Analog Stages and CDPs

[rollo]

Do we really need an analog stage. IMO its the downside of the concept. Why not just use a balanced transformer to match the impedance of the Pre or an internal TVC to do so. The analog stage of a CDP is redundant and sholud be eliminated. If its a gain issue why not use a TVC directly from D/A conv.
    Am I missing something here? Whats your opinion?

rollo
 

[opnly bafld]

« Reply #49 on: July 04, 2007, 12:34:35 PM »
   


As you know, if we take the source's output, we need x amount of gain for our system to reach full volume.
Traditionally part of the gain was in the active preamp (which provides switching and volume control) and part was in the amplifier.
Now many companies are putting all of the gain in the amplifier, which is now an integrated amp by definition, even though it may not have source switching or a volume control. Many companies try to lead us to believe they have eliminated a gain stage because no active (gain) preamp is required, but the gain is in the amplifier. How good is the added gain stage(s) compared to a good active pre?
Same with the source, say a CD player that puts out 2 volts, if the gain in the output is eliminated we now need an active preamp to drive the amplifier to full power, again how good is the extra gain in the CD player compared to an active preamp?

This is from the thread "Let's talk about pre-amps"
Lin

[rollo]

I understand. However I believe that a transformer can do the same thing that the active linestage does. That is providing gain. Yes, No?

rollo

[richidoo]

Voltage gain, but not current. The size must be large to reduce impedance enough to drive some inputs with low output current of a DAC. The size starts to add distortion as it gets too big. The newest opamps are very good, and cheap, I would prefer them, which can source current and drive cables and low impedance inputs with ease, still remaining below 0.0001% distortion. Designers have to be able to afford them in the budget, provide clean power and good implementation, which is another matter. Designing and using a transformer correctly is probably harder though?
Rich

[_Scotty_]

The little known fly in the opamp ointment is the fact that an opamp is essentially a power amp in miniature.
The rub is that the front end input circuitry is on the same die as the output transistors. When the output transistors try to supply current to a load they heat up just like the output transistors on full size power
amp. When the entire opamp heats up trying to supply current is does so unevenly this leads to an imbalance in the thermal tracking of the input circuitry which destroys the feedback loop accuracy. The end result of this is transient thermal debiasing distortion. At any given time as much half of what you hear from an opamp used in a conventional manner is distortion. The cure is to put a open loop buffer into the feedback loop of the opamp and use the buffer to drive the load. This in a way duplicates a properly designed power amp where the input circuit is isolated from the heat of output stage. Classe' was one of the first manufacturers to do this in a DVD player a few years ago. They screwed up and used too slow an opamp in their current to voltage stage after the DAC but thats another story. In a nutshell every opamp that is asked to drive a load and supply current should be have a buffer in it's feedback loop. An external buffer does not remove the burden from the opamp.
  My vote still goes a properly designed active circuit be it discrete or a buffered opamp over a transformer for analogue applications based on what I have heard so far .
Scotty

[richidoo]

Thanks Scotty, that is cool! Good rational for the burson discreet opamp circuit. What about some of the newest opamps (like LME49720) which are very linear and have low distortion (0.00003%?) Could they be designed with two internal dies, or maybe incorporate a built in feedback loop as you describe?
Thanks
Rich

[_Scotty_]

richidoo,most of the newer opamps have distortion figures that are so low that they are calculated and cannot actually be reliably measured. To get manufacturers to change how they make a chip would be a good trick. The problem would have to be more widely recognized than it is and dissatisfaction with the sonic qualities of the opamps as currently manufactured would have to be voiced. The neat thing about the latest generation of  super low distortion opamps is that when properly utilized with a buffer they can sound very good. The unfortunate fact is this problem exists in every product using an opamp unless they have inserted a buffer in the feedback loop, period. In many cases a good product with a very good DAC in it can be transformed by replacing the 10 cent/1000 opamp with a better one that is buffered and adding a clock along with some power supply mods. 

[richidoo]

Thanks Scotty. Being only an armchair electronics moron I think of an opamp as a buffer, although I know tubes can buffer current, but still same concept - a current source. Do you mean putting any kind of buffer, even another opamp in the open loop feedback to make a good opamp even better? What do you mean as a buffer in this app?
Thanks, this is interesting to me believe it or not, I appreciate your knowledge!!
Rich

[_Scotty_]

A buffer could consist of a FET or MOSFET inserted into the feedback loop of the opamp or it could be a chip
based open loop buffer. The key to a chip buffer is that it has no feedback loop,that is what allows it to be inserted into the opamps feedback loop. The buffer supplies the current to drive the load unburdening the opamp. The buffers distortion characteristics are controlled by the feedback loop of the opamp and the opamps distortion levels are closer to the factory specs under transient dynamic conditions. 

[rollo]

Scotty,
           I"m impressed your knowledge. Thanks for the opinion. Just a pipedream thought it might make sense. I see its much more complex than I thought.


rollo

[richidoo]

Great, thanks. Always more to learn, that's what I love about this hobby. Are you a designer?

Can you recommend any buffer chips in particular so I can read specs, and look at suggested implementation schematics? Thanks
Rich

[_Scotty_]

The demand for high speed buffer chips has fallen dramatically in the last few years. It used to easy to get a very wide bandwidth video buffer chip for this sort application. These things have turned into unobtainium in the last five years. Buffers are sold by Texas Instruments/BurrBrown but they are not open loop devices and therefore are not applicable. At this time discrete buffers designed around  FETS or MOSFETS are what you have to use. I wish I was a designer, but this is a depth of knowledge I do not possess. Most of what I have learned has been from phone conversations with Stan Warren over the last two decades. I get just so far in some of my conversations with him and I have to have him stop and re-explain something. Math is my weak point and designing my own circuits would require a lot of it just to start. It is easy to build a an amplifying circuit and hard to design one that doesn't loose information along the way and sound bad as well. The cool thing is Stan listens pretty effectively and the bad sounding dead ends don't get out the front door.
Scotty   

[richidoo]

Thanks Scotty. You have given me enough to chew on for a while! I found Stan's website too, will read up on his stuff.

[steve]

I understand. However I believe that a transformer can do the same thing that the active linestage does. That is providing gain. Yes, No?

rollo

Not really. They appear simple, but are quite complex.

These problems including insufficient gain, frequency balance problems, more reactive than a good wide bandwidth preamp, hysteresis distortion (especially at low signal levels), primary to secondary coupling problems, resonance problems and more. One of the weakest links in a  tube type amplifier is the output transformer. A tube amp would be much better off without them, except they are required for coupling from a high impedance source to low impedance load.

Hope this helps.

[rollo]

Steve,
            I was thinking that a TVC could be used in lieu of the analog board in a CDP. I personally did not experience any problems with the TVC as far as gain, use of long ICs, freq. bal issues. Resonance problems yes. When the Trannie was mounted on Ebony and the sides treated with BHP they seemed to go away. The quality of the trannie is key. A balanced double "C" core or "R" core done right should be just fine. What I do not understand is how to get the voltage required to drive the Amp directly from the CDP I described.

rollo