Sound Quality Vs. Measurements

[Steve Eddy]

Originally Posted by john curl

We went around the Moon, perhaps, in '68?

Yeah, first manned orbit of the moon was in '68. Easy peasey compared to landing on it, traipsing around for a bit, launching back off it, and returning safely.

se

[tvrgeek]

Originally Posted by revboden

Hey tvrgeek,

I keep thinking about all the combinations of stuff that we use. This speaker1 sounds great with that amp1 but with amp2 it sucks while amp2 sounds great with speaker2... if the amp is not designed to drive something specific, it really is a craps shoot.

Can there really be one amp to rule them all? i doubt it.

good luck,
revb.

Exactly, it is a system. Unfortunately, the industry is split forcing electronic designers and speaker designers to work with a wall between them pointing fingers at each other. Pro sound does a lot of integrated systems, but from what I hear of them at clubs, they are still not really systems, just amps stuffed on speakers. I don't know what it will take to get past the current stalled condition. The market is demanding portability not quality. What has hit the market are really poor powered subs, powered monitors with 10 cent amps, or over the top price that has no market.
A few have tried. Mat Polk was smart enough to know $300 speakers were going to be played on $200 receivers, so he tuned them assuming a high Re. Theil and Keff both tried line level eq boxes and failed in the market. Meridian and Paradigm have tried powered monitors but were out of price range. If I were a venture capital guy, I would not put my money there either, even though I know it is the correct engineering solution.

What it will take is one of you amp guys who understands there is a difference in amps team up with a speaker guy who also understands any old amp won't do. Then bring in a product at a price point that could sell. I won't hold my breath.

Besides, hard for guys to brag how many kazillion watts a system has when the speakers are all "sufficient" and it would remove the excuse for magic wires that cost more than the speakers.

[ThorstenL]

Sy,

Originally Posted by SY

Propagation delay is essentially zero for any practical amplifier; it exists mostly as audiophile mythology.

Propagation delay in typical Op-Amp type structures is several microseconds, even fully uncompensated comparators usually have propagation delays in the 100's of nanoseconds.

This emphatically is not what I would call "essentially zero", but rather "essentially and significantly non-zero". As such propagation delay clearly exists and is real, I think calling it an "audiophile myth" is rather disingenuous.

IF this propagation matters for the subjective sound quality of audio-circuits is a matter that may be debated, as the propagation delay is just another way of expressing other factors that can have an impact (but are also heavily disputed) one may suggest that low "propagation delay" may be used as an indicator of quality, but I suspect the correlation will be somewhat iffy...

Ciao T

[john curl]

Tvr, it is a lot more complicated than that, but you make a good point.

[ThorstenL]

Hi,

Originally Posted by tvrgeek

So I understand, if you restrict the feedback to within each stage, you also don't have to deal with the phase/propagation delay of global feedback that was the source of some distortions in early SS amps?

I think there are several questions here.

First, one would would be: In circuits that have looped feedback, spanning multiple stages, are distortions present or possible that are not found in circuits that omit looped feedback and if so, what are they?

Another would be: Is looped feedback around several stages the exact and complete equivalent of using the same amount of degeneration equally distributed across stages in the loop instead?

Still another would be: Is looped feedback around a single stage the exact and complete equivalent of using the same amount degeneration?

Another one would be: If our first question is answered affirmative (incidentally, it is, for around 4 decades), will reducing reducing global looped feedback and increasing local looped feedback or local degeneration reduce the distortions caused by (mis-) applying global looped feedback?

Ciao T

[ThorstenL]

Hi,

In my previous post I vented a bit about the (un) reasons causing these little spats we have all become so fond of watching, now for something a little more constructive...

Originally Posted by tvrgeek

So, we have audio amps with incredible bandwidth, stable on loads that are stupid, DF that is no more than a couple inches of wire, THD, IM, TIM, SID, "LBJ took the IRT down to 4th street USA...." * and yet, there is something that may be known, but I can't point to, that causes some amps to accentuate/create something in reproduction of a trumpet that does not occur live that makes my wife's teeth grate. Considering all the terrible things that happen to a signal from mic to speaker, it baffles me that this last stage has something audible. I guess we just don't know.

If you wish to try a few things, do try to take an amplifier that misbehaves in the way you describe and work your way through the schematic, step by step and systematically increasing the local degeneration and stage current (starting in the input stage) and opening the miller compensation suitably to retain similar stability margins (this means you can cannot just reduce the compensation cap tenfold if you reduced OLG by 20dB).

Once you have converted most (if not all) of the global loop feedback into local feedback try listening again. IF you find that the objectionable quality is reduced or gone, we have at least established an empirical link between levels of looped feedback and this peculiar acoustic aberation, or if this quality is unchanged we have established that global looped feedback is not to blame...

It does of course tell us little about the mechanisms, but it leaves us at least with a mitigation strategy. Of course such actual science is not appreciated by some, so be cautious about relating results.

Ciao T

[wahab]

Originally Posted by ThorstenL

Hi,



I think there are several questions here.

First, one would would be: In circuits that have looped feedback, spanning multiple stages, are distortions present or possible that are not found in circuits that omit looped feedback and if so, what are they?

Another would be: Is looped feedback around several stages the exact and complete equivalent of using the same amount of degeneration equally distributed across stages in the loop instead?

Still another would be: Is looped feedback around a single stage the exact and complete equivalent of using the same amount degeneration?

Another one would be: If our first question is answered affirmative (incidentally, it is, for around 4 decades), will reducing reducing global looped feedback and increasing local looped feedback or local degeneration reduce the distortions caused by (mis-) applying global looped feedback?

Ciao T

Basicly , reducing gain with local feedback , i.e , degeneration ,
will be far less efficient than using global feedback to reduce
the gain in the same proportion.

Degeneration waste gain uselessly as it doesnt allow for as
good linearization as global NFB.

[ThorstenL]

Hi,

Originally Posted by wahab

Basicly , reducing gain with local feedback , i.e , degeneration , will be far less efficient than using global feedback to reduce
the gain in the same proportion.

I am not sure what you mean with "efficient gain reduction"...

Originally Posted by wahab

Degeneration waste gain uselessly as it doesnt allow for as good linearization as global NFB.

That is an interesting assertion, but is it true?

Let us a perfectly theoretical amplifier.

The input stage contributes 1% HD open loop at a given level and has gain of 100.

The VAS contributes 1% HD open loop and has a gain 1000.

The Output Stage is Fet driven (to eliminate loading effects) and contributes 0.3% HD open loop.

So the whole Amplifier open loop will show around 1.45% THD with a fairly rich upper harmonics spectrum, as we have two stages producing "distortion of distortion". And the input stage will have a miniscule linear range as well...

With 10MHz GBP we now find we need a 100Hz open loop bandwidth so we already loose 20dB overall loop gain at 1KHz, though this gain is turned into 20dB local loop feedback IF we use miller compensation. DC Gain is 100dB.

Closed loop gain is 20, so we get around 0.002% HD at 1KHz, but again, heavily weighted towards upper harmonics and poorer linearity at upper frequencies.

Now let us degenerate the Input stage by 10dB and the VAS Stage by 10dB. So our 1% HD becomes 0.33% HD for each stage and we have 0.57% HD open loop and for 10MHz GBP we have 1Khz open loop bandwidth. Our Open loop gain from DC to the open loop bandwidth is 10,000. Once we close the loop we have 0.001% HD...

Of course, the above is merely a "milkmaid" calculation and things are more complex in reality, but it seems for a classic SS Amp judiciously adding local degeneration can actually reduced the final THD, so one may say that there are circumstances where local degeneration is more efficient at reducing distortion than global looped feedback, while at the same time improving transient distortion behaviour.

Of course, anyone who bothered to read Self, Cordell, Ottala, Gilbert et all knows that anyway, so apologies to most here for belabouring things this much.

Ciao T

[SY]

Originally Posted by ThorstenL

Sy,

Propagation delay in typical Op-Amp type structures is several microseconds

Cite?

[wahab]

Originally Posted by ThorstenL

Now let us degenerate the Input stage by 10dB and the VAS Stage by 10dB. So our 1% HD becomes 0.33% HD for each stage and we have 0.57% HD open loop and for 10MHz GBP we have 1Khz open loop bandwidth. Our Open loop gain from DC to the open loop bandwidth is 10,000. Once we close the loop we have 0.001% HD...

This , is where your logic stop being relevant as you re assuming
that degeneration has the same properties as GNFB , wich is not true.

Reducing a stage gain by a factor of 10 dB by increasing
degeneration will not reduce this stage distorsion by the same
factor , as is the case with GNFB , wich is what i pointed
in the post above.

Also , about miller compensation , you re downplaying the fact
that not only it reduce the open loop gain bandwith but that
at the same time this lost gain is in fact used to create
a local feedback loop that will inherently reduce the distorsion
of this stage.