A chip-amp to rival Hi-End - design advice

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
As far as I can see, you have respected him in your writing. He certainly seems to have taken offense, but you haven't given any that I can see.
I think the difference here is because of our different goals: my aim is to have as good a system as possible, and that everyone else also has as good as possible system - this is simply because I love music and sound, so I want the best for all; Thorsten's aim is a bit more egocentric because he also wants money for himself - which is only natural, and I understand this.

Of course one could argue that Peter Daniel also makes money from his amplifiers but he always shared his findings, and would help anyone who asked, as his time allowed... The simple truth is that he has a unique mixture of knowledge, skill, experience and equipment - so no one will be able to copy his amplifiers anyway. But they can try, and perhaps discover something great for themselves along the way - like I did :)
For this we can do a quick sum. A 2m length of wire could have an inductance of 2uH. At 1MHz, such a wire will look like an impedance of j12.5ohms. I'd say probably it isn't worth it, but then I don't know yet which RF frequencies are the worst offenders in degrading audio quality. My initial guess is in the band 1MHz - 30MHz - your earth wire will look too high impedance in that bandwidth. To deal with RF in this band, I'd say its better to look to ferrites, which absorb the energy and turn it into heat. Also fit a mains input filter which will contain a common-mode inductor.
To be honest, I find it a bit hard to comprehend how can RF noise influence the sound anyway... Logical reasoning would suggest that 1Mhz noise will result in distortion at the same frequency (or its harmonics). So how would this degrade the audible range? Especially that RF noise is usually very low level and wouldn't waste any significant amount of amplifier's power (I think?...).

My other question is - putting a choke on your mains cable filters out some of the RF noise - but what is the trade-off? From what I learned so far, there is hardly a thing in audio world that adds something without taking away something somewhere else.
 
To be honest, I find it a bit hard to comprehend how can RF noise influence the sound anyway...

Yeah, I know the feeling - for a long time I knew that RF was a potential problem in hifi (from an article many years back in HFN+RR I think) but only recently did I listen to sound changes which I can only explain by appeals to RF interference.

Logical reasoning would suggest that 1Mhz noise will result in distortion at the same frequency (or its harmonics).

Well the first thing about RF noise is its broad-band, not just one single frequency. RF is known to cause shifts in the offset voltage of bipolar opamps - Walt Jung found the AD797 gave erratic results in one of his low noise power supplies, which he laid at the door of RF. Changing to a JFET input opamp fixed that particular problem. More recently, National Semi has introduced some CMOS opamps which are marketed as 'EMI hardened' (LMV861 for example) and various application notes I've found show opamps outputs shifting from just the application of an RF carrier.

I haven't yet worked out what happens when music is present but I'm at present guessing its intermodulation. How it sounds is it makes higher frequencies more harsh, strident, sibilant. I particularly notice it on certain female vocalists - Cara Dillon is my torture test at the moment. I recently changed an amp PSU from linear to SMPSU and found her voice became too sibilant.

So how would this degrade the audible range? Especially that RF noise is usually very low level and wouldn't waste any significant amount of amplifier's power (I think?...).

Yes its low level but the total energy involved must be fairly significant - not from the point of view of wasting amplifier power, but for producing audible intermodulation products.

My other question is - putting a choke on your mains cable filters out some of the RF noise - but what is the trade-off? From what I learned so far, there is hardly a thing in audio world that adds something without taking away something somewhere else.

Its a very good point, and yes there is a trade-off. Of course a CM choke increases the impedance that the transformer's primary sees looking into the mains. So regulation suffers, but hopefully this is minimal. Can't think of any other downsides except for the potential problems caused by running higher leakage currents through the safety earth - if the earth is already quite a high impedance, adding a filter has the potential to make things worse.
 
I have been doing further research into the matter of audio cables for the last couple of weeks. Here is one of the files I stumbled upon. I think it is very valuable, because it's based on real-life experimentation.

In short, it further confirms that thin rectangular cross-section (ribbon) is the optimal conductor shape. It also hints at the actual ribbon dimensions that were found to (subjectively) sound best.

It also confirms experimentally what I felt intuitively: that natural insulation materials (silk being recommended in particular) are superior to plastics in audio applications.

