Handheld meter robustness testing

[Fungus]

For CAT II, they say less than 10kA typical.   CAT III, less than 50kA.   CAT IV, much greater than 50kA.   Wow, 50kA!!   Dave's ASTM 11A fuses are marked 30kA.  I wonder if these would be fine for a CAT IV environment. 

Hence the fuse in the (CAT IV 1000V) Amprobe HD160C:

[joeqsmith]

That's my guess.  The manual for the BM869s (CAT IV 1000) shows 10kA for the 400mA fuse and 20kA for the 11A.  I'm not sure what it actually has in it.  I've not had it apart since I made the repairs. 

[joeqsmith]

I think Cliff wiped out an interesting discussion between a YT'er and myself.  Basically this person was posting about how there was no value in these tests and how I had ran the Fluke 87V far above it's ratings.  When I asked them about it, they responded how the meter was only rated to some KVRMS.   I attempted to try and find out how they came to that conclusion and they responded with DC is the same as RMS and because I was just charging up some caps with DC then putting that across the meter, it was DC.   I think they believe it is something  like randomtronic's Mr Joules.  It made no sense that anyone with any sort of electronics or electrical aptitude would come to this conclusion.   

I was hoping they would respond but it appears YT'er or Cliff may have pulled the thread.  Not a big deal, except I now wonder how many other people think that this is what is going on. 

[tautech]

Ignore them Joe. You've got approaching 400 thanks for your efforts and I'd bet the vast majority for fine work in this thread. Those that can't be bothered taking some time to read and watch why you embarked on this mission need not be worried about.
Drag them onto the forum and into this thread, we'll sort them out. 

[joeqsmith]

I have been able to help a small percentage of people understand what was going on.  If they get to be too much of a drag, I ban them.  Pretty rare that happens. 

When I was looking at that Gossen, someone had asked me about the capacitance accuracy when testing bad caps.  The were concerned with high ESR.   I don't have a supply of bad caps so I put some series resistance in-line with a good cap and tried it.  Meter did quite well.

Now that I have been playing with this pre-production 121GW, I decided to look at the capacitance measurement in more detail. I seldom use this feature and just assumed meters were not that accurate.  Much to my surprise, the meters (eight so far) I looked at all tracked fairly close.   The speed that they lock in varies a lot and so does the number of digits.   The UT61E shows off an extra digit.  Not suggesting it has any meaning but it's there.   

I can't see replacing my VNA or RLC meters just yet.   

[joeqsmith]

Figured I would post the data I collected from my capacitor test.   The first two columns are my BK RLC meters. 

The Fluke 115 and HIOKI DT3252 have the lowest res of all the meters.  1nF is it.  The CEMs and UT61E have the highest res.  Surprised at how well the Fluke 101 did compared with ever other Fluke. 

Keep in mind, again most of these meters have been damaged and repaired by me.  They were not realigned and may no longer be in calibration.  The 121GW being pre-production and Dave had stated it may not have been in cal, does fairly well for being damaged twice now.  Poor thing.   And while I have not ever damaged the UT61E, that thing is far from factory original.   Was a little concerned the shield I had added to the Gossen may have hurt it but seems to have had no effect. 

Speed is all over the place and is range dependent for some meters.   For example, taking the 2200uF capacitor, shorting it, then connecting the meter to it. 

Gossen Ultra required 17.3 seconds 
CEM DD9939 required 16.1 seconds
EEVBLOG 121GW was 8.19 seconds
UT61E was 2.8 seconds

Fairly wide spread.  For me, the resistance auto-range would be far more important. 

[iamjanco]

I have been able to help a small percentage of people understand what was going on.

Thanks, Joe. As an albeit old-timer, but noob here looking for a good, reliable logging meter that won't leak all over me nor suffer my static cling, I found the information in both this thread and your videos invaluable, as well as very well done (that's coming from someone who's fairly anal).

Feel free to up that percentage by whatever decimal place you feel appropriate. Keep up the great work!

[joeqsmith]

I have never bought the software and cable for the BM869s.  I really have no idea how it would work or if you could use your own software with it.   

[IanB]

I have never bought the software and cable for the BM869s.  I really have no idea how it would work or if you could use your own software with it.

I have bought the PC interface cable for this meter. The supplied software is OK and does the job. You get a projection of the meter display, a graph, and a saved copy of the history. All the standard things you would expect the software to do.

Unfortunately the interface does not create a virtual comm port, it instead uses an HID interface. Therefore to use your own software you have to create a suitable HID interface driver, which compared to programming a comm port is rather inconvenient and messy. It is something I have started to investigate but have not actually done.

I am rather disappointed that Brymen chose to design the interface cable this way. I would much rather have an RS232 type serial interface that is easy to program and universally accessible by any hardware.

