Author Topic: Speaker fuse nonlinearity at rated current?  (Read 3800 times)

0 Members and 1 Guest are viewing this topic.

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Speaker fuse nonlinearity at rated current?
« on: July 11, 2019, 05:37:09 pm »
Again, a question which might work elsewhere but I chose to put it here because perhaps the discussion might be pertinent...  As is probably known, there are dodgy companies which sell "audiophile" speaker fuses that are supposed to "improve" the sound.  This goes hand-in-hand with the mods often sold to bypass speaker fuses entirely, at the risk of course of damage or destruction to the speaker.

First, let me state for the record that I don't believe a properly rated speaker fuse will cause any audible artifacts in an audio signal.  But it is admitted by the manufacturers (such as Littelfuse) that the behavior of a fuse at or near rated current is nonlinear in resistance due to the heating of the fusible element. Nonlinearity in the signal path, of course, does lead to distortion - but how much?  Sadly, no one seems to have published any hard data as to the resistance vs. current passed through a fuse.  It seems that lower rated fuses would probably show a greater % variance as you approach rated current, because they're higher resistance to begin with; that much is shared by the manufacturers.  But then you're more interested in the behavior at a current which is typical of actual use, which for speaker fuses is somewhere around 1/2 - 2A, which is roughly a range of cold resistance somewhere between 40-250 mΩ.  Making the fuse larger in order to minimize any putative effect will of course risk damage to the speakers, which is to be avoided. 

Has anyone found, or measured, data showing a change in resistance with applied current for a moderate-sized fuse (say 1A fast blow)?  In other words, has it been characterized numerically?  The only thing I have been able to find was a graph purporting to show distortion vs. applied current, and a vague reference to Bob Cordell's book on designing audio amplifiers - neither of which have been traceable to actual test conditions or discussion.  I don't really want to buy the book and then find that the data was taken out of context or that the graph was mis-attributed.  I assume that it's probably in a circuit with an amplifier and a ~8Ω speaker, but that still doesn't provide more than vague data (and no analysis) to justify the alleged results.  If the data is real, then it would seem unlikely for a fuse to contribute audible distortion just based on the facts that most speakers generate far more distortion over the frequencies shown, and the masking effects of musical signals are quite strong.

Anyone with hard data, a copy of the book, or who has actually done a test and can share?  It would be nice to put this one to rest.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14071
  • Country: de
Re: Speaker fuse nonlinearity at rated current?
« Reply #1 on: July 11, 2019, 06:13:56 pm »
I have not see a fuse directly in series with a speaker for a while. If at all I would more consider a slow blow fuse. In a slow blow fuse there would be considerably slower heating of the wire.
For the fuse wire resistance one could assume resistance proportional to kelvin temperature as a first approximation, just like for the copper coil in the speaker.  The speaker resistance should be considerably higher than the fuse resistance. Heating of the speaker coil and thus an increased resistance and this less amplitude after loud parts is a known effect that can be a problem. The self heating within a single cycle is usually less of a problem.

As one usually does no use a speaker near it's rated power, I would not worry that much about the distortion.  It could be a problem with high power speakers e.g. for PA use, that do get quite a bit warm and may need protection.
 

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #2 on: July 11, 2019, 08:50:51 pm »
Some good points to ponder; thanks.

Yes, most modern amps don't use direct inline fuses - true.  I have a ton of older (I refuse to say "vintage"  ;D ) equipment which does, however.  And a surprising number of modern speakers have their own fuses, with good reason: as an example, the Magneplanars (like mine) use ribbon tweeters which are extremely delicate and can be damaged when overdriven.  Good or bad, it's a fact of life with them, and a damaged driver is time-consuming, expensive or not replaceable at all depending on the model.  The speaker impedance is much greater than the fuse (perhaps 20-100 times when cold) or else there would clearly be other issues to deal with.  The problem with a first approximation of linear increase in resistance is that I'm pretty sure it is inaccurate when the fuse is run near its ratings for any period of time.  Again, this is something I can't find direct data on, though the fuse companies acknowledge that it happens.