The author also recommends some other techniques, like gold-plating, to further improve the qualities of the cable, but I feel this is going too far into the "very costly and very difficult to do" territory. Too far for me anyway. Gold-plating isn't a difficult process, it's doable even in household environments. But since the thickness of the plate is suggested to be a significant factor, I feel it would be extremely difficult to do properly and the price of a mistake would have been disproportionally high compared to the potential gain. - Still something I'd love to try one day ;)

Ribbons have also one very convenient quality - they don't need any termination, other than making a suitable cut-out at the end of the ribbon, which then will work the same as a spade plug - but without the disadvantage of its soldered/crimped connection.

That leaves only one important question unanswered: how should the ribbons be arranged in a pair, relative to each other? Closely spaced or far apart? Parallel width-wise or perpendicular?
- Thankfully, this should be easily and cheaply answered by experimentation. And then we should arrive at something very, very good; Expensive, yes, but possibly on par with Silversmith Audio's multi-thousand-dollar cables - good enough for me ;)

Of course, this is not the end of the road: we could try Silversmith's "air dielectric" approach, we could try plating or alloying silver with palladium, try adding insulating/dampening materials etc. etc. Some people swear that cable stands improve sound - seems silly, but hey, it's cheap to try and make some wooden stands, so why not try. Then there's all sorts of cable conditioners and lacquers - but that, to be honest, is going a bit too far into "audio voodoo" even for me...

What are your thoughts on this? :)
 
Sorry for a really late reply... :(

I don't going to argue the meaning of damping factor again. If you think it is meaningless, fine I think different.
sorry, I answered your post before reading the rest of the thread so I had not seen you had already been arguing about that with Thorsten.

Nevertheless, let me point out a few things. At least for correctness wrt those forum readers who do not have enough knowledge on the subject and may be confused.

This is not a matter of "thinking different". Whether or not a given variation of the so called "damping factor" may affect the actual acoustic output from a given loudspeaker is not an opinion! It is a matter of simple math based on fundamental laws of physics (and electronics).

First of all, the name "Damping Factor" (referred to an amplifier) is at best a misnomer. In fact, it is just a disguised, convoluted way of expressing the amplifier output impedance (and only its module, to be exact). Nothing else.

It has really little to do with whatever (real) loudspeaker (resonance) damping you can get in practice. For that will depend much more on the speaker itself than on the amplifier.

So, I state that the "amplifier DF" (as commonly intended) is just a marketing buzzword (and a technical nonsense).

Let see why. Consider the output loop (circuit). You have a simple closed circuit which contains a few elements in series with each other:

1) amplifier output impedance
2) cable resistance (and series inductance)
3) contact resistance of the various connections
4) cross-over equivalent series impedance
5) loudspeaker voice coil (DC) resistance

for the sake of simplicity, let ignore the contributions from 2), 3) and 4) and focus on just 1) and 5). That is, consider a simplified model with a single driver directly connected in an idealized way to the amplifier output.

As you can easily verify, for most loudspeaker drivers the DC resistance of their voice coil is of the same order of magnitude of their "nominal impedance". That is, for an "8 ohm" driver it is likely several ohm.

Again for semplicity, let assume it's 4 ohm (an already rather low value).

What is the maximum effective (real) "damping factor" that you can get with such a speaker?

Even if we place a perfect short circuit between its terminals, (if we keep referring to its "nominal impedance") we can do no better than DF=8/4=2 !

That is, even using an ideal, perfect amplifier which have an output impedance of exactly 0 ohm (and thus would show a DF = infinite on its spec sheet), the effective DF would be no greater than a mere 2 !

Now consider an ipotetical amplifier with a really very high DF, let say 32,000 (that is an output impedance of 0.25 mOhm).

For sure, 32,000 is way much less than infinite. So we should expect a much worse result. Should we? Let' see. Effective damping now would be:

DF = 8/(4+0.00025) ~= 1.999875...

that anyone would still round to 2.

We have made an infinite (!) worsening of the DF, yet we got a completely negligible difference in effective damping!

What if we use some real amplifier with a much lower DF, let say of "just" 32 (that is, an output impedance of 250 mOhm) instead?

Well, effective damping would change from the previous ~ 2 to DF = 8/(4+0.25) ~= 1.88

OK, now we begin to see some difference. Yet, it's a mere ~ 6% difference in effective damping for a huge (3 orders of magnitude!) change of DF.

But wait, we're not done yet. Because also these numbers are in fact meaningless!

Here comes the other nonsense of the "DF" (as a measure of real damping).

The "damping factor" refers implicitely to the capability of electrically damping the (mostly mechanical) resonance(s) of the loudspeaker.

Of course, speaker impedance are by no means like a flat, resistive value equal to their nominal impedance (if they were, there would be no resonances, thus no damping to speak of).

