Keysight 34460A/34461A/34465A/34470A Maximum Allowable Voltage change notice.

[Alex Nikitin]

This issue was reported by TheSteve in another thread but it is not limited to the 34461A only and due to it's importance deserves a separate thread in my opinion.

Here is the link to the notice.

Here is the essence of it:

Input values exceeding the maximum allowable measured voltage of 600 VDC or
440 VAC could, over the lifetime of the product, result in damage to the
instrument.

Cheers

Alex

P.S. It would be good to discuss possible implications. For example, the calibration procedure has to change too. If a failure occurs would it be a catastrophic one, taking half of your lab down?
P.P.S. If you've bought a car specified for 120 mph maximum speed only to discover later that the car can actually break down if you exceed 80mph    what would you do?
P.P.P.S - Why I, as a Keysight customer with a registered 34465A meter, did not receive this update immediately on it's release on 17th of March, but learned about this critical issue here on the forum almost by chance?!

[alm]

Based on the very brief notice we can only speculate, but since they only mention the highest measurable voltage, and not the safety ratings, I would expect the worst case damage to be to destroy the meter. Not explode and take down half your lab.

Since they talk about 'could, over the lifetime of the product, result in damage [...]', it could be something like a spark gap arcing over below 1000 V, gradually degrading it. If the meter is specified to measure DC voltages only up to 600 V, then I would argue the calibration procedure should not exceed that either.

It would not surprise me if they supply a firmware update that shows overload for voltages over the new limits. Some Prema meters do the same when their maximum voltage is limited by an optional scanner card.

The fact that they issue the change notice, and not a recall, suggests the issue is not limited to particular batches, and that they are not likely to offer modifications to existing units. Probably they think this change will not be critical for the majority of their customers. I would expect them to be working on a 34460B etc to fix this issue.

[0xfede]

I just finished the call with keysight and they confirmed it is an hardware issue.
I'm waiting for an answer from keysight to my request of changing the multimeter.
For now they leaked that on December 2017 there will be a new improved version of the 344XXX multimer line that should respect the advertised values.

Let's see what happen.

0xfede

[Dr. Frank]

That max. rating change is a severe problem, and KS should have informed all the customers about that, directly.

The background is not clear at the moment, so 'damage' could mean everything, from slight component damages, until complete wreckage of the instrument.
Also this term '.. over the lifetime of the instrument' is totally fuzzy.

What happens if I use the instrument anyhow at 1000Vp, and a damage occurs, will  KS repair / replace on their own cost?

Will they fix this problem on already sold units?

I started a request on the Keysight DMM forum, already.

During the calibration process, 500VDC and 500VAC were proposed as standard values for the 1kV ranges.
For DCV, 450V...1000V, and for ACV, 400V ... 750V would be accepted by the firmware.


Frank

[Alex Nikitin]

That max. rating change is a severe problem, and KS should have informed all the customers about that, directly.

The background is not clear at the moment, so 'damage' could mean everything, from slight component damages, until complete wreckage of the instrument.
Also this term '.. over the lifetime of the instrument' is totally fuzzy.

What happens if I use the instrument anyhow at 1000Vp, and a damage occurs, will  KS repair / replace on their own cost?

Will they fix this problem on already sold units?

I started a request on the Keysight DMM forum, already.

During the calibration process, 500VDC and 500VAC were proposed as standard values.
For DCV, 450V...1000V, and for ACV, 400V ... 750V would be accepted by the firmware.


Frank

Thank you. It is 1000V DC and 750V AC (1kHz) on the performance verification routine (and on the calibration certificate) though.

Cheers

Alex

[HighVoltage]

This is really impacting my lab work !

I have three 34461A and one 34470A meter and I am using them all a lot in the range just under 1000 VDC, as I have done with the older 34401A meters as well in the past. The 34461A is suppose to be a full replacement for the 34401A.

Limiting the meter now to 600 VDC will change a lot for me and for my customers who also have the 34461A meter in use now, integrated in to some test stands I have built for these customers. Now I need to notify my customers!

And the 34401A, 34410A, 34411A meters have been taken out of production since 1/2017
Which precision Keysight meter can I use now for 1000 VDC.

This is really horrible news for me and I would like to know from Keysight:

1. What is really the hardware problem?
2. Why was I not notified of this?
3. The solution to just amend the datasheet is not acceptable!

