why not use 1 OPAmp for amplification and buffer ??

[Wavebourn]

It appears to be a rather unique design--at least to my eyes.

Not so unique. Similar approaches were used long time ago in a precision measurement equipment. However, I never saw combination of both (bootstrap and modulated CCS).

I don't want you to buy me a lunch; I want to listen to as clean as possible sound. An I believe the author of the original question wants the same. That's why I offered my humble opinion that don't agree with your educated scientific one.

 

[KSTR]

While I like basically like Anatoliy's buffer, I fail to see why the modulated (by input voltage) CCS would yield a constant drain current -- which would be a basic requirement for minimum distortion (constant Vds and constant Id). EXCEPT for a known, specific load where it would cancel the current variation in the main FET. For a known resisistive load the proper plain voltage-derived modulation would do the trick, but what about unknown loads, presumably reactive? Here the CCS might work correcting the current either in the right direction or in the wrong direction (including a freq. dependent phase offset with reactive loads).

- Klaus

 

[Wavebourn]

While I basically like Anatoliy's buffer, I fail to see why the modulated (by input voltage) CCS would yield a constant drain current -- which would be a basic requirement for minimum distortion (constant Vds and constant Id). EXCEPT for a known, specific load where it would cancel the current variation in the main FET. For a known resisistive load the proper plain voltage-derived modulation would do the trick, but what about unknown loads, presumably reactive? Here the CCS might work correcting the current either in the right direction or in the wrong direction (including a freq. dependent phase offset with reactive loads).

Very good point, Klaus. Speakers are far from ideal loads, so using a modulated CCS instead of just a constant current source we fall between 2 minimums: a minimum of distortions caused by current modulations, and a minimum of power dissipation needed by class A stage to deliver required output power, when I select an idle current equal to a peak current on a nominal load resistance. However, I would never use such an amp with a speaker that has lower than two times of a nominal impedance less on any frequency.

 

[KSTR]

So with a known load, even when it is a speaker, we could model the current-to-voltage relationship (by a mimicking network) and derive the modulation from that. Then it would be steered modulation, as opposed to a controlled one (by "measuring" the current and adjusting it with a control loop, aka feedback of sorts).

- Klaus

 

[Wavebourn]

So with a known load, even when it is a speaker, we could model the current-to-voltage relationship (by a mimicking network) and derive the modulation from that. Then it would be steered modulation, as opposed to a controlled one (by "measuring" the current and adjusting it with a control loop, aka feedback of sorts).

Sure, but what about delays and other errors of such a "measurement"?
I intentionally avoided any additional negative feedback, especially around more than one stage. In one version of a Tower - II I used about 20 dB of global feedback to compensate this errors, but it sounded worse than without it (less real!). I assume, because of higher order distortions caused by straightening of a transfer function such a way.

Edit: let's go to the original topic to discuss my amp? It's an off-topic here...

 

[Oystein]

I see there is quite a debeate going on which solution is best.

I think most wil agree that a discrete solution is better than most Op-Amps.
Some claim that the new National Op-Amp is the best out there and it should be better than any discrete design.
This might be true but maybe discrete is still better .... anyway .. any difference will be small.
The chance an Op-Amp betters a discrete design is small and for me not worth to take the risk.
As I only need 1 stage per channel I do not need a DIL-8 containing 2 Op-Amps as I am planning to build the pre-amp dual mono.

The Burson discrete Op-Amps I can use as a single module, they are burnt-in, transistors matched an set properly.
I am confident this solution (which I already used with great result in my cd-player's output - as dual Op-Amp - I/O and buffer) will give much better performance than most (any ??) of the Op-Amps out there.
In case the LM4562/LME49720 is really as great as said here the gain in quality will probably be marginal so I will be happy with the Burson unit.

Of course you are all welcome to continue any discussion here but I got what I came for ... an answer to my original question

Oystein

 

[homemodder]

Oystein, you are going the right direction, dont fear, the National chips will probably not touch the burson in sound quality, I havent heard the bursons but built a copy of a design identical to that designed by a gentleman from china and no opamp Ive tried so far matches it. There was a debate going on about who designed it originally .

There is a design here on this forum by euvl I think his nick is, very simple and tiny, only slightly bigger than a chip opamp with few parts thats great also.

 

[jcx]

the Burson is claimed to be a copy or rebrand of the Audio-gd Earth which gets panned in this detailed op amp comparison:

http://www.sg-acoustics.ch/analogue_audio/ic_opamps/pdf/opamp_distortion.pdf
pg 81

at least this discrete op amp falls short of the better monolithic competition on objective technical measurements that appear somewhat relevant to quality audio

 

[Oystein]

The Audio-DG looks a lot like the old Burson unit.
Now Burson is talking about their second generation units which was developed in collaboration with some of the world’s most passionate musicians, sound engineers and recording artists.... as stated by themselves.
http://www.bursonaudio.com/burson_opamp.htm

The new units looks different but who knows .. maybe they only put the components on the outside instead of sandwich in between.
The models I used in my cd-player are the new design.

Their pre-amplifiers and Op-Amps are reviewed very positive so somehow they seem to do some good design.

 

[Lumba Ogir]

Wavebourn,
the modulation approach does not accomplish much, the goal should be constant Vds (preferably by means of grounded base cascode), controlling the current in order to lower (switching) distortions is not a possibility.

 

[Oystein]

To continue on the debate which op-Amp is best ( IC or discrete ) I found a test in the italian magazine PC Tuner:
http://www.pctuner.net/articoli/sch...a-scelta-migliore-per-la-nostra-scheda-audio/

You can select english language if needed ;-)

Here both the Burson, Audio-GD and some IC are compared with the Burson as clear winner.
So it seems they are not equal as suggested earlier.

I will keep my preference for the Burson unit in my design

 

[AndrewT]

have a look at discrete opamps. In general they are amplifiers that include a buffer as part of an integrated design.

They do all that you ask.

 

[Lumba Ogir]

Oystein,
my answer to your original question regarding how many Op-Amps is: the fewer the better.
Build discrete circuits to have control over all details, carefully choosing topology, devices for each location, biased according to your preference and needs.
Op-Amps represent disputable design concepts.
Op-Amps are made up by components significantly inferior electrically and sonically to the best discrete devices available.

 

[Wavebourn]

Wavebourn,
the modulation approach does not accomplish much, the goal should be constant Vds (preferably by means of grounded base cascode), controlling the current in order to lower (switching) distortions is not a possibility.

I keep Vds bootstrapping the follower by one more follower, Lumba. Variations of load impedance cause variations of the 2'nd order distortion that is already very small since Gfs is huge, and reduced current variations cause small Gfs variations, that don't bother me at all, when the 2'nd order distortions change from zero to -80 dB depending on load impedance. It's inaudible.