U1241B multimeter woes

[najrao]

On opening my Agilent U1241B which would not read ohms, I discovered that the rotary switch contact pads were burnt out and the board scorched badly in the area. I failed to take a picture then, but did one after some restoration.
Analyzing what could have caused such a disaster in an expensive instrument made by a highly acclaimed maker, I discovered to my horror that the switch, with an interpad clearance of just 0.8mm has to break 1000V rms, and remains exposed to the full voltage at the VOLTS input terminal at all times. The VOLTS selector itself has much better a clearance of 3mm, and uses a double width shorting contact.
Much correspondence with Keysight has produced no worthwhile explanation. They keep harping on how the 'incident' is unique in their databank,  is 'unfortunate', and continue to assert that their product is free of defect and complies with all requirements of certification. Critical questions are just ignored: e.g., " How do the Ohms/Diode/Capacitance ranges withstand the claimed 1000Vrms with just the 1k resistor and 22ohm PTC as protection elements? The tracks on the measurement side of the switch are spaced 0.25mm apart, and the various extremely narrow plated vias would vaporize with the ensuing current."
They have made a peace offering: to replace the damaged instrument!
These problems must exist in every other 1241, and may extend to many other models as well. I want to warn users of the serious design defects which the maker has let pass -- but won't admit to.
This post is necessarily very sketchy, but would have to be many pages long to convey the full implications. Interested users please contact me for much more detail.

[bd139]

Well I was about to grab a 1241C. Certainly not going to now. Thanks for your time to detail this.

Anyone care to suggest a safer alternative for around the same price point?

[Terry01]

Why didn't you pack it up and send it straight back if it was so badly damaged? Sounds like someone else has used it before you if it arrived in that state.

How can they argue if the meter arrives wrecked like that.

[Fungus]

I discovered to my horror that the switch, with an interpad clearance of just 0.8mm has to break 1000V rms, and remains exposed to the full voltage at the VOLTS input terminal at all times.

I'm not sure what clearance you're talking about, but it doesn't matter what the clearance of an individual thing is, so long as there's a bigger clearance further down the line.

How do the Ohms/Diode/Capacitance ranges withstand the claimed 1000Vrms with just the 1k resistor and 22ohm PTC as protection elements?

What "1k resistor"? Have you got a schematic?

Those meters are third party certified by CSA. If it says "CAT III 1000V" on the front then it's CAT III 1000V.

(although CAT rating doesn't mean it has to survive, it's allowed to fail safely if that manufacturer wants it to)

Interested users please contact me for much more detail.

Why the big mystery? Post the details here. Plenty of meters have been roasted in these forums.

eg. What conditions was it exposed to?

[najrao]

Thanks everybody  for the responses.
First, I attach the photo, now compressed, hope it won't be rejected for size. Note this was taken after my repair.
The 'switch pad clearance' is the spacing between the two inner tracks. The main hot input terminal is connected to the outer of the two through the protection network. The moving contact connects it to the inner track, for each of three switch positions: Ohms, Diode, and Capacitance. This switch remains open in all other positions. The inner pad/s of this switch connect through (at least) one very small pth via, visible close to the central guide hole. Since intertrack clearances beyond this point are very small indeed, less than 0.25mm,  I conclude that potentials here are at most a few volts.
Note that ANOTHER identical protection network is present, connected to the hot input again. This feeds the voltage measurement circuit, switched on the outer tracks and offering  much larger an intertrack clearance. As much as 3mm. The shorting contact for this is also different, at twice the pitch. It is clear that the designer is entirely aware of the need to space tracks wider apart, but fails to implement this in the Ohms circuit even though the voltage to be stood off is the same 1000V.
The protection network in each case has a 1k, 1 possibly 2W series resistor; it's speciality if any is unknown, but I presume it is a special 'fusible' resistor a la Fluke. Its thermal capacity is still far too high in my opinion to limit the energy let through into the inner measuring circuit, unless this latter can clamp its voltage to under a few volts, limiting the power, while the protection resistor passes up to 250mA before opening up.
The two spark gap 'diodes' cannot and should not have a role to play for impressed voltages of up to 1000V rms,  more if the permitted crest factor is taken into account. The PTC thermistors would act to curtail the fault current, but they too have large heat capacity and would need time.
All these protection components are soldered in place, unlike the two big  fuses for the current measurement circuits. Operation of any one or more soldered component would render the instrument unusable, or in any case require factory service. Thus, the unit can not be said to 'withstand 1000V rms', even if the Cat III certificate would appear to say so.
I have explained all these points threadbare to Keysight's 'site assistance'. To summarize their answers:
1. Never happened before, unique case in our data bank; unfortunate
2. The U1241B complies entirely with its declared specification and is certified  to CATIII
3.  Give us the failed unit, we will investigate further
4.  All manufacturers made design changes in 2011 to comply with revisions to standards; my unit is earlier. Will replace if I so want
5. All U1241B's will withstand 1000V rms at the hot input terminal with the switch in any of the positions measuring Volts, Ohms or Capacitance.

