Passive IM Distortion in Cables and Connectors

[hifiZen]

Here's an interesting paper about the effects of PIM (pasive intermodulation) as it pertains to RF applications. Note in particular the distortion levels being discussed (-153dB!) as well as the suggested remedies.

Tests have shown that nickel plate under gold on the center contact will typically result in a 40 to 50 dB increase in PIM. Stainless Steel in the body will usually give a 10-20 dB increase in PIM.

This paper provides some interesting validation for the claimed subjective differences between various cables and connectors, and at the distortion levels being discussed, suggest a reason why these effects have not been easily observed in many attempts to measure such differences.

Here's the link: http://www.amphenolrf.com/simple/PIM%20Paper.pdf

[Jonathan Bright]

Thanks for the link=interesting.

[Steve Eddy]

Originally posted by hifiZen
This paper provides some interesting validation for the claimed subjective differences between various cables and connectors, and at the distortion levels being discussed, suggest a reason why these effects have not been easily observed in many attempts to measure such differences.

Interesting paper.

But I don't see how it validates anything along the line of claimed subjective differences. Seems to me those sorts of calims can really only be validated by demonstrating actual audibility.

se

[hifiZen]

Well, perhaps 'validation' was not the right word. 'Correlation' is more along the lines of what I meant.

I am by no means taking sides in this issues. I have always been, and shall remain, skeptical of any subjective claim which doesn't at least have a plausible theory to explain it, or better, some strong scientific evidence. But, I also try to be open minded about such things. This paper is obviously just one small piece of the picture, but I think it provides some interesting food for thought.

[TNT]

The frequency band where this is discussed is far from the audiable range. I wonder what it looks like in the 0-100khz region?

/

[Steve Eddy]

Originally posted by hifiZen
Well, perhaps 'validation' was not the right word. 'Correlation' is more along the lines of what I meant.

I am by no means taking sides in this issues. I have always been, and shall remain, skeptical of any subjective claim which doesn't at least have a plausible theory to explain it, or better, some strong scientific evidence. But, I also try to be open minded about such things. This paper is obviously just one small piece of the picture, but I think it provides some interesting food for thought.

I keep an open mind as well. Just that the way I see it, until actual audibility is established, all this other stuff is meaningless except perhaps in an academic sense.

People have been spinning theories and speculations for decades. But until audibility is established, you can't really put those theories to the acid test. But then I think that's precisely why some sit around and do nothing but spin theories.

se

[FLZapped]

First of all, look at the levels being discussed. Also look at the depth of the conducting surface in the frequency range mentioned. A lot of power in very little material. 0.002 mm as compared to 0.5mm at 20kHz. That's a 250:1 disparity. Add to that the fact you must maintain strict impedances within the connectors and you have a compounded problem because you cannot just randomly increase the contact surface area as you can in audio.

Second, the concern here is interference to other services. When you have a 100 watt transmiter(+50dBm) and a co-located recever with -120dBm sensitivity, it isn't hard to figure out that a -150dBc spurious signal is at -100dBm absolute level, or 20dBm above the base sensitivity of the receiver! The receiver becomes nearly useless if that spurous signal falls in its channel bandwidth except for very strong signals. Not too likely with cellphones transmitting only a couple hundred milliwatts. (+20dBm for 100mw) (Provided they even radiate all their power, which they don't by a long shot)

This is a real and defined problem, especially in multi-channel trunked sites and multi-channel cell sites that have their signals all combined onto a single antenna. Any non-linearity outside the isolation system and you have the potential for a serious nearly unresolvable interference problem. All non-linearities we are concerned with in audio are inband, unlike rf.

Sorry, but there is no corellation to the audibility question with regards to audio systems.

-Bruce

[john curl]

I think that this input on PIM is important. I learn from it, but then I design hi end audio equipment for a living, so I don't easily dismiss distortion generation from any potential source. It is attention to details, such as whether you use a cheap audio connector that is made of magnetic materials and / or has nickel plating just under the gold plating. We have evaluated components at this level, for many years. In fact, we can change the sound of an audio component by just changing the connectors. Can't measure much however. Probably, it is because of limitations in our test methods and equipment.

[janneman]

It is indeed an interesting paper. To put it in perspective, I note the following:

The problem of PIM in passive components is increased by the 50 Ohm impedance levels in the RF bands discussed. If those impedance levels would be much higher, the impedance variation effects that lay at the root of the problem would me much less also, decreasing greatly the PIM. That would mean for audio that the problem could be present in the power amp/loudspeaker interface, but not in signal level interfaces.

If I understand the paper correctly, the problem is also acerbated by mismatch caused by the contact variations, which causes termination reflections and extra PIM. In audio, this will not happen as the cabling length is always very much less than the signal wavelength so impedance matching is a non-issue. For the same reason the skin-depth in audio is much greater than in cell phone equipment, which seems to imply that microscopic contact variations have less overall impact than at rf with very low skindepth.

The levels discussed (some -150dB below the signal) are at least a factor 100 to 1000 below the levels of known distortion mechanisms in contemporary audio technology. It seems therefor extremely unlikely that this is a problem in audio. Nevertheless, it is another lesson that contact quality and reliability is important. We should all clean and re-seat our connectors at least a few times per year, especially in dusty and/or industrial areas and/or smoking venues.

Jan Didden

[dimitri]

some thoughts
http://www.angelfire.com/ab3/mjramp/golopid/wire.html

Here you can find measured nonlinear characteristic I-V for a coaxial switch
http://www.estec.esa.nl/conferences/...apers/a022.pdf


Some simple design rules from http://www.astrolab.com/13.asp

Use of semi-rigid cable with a seamless outer conductor in place of flexible cable.
Use of a solid center conductor in place of a stranded center conductor.
Directly attach the outer conductor to the connector body by soldering or clamping in lieu of crimping.
Limit the number of parts in the current path.
Eliminate contaminants in the current path.
Use high quality machining in the connector parts with a smooth surface finish.
Avoid contaminants in the plating solutions.
Ensure adequate and uniform plating thickness.
Avoid use of magnetic materials in the current-carrying-path.
Ensure adequate contact pressure.
Contact surface of female contact fingers should cover as close to 360° as possible (i.e. narrow slots or slotless).
Use connector interfaces with radial dimensions as large as possible