The speaker impedance @ resonance is much higher than its "nominal" value. How much basically depends on the (mechanical+electrical) "Q" of the whole system. That is it depends on how much any given speaker is "self-damped".

(BTW: that "Q" is exactly what you aim to alter adding the output impedance of your amplifier to the mix).

Needless to say, the higher the "Q" (and thus the impedance @ resonance), the more "external" damping you'll need to flatten the response.

BUT, the effective, real "damping factor" actually depends on that impedance. As the speaker impedance @ resonance increase, the required "damping" impedance (from the amplifier) increases too!

In practice, speaker impedance @ resonance usually range from twice to several times the "nominal" value. It can be as high as over 100 ohms. If we now consider the previous example, but rather than using a value of 8 ohm we use some more realistic one, let say 32 ohm, this is what we get:

short circuit (DF = infinite):
DFe = 32 / 4 = 8

Zout = 250 uOhm (DF = 32,000):
DFe = 32 / (4 + 0.00025) ~= 7.9995... ~= 8

Zout = 250 mOhm (DF = 32):
DFe = 32 / (4 + 0.25) ~= 7.53

Draw your own conclusions...

As already stated also by Thorsten, IMHO it is more than evident that the reason(s) behind your perceived differences in bass response can not be due to the difference in "DF".

Besides, given that we're talking about AF and resonances, you should know that there is not only the module to consider... phase can be as much (or even more) important.

BTW, this is yet another reason why the amplifier "DF" figure is basically meaningless: it completely ignores phase, i.e. the fact that in real amplifiers the output impedance is by no means purely resistive (particularly in high-NFB circuits such as the chip amps, where the dominant pole compensation makes the amplifier output impedance inductive).

If you choose inverting configurations, you also need higher resistor values
true. There are also other ways to work around that, though.

I am glad you share your enthusiasm for Mauro's design,
well, not quite. I like a lot many of the fine though, research, ideas and principles behind it. It's a very clever design. Way much more complex and refined than it's apparent semplicity may suggest. Yet I can't say to be particularly enthusiast about it as it is. When (a long, long time ago...) I was studying EE, I was told that good circuits should not depend on critical values and/or selection of parts. And I still believe that it was a very good advice. For that reason I can not like the many critical and/or matched parts (compensations, balanced "bridge", ...) which are required by Mauro's clever design to operate properly.

On the other end, I've heard at least one well implemented and optimized build which is working & sounding pretty well. Surely much better than any other "conventional" chip-amp I've heard so far (including some ridicously expensive commercial offers).

But I can say, that personally I never liked the many compensation poles needed in his design.
well, although those compensations are making the amplifier overly critical with respect to parts choice and selection, they are also the key to some of the most important and interesting feature that makes that design almost unique!

Look at the almost ideal "reverse driven" response (that is, the amplifier output impedance module & phase vs. frequency - call it "complex DF", if you like) of the "rev.C" version. It is almost perfectly flat and resistive over all of the audio band. Now have a look at how does it looks like in a "normal" chip-amp.

Given the importance you give to "DF", than the "reverse driven" behavior should be absolutely paramount to you.

(on the other end, personally for audio I mostly like tubes... and am more than happy with output impedances of the order of a few ohms).

Then you should also read jneutron's debunking of that paper.
...
"Good John Curl Article on Capacitors"
I would not call "debunking" what is no more than an unproven, anonymous rumor!

If someone really wants to prove that Hawksford was wrong, than he should at least take the time to show what he's talking about. He has to redo all the math using what is supposed (or he suppose?) to be the more proper (?) way to apply Maxwell's equations to the case in question and eventually show that the results brings to different conclusions. And that's not enough, 'cause also the second work need be peer-reviewed to check for errors. (who tells which one of the works is right?)

Until someone will do that - not having enough knoledge and skill on the subject to try and judge myself - I'd rather trust more a paper from a well-known and respected audio researcher (and professor) than some unknown, anonymous and vague critics to it. Wouldn't you?
 
To be honest, I find it a bit hard to comprehend how can RF noise influence the sound anyway... Logical reasoning would suggest that 1Mhz noise will result in distortion at the same frequency (or its harmonics).
ever heard about intermodulation?

non-linear systems behaves quite differently from linear ones... for such systems the usual basic math does not holds true! 1+1 =/= 2 !

Especially that RF noise is usually very low level
not true either. When you have a lot of NFB, the error signal (what you actually amplify) is extremely small (consider the output voltage divided by the gain of your chip amp... which is typically in the order of billions!).

