EEVblog #933 - Keysight U1272A EMC Issue

[Berni]

The EMI Problem is no unique to this DMM. I have an old analog meter (batteries just for ohms), that also gets crazy with RF voltages. Something like 5 V at 100 kHz gave a full scale reading on essentially all ranges, including 1200 V DC. Checking a switched mode regulator can get really scary with this.

Hmm lets check the 3.3V rail....and its....1200V ?!?!

[Cerebus]

This is disturbing. A while ago the grey beards told the young engineers that you can't go wrong with HP. And now we have to wait a few years for the early buyers to do the beta testing?

Well you can't go wrong with HP, these however are made by Keysight. Every name change drives the quality down by a certain percentage.

[gamalot]

Any results with the 1250 series?

Seems like my U1253B doesn't have this issue.

[tchicago]

Could anyone test U1232A or U1233A? I almost bought it, but now I have doubts.

[dentaku]

Hmmm... A little unknown "EEVBlog blue" meter in the background?

[bitseeker]

Did a quick check on some meters here:

Agilent 34410A: OK
Keysight U1252B: OK
Greenlee DM-820A: OK
Harbor Freight "free" DMM: OK

[joeqsmith]

Good video.   Can't say I am too surprised.   Shown is the EEVBLOG branded Brymen with the single pin attached to the current input. 

[lowimpedance]

Could anyone test U1232A or U1233A? I almost bought it, but now I have doubts.

Tested a U1232a and it is affected by 10vp-p @5Mhz square wave with the same amount when applied to either the A or COM jacks, (Only the one lead no return path!) . And the reading goes from approx. -135mA to 350+ mA when I place my hand on or near the casing. ! Quite sensitive actually but not sensitive to just a wire dragged around the case, ie not physical connection.

Also tested an U1252A meter and it IS NOT affected by the same signal at all on any input jack. Rock solid zeros.

And for kicks I tested an old Escort meter (the company Agilent acquired to make handhelds!) and it was NOT affected either.

EDIT: Just tested a fluke 8020a and it does have an effect on the reading when the lead is plugged in the amp and com jacks, but no issue when just dragged around the casing.

[fuzzoli]

Same thing happens on my U1273A.   I did some tests and noticed the Low Pass Filter makes a difference.  Here's the data:

Level	LPF On	LPF Off
500mv	0.0020A	-0.275A
750mv	0.0037A	-1.026A
1.00v	0.0064A	-3.306A
1.25v	0.0096A	-6.140A
1.50v	0.0138A	-9.820A
1.75v	0.0180A	-5.420A
2.00v	0.0228A	-5.410A
2.25v	0.0360A	-6.090A
2.50v	0.0440A	-7.480A

The above readings were taken on my U1273A DMM, reading a 10Mhz 2.5v square wave signal generated from the waveform generator output of an Agilent DSO-X 3034A scope.  I also moved the meter as far away from me as the test lead would allow to prevent stray pickup when moving my hand from the scope to the notepad (it did make a difference as noted in some of the earlier posts).

So it seems the LPF has a pretty big impact on the readings.  As a bonus, I turned the filter on and took some additional readings.  Below 500kHz, the reading barely budged above 0.001A (this is interesting since the manual says the filter's cutoff frequency is 1kHz).  From 1MHz to 10Mhz (in 1Mhz steps), the reading went from 0.0016A to 0.0360A, so although the filter seems to have done it's job, it wasn't perfect.

Side Note:  Trying to put blame on the software guys (I'm allowed to as I am one), I looked at the firmware updates.  Unfortunately, going from v1.95 (mine) to v3.01 (latest) were only these two items:

            - Fix incorrect average range in autorange feature
            - Keysight rebranding, change identity to Keysight and Keysight Logo

...which is really funny since my meter is branded Keysight, but boots Agilent!       Maybe the true issue is the meter's dual personality, and ghosts of Agilent-past are wreaking havoc on random readings.   

...the mystery continues!

-Frank

[joeqsmith]

I also have the UNI-Trend UT61E that Dave shows.   Keep in mind that the one I have has been modified in my attempt to improve its robustness and adding an adaptive back light.  So my results from this test may not be valid.

41MHz, 20dBm, single wire to Amps showing a 2ADC offset.  Much worse than the EEVBLOG Brymen meter.   

[Someone]

I guess the engineer who signed the EMC test report is getting nervous now. This is the standard:

http://www.etsemc.co.uk/wp-content/uploads/EN61326-1.pdf

For the basic enclosure immunity requirements it says for EM radiated field: 80 MHz - 1 GHz, 3 V/m (I guess lower frequencies are not that important, because e.g. at 5 MHz the wavelength is already 60 m?). Industrial immunity requirement is even 10 V/m. I have no idea, is this high for an EM field? Can anyone do a back-of-the-envelope calculation how strong is the field of the harmonics at these frequencies with Dave's 5 MHz signal, using just a wire as an antenna?

Depends what the pass/fail criteria are, performance criteria "C" requiring user interaction is considered a worse failure than "B" which is self recovering (Dave puts his preference the other way around in the video) and both are less desirable than performance criteria "A" where the device functions within specifications during the test. And we're not sure how the intensities dave shows in the video compare to the commercial immunities claimed in the U1272A manual.