The only problem with a slow blow fuse is that it requires long periods of current in excess of the rating in order to blow.  If we're protecting the speaker, that may be a bad thing.  You would have to purposely use a lower value in order to ensure that the fuse blows just above the desired value after a short enough period to protect the speaker.  This, paradoxically, would result in the fuse being driven in excess of the rated current for a longer period of time and getting into the area I'm exploring here of nonlinearity.  I'm not disagreeing so much as dubious about the desirability of a slow-blow fuse for speaker protection.  Most manufacturers spec fast-blow fuses for this reason, I believe.

And also true, you wouldn't want to run the speaker near rated power intentionally.  The intent of the fuse is to blow when a large sudden transient or sustained overdrive is applied without hindering operation otherwise.  Most speakers can handle fairly large transients (which is in fact what their rated power usually refers to, rather than continuous input)  It's a somewhat contradictory requirement for a fuse, I admit; but it seems to be a fair compromise with an amplifier which doesn't have sophisticated protection circuitry.

I realize I'm looking at an edge case here; but it would be nice to have some answer for the audiophools who are paranoid about sound to the point that they risk damage to their expensive speakers - and recommend that others do the same.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 6923
  • Country: ca
Re: Speaker fuse nonlinearity at rated current?
« Reply #3 on: July 11, 2019, 09:06:41 pm »
It's not an audiophool issue. Audio fuse distortion was discussed by Bob Cordell and Douglas Self. Quote from Designing Audio Power Amplifiers - Bob Cordell. I can dig for Doug Self's discussion of it.

Fuse Distortion
"When high current passes through a fuse, it heats up the fuse element by causing a small voltage drop that leads to power dissipation and temperature rise. The fuse element must therefore be resistive for this process to take place. The fuse blows when the temperature rises to the melting point of the fuse element. The fuse element has a positive temperature coefficient of resistance, so the process is accelerated as more current flows through the fuse.
At low frequencies the audio signal can heat up and cool down the fuse element within a single cycle, causing the resistance of the fuse to vary as a function of the signal amplitude. This leads to distortion because the attenuation of the fuse resistance against the load impedance changes as a function of signal swing [5].  Fuses are often undersized with respect to the peak audio current they may be called on to pass, recognizing that a smaller fuse will provide relatively more protection and that with normal audio signals, such high currents are brief events much shorter than the time constant of the fuse element. The cold resistance of a 2A 3AG fuse was measured to be 78m \$\Omega\$, while its resistance when passing 2A DC was 113m \$\Omega\$. This represents a 45% increase in fuse resistance.
The distortion of a fuse can be measured by looking at the voltage across the fuse with a sinusoidal  signal current passing through it. The fuse under test is put in the ground leg of an 8 \$\Omega\$ load resistor so that the signal voltage across the fuse can be easily analyzed.  This  technique  largely  takes  the  distortion of  the driving source out of the picture. Figure 13.4a is a plot of fuse distortion versus frequency when a 2A fast-blow 3AG fuse is passing a 2A RMS sine wave signal. As expected, fuse distortion increases dramatically at low frequencies. Signal voltage across the fuse was 250mV. Amplifier THD (due to the fuse) is calculated by normalizing the fuse distortion voltage to the amplifier output voltage. The resulting amplifier distortion is shown in  Figure  13.4b. Amplifier distortion is lower than fuse distortion by a factor of 64 because of the small voltage across the fuse compared to the total signal voltage. At 20 Hz, amplifier distortion due to the fuse is calculated to be 0.0033%."
« Last Edit: July 12, 2019, 07:25:45 pm by floobydust »
 

Offline cvanc

  • Frequent Contributor
  • **
  • Posts: 672
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #4 on: July 11, 2019, 09:43:25 pm »
Way back in the day didn't some amp makers incorporate a fuse at the output but inside the feedback loop?
 

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #5 on: July 12, 2019, 12:12:01 am »
Yes!  At last some details.  That's excellent; thanks for quoting it.

Too bad they didn't show how the fuse resistance changed incrementally with current; as it says "the process is accelerated as more current flows through the fuse".  Based on the calculations for heating of copper wire, plus resistance change vs. temperature, I'd expect a roughly square-law relationship of resistance wrt current.  Given that the graph shows an inverse variance of distortion to frequency, it seems that you're getting inaudible distortion at the point the fuse is sustaining maximum design current at 20 Hz; 10x less at 200 Hz, etc.

