RF & Audio

[jkeny]

I'm prompted to start this thread by something that has been rattling around in my brain for a while (yes, before anybody else says it, there's a lot of empty space in there to allow rattling ) & I don't have the experience or knowledge to answer it myself. Maybe some of the more experienced could pitch in.

It was recently prompted by this post from DF96 on the Bybee thread "..... Stopping RF from getting into audio components can improve the sound, yet nothing will show up in a purely audio test."

I wonder how many agree with this? Can it be explained & what is the underlying mechanism?

 

[abraxalito]

Its been my hobby horse for quite some time - many years in fact since reading an article in a hifi magazine back in the 1980s - probably it was HiFi Answers or HiFi world. The writer concluded that RF effects on electronics was a significant part of the puzzle regarding perceived audio quality.

In my own tinkering, I find that grounding is a huge part of making a piece of electronics sound good. I have a theory that grounding affects susceptibility to RF, but its early days yet and I have very little hard data. I've been trying to capture poor audio quality by recording the output of poorly grounded kit - so far, nothing has shown up.

 

[sandyK]

I'm prompted to start this thread by something that has been rattling around in my brain for a while (yes, before anybody else says it, there's a lot of empty space in there to allow rattling ) & I don't have the experience or knowledge to answer it myself. Maybe some of the more experienced could pitch in.

It was recently prompted by this post from DF96 on the Bybee thread "..... Stopping RF from getting into audio components can improve the sound, yet nothing will show up in a purely audio test."

I wonder how many agree with this? Can it be explained & what is the underlying mechanism?

John
It may not be readily measurable at the analogue output, but I expect that when using a suitable preamp with a wide bandwidth CRO, it would show up at the bridge rectifiers, and in the case of ingress into the NFB network via long loudspeaker cables as well.e.g. "frame buzz" from nearby TV transmitters.
Alex

 

[abraxalito]

I've had this link open as a tab in my browser for months as its such a good article outlining not just the technical aspects but also mentions some audible effects of changing grounding. Its been put up before, no harm to share it again :

SoundStage! Pete's Place - JPS Labs Superconductor Interconnects: A Technical Discussion (09/1998)

'RF and cables' is the heading about halfway down. I hope this doesn't encourage the thread to degenerate into another cables slanging match though

 

[jkeny]

Thanks guys,
Yes, let's not have this thread hijacked, please. I think this is a fundamental concept & needs an open, non-judgemental discussion.

There are those that claim if it can be heard then it should be measurable on the analogue output. I can understand this logic but wonder what is the logical trap in this thinking? I'm grappling with these contradictory viewpoints!

 

[SY]

The effects of RF, when audible, are easily measurable on the output in the audio band. For example, think of an extreme case- rectification and detection- where you can actually hear radio signals breaking through.

 

[abraxalito]

There are those that claim if it can be heard then it should be measurable on the analogue output.

Well I do and don't subscribe to that view. I do in that something must be measurable, but I don't in thinking that that measurable thing must necessarily be audible - i.e. in the audible band. Its a fallacy to measure the output of say a CD player and say because there's no difference in the audio band therefore there can be no difference by the time its reached our ears.

The reason for this fallacy is overlooking downstream non-linearities. Intermodulation isn't zero in most amps/pres, especially when that's intermodulation with out-of-band signals.

 

[abraxalito]

For example, think of an extreme case- rectification and detection- where you can actually hear radio signals breaking through.

Yes, I get this and would like to figure out how to get rid of it. Do I have to adopt valves? I agree, that's totally measurable. But how about lower level effects when we're not so sure that its the GSM phone making the difference? Do you discount those effects?

 

[SY]

Your example is somewhat inapt- the effect is still measurable in the audio band, just a bit further downstream of where it occurs. Agreed that two inaudible signals can intermodulate and produce something audible, but that produced signal is absolutely measurable in the audio band.

 

[abraxalito]

Your example

Which?

 

[jackinnj]

The effects of RF, when audible, are easily measurable on the output in the audio band. .

Easiest of all is the noise from a switching power supply - or a switching ps which isn't designed with harmonics and radiation in mind.

...and if we don't have enough of them already -- wait until there are no incandescent bulbs -- replaced by CFL's and LED's with their noisy LED drivers!

 

[DF96]

If it is clearly audible, then it will be easily measurable in the audio band. I think the problem comes from somewhat lower levels of RF interference. It could be intermittent, and it will certainly vary from place to place and time to time. In measurements it might be indistinguishable from random noise, even though it is not really random. Averaging, when investigating low levels of distortion, may average away the effects of RF as it will not necessarily be correlated with the test signal.

I have put a long post on this topic in the original thread referred to by the OP. Perhaps the mods could move it here, as I had not seen this thread until after I posted it.

 

[SY]

Moving individual posts is not trivial, so as a Certified Lazy Person, I'll pass on that task.

Sorry to make you state the obvious, but your original post which started this discussion was, I think, misunderstood to mean that a measurement in the audio band of the output would not show differences, despite audibility. That clearly wasn't what you meant and I understood that, but... I appreciate your clarification.