[Someone]

I guess the engineer who signed the EMC test report is getting nervous now. This is the standard:

http://www.etsemc.co.uk/wp-content/uploads/EN61326-1.pdf

For the basic enclosure immunity requirements it says for EM radiated field: 80 MHz - 1 GHz, 3 V/m (I guess lower frequencies are not that important, because e.g. at 5 MHz the wavelength is already 60 m?). Industrial immunity requirement is even 10 V/m. I have no idea, is this high for an EM field? Can anyone do a back-of-the-envelope calculation how strong is the field of the harmonics at these frequencies with Dave's 5 MHz signal, using just a wire as an antenna?

As you note the lower limit for radiated immunity is 80MHz (and applied from a distance). This test in the video is more like the conducted immunity test which might? be appropriate for the inputs of a meter, but would be applied with them in use not the synthetic case here. So its possible a solid compliance was found to the EN61326-1 but as this test falls outside of it the particular problem was missed. Standards don't cover everything!

[Dwaine]

Nice....  I just got my new U1272A for one day and it's now completely worthless. 

I just hooked it up to my Keysight 531332a 10mhz external output.  The ma values were not as bad as in the video.  The meter does jump around a bit.  0.0153 to 0.0570ma when I move my hand to the side of the meter.

I test it out on the function generator and see if I also get the same results.  But I know I will...

Well, I should thank Keysight because when I fired up the frequency counter.  My Kenwood D710 radio fired on 144.390mhz with a reading of 144.390,000,382.  So no drift in 5 years running 24/7.  Keysight, just a hint.

[Someone]

Nice....  I just got my new U1272A for one day and it's now completely worthless. 

I just hooked it up to my Keysight 531332a 10mhz external output.  The ma values were not as bad as in the video.  The meter does jump around a bit.  0.0153 to 0.0570ma when I move my hand to the side of the meter.

I test it out on the function generator and see if I also get the same results.  But I know I will...

Well, I should thank Keysight because when I fired up the frequency counter.  My Kenwood D710 radio fired on 144.390mhz with a reading of 144.390,000,382.  So no drift in 5 years running 24/7.  Keysight, just a hint.

Worthless? How do the readings look when you're actually measuring something, with and without the artificial presence of an antagonising signal?

[Berni]

Well this only happens on the amps range and even that only with a unusually high amount of common mode AC on it. But i would imagine it does make it feel like you payed too much for it (Id assume you did pay a pretty penny for it).

I had the OLED version of it for quite a long time and i was pretty happy with it ever since. This amps jack thing is a bit of a bummer but it does not worry me too much. A real life scenario where this would happen is pretty rare.

[bernroth]

Worthless? How do the readings look when you're actually measuring something, with and without the artificial presence of an antagonising signal?

I was hitting the issue actually measuring something useful!
My Agilent PSU was supplying 1500mA of current and the meter should confirm the reading of the PSU as I had to make sure the current does not exceed the 1500mA
Yes, there have been RFID readers on the bench and yes, they were "poisioning" the nearby environment with 13.56MHz and 125kHz signals.

But: The measurement of the meter must not be off by that dimension nor showing negative readings.

Look at attached picture of my bench

[EEVblog]

Look at attached picture of my bench

Wow! 

[bernroth]

U1272A with Fluke 87V in series.

PSU and 87V are bang on

[Kelbit]

This is annoying. I bought a U1241B last year at work (when Keysight had a buy 2 get 1 free deal) and put it on our calibrated equipment list. I had trouble convincing other engineers it was any good since they had never heard of Keysight (they didn't know Agilent changed names, they thought Keysight was some dodgy Shenzhen brand). Now the series might have EMC problems?

I'll put the U1241B in our TEM cell tomorrow morning and blast it with RF to see if it suffers from the same issue. It had better work fine, otherwise I have a bunch of QC paperwork to do since it's a calibrated instrument  .

[Someone]

U1272A with Fluke 87V in series.

PSU and 87V are bang on

Great example, can't be much common mode coupling there.

[Dwaine]

Look at attached picture of my bench

I can get the meter in DC volts to jump around with it hooked up to my power supply at 10volts, with just the common lead moved around the U1272a.  Measuring with the meter has no problems with both leads hooked in.  Nothing like the previous post and picture.

I tried the same tests with my Rigol Dm3068 meter.  Did not experience the same effects like with the Keysight.

[Buzz-01]

This video proved to me that it's not me being stupid... 
It might explain some of the weird stuff going on with the U1271A I am using at work. Working on a switched-mode PSU project and at random moments the meter (VDC mode) just goes all bananas and only shows 0.000V while 5V is present at its input. I will have to switch the meter off and on a couple of times before it will show a valid reading again. I am not able to reproduce it consistently, but it happened about half a dozen times in the last month.
Other meters (Fluke 87, 189), connected in the same circuit, are working just fine.
My first experience with Agilent/Keysight, but not liking it so far! 

[Alexandre Mello]

I tested my multimeters here and all seemed to be fine. No issue.

Fluke 289
Agilent U1253B
BK Precision 2712

[dentaku]

Hmmm... A little unknown "EEVBlog blue" meter in the background?

It's not a secret anymore. Dave has posted quite some info in supporter's lounge section. You can buy membership or donate to his patreon to get access, or posting more than 1000 posts automatically gets you into that section.

He's talked about it more on the Amphour too. It looks nice and small.

[t-17]

Is there any reply from Keysight yet? I'm kind of dissapointed after buying one of these. Apparently I was wrong when I thought, "this Keysight multimeter will do the job and will have reliable readings".

I build, test and troubleshoot amateur radio stuff and now I don't know when and if I can reliably measure something with the meter anymore. That's unacceptable.

Thank you so much for bringing this to my attention! Looking forward to that explanation from Keysight.