[deflicted]

I have never bought the software and cable for the BM869s.  I really have no idea how it would work or if you could use your own software with it.

I have bought the PC interface cable for this meter. The supplied software is OK and does the job. You get a projection of the meter display, a graph, and a saved copy of the history. All the standard things you would expect the software to do.

Unfortunately the interface does not create a virtual comm port, it instead uses an HID interface. Therefore to use your own software you have to create a suitable HID interface driver, which compared to programming a comm port is rather inconvenient and messy. It is something I have started to investigate but have not actually done.

I am rather disappointed that Brymen chose to design the interface cable this way. I would much rather have an RS232 type serial interface that is easy to program and universally accessible by any hardware.

What OS? If we're talking about Linux, I might be able to help with that, since I've done similar projects at work. I don't have a BM869 to test with, but I do have the BM257, and the comm kit ordered. If the BM257 interface is similar enough, I might be able to write something that would work with the BM869. Of course, all of this is speculation at this point. I haven't read the interface spec for either meter, and I don't even have my own comm kit yet for my BM257 - it's still somewhere in Hong Kong.

If you're talking about Windows, well, then good luck. I've never written a Windows driver, so I'd have to climb that learning curve from the very beginning.

[Fungus]

Unfortunately the interface does not create a virtual comm port, it instead uses an HID interface.

Thus guaranteeing incompatibility with a large percentage of the computers in the world.

(and it will only work with the others for as long as Brymen can be bothered to update the drivers, you're on your own as soon as they stop)

[stj]

but is the HID interface in the meter, or the cable??
most meter i.c.'s just stream the lcd segments every time the display is updated.
if some smartass put a microcontroller in the cable to re-format the data, then you could build a different cable.

[IanB]

but is the HID interface in the meter, or the cable??

It's in the cable. When you plug the cable into a USB port it creates an HID device on the host computer.

most meter i.c.'s just stream the lcd segments every time the display is updated.

Indeed, this is what the 869s does too.

if some smartass put a microcontroller in the cable to re-format the data, then you could build a different cable.

Someone did post here about creating their own cable from scratch. They reverse engineered the optical protocol at the meter and transmitted the data to the computer over a serial interface.

[joeqsmith]

I wonder if they don't have a mechanical switch in there for the leakage.   Note the mechanical breaker is still present but I suspect turn off the TRIAC when the current reaches the set point (calling that their programmable breaker).

I did not transient test the killawatt I had but I did have it apart.  Cheap, like you would expect for the price.   

Joe, just for a lark, can you put a few USB wall warts ( not the cheap ones, but ones from reputable manufacturers, because we all know how the cheap ones will fail anyhow) across your line simulator and see what it takes to kill them. My guess is that almost all of them will fail 2kV, but just how badly they do this, and how much they smoke doing so. Most will probably work fine with your test voltage, being universal input devices. You will just need a sacrificial USB voltage monitor and a resistor to draw 100mA to 500mA out of them during the test. Pretty much how they behave in transient conditions, and what a simple mains transient like a motor disconnect on the same branch circuit will affect them.

I would guess they fall under the generic case and are only required to be tested at 1KV line-to-line.   

Again, that sub 20J I test with is basically nothing.  I don't see it ever causing a meter or other device to explode or put out much for smoke.   Really what you would want to do is run these on a real combo generator if you wanted to see something more like the Fluke multimeter videos where they show the cases coming apart.  This is nothing close to what I have setup.

True, but you do have a good audience, and it will be a good thing to show that cheap and nasty really is that, even with limited energy. I think the saving grace for most products these days is the thin wire they use, which acts as an impromptu fuse when the thing fails short circuit. As the fuse is often the thing "designed" out by the cost cutters, and the wire is also "value optimised" to be as thin as possible and as low grade copper as possible this probably is the only thing keeping the fire rate down.

The transient generator will not put out the transient superimposed on the AC line.  Actually, I can't generate an AC wave with it, only the full rectified 220V.   Assuming you understood this, I am not following your thoughts of having a load and monitoring their outputs.   Could you explain this?     

I had attempted recently to try and understand a trolls comment about me applying a set of capacitors directly across the meters when I test them.  I found the comment very strange as I have shown the open circuit waveforms MANY times.  Anyone with some basic skill would understand that something is causing that decay and it's obviously not the meters.   The meters have a very high input impedance so normally all that energy is actually dissipated internal to the generators output network.  So as long as the meter does not breakdown, there will be very little that happens.   It could be a 10KJ supply and it would make no difference.  This is why you never see me run the half cycle generator unless a meter breaks down with the new generator.  Nothing will happen unless something does breakdown, then we get a little punch.  I had made a video about that generator, basically I was trying to simulate a half cycle of the CAT III 600V environment. 

