EEVblog #933 - Keysight U1272A EMC Issue

[tszaboo]

Yes, you are right.

[wraper]

It's a diode after all, however strangely the PCB has Cxx reference on it, probably C45, digits are unclear, but I'm pretty sure it is C. There is another such diode on the PCB with D21 reference.

[coppice]

Those inductors which have been shorted in the revised design are one of the commonest rookie mistakes people make. When design reviewing a design with current shunt sensing, that is failing EMC tests, this is one of the commonest errors you see. I understand a rookie at Keysight making such a mistake, but I'm surprised there was no design review process to pick it up and correct it early in the design cycle. Getting to market without going thorough EMC testing is really poor.

I expect someone did their EMC tests exactly to the IEC specs, without being inquisitive enough to really flush out problems. Too many people treat "playing" with their products as undisciplined, when its actually the best way to get to the heart of a design's true behaviour.

[BrianHG]

It's not the parts on the main PCB...
Dave, look at 15:55 in your video, your finger is touching the difference in the input board. There are 2 x 2 pin connectors on the new meter where the old meter on the left has only 1 x 2 pin connector.

[Monkeh]

It's not the parts on the main PCB...
Dave, look at 15:55 in your video, your finger is touching the difference in the input board. There are 2 x 2 pin connectors on the new meter where the old meter on the left has only 1 x 2 pin connector.

It's not the parts on the main PCB you say, after replacing the ferrite beads with jumpers 95% fixes the problem?

[BrianHG]

It's not the parts on the main PCB...
Dave, look at 15:55 in your video, your finger is touching the difference in the input board. There are 2 x 2 pin connectors on the new meter where the old meter on the left has only 1 x 2 pin connector.

It's not the parts on the main PCB you say, after replacing the ferrite beads with jumpers 95% fixes the problem?

Ok, maybe I've taken Dave's ending comment that there is still a bit of interference effect remaining as too big an issue while noticing he was touching a new addition to both the input and main PCB.

[Monkeh]

It's not the parts on the main PCB...
Dave, look at 15:55 in your video, your finger is touching the difference in the input board. There are 2 x 2 pin connectors on the new meter where the old meter on the left has only 1 x 2 pin connector.

It's not the parts on the main PCB you say, after replacing the ferrite beads with jumpers 95% fixes the problem?

Ok, maybe I've taken Dave's ending comment that there is still a bit of interference effect remaining as too big an issue while noticing he was touching a new addition to both the input and main PCB.

There's a couple little filter changes (probably more than just the obvious resistors), plus likely firmware changes other than the default LPF. Just removing those ferrites solved it adequately IMO.

[komet]

Is it too late to apply for a replacement? I missed this the first time round, but do have a 2013 era U1272A.

[jolshefsky]

Probably not a big deal but I noticed R2 in the upper left corner of the first (back?) side changed value a lot from 5682(?) to 1330: ~57K to 133 ohms.

[bitseeker]

Is it too late to apply for a replacement? I missed this the first time round, but do have a 2013 era U1272A.

You'd have to check the service note for expiration, but I think it's still valid. Contact Keysight. It's worth a try. If they say no, then you can do the repair via the findings in this thread.

[cdz]

Nice article about ferrite beads and possible oscillation : http://www.analog.com/en/analog-dialogue/articles/ferrite-beads-demystified.html

[tszaboo]

Those inductors which have been shorted in the revised design are one of the commonest rookie mistakes people make. When design reviewing a design with current shunt sensing, that is failing EMC tests, this is one of the commonest errors you see. I understand a rookie at Keysight making such a mistake, but I'm surprised there was no design review process to pick it up and correct it early in the design cycle. Getting to market without going thorough EMC testing is really poor.

I expect someone did their EMC tests exactly to the IEC specs, without being inquisitive enough to really flush out problems. Too many people treat "playing" with their products as undisciplined, when its actually the best way to get to the heart of a design's true behaviour.

I still cannot believe what I'm seeing. I worked with analog a lot. A differential signal enters your multimeter. You place a ferrite bead between analog grounds, while the positive goes into the meter without the exact same filtering. You end up with a meter with really poor common mode noise immunity. Because your input filter is not symmetrical. Its simple as that. I dont think a DMM designer could ever lose scope of the input filtering.
Sorry for the analogy, but its like building cars, and then after many iterations, you end up bigger wheels on one side.