Actually, the RF noise can easily be much larger than that!

Here: http://www.dnm.co.uk/aboutamps.html

there is a simple, non-technical (but mostly correct) explanation of several of the problems you're facing.

World is not that simple...

(otherwise, we would all have the same, perfect amplifier since a long, long time...).
 
Until someone will do that - not having enough knoledge and skill on the subject to try and judge myself - I'd rather trust more a paper from a well-known and respected audio researcher (and professor) than some unknown, anonymous and vague critics to it. Wouldn't you?

No.

Jneutron (John Escallier) is neither anonymous nor unknown.

The sad thing is that what he's written about Hawksford's paper is scattered across a number of threads on at least two audio forums. I wish he'd put it all down in a single piece that could easily be linked to. He'd previously contacted Hawskford about this but Hawksford turned tail and ran away.

As for the "paper," it's never been been published anywhere except two consumer audio magazines (Hi-Fi News and Stereophile). Which I find a bit odd for someone who obviously isn't shy about publishing in professional journals. Especially regarding something as profound as the claims he makes. If he's correct, it would be of tremendous significance far beyond the world of home audio. I guess he figures that audiophiles are the only ones deserving of his discovery.

Personally I think he knows he screwed up but doesn't have the huevos to own up to it. And why should he? He's safe. Since it was never published in any peer reviewed journal, all he has to do is ignore it and he knows he can rely on others to defend him using logical fallacies like appeal to authority as you do here.

se
 
No.

Jneutron (John Escallier) is neither anonymous nor unknown.
if I've not misunderstood, in the thread I've been pointed to he says that someone else - who don't want to work for free and wants to remain anonymous - have expressed those critics.

Nevertheless:
he knows he can rely on others to defend him using logical fallacies like appeal to authority as you do here.
sorry, but this is exactly what you're doing now with John's opinion... :cannotbe:

I know very well that appeal to authority is NOT a proof. Yet, given no other/better option to judge, to whom should I believe?

Or is there a real proof, somewhere? (where?)
 
Last edited:
Or is there a real proof, somewhere? (where?)

Hawksford

Do Wires Really Sound The Same?

See posts by jneutron.

Just as an example:

Points of error (not a complete list, but enough to show the article as having fatal flaws.

1. Page 6..skin depth. The equation shown is the penetration depth equation, and is accurate for large planes of conductive material. When this equation is used for wires, it is referred to as the skin depth approximation equation. Approximation because at lower frequencies such as audio, it becomes inaccurate. The best test of it's applicability is to first calculate the skin depth at the frequency and in the wire material. If that depth is 1/5th (or more) of the radius of the wire, then the equation will not give exact results...it falls apart. At this point, had the paper been submitted to a publication I referee, it would have been called to task; since further use of an inaccurate equation cannot result in accurate results.

2. Page 7:quote:
""propagating electromagnetic wave consists of an oscillation of energy back and forth between the magnetic and electric fields, the energy in the electric and magnetic field must therefore be equal. Think of space (both in general and within the dielectric of the cable) as a distributed LC (oscillator) network.""

This is incorrect. The magnetic and electric fields are in phase. Meaning, when the energy in the magnetic field is maximum, so is the energy in the electric field. They are indeed exactly the same amount, and the field relations are dependent on the characteristic impedance of the medium.

References for my statement:

Jackson, "Classical electrodynamics" second edition, page 271, equation 7.11 and text as follows.. " ...eq 7.11 implies that E and B have the same phase"

Rojansky, "Electromagnetic fields and waves", 1979, page 390, figure 239 and 240 clearly show the electric and magnetic field as in phase.

Shadowitz, "The electromagnetic Field", page 539 figure 15-2, also depicts the fields as in phase.

And, most of all, THIS ARTICLE of Malcolm's, figure 1, even HE shows them as being absolutely in phase in his diagram. So his statement is not consistent with his own drawing, and the drawing is correct.

3. Examine figure 3. He has drawn the wire currents both in the same direction. That is incorrect. One goes the other way. This could of course be a simple typo, however in the final article which was published, he likened the fields at the surface as that of water flowing in a river, which textually was consistent to the inaccurate drawing, but inaccurate to a correct one.

The balance of the article discusses grain boundary discontinuities altering transmission, but yet there is no basis in fact presented, upon which to build these assertions.