I have even asked their concurrence for  me to test my unit with voltage applied in steps starting at say 300V.
Their replies do not even pretend to a answer the points raised. I can forward the correspondence by private mail if this is useful.

[najrao]

I must have been dropped on my head as a baby by my mother: I just can't attach the pictures, even after compressing them. The server has rejected the attachment again. What do I do?
Am trying again, with even more compression.

[bd139]

Upload them to imgur.com and insert the provided bbcode links. Works much better!

[Fungus]

The two spark gap 'diodes' cannot and should not have a role to play for impressed voltages of up to 1000V rms,  more if the permitted crest factor is taken into account. The PTC thermistors would act to curtail the fault current, but they too have large heat capacity and would need time.

If the meter can measure 1000V then none of the safety components should do anything until way beyond that.

Thus, the unit can not be said to 'withstand 1000V rms', even if the Cat III certificate would appear to say so.

I've got a $3 multimeter that has correctly measured up to 1000V. I have no doubt yours can withstand that much. The CAT ratings are for what happens beyond.

Here's the CAT ratings chart. There was a second edition to add CAT IV. Does your meter mention CAT IV on the front?



Note that CAT III 1000V is the same as CAT IV 600V. CAT III 1000V allows for an 8000V spike with 2 Ohms impedance, ie. 4000A could pass, if the device allows it to.

Note that a CAT rating doesn't mean the meter has to survive, it's allowed to die so long as it does it safely (although the big hole in your PCB doesn't look 'safe' to me).

In tests, our joeqsmith managed to kill a Keysight U1231A in a similar way:
https://www.eevblog.com/forum/testgear/hear-kitty-kitty-kitty-nope-not-that-kind-of-cat/msg852693/#msg852693

Your input protection looks similar to the U1231A. Joe managed to get arcing across the rotary switch contacts just like you're seeing but it took 5000V to do it.

I have explained all these points threadbare to Keysight's 'site assistance'. To summarize their answers:
1. Never happened before, unique case in our data bank; unfortunate
2. The U1241B complies entirely with its declared specification and is certified  to CATIII
3.  Give us the failed unit, we will investigate further
4.  All manufacturers made design changes in 2011 to comply with revisions to standards; my unit is earlier. Will replace if I so want
5. All U1241B's will withstand 1000V rms at the hot input terminal with the switch in any of the positions measuring Volts, Ohms or Capacitance.

Keysight certifies their meters independently at great expense. Their tech support people have no reason to believe a random internet user over their certification process. They don't know what the meter was exposed to, if there was contamination inside, or whatever.

You haven't told us what happened to your meter. What were you doing when it failed?

[najrao]

I can not be sure what were the conditions just at the time of failure. I was not using it,  I and one who was did not report any unusual circs. I opened it when it died on Ohms, to see what.
It is just possible someone measured a small elco without discharging it, but it cd not if bn more than 100u@150V.

I should add that there is no issue of the instrument being unable to measure 1000V on the two VOLTS ranges: there is a string of 5×2M MELF resistors coming up in series immdtly after the protection circuit, and these are located, spaced and connected up with generous clearances as should.
I agree entirely that the protection components should have NO role to play whatsoever when the unit is operated within its ratings.
The two main defects are:
1. The switch pads for the Ohms range are too close to each other (0.8mm), and the live side is energized at full voltage even when not in the Ohms range at all. Tracking and/or flashover is inevitable, and waiting to happen if not already
2. The Ohms range measuring circuit does NOT have the benefit of Megohm series resistors;  nor can it possibly. I doubt if the circuit can accept more than a few volts applied, or pass tens or hundreds of mA  for seconds on end without destruction. So any claim of 'withstands 1000V rms' on these ranges would not pass muster.
That the instrument complies with various safety standards to Cat III is irrelevant, as the user would be left with a brick after the event --- but would not need to go to the emergency room.  Small consolation that! The makers should at least notify that connecting the probes to a voltage source, as a charged cap, even accidentally while on the Ohms/Diode/Cap ranges could destroy the unit. Instead, this manufacturer continues to harp on certification,  and maintains that it is perfectly acceptable to connect a voltage source of up to 1000V! I even suggested to them that they respecify a much lower safe voltage level for the Ohms range. No engineering answers are forthcoming.
I am not the sole depository of all knowledge on the subject, and would humbly request to be told  any flaws in my arguments.
Has any body tested connecting a voltage source while in the Ohms// ranges? Even on some other units? I want to do this under controlled conditions just to see how it goes.

[bd139]

Thatd need to be a big one to do that sort of damage. Also more than the voltage rating of the device. Could have been a transient on the wrong range. Very low probability.

I just bought a 1241C in the end so will tear down for compare.

[joeqsmith]

Maybe learn to crop?  I don't care about what the meter is sitting on.   

Hard to believe you could get this sort of damage.  Did you trace out the circuit for the front end up to the switch contacts?  If so, maybe add the paths where you feel things are too tight and post it.  Hand drawing is just fine. 

I would not expect the directly discharging a 150V cap into the meter while it was in the resistance mode would do this. 

The makers should at least notify that connecting the probes to a voltage source, as a charged cap, even accidentally while on the Ohms/Diode/Cap ranges could destroy the unit.

Most of the manuals I have looked at warn against changing modes with the meter connected and what voltages the resistance function can handle.  Strange the manual for this would not cover the basics.

Page 5

– Turn off circuit power and discharge all high-voltage capacitors in the
circuit before you perform resistance, continuity, diodes, or capacitance
tests.

page 28

To avoid damaging this device, do not exceed the input limit.  ....

page 32,33

Disconnect circuit power and discharge all high-voltage capacitors before
measuring resistance to prevent possible damage to the multimeter or the
device under test.

page 34

Disconnect circuit power and discharge all high-voltage capacitors before
measuring capacitance to prevent possible damage to the multimeter or the
device under test. To confirm that capacitors have discharged, use the DC
voltage function.

Has any body tested connecting a voltage source while in the Ohms// ranges? Even on some other units? I want to do this under controlled conditions just to see how it goes.

I only looked at the one Keysight meter and I don't normally use a source with enough energy available to do the sort of damage you appear to show.  I am interested in seeing what ever test you come up with along with the results.  Be sure to make a video of it for us. 

I am very impressed that they would offer a replacement especially after someone attempted to repair it.

[Fungus]

Well I was about to grab a 1241C. Certainly not going to now.

Even though they offered to exchange an old, damaged meter for a new one?

[bd139]

That's a good point and one I realised earlier. The info provided after I posted that shows that (a) it didn't blow up and kill anyone even if it did die and (b) that damage requires some serious energy to create looking at the pictures and (c) they covered their warranty obligation. Clearly something is amiss here and I'd put it well in user error territory. If it is run within rated limits, there should be no critical failure here. It's properly certified and tested as well.

I can only see that a massive transient or high voltage ended up in this. Any high current on the wrong terminals would have blown up somewhere in the front end.

Therefore I ordered a U1241C earlier after evaluating the above.

[joeqsmith]

Well I was about to grab a 1241C. Certainly not going to now.

Even though they offered to exchange an old, damaged meter for a new one?

I doubt many companies would make that offer.  Doubt they are swapping it for another B. 

So if I hook my Gossen to a MOT can I get a free upgrade as long as I claim its a design problem?       

[bd139]

yes it's quite remarkable.

I ordered a U1168A lead kit and U1180A thermocouple kit so I'm invested in this one

[Fungus]

I would not expect the directly discharging a 150V cap into the meter while it was in the resistance mode would do this. 

I'm with joe.

Somebody's not 'fessing up to what really happened to it.

[najrao]

It was easy enough to trace the signal path from hot jack right up to the 'innards. Alas, there is no diagram or schematic. The critical tracking path is at the switch pads, now lost in the photo. More importantly, this short space is left stressed at full voltage at all times of voltage measurement even with the switch open.
I tend to believe that a moderate voltage and moisture started the tracking across this 0.8mm gap. Elsewhere, I have seen a picture of how just 12V is enough to track across an inch of phenolic substrate in tropical climate. The burning and charring is progressive over a long time.
I have offered to return the unit to the makers, and have requested to be told of their findings.
I may be guilty of not reading the instruction manual. But the claim of 1000V withstand is misleading and should have been qualified.
I am leaving their offer of replacement to their discretion.
Thanks everybody for the support.