[HighVoltage]

Datatec, Germany has already updated their website with the new specifications
Meilhaus, Germany has not updated the specifications
See screen shots below

[Dr. Frank]

If one looks into the 34411A schematics, which might be very close to the newer instruments, concerning the voltage input topology, the DCV and ACV path is identical up to one certain relay (K201 in the '411A).

As both paths are affected in the same way, the culprit will be located between the input jacks and this relay.. and there are not so many components in between.

As the components been used, seem to be the very same as in the 34411A, a layout error is most probable..
On the pictures of the PCBs from the initial 3446xA thread, there are some internal PCB traces visible, which might be too narrow to other signal lines..
So I suspect sparking inside the PCB layers, which can't be fixed other than a PCB redesign.

Frank

[Alex Nikitin]

1. What is really the hardware problem?
2. Why was I not notified of this?
3. The solution to just amend the datasheet is not acceptable!

Exactly. I think that the only acceptable course of action from Keysight would be a full disclosure of the hardware failure issue in detail and a free hardware update. I understand that it would involve a lot of hassle (including, most likely, a re-certification of the meters) but anything less would be an attempt to get away with what is essentially a mis-sale .

Cheers

Alex

[Dr. Frank]

1. What is really the hardware problem?
2. Why was I not notified of this?
3. The solution to just amend the datasheet is not acceptable!

Exactly. I think that the only acceptable course of action from Keysight would be a full disclosure of the hardware failure issue in detail and a free hardware update. I understand that it would involve a lot of hassle (including, most likely, a re-certification of the meters) but anything less would be an attempt to get away with what is essentially a mis-sale .

Cheers

Alex

In addition, especially the occurring failure mode has to be explained in detail.

Maybe the failure creates a mis-reading afterwards (instead of a real smoking damage), so it would not be possible for a high-voltage application to get aware of the problem.

In other words, users as HighVoltage may have measured erroneously in the 1kV range for a long time, already.

Frank

[HighVoltage]

As the components been used, seem to be the very same as in the 34411A, a layout error is most probable..
On the pictures of the PCBs from the initial 3446xA thread, there are some internal PCB traces visible, which might be too narrow to other signal lines..
So I suspect sparking inside the PCB layers, which can't be fixed other than a PCB redesign.

Frank

I totally agree.
But if this is the case, then a PCB revision should be a the next step and not a change in specifications.

[Alex Nikitin]

This is really impacting my lab work !

I have three 34461A and one 34470A meter and I am using them all a lot in the range just under 1000 VDC, as I have done with the older 34401A meters as well in the past. The 34461A is suppose to be a full replacement for the 34401A.

Limiting the meter now to 600 VDC will change a lot for me and for my customers who also have the 34461A meter in use now, integrated in to some test stands I have built for these customers. Now I need to notify my customers!

And the 34401A, 34410A, 34411A meters have been taken out of production since 1/2017
Which precision Keysight meter can I use now for 1000 VDC.

If the meters are used only on 1kV DCV range a possible solution would be to use an external calibrated divider 100:1 and 10V DCV range on the meters. Caddock divider arrays are not that expensive, very stable and usually available from stock.

Cheers

Alex

[HighVoltage]

If the meters are used only on 1kV DCV range a possible solution would be to use an external calibrated divider 100:1 and use 10V range on the meters. A Caddock divider arrays are not that expensive, very stable and usually available from stock.
Cheers
Alex

Yes, there are solutions, of course.
But my test stands are certified and have specific uncertainties. Adding a divider would change uncertainties and all documentations. What a nightmare ! And I don't think I will get paid for that by my customers.

On the other hand, I just checked the calibration of one of my 34461A that has been used a lot just under 1000 V DC and it is right on spot. So may be Keysight is just too cautious about this issue. That is why it is so important that Keysight will tell us what is really going on in the hardware signal input path.

[mikeselectricstuff]

My guess would be possible deterioration of HV resistors or protection devices over time.

[Alex Nikitin]

On the other hand, I just checked the calibration of one of my 34461A that has been used a lot just under 1000 V DC and it is right on spot. So may be Keysight is just too cautious about this issue. That is why it is so important that Keysight will tell us what is really going on in the hardware signal input path.

Yes, we need to know more, as it could be either (or both) a slow change in parameters (say, due to an increased leakage between tracks) or a complete failure with potentially serious implications for the meter and possibly the circuit connected.

Cheers

Alex

[EEVblog]

I suspect they have found (or had reported to them and investigated) that a high voltage component like a MOV or some such component used on the front end can deteriorate.
I'm not buying it's a PCB layout issue, it's got to be component deterioration with repeated/sustained HV.

And a company like Keysight can't just fix the problem in new units and that's that, as these are often used in critically specified systems. You can't just change things willy-nilly. So they have no choice but change the spec sheet for this model to reflect the worst case in all units sold and still in stock and production.
Once they fix it they will come out with a B model or whatever like how they normally upgrade products like this.

[Alex Nikitin]

My guess would be possible deterioration of HV resistors or protection devices over time.

That kind of a failure can be fixed relatively easily by changing the components affected. So far I am inclined to go with Dr. Frank's guess on the board layout problem (and the board material can also deteriorate over time with a continuous application of a strong electric field).

Cheers

Alex

[wraper]

My guess would be possible deterioration of HV resistors or protection devices over time.

That kind of a failure can be fixed relatively easily by changing the components affected. So far I am inclined to go with Dr. Frank's guess on the board layout problem

I'd like to see how you "Relatively easy" recall every single meter ever sold, do a component level repair, recalibrate, and send back.

[Alex Nikitin]

My guess would be possible deterioration of HV resistors or protection devices over time.

That kind of a failure can be fixed relatively easily by changing the components affected. So far I am inclined to go with Dr. Frank's guess on the board layout problem

I'd like to see how you "Relatively easy" recall every single meter ever sold, do a component level repair, recalibrate, and send back.

It is still a lot easier and cheaper than to recall every single meter ever sold, replace the main pcb and then recalibrate and send back  . There is no way around it, the meters were mis-sold. When a car manufacturer recalls 100000 cars to fix a problem for free, it does not surprise anybody. Here is essentially the same situation, s*it happens.

Cheers

Alex

[wraper]

It is still a lot easier and cheaper than to recall every single meter ever sold, replace the main pcb and then recalibrate and send back  .

Not sure about it. With U127xA they choose to just send replacement meters instead, leaving you with 2 meters.

[Alex Nikitin]

It is still a lot easier and cheaper than to recall every single meter ever sold, replace the main pcb and then recalibrate and send back  .

Not sure about it. With U127xA they choose to just send replacement meters instead, leaving you with 2 meters.

The  U127xA is most likely costs not that much to make, so a replacement could be the cheapest option, I am sure it is not an option with the top level bench meters.

Cheers

Alex

[Fgrir]

I'd been thinking about buying a 34465A for my bench, but this spec change would be a deal-breaker for me so I'm glad I haven't pulled the trigger.  I'll be curious to watch Keysight's response to the issue - anything less than a free replacement for affected users would seriously damage their credibility in my eyes.

And you have to love targeted advertising...

[Alex Nikitin]

I'd been thinking about buying a 34465A for my bench, but this spec change would be a deal-breaker for me so I'm glad I haven't pulled the trigger.  I'll be curious to watch Keysight's response to the issue - anything less than a free replacement for affected users would seriously damage their credibility in my eyes.

The problem right now is that Keysight has nothing available even to replace these meters with (except, perhaps, 3458A but that would be a bit much to expect  ), so if you need to measure voltages over 600V DC and 440AC accurately, you have to look elsewhere or resort to some band-aids like an external voltage divider.

Cheers

Alex

[HighVoltage]

The problem right now is that Keysight has nothing available even to replace these meters with (except, perhaps, 3458A but that would be a bit much to expect  ), so if you need to measure voltages over 600V DC and 440AC accurately, you have to look elsewhere or resort to some band-aids like an external voltage divider.

Cheers

Alex

May be they will start production of the 34401A, 34410A, 34411A again.
I am glad I still have some of the older models.
In addition, the older models will now probably jump up in price on ebay around the world

[Tom45]

This isn't much of a problem for me now that I know about it. In the rare cases I need to measure higher voltages I now know to use a different meter than my 34465A. It would have been nice if Keysight had notified me.

But for those that need up to 1000 volts, Keysight had better get busy and take care of them.

Edit: and thanks for making this a separate thread. I wouldn't have known about this otherwise.