The chargers on the other hand are not going to be designed for industrial use.  Not that all meters are.  I like Gossen's several comments about me running an industrial meter at 10V/m which is pretty much the industrial standard.  They test at 3 which is what my toaster is rated for.       Anyway, these chargers would not handle a hit from that half cycle generator because again, this is trying to simulate a much higher levels than the chargers are designed for.    Also, the chargers by nature are a low impedance device.  After all we are trying to charge something with them.   So the generator may not put out enough energy to do anything.   It's like this people who post, I connected my Fluke to a 40KV supply and it lived.  Then you try to get them to explain the sources output impedance....    It's an endless supply of humor.

Point of going over all that again is that I am trying to see how you would propose that we actually run these USB supplies using this setup.  I have no problems running a few but want you to think about what it is you are asking. 

[SeanB]

Load them with 100mA or less, just basically to get some of the less regulated ones not actually providing a 5v output, as I have seen many that will be 6V with no load and then drop with load. You DC source of 220V will be enough hopefully to provide this 0.1W of power to the DUT, and then the single pulse will easily show how they fail at providing creepage and spike resistance. Load could be basically a LED and a series resistor, and then a 68R resistor to get to around that 100mA, though I would hesitate to connect any meter you like to the secondary side, unless it will survive the full test voltage itself, or you have a big supply of those HF free meters to sacrifice on the secondary.

[joeqsmith]

Load them with 100mA or less, just basically to get some of the less regulated ones not actually providing a 5v output, as I have seen many that will be 6V with no load and then drop with load. You DC source of 220V will be enough hopefully to provide this 0.1W of power to the DUT, and then the single pulse will easily show how they fail at providing creepage and spike resistance. Load could be basically a LED and a series resistor, and then a 68R resistor to get to around that 100mA, though I would hesitate to connect any meter you like to the secondary side, unless it will survive the full test voltage itself, or you have a big supply of those HF free meters to sacrifice on the secondary.

Interesting.  You think these would all run off DC?  Or have you actually tried this with a few of them?   I was thinking some of the lower end ones would not have a transformer in them.    The generator is programmable for the amount of current it can supply. 

I would have no concerns about hooking most of the remaining meters to the output to monitor.  It won't be any worse than what I have put them through!    

Removed video reference

[joeqsmith]

It seems like these few comments belong in this thread. 

I think it's important to also understand Joe's tests in context.
The Fluke 87V, the most trusted meter on the market, fails every single one of Joe's tests. According to Joe's tests it's one of the worst meters on the market. Yet I doubt there is a single 87V owner ever who has seen their meter die due to any ESD or pulse overload etc.

I agree.. most of the tests are a worst case scenarios (black swan events). 87v is a standard when it comes to rugged meters but has failed many of Joes tests. In fact Fluke 101  has passed more tests then any of the more expensive meters.

For me it's simply a matter of has a meter passed independent safety testing (UL, ETL etc). If so then it's good enough to recommend and use it on anything it's rated for.
Sure, if a meter is failing ESD testing or something that could potentially be common place, then that may be a cause for concern, but even the Fluke 87V has shown no sign of doing that in practice for the 13 years it's been released as the V series, apart from Joe's test.

So let's look at the spreadsheet.  If it's not clear, NT - Not Tested.   I had attempted to run an 87V at the same transient level that the Fluke 101 had survived to.  13KV peak, 100us FWHH and the meter failed.    I think I lost an opamp and a few clamps from that test.   Lot's of confusion about the result.  Back then I was not doing anything for the AC line, HV DC/AC or ESD.  The tests evolved after running those first set of $50 meters. 

Later I ran an 87V a second time.  Again, I did not run it on the high voltage AC/DC supply.  I also did not run the rectified AC test or try any ESD test with the grill starter.   I ran it at 1KV and it passed then switched to 1.5KV where it was damaged.   So I'm not sure what all this stuff is about it failing every single one of Joe's tests.  Again, spreadsheet and videos are on-line, you just need to look at the spreadsheet to see what was done. 

The 87V is a very old design and failed at the lowest levels of every Fluke meter I have looked at.  Maybe they learned a few tricks over the years to improve their products robustness.  Not working for Fluke, I really have no idea.  One day I may try to buy a brand new off-shore one and see if it does any better. 

I am not sure how to respond to the comment, "According to Joe's tests it's one of the worst meters on the market".  I guess that would be Dave's interpretation of the data, which based on the previous comment, I'm not sure he understood or even looked at.   Personally, my own view of the data I have collected is if your meter fails that grill starter, I don't care what it costs, those are the bottom of the barrel. 

I don't work for Fluke and have no idea about the number of returns for the 87V or the root cause of failures. 

One way to justify the level of performance is to base it off another perceived higher end product with roughly the same performance.  Fluke has really seemed to step up their own game with the newer designs I have looked at rather than holding the bar.  IMO, this is good on Fluke as a whole they are making some of the most robust meters out there today. 

Many times I have stated that for my home hobby use, I am not too concerned with the safety standards because I don't use handheld meters like these in a CATIII and up environment.   For me, passing the 61326 EMC standards is far more important.   The pre-production 121GW I have calls out the safety but not the EMC standard.   

In fact Fluke 101  has passed more tests then any of the more expensive meters.

Sorry, I missed responding to this one.  Again, the following is right off the spreadsheet:   

The Fluke 101 cost $48 and has been tested to 12KV 50us FWHH and to 13KV 100us FWHH and was not damaged. 
The Fluke 107 cost $117, was tested to 14KV 100us FWHH and was not damaged.
The Brymen BM235 cost $130, was tested to 12KV 50us FWHH and was not damaged.
The Fluke 115 cost $132, was tested to 12KV 50us FWHH and was not damaged.
The HIOKI DT4252 cost $150, it was tested all the way up to 14KV 50us FWHH after I added some plastic to extend the length of the barrier.  Again this meter was an oddball as the plastic had started to melt.  It was not an electrical failure like most.

Almost forgot about the most expensive handheld sitting here
The Gossen Metrawatt Metrahit Ultra Bluetooth M248B cost $850, , was tested to 12KV 50us FWHH and was not damaged.


So that's three four meters that cost more and are hanging right there with the 101.  That IS the fact!

[Cliff Matthews]

Quote from: joeqsmith on June 29, 2017, 10:10:26 PM

I think Cliff wiped out an interesting discussion between a YT'er and myself.  Basically this person was posting about how there was no value in these tests and how I had ran the Fluke 87V far above it's ratings.  When I asked them about it, they responded how the meter was only rated to some KVRMS.   I attempted to try and find out how they came to that conclusion and they responded with DC is the same as RMS and because I was just charging up some caps with DC then putting that across the meter, it was DC.   I think they believe it is something  like randomtronic's Mr Joules.  It made no sense that anyone with any sort of electronics or electrical aptitude would come to this conclusion.   

I was hoping they would respond but it appears YT'er or Cliff may have pulled the thread.  Not a big deal, except I now wonder how many other people think that this is what is going on. 

>

Ah.. For the record, I deleted my first post (re: corona doping in the 121GW) because I was pissed at the (nameless for now) idiot calling you a retard. Clearly, if he had any clue of the kind of work you've done on the channel, he needed his ass kicked - enough said.

[joeqsmith]

If we had saved that one comment, I would post the whole thread here, unedited.  Not to call them out for bad choice of words or for the personal attack but because I find it interesting that someone would conclude from everything I have posted that I am directly attaching charged capacitors across these meters.    

Sure, if a meter is failing ESD testing or something that could potentially be common place, then that may be a cause for concern, but even the Fluke 87V has shown no sign of doing that in practice for the 13 years it's been released as the V series, apart from Joe's test.

But again, even Dave suggests that I ran some sort of ESD test on the 87V and it failed.  Again, nothing of the sort ever happened.

[joeqsmith]

AC and DC safety from the ARRL manual. 

[joeqsmith]

The RadioShack, catalog number 22-813 

https://www.youtube.com/watch?v=r3cKTbtOdfg&feature=youtu.be

[mikeys]

How about testing these two rebranded Brymen pocket meters Joe? They seem to have decent input protection for the form factor. Might be a bit of a hassle for you since they don't have sockets, though. Cat III 300v rated.

http://isswww.co.uk/amprobe-pm51a-pocket-dmm-w-freq-and-capacitance

http://isswww.co.uk/amprobe-pm55a-automatic-precision-pocket-dmm

[EEVblog]

But again, even Dave suggests that I ran some sort of ESD test on the 87V and it failed.  Again, nothing of the sort ever happened.

Ok, I stand corrected.
I could have sworn that you said this in a comment somewhere, that's what I was going off.
And above you say the 87V died at 1.5kV, but that's not on your spreadsheet?

[Fungus]

And above you say the 87V died at 1.5kV, but that's not on your spreadsheet?

Note that there's two Fluke 87s on the spreadsheet.

[floobydust]

joeqsmith, I'm curious about what kind of capacitance/series resistance is behind your transient generator?

I have used Teseq surge generators but man are they a fortune; over $26K USD plus CDN and yearly calibration was few more grand. Just super expensive, so I can see no companies spending this much on equipment for a single test.
I use lower (i.e. 12ohm vs 2ohm 61000-4-5) series resistance until I know the design has a chance.