In the final published version, he actually tested a wire pair and presented a scope waveform which verified his claim. Unfortunately, if you examine closely, you will note that he used STEEL wire for the test, as opposed to his discussion of copper throughout the article. John Curl asked Malcolm, and reported on this forum, that the steel had a permeability of about 100, meaning that the wire itself presented a rather large inductance to the test configuration, but yet this inductance was ignored. (the internal inductance of a wire is 15 nH per foot times the permeability of the wire..this means 1.5 uH per foot per conductor was ignored). This presents a lumped inductance which is quite consistent with the test results that were attributed to a "loss component".

As I stated in the past, peer review would have prevented this article from being published, as it has several fatal flaws. I wish everybody would allow this article to die a reasonably quiet death, as Malcolm deserves that much.


se
 
I find the whole discussion about Hawkesford's paper interesting, but largely irrelevant :)
That cables do sound different is a fact.
If Hawkesford's paper - be it wrong or not - led to development of better cables then so much the better for all of us!

Personally, I intend to focus on the more practical side of cable science... Well, I will do as soon as I gather enough courage to spend a few hundreds of pounds on solid silver ribbons... :rolleyes:

UnixMan said:
ever heard about intermodulation?

non-linear systems behaves quite differently from linear ones... for such systems the usual basic math does not holds true! 1+1 =/= 2 !

When you have a lot of NFB, the error signal (what you actually amplify) is extremely small (consider the output voltage divided by the gain of your chip amp... which is typically in the order of billions!).

Actually, the RF noise can easily be much larger than that!
Thanks UnixMan, this makes perfect sense. I've just ordered a telephone extension cable and I'm going to move my modem and wireless phone docking station to a room across the house - away from my HiFi! ;)
 
Of course, there is the matter of subjectivity as well.

But I don't agree with your point.

If I say "It is a fact that a trumpet sounds different from a french horn", and someone else says "They both sound the same to me" - it's not the physical differences in instruments that are in question here. It's the inability of one individual to discern a difference which is plainly obvious to another. That does not, in any way, change the fact that a trumpet IS different from a french horn, and DOES sound differently.

The analogy is clear - if one cable is physically different from another, there has to be a difference in their sounds too. Whether one is able to tell that difference is beside the point. (although whether the difference is significant certainly is of interest)
 
Besides, out of curiosity - what cables are you using? And why? Have you tried using, say, 20m of neatly coiled bell wire instead? Or do you claim that it would make no difference to the sound anyway?

Certainly one can make a cable bad enough as to alter the signal enough to actually be audible. But not every physical difference will be.

se
 
plese quit batting around those strawmen - some of us have allergies

64 Kb mp3 clips are sufficient for most to distinguish families of musical instruments

we can easily point to harmonic structure in the waveforms - and compare against masking thresholds to predict that most people will hear those differences


but I don't see cable believers trading clips at any high resolution to demonstrate the differences/flavors they hear in reasonable interconnect - seems like a major missed marketing opportunity - can't you just see it: Cu_pvc.wav, Ag_ptfe.wav, Au_cotton.wav...

after all many subjectivists when faced with objective measurements or engineering calculations showing the effects they are ascribing perceived "audible" difference to are way smaller than other errors of similar character in their signal chain resort to claiming they can "listen through" the other defects

right up to the point they decide to flip-flop to "your system isn't resolving enough" - somewhere around when blind test protocols are introduced to the argument



some choose the informed electrical and psychoacoustic engineering based argument that the inevitable differences between any physically distinct systems, in the case of audio interconnect cables, can be orders of magnitude below worst case estimates of human auditory resolution for technically competent interconnects

and await those (so far missing) properly blinded, controlled listening test reports showing cable differences
 
Last edited:
I know very well that appeal to authority is NOT a proof. Yet, given no other/better option to judge, to whom should I believe?

Since you're asking here - you shouldn't believe any person. You should look at the arguments only and not pay attention to the person making them at all. Paying attention to the person would mean you're giving weight to authority (or lack of it).

So, when you said:

Until someone will do that - not having enough knoledge and skill on the subject to try and judge myself - I'd rather trust more a paper from a well-known and respected audio researcher (and professor) than some unknown, anonymous and vague critics to it. Wouldn't you?

You were obviously paying attention to the persons of Hawksford and jneutron. That's just intellectual laziness. Your knowledge and skill on a subject are your own responsibility - you're free to put in the effort to increase them. The person you were hoping who would help you would help only by showing where Hawksford made mistakes, and jneutron helpfully does indeed do this. Then its for you with your skill and knowledge to decide if he makes errors in his showing.

