EEVblog #933 - Keysight U1272A EMC Issue

[joeqsmith]

Just curious.....

Presuming the schematic isn't generally available, has anybody tried reverse engineering the front end of the troublesome meter to see where the problem might be?

... and to then, perhaps, compare this to a comparable meter without such problems?

And this is why I asked about is this a money or a technical problem.  Because if Keysight's EMC experts can't come up with a fix, it would very funny to pick one of these up and take a crack at it!   

Who said Keysight's EMC experts can't come up with a fix?

None but I have asked the question a few times and explained my reason for asking.   

[julian1]

Has anyone opened one up yet - to try and identify the front end op-amps that might be at fault?

There's an interesting issue - given that the design and schematics are not open-source - and that the forum is left to speculate about the function of their own instruments.

Hardware has gone backwards since the cold-war when vendors were required to disclose the details of their designs.
 

[Cerebus]

But as Cerebus said, that whole directive crap was written by lawyers basically... So it's not human readable... That is why people pay so much for certification... Sometimes it's more complicated to figure out what should be tested than to test it...

To be fair, I suspect part of the problem is in writing something that can be translated to about 10 different languages and still keep its meaning intact but there are a lot of typical 'a lawyer wrote this' things going on too. Like, (made up example) to understand a paragraph on page 27 you have to re-read paragraphs on pages 2, 9, 19 and 24. And trust me, the EMC directive really is one of the clearer ones. I don't think anybody, ever, has really understood the working time directive.

[Fungus]

As mentioned in our earlier post, we asked our EMC technical experts to look into this – and they were able to replicate the symptom. Keysight engineers found the majority of U1272A measurement modes and ranges meet published specifications in the presence of an RF field, but the product is susceptible to RF fields in certain measurement modes.

Please tell me how to fix this problem.
We have many of the U1272A meters for teaching purpose and if this EMC problem is not solved for us, we have to replace all of them.

Seems to me like a good opportunity for some real education.

[joeqsmith]

As mentioned in our earlier post, we asked our EMC technical experts to look into this – and they were able to replicate the symptom. Keysight engineers found the majority of U1272A measurement modes and ranges meet published specifications in the presence of an RF field, but the product is susceptible to RF fields in certain measurement modes.

Please tell me how to fix this problem.
We have many of the U1272A meters for teaching purpose and if this EMC problem is not solved for us, we have to replace all of them.

Seems to me like a good opportunity for some real education.

I was thinking that when I saw that post.  Imagine learning some practical RF skills in school.   If successful, maybe Keysight would hire the class once they graduate to be their next EMC experts. 

[FrankBuss]

Good idea, the device could be used as an example what can go wrong when designing a multimeter, and that strong RF fields can cause strange things. And give the students a printout of the Keysight posting (and the reactions after it), as an example how you should not response to the community, unless you want to become a lawyer

But really, is it such a big problem? I don't measure current often and usually there is no strong RF field (probably stronger than the EMC standards require) near the multimeter. Not a reason for replacing it in my opinion, if there are no other problems.

[alouko]

But really, is it such a big problem? I don't measure current often and usually there is no strong RF field (probably stronger than the EMC standards require) near the multimeter. Not a reason for replacing it in my opinion, if there are no other problems.

It is a real problem.  For several reasons:

• No other DMM except Keysight U127x and in a lesser degree U124x and even the newest U128x, suffer from this fault.
• It also affects AC current measurement.  I just quickly tested and got spurious readings.
• RF fields in 10MHz range in volts amplitude are common.

Basically, you cannot trust current measurements with this DMM.

[EEVblog]

Interesting stuff.  Would really enjoy a video where Dave walks us through the process of having his new meter certified.   

It goes off to a lab somewhere on the other side of the planet and hopefully comes back certified 

[bernroth]

It is a real problem.  For several reasons:

• No other DMM except Keysight U127x and in a lesser degree U124x and even the newest U128x, suffer from this fault.
• It also affects AC current measurement.  I just quickly tested and got spurious readings.
• RF fields in 10MHz range in volts amplitude are common.

Basically, you cannot trust current measurements with this DMM.

ACK

It can be very dangerous as well:

I was measuring the charging current of a LiIon battery. If the charging current is too high, the battery may explode and cause real damage.
Fortunately I had more confidence in my trusty old HP/Agilent power supply as in the DMM 

Keysight should issue a public warning message telling that this DMM is crap should not be used measuring currenty in noisy environments
Just my 2 cents...

[EEVblog]

But really, is it such a big problem? I don't measure current often and usually there is no strong RF field (probably stronger than the EMC standards require) near the multimeter. Not a reason for replacing it in my opinion, if there are no other problems.

It is a real problem.  For several reasons:

• No other DMM except Keysight U127x and in a lesser degree U124x and even the newest U128x, suffer from this fault.
• It also affects AC current measurement.  I just quickly tested and got spurious readings.
• RF fields in 10MHz range in volts amplitude are common.

Basically, you cannot trust current measurements with this DMM.

Yep, bang on all those points, it's a big problem and Keysight know it. They just don't want to accept the ramifications.
People have reported dodgy readings in typical bench measurement scenario's, s it's not some oddball niche thing like the Fluke GSM issue.

[Tom45]

I get exactly the same phenomenon with Dave's uCurrent.  5MHz square wave to either current input posts and there is full negative DC reading in the voltage output!  So I think it is a chopper opamp in U127x, too.  Possibly MAX4239 in that or a similar one.

Since the uCurrent apparently has the same problem, it should be easier to start by analyzing the uCurrent than by reverse engineering the U1272a.

[alouko]

I get exactly the same phenomenon with Dave's uCurrent.  5MHz square wave to either current input posts and there is full negative DC reading in the voltage output!  So I think it is a chopper opamp in U127x, too.  Possibly MAX4239 in that or a similar one.

Since the uCurrent apparently has the same problem, it should be easier to start by analyzing the uCurrent than by reverse engineering the U1272a.

The chopper opamp is the culprit in uCurrent, I think.  It has an internal oscillator and I think the offset voltage zeroing goes north (as the output voltage goes south :-).

As far as I know, there is no separate chopper opamp in U12 DMMs, but perhaps Dave could make an educational vblog on the subject?

[bernroth]

https://www.amazon.com/gp/customer-reviews/R1A3O2ESRECTDH/ref=cm_cr_arp_d_rvw_ttl?ie=UTF8&ASIN=B004URG9EU

Another problem is the AC mA mode, if using LED lamps which is very common now a days, the meter tells you the lamp is using double current, and this, of course when you do the math, will give you the wrong wattage. Using traditional incandescent lamps, seems to be OK.

[2N3055]

Are you guys seriously discussing comparison of susceptibility to EMC of  uCurrent in uA range with multimeter on a 10 A range...   

[alouko]

Are you guys seriously discussing comparison of susceptibility to EMC of  uCurrent in uA range with multimeter on a 10 A range...   

Yes. Why not?

We are discussing the mechanism of this fault and uCurrent is fully relevant as its upper current range is up to 1.5 amps, or even higher with 3 AAA batteries.

[madires]

https://www.amazon.com/gp/customer-reviews/R1A3O2ESRECTDH/ref=cm_cr_arp_d_rvw_ttl?ie=UTF8&ASIN=B004URG9EU

Another problem is the AC mA mode, if using LED lamps which is very common now a days, the meter tells you the lamp is using double current, and this, of course when you do the math, will give you the wrong wattage. Using traditional incandescent lamps, seems to be OK.

Very nice of Keysight to sell an EUR 460 DMM without an IR-USB cable, which is 45 bucks extra. My inexpensive (non-Keysight) LCR meter for EUR 120 had one in the box. So I have to adapt what I would pay for the U1272A from 200 down to 155 

[2N3055]

Are you guys seriously discussing comparison of susceptibility to EMC of  uCurrent in uA range with multimeter on a 10 A range...   

Yes. Why not?

We are discussing the mechanism of this fault and uCurrent is fully relevant as its upper current range is up to 1.5 amps, or even higher with 3 AAA batteries.

Because uCurrent documentation specifically states it has no input protection of any sorts to achieve lowest burden voltage..  And since we don't know schematics of Keysight, you can't compare anything..
I'm not saying what you say cannot be true, but seems like a bit of stretch to compare it something you didn't analyze or have no detailed knowledge.... That's all mate, I'm just saying a bit more of data would be good before making a theory..  No offense meant!!

Have a good one!

[dr.diesel]

Unfortunately, further evidence of lost quality for each name change.

[Zbig]

Blah, I was getting my U1273A on the premise of overbuying a bit in order to have something solid, dependable, to "grow into"; not having to think about DMM for at least few years to come. Turned out to be everything but. First, I've learned the OLED window is apparently made out of sugar and gets scratched if you look at it funny (there's my thread on this issue somewhere here). So much for industrial quality and robustness. And now, this. Screw this, that's enough. I think I'll get rid of mine just out of pure disgust of how this was non-handled by Keysight. Guess what brand my next one won't be?

@Keysight Technologies Rep: is that what you were after posting your lousy non-answer?

Does anyone have any experience creating Reddit threads? Those tend to go pretty viral and get good exposure.

[Barny]

I got mine during the discount buy a lab-suply get the U1272A extra.
That was the first HP / Agilent / Keysight / .... measuring equipment I bought new and after the behaviour of the PR and the problems with my DMM probably the last one.

In future measurements, I'm using my cheap DMM (retrofitted with a HRC fuse) instead of my U1272A.
And this is really sad.

[Tom45]

Are you guys seriously discussing comparison of susceptibility to EMC of  uCurrent in uA range with multimeter on a 10 A range...   

When the U1272A was first announced, I think someone commented that because of its 300 microamp range it might have adopted similar technology to Dave's uCurrent.

[alouko]

When the U1272A was first announced, I think someone commented that because of its 300 microamp range it might have adopted similar technology to Dave's uCurrent.

Well, my $100 Fluke 17B+ (Chinese one, not officially available outside China, but for sale in Aliexpress and elsewhere) has also 300uA range, with 100nA resolution.  And it doesn't suffer from this "symptom".

[TheAmmoniacal]

Offering firmware fixes is pretty shitty when you have to purchase the interface cable separately. In this country the U1272A is $638 USD and the USB cable is $42 (+ $10 shipping).

Definitely not worth it.

[FrankBuss]

Offering firmware fixes is pretty shitty when you have to purchase the interface cable separately. In this country the U1272A is $638 USD and the USB cable is $42 (+ $10 shipping).

Definitely not worth it.

Agilent gear is a bit more expensive, because you get good support for it, for example updated datasheets

[D3f1ant]

Samsung was most likely using a Keysight U1272A when testing the Note 7 charger.