John Curl's Blowtorch preamplifier part II

[john curl]

Wayne, what do you pay for a cup of coffee these days? '-)

[RNMarsh]

Originally Posted by john curl

Related to old Motorola jfet #'s from the '60's. Ask Richard Marsh.

I cant remember the details that long ago. a diff pair? quad dip bjt compliments were by Mototrola... :-( RNMarsh

[john curl]

Richard, look to the input stage devices of your headphone amp. Would you like to put the numbers up here for us to discuss?

[wayne]

Originally Posted by john curl

Wayne, what do you pay for a cup of coffee these days? '-)

Depends if it is medicinal or for taste just black though. Good point on the price of the semis they are minor compared to the toroids and chassis. And you could spend $50.00 on a Muse Op amp if you wanted.

[EDDELARUE]

Originally Posted by wahab

Nice , but +-5.5V max supply voltage.....

i found this article useful concerning on just output, we can to use this approach to increase voltage excursion.

http://www.edn.com/contents/images/45890.pdf

[john curl]

We do this sort of voltage extension on a regular basis. This way we can take 5-15V IC's and make them perform like 60V IC's. There are several approaches.
However, it does complicate the design, partially removing the advantage of using IC's in the first place. Then discrete design often looks just as good.
The REAL problem is WHY we would go to such a length to put a BLINDINGLY FAST IC in a circuit, and find that it sounds better than a moderately fast IC.
The 'irony' here is that I have one 'critic' stating here that 2V/us is all anybody really needs, in the form of a 4558, and an equal 'critic' states that 1000V/us is very worthwhile. Go figure everyone, what is necessary to get the best sound possible?
Now, I am somewhat in between, with 20V/us a reasonable minimum, and 100V/us optimum FOR A PREAMP. This is not derived only from slew rate related TIM, but slew rate related PIM, that Dr. Otala determined about 35 years ago must be another important cause of sound 'corruption'. It has been a long hard road that we have traveled in order to get really higher quality audio designs, bitten along the way by 'critics' who are like mosquitos, biting at every opportunity.

[Esperado]

Originally Posted by john curl

The 'irony' here is that I have one 'critic' stating here that 2V/us is all anybody really needs, in the form of a 4558, and an equal 'critic' states that 1000V/us is very worthwhile.

John, once again, it is not a question of numbers, it is a question of *context*.
We do not have to care of slew-rate when no feedback is applied. The bandwidth and the static distortion is all we have to worry about.

If we use global feed-back, it is obvious that, like all servos, the accuracy will depend of the speed the servo can react (correct) to errors. We are in a dynamic mode and a more complex system.
If the amplifier makes no errors in open loop, we don't have to care neither to the speed of the loop, because it has nothing to correct, ok ?
More It makes errors in open loop, more we will need feed-back, to correct them. More we will have feed-back, more we will need to minimize the delay, ie increase the open loop bandwidth = slew-rate.
So we cannot say 100V/µs is enough. All depend of the context.

As far i'm concerned, and because there is never a limit and best is better, i look for the maximum slew-rate, and a minimal distortion as the two are often correlated.

Take the same amp in voltage feedback and in current feedback. If you set the slew rate identical in the two amps, the voltage feed-back will have less distortion, because the differential input stage cancel a little the distortion. Take advantage of the current-feedback slew-rate ? The current feed-back wins.
May-be we are mosquitos, but mosquitos with a brain, living today, with technology and understanding of today, not 35 years ago..

Looking inside the loop of any amp with global feedback, it looks they are not yet fast enough, as the response curve begin to increase around 1Khz (3 Khz for my 1200V/µs one).

[magnoman]

Do these 1000v/us CFA enter class AB operation on their input pairs when slewing this fast?

Thanks
-Antonio

[kgrlee]

Originally Posted by john curl

The 'irony' here is that I have one 'critic' stating here that 2V/us is all anybody really needs, in the form of a 4558, and an equal 'critic' states that 1000V/us is very worthwhile. Go figure everyone, what is necessary to get the best sound possible?

The real arbiter is properly conducted Listening Tests.

There are also technical measurements like Hirata & Quan that are easier to conduct than a pukka Double Blind Listening Test and are said by self styled gurus to be important indicators of sound quality.

But these same gurus are designers who refuse to carry out Listening Tests and even measurements on their own designs, falling back on their "billion years of experience and zillion awards".

These qualifications excuse them from such trivial tasks and when they do carry out tests, they refuse to release the results on the grounds that us unwashed masses have insufficient intelligence to interpret them properly.

These gurus take every opportunity to avoid ALL tests that may make their supa dupa designs appear the same as evil 4558 and their $zillion ultra supa dupa designs appear worse.

[gerhard]

Originally Posted by Waly

What exactly means "all class A"? And why output stage bias > 20 mA and not > 5mA or > 50mA?

That bias thing is easy to understand. Take a vacuum cleaner as an example.
A lesser v.c. might need only 1000 Watts to operate. It probably sucks not
too bad, but a 2000W unit simply must do it twice as good. Just imagine a
crumb. Could it resist the power of _ADDITIONAL_ ONE THOUSAND Watts?
Never ever!
Ok, the 2000W unit could double as an efficient 2KW heater blower.
But one thing is clear: No additional 1000 Watts, no terrorized crumbs.

So, that is as plain as a pikestaff: 20 mA bias burnt is better than 5.
No question. Spray more - get more. Ooohps, that was AXE.

Gerhard.