[Kas.Lewis]

I have seen other "issues" I put them in potentiates because I am not sure if its expected behaviour or not. One is when moving the test leads around on both my U1272A and U1282A the display goes wild, this happens on resistance and continuity modes (might be others but I don't recall for surer at teh moment). The other issue I noticed was when doing a sand test on the U1282A the display also went wild (Video here). Another issue I brought up with Keysight during the review I did of the U1282A was the inconsistency between the PWM output from the U1282A and what the U1272A was reading the PWM values where. They told me to do a factory reset...

Anyways just some observations of mine.

Kas

[Zbig]

I have seen other "issues" I put them in potentiates because I am not sure if its expected behaviour or not. One is when moving the test leads around on both my U1272A and U1282A the display goes wild, this happens on resistance and continuity modes (might be others but I don't recall for surer at teh moment). The other issue I noticed was when doing a sand test on the U1282A the display also went wild. Another issue I brought up with Keysight during the review I did of the U1282A was the inconsistency between the PWM output from the U1282A and what the U1272A was reading the PWM values where. They told me to do a factory reset...

Anyways just some observations of mine.

Kas

What the hell did I just watch... What was the point of your video, might I ask? To prove that you can throw some sand on your meter and then shake it off? Seriously?

[FrankBuss]

The sand test makes sense. Imagine you are on holiday at the beach and suddenly feel the urge to do some electronics work

[HighVoltage]

The other issue I noticed was when doing a sand test on the U1282A the display also went wild
Kas

How about a "sand test" on your other test equipment, may be a nice scope?
This must be the weirdest EMC issue test I have ever seen?
I feel really bad for your DMM.

This was part of your official review for Keysight?

 

[Bud]

The sand test makes sense. Imagine you are on holiday at the beach and suddenly feel the urge to do some electronics work

No, it is in case your boss sends you to install cell tower in Sahara desert. You want to be safe.

[hscade]

Hello,

after watching Daves review to the EMC fixed version of the U1272A, I wrote immediately a email to Keysight.
They offer me to replace it with a fixed U1272A (not Agilent branded  ) or one from the U1280 series (Dave did a review of the U1282A in the past).

Question: Is worth to step to the U1280 (I would take it only if it is the U1282A!)?

I was happy with the U1272A except of missing data logging(which is an option, i know) and the EMC issue which was only a concern because ... its Agilent and it was pricy.
Features like low impedance, auto hold  and LPF are "not needed", at least I've not used them very often in the past.

With the U1280 I would get the data logging capability without paying extra but would sacrifice the setup menu.

Mhhh ... decisions must be made ... but at what cost!!! 

[bitseeker]

It's certainly a tradeoff in features, shape, and size. Ultimately, you'll have to see what matters most to you. I went for the 1280, but it's a really big meter. Once the dust settles on the changes to fix the 1272, I'll probably mod mine.

[HighVoltage]

I also received the U1282A as a replacement and was surprised how large it is compared to the U1272A.
However, I do like it a lot.

[lowimpedance]

after watching Daves review to the EMC fixed version of the U1272A, I wrote immediately a email to Keysight.
They offer me to replace it with a fixed U1272A (not Agilent branded  ) or one from the U1280 series (Dave did a review of the U1282A in the past).

Well.... if the Keysight brand is not up to the task then you could always swap around the main PCB so the latest Rev 6 resides in the Agilent case !   .

Or just get the 1282 and do the mods to the 1272, then you have both to cover what ever need arises.

[bambadoo]

Filled out the Keysight form and sent it some days ago. Havent got a reply from Keysight yet. Does it take long?
Havent got a automatic "receipt" or "thank you for filling a form" mail either but its probably like that. 

Best regards.

[TheAmmoniacal]

Filled out the Keysight form and sent it some days ago. Havent got a reply from Keysight yet. Does it take long?

Best regards.

Took about a week for must of us I believe.

[LADmachining]

I completed the form last Friday. Have just received an acknowledgement email stating a representative will contact me  to confirm the delivery details.

[Kas.Lewis]

The point, for those who have read the IP67 spec and gone over the marketing material, was to verify the claims made by Keysight. They claim in both their marketing material as well as giving it an IP67 rating that sand should have no adverse affects on the unit. Unfortunately this was not the case. The dial started to wear down and as you can see electronically there appears to be something going on.

Kas