Author Topic: Some thoughts about multimeter input protection  (Read 798 times)

0 Members and 1 Guest are viewing this topic.

Offline deuteron

  • Newbie
  • Posts: 2
  • Country: de
Some thoughts about multimeter input protection
« on: September 24, 2020, 10:03:50 am »
Hello alltogether,
first of all I'd like to say hello to all the participants in this forum. I came across this forum due to the work of joeqsmith and its very usefull multimeter testing videos on youtube. The reason why I'm interested in this stuff is because I'm looking for a couple of new multimeters for our physics lab at the university of Bonn, Germany. I got a lot of informations from youtube review videos stumbling also about this very nice Uni-T ut181a. Some of you are certainly familiar with the input protection problem this meter has. Once again many thanks to joeqsmith for his extensive work not only regarding this meter.
Since I'm a scientist I'm not able to let problems go when they once showed up. So I started may own research into this problem, but this of coarse rather theoretically so far. Before I invest the money for a test meter and have to throw it away in the end, I would like to get some feeling whether I'm able to fix the problem by my own (perhaps with some help of the electriconics department of the institute). During my research I was confronted with three questions I was not able to clarify so far. Therefore I'll put these questions here and I very much hope to get help from the experts in this forum:
1) To my understanding the biggest problem with ESD transients are their very high voltage (up to several tens of kV) together with the very short rise time of these signals in the order of 1 ns and below. This means that the clamping device must have a very short response time. Therefore fast acting MOV's are very often used to do this job as a primary protector. On the other hand also GDT's are also often found at that place. These are very much slower in their response as Joe demonstrated in one of his videos. Nevertheless also meters with GDT's in the frontend section are surviving Joe's tests, strange !
2) I learned about MOV's that they have a considerable capacity of around 1 - 2 nF for low and around 100 pF for high frequencies (1 GHz). Thus the MOV together with the input resistors (e.g. PTC's ~ 1K) form some kind of a low pass filter. For the high frequency components of the transient signal this would lead to a rise time of about 0.1 us and to a cut off frequency (3 db) around 1.6 MHz. This would mean on the one hand that the transient signal seen by the MOV has (instead of what has been said earlier) a much longer rise time compared to the original signal, because the fast components have been cut away due to the low pass filter. Can that be true ? On the other hand I wonder (in that case) why the bandwidth of some meters protected in that way extend to the 100 MHz region for frequency measurents.
3a) As a consequence of (2) I also get a problem with Joe's argument concerning the long trace (V+ - PTC - MOV - common ground) and the inductance being introduced by that. A simple calculation results in about 100 nH for a 10 cm long path, of course a very rough estimation. But let's see what that is doing. With highest frequency of around 1 GHz for the transient, we get a impedance of 630 Ohms. This would result in a potential difference along this path of 630 V per Ampere of current. This is certainly high enough introducing further problems into the upstream electronics. But if thought (2) is correct the impedance would drop to 1 Ohm and thus to 1 V/A which can be neglected in my opinion.
3b) I studied more or less all of Joe's transient testing videos. Interestingly there are at least two meters with similar long path lengths, the Fluke 17b+ and the Hioki DT4252. Both seem to be very robust meters with no problems concerning the 'grill starter test' and further transient testing.
So, whats going on here ? Has somebody any idea ? Are my arguments wrong ? Do I make a serious mistake ?
Many thanks for your thoughts,
Stefan
PS: Perhaps Joe is willing to make some more simulations with his test boards. E.g. one could make a test board with long ground traces and measure the voltage drop along ground. Is the PTC/MOV combination really behaving as a low pass filter ? Also this could be measurable.
 

Offline MosherIV

  • Super Contributor
  • ***
  • Posts: 1526
  • Country: gb
Re: Some thoughts about multimeter input protection
« Reply #1 on: September 24, 2020, 12:42:18 pm »
Hi

Welcome to the forum.

You are over thinking things!
Just get dmm from reputable brands with independantly verified CAT rating.
In no particular order of preference : Keysight, Fluke, Gosen, AVO, Brymen to name a few.
They all have very good input protection. If they cannot withstand a esd, warranty will cover a replacement.

Yes, input protection will limit the freq response of the dmm. Since there primary use id for low freq measurement, high freq response should not be an issue.
Yes you can used the freq measurement function on dmms BUT there are dedicated instruments that will be better at that function.
 

Online joeqsmith

  • Super Contributor
  • ***
  • Posts: 6864
  • Country: us
Re: Some thoughts about multimeter input protection
« Reply #2 on: September 24, 2020, 01:15:33 pm »
The only way I see to solve it is for UNI-T to get involved.  That may happen if enough people provided them with feedback.   