[joeqsmith]

I was not clear with my hand drawn comment.  I am suggesting that YOU take the time to draw up the handful of parts making up the front end and then adding where you feel there is a clearance problem. 

Too bad you lost your high res pictures. I guess you could take new ones or was the camera lost as well?  Or was it a case where in your attempt yo solve a murder, you destroyed the body before collecting the evidence? 

It's common for meters to have the full voltage at various parts but normally it is not a direct link.  Some like those really cheap Aneng meters people keep going on about, put what little current limit they have after the switch.  The switch in that case breaks down and can actually shunt the entire transient.  Funny is some people hail that as a good thing.  Anyway, I doubt your meter is like that.   I assume if the problem was progressive over a long time as you are now suggesting, the meter has been having problems for a long time as well and it went ignored?  Seems strange if you are working with KV levels.

It sounds now like you may be thinking that the high voltage started this area to have a low resistance and then someone connected a large capacitor charged to 150V and that's when it went.  Assuming the 1K resistor is in series with the PTC before it gets to the pad/s in question, I don't buy it as the initial current would be fairly low and if the resistance were low enough to cause the PTC to heat up, the current will fold back even further. 

Personally, it would have been much better had you documented every detail rather than posting all your feelings about what you felt happened and dragging Keysight through the mud.   Asking for people to write you for more details is very strange.  At first, I just assumed you were phishing.  I've damaged a lot of meters but I always attempt to show the details upfront.

[najrao]

I am not familiar with photos and sketches to attach, hence the glitches. Now see a diagram.
I was NOT phishing, nor seeking any kind of compensation from  Keysight. I did not even remotely suggest it to them. Even now, I hv left the replacement offer to their discretion; and will give them the damaged unit unconditionally.
My interest is in obtaining a scientific explanation of how this instrument can withstand the claimed 1000V at the input terminal, given the small clearance/s in the internal circuitry.  It now appears that the English word "withstand" does not include unmolested survival after the event,  and the instrument "withstands" as long it does not blow up in the user's face.
Tracking at  low level of voltage stress is not very common on fiber glass substrate, but can occur if there was a surface defect or inclusion.
Keysight do not suggest that I proceed to test the Ohms input at voltage, as the switch is damaged.

[joeqsmith]

Thanks for taking the time to post this.  On the backside of the contacts that burned, I wonder if there is another series resistor and then the clamp, or just the clamp?  This will be there to provide some protection for the driver.   Could you please take a close look at the two GDTs and provide any markings that are on them?

****************

!!! Please recheck the PTCs !!! Are these really 22 ohms?  If they indeed measure this low, could you please cut away any shrink tube and provide any markings on these as well.  Thanks.   
 
****************


Maybe we can make some sort of model of the meter for the fun of it.

I have started to life cycle test a few meters function switches.  As the parts wear, the metal particulates could very well end up in the area between the pads. 

[HighVoltage]

Besides a few other Keysight handheld multimeters, I also have a U1241B and have used it many times for up to 1000 V DC, without any Problems at all.

It is rated CAT III 1000V / CAT IV 600V and has fulfilled all my needs for this handheld.

Something awefull must have happened to yours, far outside the normal range of operations and it should not be Keysight's responsibility to replace it.

 

[PA0PBZ]

It's more than 0.8mm, like 1.1-1.2mm:

[Fungus]

Thanks for taking the time to post this.  On the backside of the contacts that burned, I wonder if there is another series resistor and then the clamp, or just the clamp?

It's weird that there's no resistor ladder as part of the input protection. I'm guessing there's one on the other side of the PCB.

[PA0PBZ]

It's weird that there's no resistor ladder as part of the input protection. I'm guessing there's one on the other side of the PCB.

No, nothing. Also, the left resistor in OP's drawing is only 275 Ohm in my meter.

[Fungus]

It's weird that there's no resistor ladder as part of the input protection. I'm guessing there's one on the other side of the PCB.

No, nothing. Also, the left resistor in OP's drawing is only 275 Ohm in my meter.

Joes reverse engineering (in his video) shows similar values.



Low values allow the PTC to heat up nicely. The real protection for the rest of the meter should be in the resistor ladder after the PTC.