The "audiophool" issue to me was not whether fuse distortion was illusory, but whether it was audible.  This seems to be a strong data point which indicates it is not.  I subscribe to the Bob Carver doctrine which says that distortion of a few tenths of a percent is barely audible on a pure sine wave at the ear's most sensitive frequency.   8)  Rigorous testing of MP3 codecs suggests to me that much higher levels than that can easily be masked in music.

I know Self talked at some length about the fallacy of "thermal distortion" in semiconductors; I was not aware that he also discussed fuse distortion.

Way back in the day didn't some amp makers incorporate a fuse at the output but inside the feedback loop?

I've seen a couple of designs suggested, but as far as I know they were considered impractical or unreliable to implement.  If any were used in production amplifiers, I don't know of them.  Maybe some of the Haflers?  The main issue of course would be designing the amp so that the feedback loop didn't go open circuit / unstable when it blew.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 6923
  • Country: ca
Re: Speaker fuse nonlinearity at rated current?
« Reply #6 on: July 12, 2019, 01:18:29 am »
From Audio Power Amplifier Design Handbook, Douglas Self, Ch. 17

"Fuses running within sight of their nominal rated current generate distortion at LF due to cyclic changes in their resistance caused by I2R heating; the THD would be expected to rise rapidly as frequency falls, and Greiner [6]states that harmonic and intermodulation distortion near the burn-out point can reach 4%. It should be possible to eradicate this by including the fuse inside the global feedback network, for the distortion will be generated at low frequencies where the feedback factor is at its greatest, but there are problems with amplifier behavior after the fuse has blown."
[6] R. Greiner, Amplifier – loudspeaker interfacing, JAES 28 ( 5 ) (May 1980) pp. 310–315.
 
The following users thanked this post: GregDunn

Offline blacksheeplogic

  • Frequent Contributor
  • **
  • Posts: 532
  • Country: nz
Re: Speaker fuse nonlinearity at rated current?
« Reply #7 on: July 12, 2019, 02:53:07 am »
Fortunately, I was born with ears that are not that sensitive. I image that if you can hear the distortion a fuse is causing under normal operating conditions there's a myriad of other equally annoying problems spoiling your listing pleasure.

 
The following users thanked this post: Ysjoelfir

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16544
  • Country: us
  • DavidH
Re: Speaker fuse nonlinearity at rated current?
« Reply #8 on: July 12, 2019, 03:08:19 am »
From Audio Power Amplifier Design Handbook, Douglas Self, Ch. 17

"Fuses running within sight of their nominal rated current generate distortion at LF due to cyclic changes in their resistance caused by I2R heating; the THD would be expected to rise rapidly as frequency falls, and Greiner [6]states that harmonic and intermodulation distortion near the burn-out point can reach 4%.

I would have cited that Douglas Self if you had not.

Analog integrated circuits suffer from the same problem as heat washes across the die from the output transistors to the input transistors.  In operational amplifiers, this limits open loop gain at DC and low frequencies.  It can also determine settling time.  Symmetrical layouts are used to minimize this in precision, low distortion, and fast settling time parts but this is why it is a good idea to unload operational amplifier outputs with external buffers to minimize changes in power dissipation with signal level if these things matter.  Circuits made from discrete parts do not suffer from this of course because the transistors are effectively thermally isolated from each other.
 

Offline sibeen

  • Frequent Contributor
  • **
  • Posts: 271
  • Country: au
Re: Speaker fuse nonlinearity at rated current?
« Reply #9 on: July 12, 2019, 03:58:35 am »
From Audio Power Amplifier Design Handbook, Douglas Self, Ch. 17

Hmm, I was certain I had Self lying around on my bookshelf somewhere and was eventually able to locate it. Went looking for chapter 17 and found, to my dismay, that my version only had thirteen chapters. It seems that this book may have had a few revisions since I procured my copy. Another bloody downside to this getting old shit :)
 

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #10 on: July 12, 2019, 05:49:38 am »
Fortunately, I was born with ears that are not that sensitive. I image that if you can hear the distortion a fuse is causing under normal operating conditions there's a myriad of other equally annoying problems spoiling your listing pleasure.

No one is born with ears that sensitive.  Many audiophools like to flatter themselves that they can hear it, even when it demonstrably doesn't exist; that's why they are not fond of moderated double-blind tests and other scientific methods...
 
The following users thanked this post: Ysjoelfir

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 6923
  • Country: ca
Re: Speaker fuse nonlinearity at rated current?
« Reply #11 on: July 12, 2019, 05:57:20 pm »
I enjoy the two books (Self, Cordell) because the authors measure distortion in fuses, relays, binding posts, electrolytic capacitors etc. Instead of the speculation, religious wars, and audiophoolery, they apply some science.
It's also relevant because in designing commercial audio power amplifiers, you want the lowest THD number for specmanship, to compete.

The unexpectedly high THD in production due to chinese gold-plated steel (magnetic) output binding posts was interesting, who would think that's a legit issue?
But we have conquered THD, it's in the ppm level now on many amplifier designs.
Recording engineers have very sensitive hearing, they earn a living every day with their ears. They notice distortions it took me while to track down doing repairs for them.

Douglas Self 5th ed. almost doubled in size compared to 4th ed. He added much more material. It's a good read for those 8hr plane flights.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 14071
  • Country: de
Re: Speaker fuse nonlinearity at rated current?
« Reply #12 on: July 12, 2019, 06:42:50 pm »
The low voltage drop at the fuse reduces the problem quite a bit and there is a similar effect of heating the coil in the speaker - the part that should be protected. With the fuse at something like 1/50 the coil resistance the heating of the coil would be the bigger problem. At maximum power the coil heats up by something in the order of 100 K. So to get a similar resistance change the fuse wire would need to heat up by some 5000 K.

If really a problem it would be relatively easy to have fuses with a wire (e.g. brass) that does not change resistance that much.

Anyway one usually does not hear music anywhere near the power handling capacity of the speaks. At such an amplitude the mechanical / magnetic nonlinearity of the speaker is likely the larger problem.
 

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 8972
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
Re: Speaker fuse nonlinearity at rated current?
« Reply #13 on: July 13, 2019, 03:12:46 am »
Yes, most modern amps don't use direct inline fuses - true.  I have a ton of older (I refuse to say "vintage"  ;D ) equipment which does, however.  And a surprising number of modern speakers have their own fuses, with good reason: as an example, the Magneplanars (like mine) use ribbon tweeters which are extremely delicate and can be damaged when overdriven.  Good or bad, it's a fact of life with them, and a damaged driver is time-consuming, expensive or not replaceable at all depending on the model.  The speaker impedance is much greater than the fuse (perhaps 20-100 times when cold) or else there would clearly be other issues to deal with.  The problem with a first approximation of linear increase in resistance is that I'm pretty sure it is inaccurate when the fuse is run near its ratings for any period of time.  Again, this is something I can't find direct data on, though the fuse companies acknowledge that it happens.
Wouldn't the tweeters be operating at a frequency range far above the point where the thermal time constant of the fuse allows it to vary its resistance in time to the signal?
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #14 on: July 14, 2019, 10:01:51 pm »
Yes, most modern amps don't use direct inline fuses - true.  I have a ton of older (I refuse to say "vintage"  ;D ) equipment which does, however.  And a surprising number of modern speakers have their own fuses, with good reason: as an example, the Magneplanars (like mine) use ribbon tweeters which are extremely delicate and can be damaged when overdriven.  Good or bad, it's a fact of life with them, and a damaged driver is time-consuming, expensive or not replaceable at all depending on the model.  The speaker impedance is much greater than the fuse (perhaps 20-100 times when cold) or else there would clearly be other issues to deal with.  The problem with a first approximation of linear increase in resistance is that I'm pretty sure it is inaccurate when the fuse is run near its ratings for any period of time.  Again, this is something I can't find direct data on, though the fuse companies acknowledge that it happens.
Wouldn't the tweeters be operating at a frequency range far above the point where the thermal time constant of the fuse allows it to vary its resistance in time to the signal?

Absolutely.  The problem is that the people who believe unconditionally in the "fuse distortion" meme without understanding the mechanism are more likely to bypass the tweeter fuses because of some audiofool contention that the tweeters will reveal the distortion more readily.  They will fry their tweeters on a transient or overload without even getting any benefit in the meanwhile.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 6923
  • Country: ca
Re: Speaker fuse nonlinearity at rated current?
« Reply #15 on: July 15, 2019, 03:38:12 am »
Modern loudspeakers use PTC polyfuses for the tweeter protection. Cheap speakers use light bulbs in series.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16544
  • Country: us
  • DavidH
Re: Speaker fuse nonlinearity at rated current?
« Reply #16 on: July 16, 2019, 03:52:06 am »
Fortunately, I was born with ears that are not that sensitive. I image that if you can hear the distortion a fuse is causing under normal operating conditions there's a myriad of other equally annoying problems spoiling your listing pleasure.

No one is born with ears that sensitive.  Many audiophools like to flatter themselves that they can hear it, even when it demonstrably doesn't exist; that's why they are not fond of moderated double-blind tests and other scientific methods...

I agree especially given how much distortion the speaker creates in comparison.  But the "blameless" amplifier is still worth studying for the principles involved.

Other amplifier behaviors like transient intermodulation distortion are more interesting.
 

Online Berni

  • Super Contributor
  • ***
  • Posts: 4922
  • Country: si
Re: Speaker fuse nonlinearity at rated current?
« Reply #17 on: July 16, 2019, 07:22:28 am »
Yep i seen the in series lightbulb trick in big PA speakers. Apparently high power automotive bulbs work well for that because these low voltage halogen bulbs have very low resistance when cold, but at some point they hit a knee where they warm up enough for the resistance to rise and that kicks off a runaway where the bulb takes up more and more power until most of the power is on the bulb and it stays lit until the power is reduced.

But to have the whole speaker on it and not just the tweeters would require a ridiculously big light bulb.

In terms of amplifier side protection a good way to do it is a relay. Its done in some amplifiers where a circuit monitors the output and if its not happy with what its seeing it clicks a relay that shorts the output to ground. The amplifier is limited in terms of power by its supply rails anyway, so what these look for is usually just excessive DC bias because that's dangerous to speakers and might not even blow a fuse. And by using a relay to just short the output instead of switching it means that the relay is out of circuit in normal operation.
 

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #18 on: July 19, 2019, 08:43:29 pm »
Fortunately, I was born with ears that are not that sensitive. I image that if you can hear the distortion a fuse is causing under normal operating conditions there's a myriad of other equally annoying problems spoiling your listing pleasure.

No one is born with ears that sensitive.  Many audiophools like to flatter themselves that they can hear it, even when it demonstrably doesn't exist; that's why they are not fond of moderated double-blind tests and other scientific methods...

I agree especially given how much distortion the speaker creates in comparison.  But the "blameless" amplifier is still worth studying for the principles involved.

Other amplifier behaviors like transient intermodulation distortion are more interesting.

I am a big fan of Self and his approach to amplifier analysis.  He has taken a lot of the voodoo out of the design of power amplifiers, and shown that you don't need insanely complex or expensive topologies to provide a completely transparent amplifier.  Unfortunately, confirmation bias will continue to sell a lot of high end gear which is no better audibly.

I actually corresponded with Walt Jung and built one of his preamps - which is in fact a solid and simple piece of gear.  His assumption that audio gear like RIAA preamps and the like actually need to deal with 20V full bandwidth signal peaks, though, was misplaced.  TIM is a real phenomenon which almost cannot occur in a properly designed amp receiving normal music signal levels.  Jung artificially introduced massively high signal levels in his tests which pushed the amp well beyond what it would see in handling musical input, and used that as justification for claiming TIM was a prevalent phenomenon in audio gear.  His preamp is excellent and has a very low component count and cost, so I can forgive him his zeal.   :)  I am in fact using it today.
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16544
  • Country: us
  • DavidH
Re: Speaker fuse nonlinearity at rated current?
« Reply #19 on: July 19, 2019, 11:06:29 pm »
I am a big fan of Self and his approach to amplifier analysis.  He has taken a lot of the voodoo out of the design of power amplifiers, and shown that you don't need insanely complex or expensive topologies to provide a completely transparent amplifier.  Unfortunately, confirmation bias will continue to sell a lot of high end gear which is no better audibly.

The part I found interesting is how similar the design and analysis is to operational amplifiers.

Quote
I actually corresponded with Walt Jung and built one of his preamps - which is in fact a solid and simple piece of gear.  His assumption that audio gear like RIAA preamps and the like actually need to deal with 20V full bandwidth signal peaks, though, was misplaced.  TIM is a real phenomenon which almost cannot occur in a properly designed amp receiving normal music signal levels.  Jung artificially introduced massively high signal levels in his tests which pushed the amp well beyond what it would see in handling musical input, and used that as justification for claiming TIM was a prevalent phenomenon in audio gear.  His preamp is excellent and has a very low component count and cost, so I can forgive him his zeal.   :)  I am in fact using it today.

I think where designs go wrong yielding problems with TIM is controlling bandwidth with the negative feedback network which increases overload recovery time.  This also explains why some designs which include a lot of negative feedback sound bad; they just suffered from excessive TIM.
 

Offline floobydust

  • Super Contributor
  • ***
  • Posts: 6923
  • Country: ca
Re: Speaker fuse nonlinearity at rated current?
« Reply #20 on: July 20, 2019, 01:50:03 am »
There are some AES (paywall) papers and JBL marketing on loudspeaker "power compression" where the voice-coil heating causes losses and distortion. Some effects might be similar to that of the lowly fuse.

OT Douglas Self's Blameless amp I have not built but it is very popular as a modern amplifier with great numbers (extremely low THD). I did not agree with his approach of piling on open-loop gain and using high feedback factor with two-pole comp to keep it stable. It also needed some polish; VAS current-limiting and a Baker clamp to keep it behaved.
It was a strategy in the 1980's where a lower THD number was a figure of merit for a better amplifier, the one to buy. The Japanese amplifiers then most were unstable and blew up, Sansui especially, in fighting for specmanship. I hated doing repairs on them, they were very picky about the replacement transistors characteristics.
That approach didn't work out, THD is not that annoying as it's harmonically related verses TIM, so 0.02% or 0.002% THD I doubt is audible is worth pursuing. It's the dynamic distortions that stand out.
 

Offline GregDunnTopic starter

  • Frequent Contributor
  • **
  • Posts: 725
  • Country: us
Re: Speaker fuse nonlinearity at rated current?
« Reply #21 on: July 20, 2019, 02:47:57 am »
I am a big fan of Self and his approach to amplifier analysis.  He has taken a lot of the voodoo out of the design of power amplifiers, and shown that you don't need insanely complex or expensive topologies to provide a completely transparent amplifier.  Unfortunately, confirmation bias will continue to sell a lot of high end gear which is no better audibly.

The part I found interesting is how similar the design and analysis is to operational amplifiers.

All good power amps since about 1972 look like op amps, because they make use of the same series of building blocks that Self describes so effectively in his book.

Quote
Quote
I actually corresponded with Walt Jung and built one of his preamps - which is in fact a solid and simple piece of gear.  His assumption that audio gear like RIAA preamps and the like actually need to deal with 20V full bandwidth signal peaks, though, was misplaced.  TIM is a real phenomenon which almost cannot occur in a properly designed amp receiving normal music signal levels.  Jung artificially introduced massively high signal levels in his tests which pushed the amp well beyond what it would see in handling musical input, and used that as justification for claiming TIM was a prevalent phenomenon in audio gear.  His preamp is excellent and has a very low component count and cost, so I can forgive him his zeal.   :)  I am in fact using it today.

I think where designs go wrong yielding problems with TIM is controlling bandwidth with the negative feedback network which increases overload recovery time.  This also explains why some designs which include a lot of negative feedback sound bad; they just suffered from excessive TIM.

True, they do exhibit more TIM - but not with real world music signals.  If you aren't overloading the amp, and it has enough slew rate for the signal, you won't see a problem.  Jung only had issues because he was pumping 20 KHz 20V square waves into the circuit, and he felt that was something the amp would see on a regular basis.  Except if you're using the amp way out of its intended envelope, it does not.


It was a strategy in the 1980's where a lower THD number was a figure of merit for a better amplifier, the one to buy. The Japanese amplifiers then most were unstable and blew up, Sansui especially, in fighting for specmanship. I hated doing repairs on them, they were very picky about the replacement transistors characteristics.
That approach didn't work out, THD is not that annoying as it's harmonically related verses TIM, so 0.02% or 0.002% THD I doubt is audible is worth pursuing. It's the dynamic distortions that stand out.

The problem was that they didn't know how to properly compensate the circuit; they just piled on the feedback and didn't consider stability or overload capability of the input stage(s).  I like Self's approach of analyzing each section of the amp separately in addition to looking at the output; current starvation and compensation of individual stages can cause internal issues even though the output of the block looks OK due to overall feedback.  No doubt there were some bad actors in that era, though I remain unconvinced that any of them were bad enough to actually cause audible deterioration of musical signals.  Maybe a poorly-designed phono preamp might have generated some anomalies, but it would take some very sloppy design.  Power amp circuits may be more susceptible to it, but again, actual music doesn't make severe demands of an amp's slewing capability.  You might see it on test tones, but that's it.

Pass and Baxandall both quoted observed music slew rates which, translated to a ~100W amplifier, came out somewhere between 1-2 V/µS.  Even my old Dynaco 400 will comfortably handle that; in fact, even if we assume a 5x safety factor it's fine. 
 

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 16544
  • Country: us
  • DavidH
Re: Speaker fuse nonlinearity at rated current?
« Reply #22 on: July 21, 2019, 02:41:39 am »
I am a big fan of Self and his approach to amplifier analysis.  He has taken a lot of the voodoo out of the design of power amplifiers, and shown that you don't need insanely complex or expensive topologies to provide a completely transparent amplifier.  Unfortunately, confirmation bias will continue to sell a lot of high end gear which is no better audibly.

The part I found interesting is how similar the design and analysis is to operational amplifiers.

All good power amps since about 1972 look like op amps, because they make use of the same series of building blocks that Self describes so effectively in his book.

They look like the *simplified* models which have not reflected actual designs since the 324.  (1) For instance I have yet to see a discrete audio amplifier design which takes advantage of any transconductance reduction method other than resistive emitter/source degeneration.  Everybody apparently bases their discrete amplifier on the 318 operational amplifier.

Since the more advanced transconductance reduction methods require matched transistors, I suspect a more modern design, that is a design not limited by preconceived notions from the simplified operational amplifier model, would either use an operational amplifier as an input stage, or insert a current feedback gain stage in place of or in addition to the VAS.  (2)

In practice as mentioned above, none of this matters because common audio power amplifiers are now "good enough" unless you want low output noise.

(1) The real 324 schematic is shown below.  Pay close attention to multiple collector transistors Q18 and Q20 and the unusually low compensation capacitance marked 5.0 picofarads.  Most 324 schematics do not show either.

(2) Or combine the two by abusing the operational amplifier by using its output as a current feedback input and taking the outputs from the supply pins.  This may not be a good however because of thermally induced distortion.  But that would not apply to a discrete design.
 

Online Siwastaja

  • Super Contributor
  • ***
  • Posts: 8106
  • Country: fi
Re: Speaker fuse nonlinearity at rated current?
« Reply #23 on: July 24, 2019, 01:31:17 pm »
The mechanism of distortion from a fuse is obvious. Does it matter? Does it matter if it matters? The real question is, why have the fuse there to begin with. A properly designed system has zero need for a fuse at the output. The DC input of the amplifier stage can be fused, instead. If output short circuit protection is needed, it's unlikely to be achievable reliably with a fuse (and in any case, it would again work on the DC input side), and active protection controlling the transistor drive is needed instead, to react quickly enough.

When the fuse is in the DC input, the small fluctuations in voltage are reduced by the amplifier's PSRR.
 

Offline SiliconWizard

  • Super Contributor
  • ***
  • Posts: 14297
  • Country: fr
Re: Speaker fuse nonlinearity at rated current?
« Reply #24 on: July 24, 2019, 07:08:38 pm »
A properly designed system has zero need for a fuse at the output. The DC input of the amplifier stage can be fused, instead.

Agree. I've designed a few audio amps. None of them ever had fuses at the outputs. Fuses on the power supplies, yes. As you said, their impact was negligible.

My later designs would include active monitoring of the average DC voltage (over a period of a couple hundred ms or so) at the outputs to protect speakers.
If the absolute value of the DC voltage exceeded a given threshold (pretty small AFAIR), it would just shut down the output stage. That proved pretty reliable and didn't have any measurable influence.

 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf