of cycles per year, and two years, I would expect it to be a little less clicky.  Maybe he will show you the insides as part of the review he mentioned doing.

[joeqsmith]

Finally managed to get the profilometer long enough to do a scan of a 8x8mm patch near the centre of the switch, but the nickel coating is a bit too reflective it would seem. (It registers as lower than the soldermask )  Didn't have time to redo it as a dual scan with two modulation thresholds so here's one to begin with:

Thanks for looking at it. You can clearly see the groves.   I'll try to scan part of that H-PAK meter.  With that wiper contact removed, we should get a pretty good contrast between the pads that were cycled and the ones that were not. 

[rsjsouza]

When he grabs it and twists the knob it doesn't sound very clicky, not like in your "intro" video.

Would it kill him to open it and take a look instead of just opining?

He did say tens of thousands of cycles per year, and two years, I would expect it to be a little less clicky.  Maybe he will show you the insides as part of the review he mentioned doing.

Hopefully. I too agree that he could have shown more of this particular meter, especially because he mentioned having six of these meters and I can't help but wonder if the tens of thousands could be a collective estimate.

Regarding the "clickiness", I notice a slight difference in the feel between my 4yo NOS U1273A and the brand new U1282A - the latter feels more rubbery. At a certain point I had a U1233A but I recall it had a similar feel as to the U1273A.

Of all the meters I have/had, nothing beats the Fluke 179 in terms of robustness feel. At 11:56 of this video I switch both the 179 and the U1233A and you can hear the sound.
https://youtu.be/0asZe5r-hlg?t=11m56s

[Fungus]

it doesn't sound very clicky, not like in your "intro" video.

He did say tens of thousands of cycles per year, and two years, I would expect it to be a little less clicky.

There's four clickers altogether so the meter would still click a bit even if three of them are broken.

[joeqsmith]

One thing to keep in mind is when I life cycle them, the forks are not real tight fit.  The setup has a fair amount of play, or slop in it.  This was on purpose as I am trying to avoid putting the switches into a bind.   Because of this, as the switch approaches the next detent, it's free to snap into place.   If you are holding the meter and turning the knob by hand, the sound level will be dependent on how tight you are holding it.   

I changed the software slightly so when the test is started, the knob is rotated back and fourth as the current is reduced.  The minimum current required to turn the knob is recorded.   I have no idea how this would equate to torque.  The Keysight meter required a fair amount of current to rotate over the Fluke.

[Paul Moir]

I'm just wondering what kind of job you have that involves turning the knob on a U1231A every 30 seconds, and why their boss hasn't bothered to optimise their task.
EDIT:  Ha, off by an order of magnitude.  But still every 6 minutes seems like a lot.

[Fungus]

I'm just wondering what kind of job you have that involves turning the knob on a U1231A every 30 seconds, and why their boss hasn't bothered to optimise their task.

That's not what bosses do.

EDIT:  Ha, off by an order of magnitude.

I'm guessing his numbers are, too. He wasn't exactly rigorous in his video - didn't even take the back off and look.

Mostly he just said "I don't believe it!", I guess that's why he's a boss.

[joeqsmith]

I'm just wondering what kind of job you have that involves turning the knob on a U1231A every 30 seconds, and why their boss hasn't bothered to optimise their task.
EDIT:  Ha, off by an order of magnitude.  But still every 6 minutes seems like a lot.

He did post a bit about what they do. It's not just about twisting knobs. 

I'm guessing his numbers are, too. He wasn't exactly rigorous in his video - didn't even take the back off and look.

Mostly he just said "I don't believe it!", I guess that's why he's a boss.

 

It's not uncommon for people to not buy into the benchmarks I have been showing.  In this case, he went though the trouble to make a detailed video proving what I show is flawed and as a bonus will do a full on review of it later.  He's put out a fair number of technical reviews.  You just have to wait and see. 

[saturation]

Rotating the switch is part of accelerated life testing and what that means.  You have to prove your model represents real life otherwise skeptics can raise questions.  In toto Joe's video is simply data and viewers need judge what it means.

For mechanical models, if the test cycles generate heat, failure can be premature, because heat generated from friction has to be added to the model projection, as real life usage have periods of long downtime, consider as work only 8 hours of a 24 day as well as real life environmental considerations, field users in Canada are exposed to colder conditions than field users in Florida etc., UV exposure, humidity etc., which affects plastic rate of failure.

If the video data true as quoted below, and a cal certificate is part of the operational costs, the cost of mentioned calibration is high, it even pays to buy a new meter with papers rather than send it out for calibration.  It makes the Keysight meters more attractive and economical with an expected working life limited by the calibration date, at which time its replaced, regardless of wear.

I did ask him to run his own test on the Keysight meter.   This one is 2 years old and gets 10s of thousands of cycles yearly. 

https://youtu.be/_EQdxZK0yHw

[joeqsmith]

I would guess skeptics will be skeptics no matter the amount of data or how it is presented.  Did we go to the moon, is the Earth really flat, did man  make dinosaur bones and bury them to dig up later.  The best I can do is fully disclose what it is that I am doing during these test and as you you wrote, let the viewers determine what value it has to them.  I can't agree with you more that if I were working for a design/manufacture and our goal was to make a model to determine the life of a new product, this is not a trivial thing to do.   That's far outside the scope (and budget) of anything I would ever attempt at home.     
 
I have mentioned a few times now that during my very first attempts to sort out this test, I cycled one fast enough that the heat indeed caused the plastic to yield.  The few I have looked at do not seem to build more than a couple of degrees when cycling them in the three to five second range.  Beyond the real life conditions you mention, for my own use the meters can get exposed to some pretty bad chemicals as well.   Others have also mentioned the connector failure rates.  Someone posted about the leaking batteries causing early failures as well.   

Based on my own experience, I doubt many companies will opt to have an outside cal house generate reports as part of their calibration cycle for their handheld meters. I would assume from his video they do this for every meter on a schedule as part of their ISO but I have no clue what they gain by tracking the drift of their low end handheld meters.