 

[jkeny]

OK I've copied it here from the Bybee thread

There is a hidden assumption in most audio testing: whatever comes out was either in the input (from my signal generator), or was generated purely internally by the device under test. There is another source of input signal: the environment. By this I mean the mains supply and surrounding EM fields. These do not necessarily enter at the input socket, although they may do this. If the environment generates a large signal this will be noticed by a tester. If, as is more common, it generates a small signal the tester might not notice or may attribute problems to the DUT or the surrounding equipment.

Many, but not all, audio designers know nothing about RF. You can tell this from the fact that they either never mention it, or spout nonsense about it. This means that their amplifiers may have good or bad RF performance, as a matter of luck. An amplifier with good RF performance can cope with interference, a bad one not. The RF getting in will depend on the environment (e.g. nearby base stations), and the cabling. An amp with poor RF performance may be more fussy about cabling, as cables can differ greatly in their RF pickup. As a result the poorer amplifier may be regarded as having "better discrimination", because it is fussy about cables and prefers those with low RF pickup. (In my opinion, for this and other reasons, cable fussiness is a sign of poor electronic design - many people sincerely believe the opposite!).

We can't hear RF, but device distortion or overloading can convert RF into audio. In listening this might be heard as noise or vague fuzziness. Instruments might not always see it, especially if averaging is used, as the RF-induced noise is not correlated with the sig gen but neither is it random. Two people testing the same equipment in two different locations, or at the same location at different times, may get different listening results which may or may not show up in measurements. When people have gone looking for RF in audio equipment they usually find it, although levels vary greatly between different designs.

RF interference, unless severe, will not show up in audio tests. Even when visible in a measurement it might be ignored, unless the tester is aware of it as a possible issue. Perhaps audio test results should always be accompanied by a wideband RF spectrum from a standard antenna in the same room, so we know what EM soup the amplifier was sitting in.

 

[DF96]

Thanks. I too am a CLP!

I remember reading a couple of articles many years ago (in Wireless World or Hi-Fi News, I forget which) on this. The author measured RF at various places in a couple of commercial amplifiers. It was being picked up on the speaker leads, then going via the feedback loop to the input LTP. In one or two cases the level was high enough to be a potential problem. Ever since I read these I have thought about RF in audio. Valve circuits are probably more resilient on average. Zero feedback designs might avoid some problems, although they may still suffer from RF getting in via the input socket.

I am still thinking about how best to detect (sorry!) this problem via measurements.

 

[jkeny]

Moving individual posts is not trivial, so as a Certified Lazy Person, I'll pass on that task.

Sorry to make you state the obvious, but your original post which started this discussion was, I think, misunderstood to mean that a measurement in the audio band of the output would not show differences, despite audibility. That clearly wasn't what you meant and I understood that, but... I appreciate your clarification.

....... but your original post which started this discussion was, I think, misunderstood to mean that a measurement in the audio band of the output would not show differences, despite audibility. That clearly wasn't what you meant and I understood that, but... I appreciate your clarification.

If you mean this SY

"..... Stopping RF from getting into audio components can improve the sound, yet nothing will show up in a purely audio test."

No misunderstanding - it is clearly stated.

 

[abraxalito]

If you mean this SY No misunderstanding - it is clearly stated.

I took that to mean that the normal kind of audio measurements (you know, single tone, modulus only, multiple averaged FFTs ,THD and such like ) would detect nothing amiss. Whereas in principle the effect is measurable, just we haven't yet devised suitable measurements which will reveal it clearly.

 

[jackinnj]

Analog Devices has written a number of papers on interference with precision measurement. I think that one of the techniques they discussed was just using an am radio as a sniffer, another was the French engineer using his hand-held ham radio transceiver to bounce a few watts around. I like the term "unexplained DC offsets" -- which in audio land cause distortion from one side of an amplifier to "miss" 0 V DC longer than anticipated.

 

[DF96]

Perhaps I should have said "nothing may show up in a purely audio test" i.e. it may show, it may not. I apologise if I have sent people off on a wild goose chase.

My point was that the effects of RF may not be recognised. In the original context (fancy cable-like devices) a device test using purely audio signals might not show anything, if the actual mode of operation of the device is to stop RF. You would need to do RF measurements to see anything.

If a whole audio system is measured, then it ought to be possible to measure anything that can be heard (whether due to RF or no) but first you have to work out how to measure it. This is because it is neither random nor necessarily correlated with the audio signal, so the tricks used to enhance measurements may hide the effect. The fact that something is in principle measurable does not guarantee that it will be measured!

 

[Max Caliber]

My point was that the effects of RF may not be recognised. In the original context (fancy cable-like devices) a device test using purely audio signals might not show anything, if the actual mode of operation of the device is to stop RF. You would need to do RF measurements to see anything.

I don’t think using RF measurements on the output of an audio device would be of much use as the frequency of the interference would no longer be in the RF domain but would more likely be de-modulated into the audio frequency domain.

Although I haven’t tried it (yet), I believe that using the Audio DiffMaker software, one would hear the effects of RF interference as an elevated background noise.

See the following sites:
http://libinst.com/index.htm
http://libinst.com/Audio%20DiffMaker.htm

Cheers

Philip