At the end of the day, its only a person's knowledge and skill which matters. An intellectually lazy person will trust another's statements of authority and not be overly concerned that that person doesn't show why he's right. So learn to discern the difference between 'showing' and 'telling' - the former is art, the latter, propaganda.
 
The analogy is clear - if one cable is physically different from another, there has to be a difference in their sounds too. Whether one is able to tell that difference is beside the point. (although whether the difference is significant certainly is of interest)

Looks to me like your reasoning is flawed. Sound really does mean perception by the ears. If someone hears no difference between cables, then there really is no difference in the sound of the cables. What a person is able to 'tell' about the sounds he or she hears is quite secondary.

So no, not all cables do sound different, even though they may be physically different (and therefore must measure different).
 
but I don't see cable believers trading clips at any high resolution to demonstrate the differences/flavors they hear in reasonable interconnect - seems like a major missed marketing opportunity - can't you just see it: Cu_pvc.wav, Ag_ptfe.wav, Au_cotton.wav...

Its a really cool idea, I'd love to capture the differences I hear between cables (hypothesized to be differences in common mode RF levels at the receiver) in a .wav file. Maybe one day I will be able to post these up online:D

after all many subjectivists when faced with objective measurements or engineering calculations showing the effects they are ascribing perceived "audible" difference to are way smaller than other errors of similar character in their signal chain resort to claiming they can "listen through" the other defects

Yeah, its because the 'defects' are of quite a different form. Just like amplifier defects are quite different in form to speaker ones. But I'm not of the view these (assuming we are indeed talking about the same things) audible defects are small ones - I can hear them when not even in the same room, I hear them with just one speaker powered and listening around the back of it.

right up to the point they decide to flip-flop to "your system isn't resolving enough" - somewhere around when blind test protocols are introduced to the argument

Its true that if a system is already plagued with RF hash, then small differences in CM RF at one stage are going to be swamped.
 
<snipped>

To be honest, I find it a bit hard to comprehend how can RF noise influence the sound anyway... Logical reasoning would suggest that 1Mhz noise will result in distortion at the same frequency (or its harmonics). So how would this degrade the audible range? Especially that RF noise is usually very low level and wouldn't waste any significant amount of amplifier's power (I think?...).

<snipped>

Sorry for turning back to a previous topic. But I don't think it has been covered well-enough in this thread, yet. And it's fairly important.

The main problem with getting RF into an audio amplifier is not the "mixing" or intermodulation between the RF and the audio frequencies, which would just give sum and difference frequencies, which would still be out of the audible range.

The main problem with RF, in lower-frequency circuits in general, is that it gets rectified by every P-N semiconductor junction that it encounters.

That can manifest itself in lots of ways, none of them good (in non-RF circuits, at least).

Here are just a few quasi-examples:

1) In a discrete transistor circuit, or in an integrated circuit, the DC that results from the rectification can change the carefully-designed "DC Operating Point", in transistor sub-circuits. That can de-optimize a whole range of original design parameters (including linear operating ranges, for example) and basically make the circuit work "not quite right", depending on exactly where the RF has an effect. (At an extreme, with strong-enough RF, opamps' and transistor amps' outputs could even be pegged to one of the power rails.)

2) If the RF is simple AM (amplitude modulation), it is basically demodulated by the rectification, and the lower-frequency signal being carried by the AM might get summed with different currents or voltages at various points in a circuit. Sometimes you can even hear radio stations from the output of an audio amplifier that has no RF filtering. But the effects can be weirder, depending on exactly where in a circuit the RF has an effect.

3) RF can come from many sources. Anything that is narrow in the time domain is wide in the frequency domain (and vice versa). So a simple switch being flipped, or a relay closing, or an arc being struck, which would be very short/narrow events in the time domain, can produce a quite broad (in frequency) RF burst. If such an RF burst enters an audio amplifier, and gets rectified by certain P-N junctions, it can temporarily (burstily?) mess up DC operating points, as described above, and can even result in audible pops and cracks.

This is all more-complicated than it sounds, since ICs can have thousands of transistors, with many different purposes, and the effects are probably not usually profound-enough to be overwhelmingly obvious. And it is all compounded by the fact that for RF, everything is an input, and an antenna.

The best thing to do is have RF filters on all inputs, all outputs, and all power inputs/outputs, and also use shielding. It's all mentioned (with circuit filtering and shielding configuration examples) in ADI - Analog Dialogue | Op Amp Applications Handbook (probably mainly in Chapter 7), and in many other places.

Cheers,

Tom Gootee
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.