EEVblog #441 - How To Track Down Common Mode Noise

[nitro2k01]

[whatchitfoool]

The video linked at the end,   EEVblog #442 analog Vs Digital scopes. ,  is marked private and is unviewable.


! Private video

[BjornR1989]

It makes me wonder whether this is the cause of the false measurement on my Fluke PM3394B oscilloscope.

Less low resolution: http://shared.bjornroesbeke.be/.private/eevblog/IMG_20130322_011334.jpg

Less low resolution: http://shared.bjornroesbeke.be/.private/eevblog/IMG_20130322_011406.jpg

The input is ground coupled, yet it still doesn't draw a pure horizontal line, no matter what voltage or timebase i set it to. Even the peak-to-peak voltage measurement is way off.
I'm not sure what needs calibration, the oscilloscope or its owner.

[c4757p]

My Tektronix has a bit of similar ground noise. I don't really care about it, though, because it doesn't zoom with the signal - it is noticeable if the signal is perfectly flat, but it's never any bigger than what you're showing in that picture, even zoomed in all the way. (My guess is that it has to do with the fact that the inverter generating CRT bias voltages and the CRT signal-level circuitry are lumped together in the same shielded can.)

[Bored@Work]

The video linked at the end,   EEVblog #442 analog Vs Digital scopes. ,  is marked private and is unviewable.

Dave is probably still busy editing and uploading it.

[BravoV]

<sigh> .. another price hike in used quality probe at fleabay 

[ftransform]

This was the most exciting EEVBLOG in a long time. It felt like a detective story.

I Think I am going to start flipping breakers to determine what the fuck is causing certain readings in my house. 

[jahonen]

This noise is also described by Howard Johnson in his book High-Speed Digital Design.

10x setting gives more noise because 1/10x attenuation affects only differential mode signals. From oscilloscope perspective, the absolute noise level stays the same on both settings, but we have set the scaling differently on the oscilloscope. Remember that common mode noise is generated on the shield and not on the center conductor. A ferrite clamp around the coax might improve things.

This common mode noise is also the killer noise mode on EMC tests and usually the hardest noise type to get rid of.

Regards,
Janne

[robrenz]

This would be a great time for a contrasting follow up video showing what a high gain differential probe or amplifier can do in this situation.

[os40la]

Dave mentions about doing a high voltage probe video at 07:34 on video #441. I hope he does one on a laser power supply 40w. I would like to see how that can be done. 

[tylerl]

So in the end was the culprit RF noise being picked up by the coax? Or was the noise coming in through the power cable?

If it's coming in through the power cable, then I would think that a you could do some quick mods using ferrite beads to clamp that high-frequency noise down and quiet the signal... which *also* makes me think that with such an expensive scope, that sort of mod would already be done in the factory.

[Rerouter]

it was EMI being picked up by the transformer in the power supply, radiated noise... he described it quite well towards the end of the video,

[alm]

I didn't really see the point of putting the scope on an isolation transformer and disconnecting the safety ground. Not only is it potentially dangerous (especially for unexperienced people watching the video), but it's not very effective at the ~66 kHz noise frequency. The scope will still have a ton of capacitance to ground, since it's sitting on a grounded ESD mat and next to other grounded equipment. Which is why graybeards sometimes talk about hanging a scope by silk threads (and then tell you that you should never do it).

[Galaxyrise]

Why doesn't the PSU chassis shield the transformer in this case? 

[alm]

Maybe it's entering through the plastic front panel?

[JackOfVA]

Why doesn't the PSU chassis shield the transformer in this case?

At relatively low frequencies - and the induced signal Dave observed was at 60-odd KHz which is "low" for this purpose - magnetic shielding is difficult. At a few hundred KHz and above, Faraday type shielding is effective - surround the object radiating or receiving with a gap free conductor such as copper or aluminum. The ambient magnetic field then induces a current in the conductor that in turn produces a nearly equal and opposite magnetic field and hence the ambient field is cancelled. (Faraday's law of magnetic induction.)

As the frequency drops, Faraday shielding becomes ineffective and instead a high permeability magnetic "wrap" around the object is required that more or less shunts the ambient field around the object to be shielded. The most often used material is mu-metal and it's extremely expensive and must be handled and worked with care to avoid reducing its shielding effectiveness,

The most likely place one will run across mu-metal shields is around the CRT of an analog oscilloscope - all high performance oscilloscopes encased the CRT in a mu-metal shield to prevent nearby magnetic fields from affecting the CRT's electron trajectory.

Some very high quality audio transformers have a mu-metal shield, but it adds considerably to the cost.

You can read about mu-metal at http://www.mumetal.com/

[SeanB]

How to destroy Mumetal, bend it, dent it or curve it.  But it will do an incredible job of reducing magnetic fields inside an area provided you enclose it totally on all sides. A lot of microphone transformers use it in a strip wound inside the steel can, with a top and a bottom disk, with just a hole for the wiring. The steel attenuates the magnetic noise, and the mumetal inner does the rest. You can put a microphone transformer next to a valve output transformer and not get much feedback until you get to insane gains.

[c4757p]

The most likely place one will run across mu-metal shields is around the CRT of an analog oscilloscope - all high performance oscilloscopes encased the CRT in a mu-metal shield to prevent nearby magnetic fields from affecting the CRT's electron trajectory.

Ah! I used to have a very crappy oscilloscope that would distort when I turned on my power supply (huge transformer) nearby. Upgraded to a good one and it doesn't do that. I was wondering how they could keep the magnetic fields out.

[CHexclaim]

Is it me or when Dave eliminated the main sources of interference, the waveform on the oscilloscope varied as he spoke, as if there was some sort of interference coming out of the audio circuitry on the camera?

CH!

[CHexclaim]

Jee, I just watched the aside video comparing with an analogue scope. It is scaring to see how much knowhow and experience one has to have in order to use the measuring tools correctly. Jim Williams came to my mind.

Getting an incorrect diagnose of a problem seems inevitable at some point in life.

I was involved in the design of interference filtering so I know the before an after, but it is still scary.

CH!

[zinob]

I assume that the noise has virtually no power behind it so wouldn't adding a load of just a few milliamps cause the common mode noise to diminish tremendously? This would of cause not measure the PSU at no load, but making the (somewhat frivolous) assumption that noise is constant over the operating range of the supply this should make sense, right?

Sorry for the bump...