Let's assume that as a educator, you're smart enough not to involve your place of employment and decide to experiment on your own.  In the case of the UT61E, I had spent some time trying to explain the problems with the meter and possible ways to improve it's electrical robustness.   It wasn't my goal to provide a solution for the uneducated beginner to copy an paste but rather to show how to go about solving problems like this.   I had a few people wanting to make up parts lists and such, which IMO is being negligent as you are putting that group (I suspect made up mostly of children and young adults) at risk.   I can't control who watches the videos but I can control the content. 

Let's assume you took the time to watch the videos for the UT61E and understood what was being shown.  I talked about the MOVs, their capacitance and inductance.   I also demonstrated what the transient looks like on the IC that would normally be damaged as I made changes to the circuit.   I further explained how the changes I made were a trade off and how there may be better ways to solve it.   

As I demonstrated, the UT181A's layout could be improved.  UNI-T could also possibly change other parts of their design to improve the meter's robustness.   I'm only showing one technique.  In the end, while the meter may be electrically more robust,  I consider it less safe.  None of my changes have gone through certification.  The videos again are for educational purposes and may offer some very small amount of entertainment.

If you wanted to experiment with RF, I recommend getting one of the NanoVNAs.  I have not looked at OWOs new V2 but suspect that would be what you are after.   You would soon learn that wires are not 0 ohms.  Everything has a complex model.   I've been a professional designer for 40+  years now and the area of high speed design has always been very interesting.  I retired the day I entered the work force.       

« Last Edit: September 24, 2020, 01:19:14 pm by joeqsmith »
How electrically robust is your meter?? https://www.youtube.com/channel/UCsK99WXk9VhcghnAauTBsbg
 
The following users thanked this post: SeanB, Marco1971

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 12045
  • Country: us
  • DavidH
Re: Some thoughts about multimeter input protection
« Reply #3 on: September 24, 2020, 10:43:00 pm »
So, whats going on here ? Has somebody any idea ? Are my arguments wrong ? Do I make a serious mistake ?

I think what is going on is that the last couple decades, multimeter designers have forgotten how multimeter input protection was implemented previously, and have no idea what they are doing.

How would you implement multimeter input protection with no MOVs, GDTs, and polyfuses?
 

Online joeqsmith

  • Super Contributor
  • ***
  • Posts: 6864
  • Country: us
Re: Some thoughts about multimeter input protection
« Reply #4 on: September 24, 2020, 11:05:21 pm »
I looked at an old Fluke 77 which has some protection but nothing like a modern meter.   The UT181A the OP is asking about has some very nice parts for the front end with the typical clamp techniques you would see on most higher class meters.
How electrically robust is your meter?? https://www.youtube.com/channel/UCsK99WXk9VhcghnAauTBsbg
 

Online bdunham7

  • Super Contributor
  • ***
  • Posts: 1250
  • Country: us
Re: Some thoughts about multimeter input protection
« Reply #5 on: September 25, 2020, 02:56:09 am »
Two random thoughts that probably have been addressed elsewhere:

To my understanding the biggest problem with ESD transients are their very high voltage (up to several tens of kV) together with the very short rise time of these signals in the order of 1 ns and below.

You need halfway decent oscilloscope probes to get a 1ns-rise signal into a scope, and with 10X attenuation at that.  1X seems impossible unless you use a direct connection to a low impedance source.  How does an edge that fast get into a meter if you  are using standard test leads? 

Quote
Nevertheless also meters with GDT's in the frontend section are surviving Joe's tests, strange !

Keep in mind that MOVs will clamp at a level that represents a significant overvoltage relative to the rated maximum voltage, while the GDT will essentially short and clamp to a very low voltage.  If the downstream components can absorb a little energy and withstand the front of the transient until the GDT can start conducting, the GDT will actually do a better job controlling the overvoltage.
A 3.5 digit 4.5 digit 5 digit 5.5 digit 6.5 digit DMM is good enough for most people.
 

Online joeqsmith

  • Super Contributor
  • ***
  • Posts: 6864
  • Country: us
Re: Some thoughts about multimeter input protection
« Reply #6 on: September 25, 2020, 12:04:46 pm »
...
To my understanding the biggest problem with ESD transients are their very high voltage (up to several tens of kV) together with the very short rise time of these signals in the order of 1 ns and below.
You need halfway decent oscilloscope probes to get a 1ns-rise signal into a scope, and with 10X attenuation at that.  1X seems impossible unless you use a direct connection to a low impedance source.  How does an edge that fast get into a meter if you  are using standard test leads? 
...

I am sure the leads would have an effect but we have no way to control what leads (custom or otherwise) are being used.   So I make them as short as possible as again, my goal was to test the meters front end.   I also don't know what wave shapes would damage a given meter.   I can tell you that the BBQ starter's waveshape is nothing like the standards.  The gun I designed on the other hand is much closer.   I had planned to test with the gun using some sort of leads but by that point, I was looking at a much higher class of meter.   It's rare a meter won't survive that BBQ starter.   The last one was Dave's 121. 

Here you can see the gun being used with some longer test leads to interface with a cheap meter.   I am sure the leads would change the wave, but the meter was still damaged.
   
How electrically robust is your meter?? https://www.youtube.com/channel/UCsK99WXk9VhcghnAauTBsbg
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 12045
  • Country: us
  • DavidH
Re: Some thoughts about multimeter input protection
« Reply #7 on: September 25, 2020, 05:45:34 pm »
Keep in mind that MOVs will clamp at a level that represents a significant overvoltage relative to the rated maximum voltage, while the GDT will essentially short and clamp to a very low voltage.  If the downstream components can absorb a little energy and withstand the front of the transient until the GDT can start conducting, the GDT will actually do a better job controlling the overvoltage.

Zener (avalanche) diodes are very effective and fast but GDTs have the advantage of zero leakage which is important in high impedance circuits.

It is also important to *not* clamp a high impedance input in such a way as to produce a low impedance input because the high current will destroy the protection circuit which still counts as a failure.  So a GDT cannot just be placed across the input.

So the universal protection circuit comes down to using a series resistance to limit the current before low leakage shunt protection.  Usually the shunt protection uses low leakage diodes which are themselves clamped with a zener (avalanche) diode or other source of low impedance; this has the effect of isolating the zener (avalanche) diode's leakage from the circuit.  More protection is provided by using a series resistor followed by a GDT followed by another lower series resistance and then low leakage diode shunt protection.  The diode shunts never see any voltage higher than the GDT breakdown voltage so there series resistance is selected accordingly.

If the circuit to be protected can sustain the breakdown voltage of a GDT without damage, then only the series resistance and GDT are needed but this is rare these days.  In the past with vacuum tube input stages, the series resistance and GDT in the form of a neon lamp was all that was necessary.

Note that the first series resistance *must* be able to sustain high peak voltage without breaking down.  Common resistors are not suitable unless used in compensated arrays.  In the past carbon composition resistors were typically used but suitable high voltage thin and thick film resistors are now available.  Ceramic composition resistors may also be suitable.

To maintain AC performance, the series resistors are bypassed with capacitors but this lowers the protection rating as frequency is increased, leading to the common voltage derating versus frequency curve.

Protecting outputs is a more challenging exercise but it is very achievable.  On a multimeter, the ohms mode has a current output which must be protected connected in parallel with the high impedance voltage input.  Some old multimeters use an high voltage incandescent bulb as a fast response PTC thermister.  A high voltage depletion mode MOSFET acting as a current limiter is also useful here.  Or design the current source with very high voltage compliance which only has to go one way.
 

Offline Gyro

  • Super Contributor
  • ***
  • Posts: 6015
  • Country: gb
Re: Some thoughts about multimeter input protection
« Reply #8 on: September 25, 2020, 07:30:06 pm »
Note that the first series resistance *must* be able to sustain high peak voltage without breaking down.  Common resistors are not suitable unless used in compensated arrays.  In the past carbon composition resistors were typically used but suitable high voltage thin and thick film resistors are now available.  Ceramic composition resistors may also be suitable.

Some solutions are more suited to larger bench meters than handhelds. Here's the standard (old) Datron approach, an 88k 12W Carbon composition resistor chain which allows clamping to be achieved by a combination with low leakage jfets and zeners. It allows the meter to withstand 1kV continuous on the lowest 10mV range. The green matched set is the relay selected 10M attenuator for the high voltage ranges.
« Last Edit: September 25, 2020, 07:47:10 pm by Gyro »
Chris

"Victor Meldrew, the Crimson Avenger!"
 

Offline Wytnucls

  • Super Contributor
  • ***
  • Posts: 2962
  • Country: be
Re: Some thoughts about multimeter input protection
« Reply #9 on: September 26, 2020, 03:50:12 pm »
Here is the UT181 EMC certification (Intertek):
EMC directive/2014/30/EU

https://www.uni-trend.com/uploadfile/2018/1129/20181129043348137.pdf

For info: The contact discharge (4kV) is applied to non-conductive surfaces on the meter, not the terminals!

Check below the IEC test procedure for EMC:
« Last Edit: September 27, 2020, 08:51:10 am by Wytnucls »
 

Offline deuteron

  • Newbie
  • Posts: 2
  • Country: de
Re: Some thoughts about multimeter input protection
« Reply #10 on: October 12, 2020, 09:41:19 am »
Dear all,
thank you very much for all your replies. I've been on a quite long business trip and were not able to connect to this forum during that time.
I will study your posts come back to here.
Stefan
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf