John Curl's Blowtorch preamplifier part II

[RNMarsh]

Originally Posted by elektroj

If one puts together shunt type attenuator with multiple relays, only one contact will be 'on' at a time. Such attenuator won't be cheap, but, so what? We're after the best here, aren't we?

Motorized pot -->- motorized rotary switch for shunt type atten.?

[jcx]

Originally Posted by RNMarsh

It isnt a problem if you design properly for it. His problems doesnt have to be others problem unless it is an unsolvable issue. Two power amps as a single mono-block? Then see how it does in thd and sound. The thd at highest freq would be interesting. Maybe a flat thd vs freq without the usual rising thd as freq increases.

I can see other apps also; A S.E. input to balanced output circuit that has low thd for line levels.

Either topology might be used with IC Opamps of lower cost to get very low thd.

Easy way to have super low thd in oscillator circuits. Instead of just higher gain/fb approach.

No good reasons yet to not try it and see the thd curves resulting.... what about those abnotious upper middle/high freq everyone talks about -- gone?

Has anyone done much on these circuit approaches -- got articles?

Thx-RNMarsh

I have a composite amplifier thread - includes a feedforward sim with one of Cordell's Book circuits as main amp

High loop Gain Composite Op Amp Circuits

feedforward sim in post #10, I also give the VanderKooy Lipshitz Error Feedforward paper ref: http://quad405.com/jaes.pdf (looks like archive.org scan is corrupted)


but I've only built high loop gain multiloop op amp composites

100+ dB global loop gain @20Khz is easy with 2 modern op amps in a multiloop

using output CFA DSL driver op amp, specs working alone while driving 25 Ohm load of -100 dB THD <100 kHz, -80 dB @ 1MHz

I don't think there's much chance of measuring distortion rise in upper middle/high (audio) freq everyone talks about - certainly not above noise floor with human hearing scale integration times

[PMA]

Originally Posted by BV

Catalogue data are quite nice, but try to measure e.g. PGA2310 at -6-20dB gain (not 0dB..) and input signal about 2V RMS.

What was the problem? A constant, visible level of higher order distortion components? Like cross-over distortion?

[kgrlee]

Originally Posted by jan.didden

My own (humbling) experience exactly. I've written it up here somewhere, but we tried a blind test of an FM multiplex 15kHz filter. Ruler flat to 15kHz, then a steep roll-off of more than 60dB at 19kHz. Lots of phase shift of course.

Of four participants, one stepped out because he admitted not hearing any difference. The remaining three, yours truly among them, all identified correctly the two different situations but inverted: we thought that the situation with the filter switched in sounded the best! We also had similar descriptions of that 'best' situation: brighter, more attack, more transients. As I said, humbling.

Why be humbled? Just shows you and your two friends aren't deaf like the pseudo Golden Pinnae.

I would rather say understandable - you have used a poor audio system for your test. 95% or maybe even more audio systems are poor, including so-called highend.

Pavel, rather than misquote Sturgeon, I'll ask what system you recommend to investigate if a brickwall filter improves the sound.

Player, preamp, power amp, speakers. A serious enquiry. But please don't specify Blowtorch as I don't think its transparent enough.

In case, its not obvious, I want to see if the State of the Art is such that distortions (in the widest sense) have been reduced sufficiently since 1945. The BBC, Jan's, my tests and some others which I can't remember .. suggest not enough.

There is a serious problem investigating this as most sources today are evil digital and hence severely band-limited. Its now obvious that most SACDs are made from 44.1kHz masters. IIRC, Pavel suggested a few Telarcs but this is still insufficient.

One of the most important tenets of my Double Blind Tests is that the subject chooses the programme material. If he is unfamiliar with the Telarcs and especially if he doesn't listen to them for enjoyment, the test is much less valid to answer the question, "Does it sound better?".

I'm rather disappointed that even in this august company, including aficionados like JC who spurn all evil CDs, there aren't more suggestions for good music with extended supa HF.

It may be we can only test the utility of a brickwall filter on bandlimited programme. So this really only tests how poorly the evil preamp & power amp reacts to supersonic crud from the evil player.

If that's the case, maybe Blowtorch is a suitable candidate for the ancillaries to this test after all.
_________________________

How does terminating a cable at the speaker terminals improve Group delay? What would it take?

... but there's *always* priors in electronics. Usually Blumlein or Baxandall!

It takes a Zobel network at the speaker ... and Yes, it does make for prettier square waves at the speaker terminals if that's your thing.

Great Guru Baxandall would not have agreed with the need ... but he tells you how to do it here Microphone Engineering Handbook Chap 8 and as a bonus, "how to design small audio transformers". Grab this before this knowledge disappears forever.
_________________________
Why do EQs sound different?

Thanks for this Electroj. Although I have the greatest respect for Guru Gerzon, (and my pseudo guru status in certain circles is based solely on regurgitating his Word) here, Michael is very naughty in quoting Harwood. What he DOESN'T mention is that Harwood also showed that for the speaker distortions he was looking at, Qs around 0.5 were in fact the MOST audible. Higher Qs were much less of a problem.

Peter Fryer confirmed this in Intermodulation Distortion Listening Tests (this paper is about a lot more than Intermodulation) and I believe so did the false prophet Floyd and his acolytes much later.

[PMA]

Originally Posted by kgrlee

Pavel, rather than misquote Sturgeon, I'll ask what system you recommend to investigate if a brickwall filter improves the sound.

Player, preamp, power amp, speakers. A serious enquiry. But please don't specify Blowtorch as I don't think its transparent enough.

Richard,

it is extremely difficult to recommend anything that is not under own personal control, this especially applies to the audio chain as a whole and not divided into single components.

But I have one non-recommended component, and it is the Behringer active crossover. If this one was used in the test, I cannot take the results serious. This is based on my personal experience.

Regarding the line stage and transparency, my recommendation was expressed in one of my yesterday's posts.

Sincerely,

[Bonsai]

Originally Posted by jcx

I have a composite amplifier thread - includes a feedforward sim with one of Cordell's Book circuits as main amp

High loop Gain Composite Op Amp Circuits

feedforward sim in post #10, I also give the VanderKooy Lipshitz Error Feedforward paper ref: http://quad405.com/jaes.pdf (looks like archive.org scan is corrupted)


but I've only built high loop gain multiloop op amp composites

100+ dB global loop gain @20Khz is easy with 2 modern op amps in a multiloop

using output CFA DSL driver op amp, specs working alone while driving 25 Ohm load of -100 dB THD <100 kHz, -80 dB @ 1MHz

I don't think there's much chance of measuring distortion rise in upper middle/high (audio) freq everyone talks about - certainly not above noise floor with human hearing scale integration times

jcx, I think the bridge type connection I show does not have the high loop gains of the composite connection because it actually operates in pararallel to the main feedback loop.

Maybe we should split this discussion off into a new thread - I am also thinking about reconfiguring so you get genuine feed forward. Anyway, I would be interested in your thoughts.

[fas42]

Originally Posted by elektroj

If one puts together shunt type attenuator with multiple relays, only one contact will be 'on' at a time. Such attenuator won't be cheap, but, so what? We're after the best here, aren't we?

I come from the "chain is only as strong as its weakest ..." line of thinking: chain A is made from 99 links of strength 100, 1 link of strength 50; chain B is made from 100 links of strength 50 - which chain is stronger? In the audio world everyone will put up their hand to say chain A is obviously superior; everything in my experience says it works just that real, hard metal "rope" - that single, lesser link sets the standard for everything else.

Not very palatable, but that's how it works, IME ...

Frank

[RNMarsh]

Originally Posted by Bonsai

jcx, I think the bridge type connection I show does not have the high loop gains of the composite connection because it actually operates in pararallel to the main feedback loop.

Maybe we should split this discussion off into a new thread - I am also thinking about reconfiguring so you get genuine feed forward. Anyway, I would be interested in your thoughts.

The composite with super high gain is the basis of at least one oscillator for obtaining super low thd. It isnt the same thing as the bridge circuit, of course.
Let go somewhere else and investigate it's potential -- pro and con- further.

[Oh, and zobals are not what i am talking about... group-delay.]

Thx-RNMarsh

[kgrlee]

Originally Posted by RNMarsh

[Oh, and zobals are not what i am talking about... group-delay.]

If you want to control group-delay at the end of your speaker cable, you will find Zobel networks very useful indeed.

Of course, they have to be matched to the amplifier, speaker & cable on an individual basis. That's what Guru Baxandall's stuff is about. The matching.
____________________

it is extremely difficult to recommend anything that is not under own personal control, this especially applies to the audio chain as a whole and not divided into single components.

Pavel, could you recommend a complete chain for this purpose? Specifying anything important that I might have left out.

Commercially available if possible.

If nothing is commercially available that is good enough NOT to benefit from Bandwidth Limitation, that's an important piece of info too.

[BV]

What was the problem? A constant, visible level of higher order distortion components?

No , but higher level (about 0,007-0,02%) of low order distortions (only 2. and 3. ) , frequency independent. Too much for my taste..No , but higher level (about 0,007-0,02%) of low order distortions (only 2. and 3. ) , frequency independent. Too much for my taste..Higher harmonic burried in noise background (under -120dB)