New UNI-T UT892 2000V AC/DC High Voltage Multimeter

[Kleinstein]

The live terminal is supposed to be for some non contact testing. Normally that function is more inside the case and not with a terminal.

The PTC for the high voltage input is odd: it would hardly help. I very much doubt it could withstand a higher voltage that chain if 2.5 M resistors. The main means of the protection are the resistors and they could have used more of those instead of the PTC. The PTCs are already a bit boarderline at 1000 V.

For the more normal meters there usually is a PTC (or multiple in series) for the protection of the ohms current source. There may also be one to protect the voltage sense path for the ohms function that can be shated with a mV function. The point here is that the function would go through the switch and there is a chance for arc over at the switch (or extra spac gaps) that may need extra protection. A decent CAT rating may want some protection for the case of arc over at the switch case, at least as a backup for transisents.
The big simplifcation from the extra terminal for just high voltage but no ohms is that the signal does not have to go through the switches.

There is some sense in not reusing the 10 M from the normal range: the normal range does have some connection to the switch for the ohms mode and this adds a little bit of leakage and capacitance. So A totally seprate chain of resistors makes sense for better accuracy and the resistors look relatively cheap, even if they would use 20x1 Meg instead.

[Fungus]

I'd want a much higher input impedance for that.

CAT rating does not mean higher input impedance, it means higher clearance/creepage and higher power rating of passives.

I don't want it from a safety point of view, I want it from a measurement point of view. eg. Measuring the voltage at the output of a Cockroft-Walton generator can be delicate work.

[Gyro]

It looks as if there would be sufficient space to insert a strip of insulating material down the side of the PCB, retained by the front and rear, to bolster the minimal isolation of the case flange. I think the 2.5M resistors are probably still above their element limiting voltage though.

[PartialDischarge]

It looks as if there would be sufficient space to insert a strip of insulating material down the side of the PCB, retained by the front and rear, to bolster the minimal isolation of the case flange. I think the 2.5M resistors are probably still above their element limiting voltage though.

The case has a big lip + the outer silicone cover. If you are worried about that I’d simply put silicone over the string. Also the mod that I did makes it safer since now there is 800kohm right after the jack.

[Gyro]

I don't want it from a safety point of view, I want it from a measurement point of view. eg. Measuring the voltage at the output of a Cockroft-Walton generator can be delicate work.

That's when it's a good time to build an external 100M+ divider (that's what we in the 20K Ohms-per-Volt club would do!)

[Edit: Yes, I know 20k OPV at 2kV is only 40M]

[Gyro]

The case has a big lip + the outer silicone cover. If you are worried about that I’d simply put silicone over the string. Also the mod that I did makes it safer since now there is 800kohm right after the jack.

No critisism, just interested in how / why they've implemented things.

[Fungus]

The case has a big lip + the outer silicone cover.

Yep, you're in much more danger from holding the probes than from anything that might jump out of the case.

As noted though, it's an obvious fix and would have cost them nothing.

Measuring the voltage at the output of a Cockroft-Walton generator can be delicate work.

That's when it's a good time to build an external 100M+ divider (that's what we in the 20K Ohms-per-Volt club would do!)

You still need a very high impedance meter to measure the voltage.

(eg. A Fluke 8060A - is there anything that meter doesn't do well?)

[Mechatrommer]

So I decided to modify the meter, since I don't know how good those 2.5M resistors are, I replaced them with 2.4M ones that I have. Also I removed the PTC and put there the rest of the resistance minus a bit to compensate for the low 0.2% reading. Now it is measuring +-1V.

there you go a high voltage nuttery genius came to play clever than the designer... you just replaced 4 digit (more precise) resistors with cheaper 3 digit batch, how well you know your 2.4M batch? do they cost like $1 each? with rubbery stuffs surrounding their pins? and rated 2KV each? 8x of normal 0805 or 1206 is good for 1.2-1.6KV isolation theoritically (physically) https://www.vishay.com/docs/28705/mcx0x0xpro.pdf, there are more components downstream to provide extra isolations. and you just replaced a probably $1-5 PTC part that probably can save the DMM's or yours ass in case of downstream breakdown with $0.01 parts that provide no protection at all and much higher dielectric breakdown risk. you've probably screwed the factory's hygienic grade too.

btw for those who asked why those dividers are so to the side, with concern that they are near to the user's side (with lips and extra rubber holster as isolation, wouldnt that be enough?). well i can think of that... creepage on pcb is real, and moving them closer to other components will increase risk of that, since they want to cheap out by not making extra costly slots between them. so they can achieve $40 price point that beginner users love. ymmv. https://www.tempoautomation.com/blog/understanding-pcb-creepage-and-clearance-standards/

[PartialDischarge]

So I decided to modify the meter, since I don't know how good those 2.5M resistors are, I replaced them with 2.4M ones that I have. Also I removed the PTC and put there the rest of the resistance minus a bit to compensate for the low 0.2% reading. Now it is measuring +-1V.

there you go a high voltage nuttery genius came to play clever than the designer... you just replaced 4 digit (more precise) resistors with cheaper 3 digit batch, how well you know your 2.4M batch? do they cost like $1 each? with rubbery stuffs surrounding their pins? and rated 2KV each? 8x of normal 0805 or 1206 is good for 1.2-1.6KV isolation theoritically (physically) https://www.vishay.com/docs/28705/mcx0x0xpro.pdf, there are more components downstream to provide extra isolations. and you just replaced a probably $1-5 PTC part that probably can save the DMM's or yours ass in case of downstream breakdown with $0.01 parts that provide no protection at all and much higher dielectric breakdown risk. you've probably screwed the factory's hygienic grade too.

Lol, I was expecting this kind of comment. I’m not going to explain you anything cause you like to assume and it seems that already know everything, let me just tell you that I design electronic HV stuff for power distribution, and I mean up to 36kV not the 200V people usually refer to HV. You haven’t even noticed what my username means, cause well you don’t know much about HV. Well you also haven’t even noticed that I took 40k off of the resistive string and are telling me about the precision of the resistors 

[Mechatrommer]

...let me just tell you that I design electronic HV stuff for power distribution, and I mean up to 36kV not the 200V people usually refer to HV. You haven’t even noticed what my username means, cause well you don’t know much about HV.

Lol I was expecting that kind of comment. btw, what matters my guru? if its 2000V? or 2004V? will it make any difference in your VHV circuit? If i'm the one who do your mod... wait for Joe the Que Smith to do the bashing or at least lets wait for his comment on that cheapy uni-t circuit.

[Mechatrommer]

and i dont have to show you whats mine to prove anything... you've concluded...

...cause well you don’t know much about HV.

thats enough to make you feel happy. for now, i'm looking forward to hear whats mr joe have to say about this... cheers.

[joeqsmith]

and i dont have to show you whats mine to prove anything... you've concluded...

...cause well you don’t know much about HV.

thats enough to make you feel happy. for now, i'm looking forward to hear whats mr joe have to say about this... cheers.

I would have no way of knowing anyone's level of intelligence except from their posts.  That could be misleading. 

Who would trust Uni-T for safety or minimal honesty?
...  Have to see a teardown and BBQ lighter test lol.

We could get one and treat it as a special case, cause it's a special meter.  Maybe just test the 2kV input only.   I ran one of their high voltage sticks.  It died fairly early on as well. 

I’ll do that when it arrives with my Hipotronics HD140

Assume this was done before mods to meter.   I'm still waiting on results.

So I decided to modify the meter, since I don't know how good those 2.5M resistors are, I replaced them with 2.4M ones that I have. Also I removed the PTC and put there the rest of the resistance minus a bit to compensate for the low 0.2% reading. Now it is measuring +-1V.

+/-1V of reading or full scale?  What are you using to make these measurements? 

Are you planning to do any other testing with it?

It looks as if there would be sufficient space to insert a strip of insulating material down the side of the PCB, retained by the front and rear, to bolster the minimal isolation of the case flange. I think the 2.5M resistors are probably still above their element limiting voltage though.

The case has a big lip + the outer silicone cover. If you are worried about that I’d simply put silicone over the string. Also the mod that I did makes it safer since now there is 800kohm right after the jack.

If I understand your first post about your modifications, you were concerned about the original 2.5M resistors being unsafe and the reason you made these changes was to improve safety?  Your feelings that the accuracy was improved was a side effect? 

If this is correct, how did you determine that these mods are safer?   

I would like to see the backside of the PCB if you took any pictures when you made these modifications.   

[PartialDischarge]

The accuracy I did test full scale, although I quickly changed voltages and linearity did seem good, as mentioned I have not had much time. These is done with a Fluke 410B calibrated with a 3458A and Fluke 80E divider.
No, the HD140 still have not used on the multimeter since I have it at another place.

I don't think the meter is unsafe and the resistors were not changed for that reason. They were changed because not all 1206 resistors are HV and I don't know what the ppm of those resistors are. Plus that PTC is useless and the sooner there is high resistive impedance in a long PCB trace the better, less corona, less prob of "problems". So I'm happier with  a 760k resistor up front rather than with a 1.5k PTC.

I'll do more tests as time permits, but duty is first!

[joeqsmith]

Sounds good.  Look forward to seeing some additional data from it. 

[floobydust]

You can't tell if the multimeter is safe without doing a spacings analysis. Commonly there is an unexpected arc path- at the rotary switch, to a mounting screw or nearby part etc. We haven't seen the other side of the PC board.

Uni-T is doing their usual cheap ass front end - no MOV, no GDT's, no surge resistor. PTC's are way too slow 10's msec to heat up- leaving the tiny SOT-23 clamp diodes. Hello BBQ lighter lol.
Look at the PCB and see the option for a larger SOT-89 clamp transistor Q1A instead of Q1+Q2. Thanks for leaving that one out.
They are fully capable of making a decent multimeter, they do this for OEM's but not for their brand name. Just stupid.

I think MELF resistors are best for their voltage rating (operating and impulse). With 0805/1206 you run into the limitations with the pad spacing, and worse with a zig-zag because the spacing every four resistors is low.
The VCR (voltage coefficient of resistance) is surely terrible with cheap parts. I think that would be an issue but there's not a lot of information about that.

[PartialDischarge]

So here are some more tests

For the 1000V input, AC test up to 2kVrms, there is a limiting device that trips at around 1.9kVrms:



For the 2000V input, AC test up to 4kVrms, no problems, although this is with the 760k resistor up front instead of the original  PTC:
 



For the 2000V input, AC test up to 5kVrms, no problem, current meter reads 250uA as expected, I don't really see the point in going further, 5kVrms is 7kV peak.

[joeqsmith]

Thanks for taking the time to make these clips.  Would you consider getting a second meter to repeat the tests without the mods?  May be interesting to see you sweep the voltage as well.   

[PartialDischarge]

Thanks for taking the time to make these clips.  Would you consider getting a second meter to repeat the tests without the mods?  May be interesting to see you sweep the voltage as well.

I don't just need another one cause this one is fine and I'm going to use it, but I can short the PTC and test again, that'll be easier.

[joeqsmith]

No concerns about the unknown resistors?

[PartialDischarge]

No concerns about the unknown resistors?

I'm more concerned about a 2 inch trace running in the PCB from the input. Resistors if they are bad they will change their properties a lot but they are not going to suddenly short to 0.

[PartialDischarge]

So I repeated the test with the resistors at the PTC shorted and it did arc at about 4.8kV. I did repeat the test 3 times after that, and each time the arcing voltage became lower, which is expected as what usually happens is that a carbon conductive layer forms and the HV already knows where to go to the next time.

Also the humidity plays a role here, it is very dry here but in the winter or near the sea the result would have been probably 20%..30% lower.
So as I expected having an initial resistance near the input port increases the overall arcing/corona resilience.

[Fungus]

So I repeated the test with the resistors at the PTC shorted and it did arc at about 4.8kV. I did repeat the test 3 times after that, and each time the arcing voltage became lower, which is expected

Cool! 

Can you do it with the lid off and see where it arcs?

[Mechatrommer]

Cool! 
Can you do it with the lid off and see where it arcs?

yeah and then send back report to uni-t on where they should change the layout, but i think the carbon should already visible by now ymmv. its fun and tempting to some people to stress test an instrument beyond its rated operation just to find out which brand has more "air" inside

[PartialDischarge]

Can you do it with the lid off and see where it arcs?

Not really, I'd have to solder the 9V battery terminals that are on the back cover, and do the test in the dark. This Hipotronics is designed to test, not to burn or destroy so the arcing is probably going to be very dim. Also it could also happen on the back side of the board since that PTC is through hole.

Also this test was for AC but for DC which is what mostly everyone will use this meter for, the threshold is going to be much higher.

[joeqsmith]

Why does the battery need to be inserted? 

... i think the carbon should already visible by now ...

Certainly worth having a look.