Buffers only or opamps too?

[hihopes]

Hi, I am a newbie amp builder as well as a brand new head-fier. I want to take a headphone output from my passive pre-amp. I haven't tried it yet, but since the output is already quite sufficicent for my power amps, I suspect I don't really need much, if any voltage boost. From what I have been reading in this forum, I do think a buffer stage would probably be a big help. I have a 13.8v "brick type" transformer with plenty of amps to spare, so that removes the constraints of a 9v supply and miserly current consumtion.
Quality is paramount!!!!!!!
Questions:
It seems to me that more chips make better sound? Is this so? How many buf634p's would be optimal and how many would be plain wasteful? Would I get the best results from a unity gain buffer, or would I do better to precede this with a couple of op-amps?
Any and all suggestions welcome!
(PS Before you ask me what phones I am using, I am still looking around for some. Where I come from, choice is limited, prices are steep and money is a bit few.)

 

[guzzler]

It's not correct to say that as the preamp will drive a power amp, it will definately drive a set of headphones. I would suggest making an amp with a gain stage (ie, opamp), followed by a buffer. Then, experiment with the gain when you have your headphones to see what suits you. If you find you do only need unity gain, you could consider a buffer only amplifier. Stacking buffers increases slew rate and reduces THD; consensus is that 2 or 3 buffers per channel gives reasonable improvement, anything over is diminishing returns. If you find unity is fine for you, you might consider a Szekeres type amplifier, plenty more over at HeadWize

good luck!

g

 

[tangent]

You don't want to use open-loop buffers like the BUF634 without an op-amp, even if you don't need gain. They have high distortion (on the order of 1%) because they lack feedback. Even if you just wrap a gain=1 op-amp around the buffer, that's still an improvement over a straight open-loop buffer.

 

[hihopes]

Thanks for the useful input. I have been looking at the Szekeres design as an option. Is this likely to give superior perfornance to an op-amp design?

If buf634p buffers have such high distortion, wouldn't I do better not to use them, but to rather use op-amps alone? Or should I be looking at a different make /model of buffer?

 

[guzzler]

the Szekeres is outwardly simpler and cheaper, but to get the most from it you need a very good power supply and good passive parts.

The BUF634 is a current buffer, ie it provides a high input impedance and a low output impedance. Wrapping it up inside a feedback loop of an opamp makes the opamp correct the output from the buffer so it is (basically) the input to the opamp just at a higher voltage level. An opamp by itself is not likely to provide enough current to drive a set of headphones easily, hence the use of a current buffer. A "chip" design does not need such a good power supply as the discrete Szekeres, but that does not mean you can skimp there, only it doesn't need to be quite as spectacularly good as the one for the Szekeres. A lot of people use the Elpac wallwart here which, off the top of my head, is around the $30 mark

g

 

[hihopes]

Are you are saying is that when combined with an op-amp, the distortion of the buffer is corrected?

I have a 13.8v power supply which is currently only used for my active crossover, so there must be at least 4 amps excess current available from it. I would like to use this to power the amp.
Questions:
What can I do to get the power from this really smooooooooth?
There must be a way to use it as what would effectively be a +/- 12v supply br combining + and ground at ground. I have been thinking about this for days and at first I just thought of using a big resistor to keep the 2 apart at ground, but that would be practically the same as just using separtate supplies for the 2 channels - leaving 2 grounds. But now a friend says he once saw a design for this - he seems to think they used diodes, but he cant remember where he saw it. any ideas? (And if it is practical, would you consider this good enough for the Sjekeres once nicely smoothed?)

 

[tangent]

Quote:

Are you are saying is that when combined with an op-amp, the distortion of the buffer is corrected?

Yes. That's what feedback is for.

You may be asking yourself why the buffer doesn't have feedback to begin with. For one thing, having a no-feedback buffer opens up a lot of flexibility. For instance, you can trivially parallel (stack) open-loop buffers, but you can't parallel closed loop ones without cute hacks.

Quote:

What can I do to get the power from this really smooooooooth?

Add a regulator. Do a search. There are a great many regulator circuits out there.

Quote:

There must be a way to use it as what would effectively be a +/- 12v supply br combining + and ground at ground.

I'm confused. Do you have two single-voltage supplies that you are trying to combine into a dual-voltage supply, or is your "13.8V" supply actually +/-13.8V already and you want to make it +/-12V?

 

[hihopes]

Sorry, didn't mean to confuse you. It is a single 13.8v supply and I'm trying to make it into a +/- 13.8v supply. Any chance, or am I asking too much?

 

[guzzler]

yep, that's asking too much unless you want to get into switching DC-DC convertors, not the best with discretes. 13.8V will be fine to test out ideas, then if you really like you can get yourself some beefier power supply action

g

 

[tangent]

You can't make a low voltage into a high voltage (+/-13.8V is effectively twice 13.8V, or 27.6V) with linear techniques. DC-DC conversion is nonlinear, and it gives "dirty" DC power unless you spend a lot of money.

 

[hihopes]

Maybe I should be planning to go valve?

 

[hihopes]

Guzzler and Tangent - thanks for your replies. It all sounds so logical when you put it like that. I always knew I would have to build a good power supply one day. It looks like it will have to be sooner rather than later.

As to my previous question - given a good power supply (and I won't be cutting any corners here) am I likely to achieve a better quality result from the Szekeres or from a good op-amp design? ( I am not talking about the TPA6120a2 yet - I don't have the confidence for that yet, though I will be going there one day).

 

[guzzler]

Valve amps are a harder build, due to (usually) point to point wiring, more case work, difficult to source components, and are usually more expensive; not to mention the high voltages involved!

I think you'd get better quality more easily from a chip design: less sensitive to layout than discrete, better rejection of bad power, PCBs available (

). The PCB especially will help you if this is your first build as it is one less thing to worry about and means you won't make an obvious mistake. (OK, not true, we've all messed up a PCB before

If you would like to try discrete, you could try a Gilmore Dynamic which should give better results than a straight Szekeres and is available in kit AFAIK.

g

 

[hihopes]

I am not an absolute green-horn. I have built a chu moy (twice - it was harder than I expected to translate it onto stripboard the first time.) But i am still pretty much a beginner. I think for a start I will try a few ideas with chips. I like the A47 idea, but have only come across a circuit diagram so far. Is there a full article on it that someone can point me to?
I am also interested in biasing op-amps into class A using a j-fet cascode. Is it possible to do this in the A 47, or are the op-amps already doing this to each other?
Perhaps I can add a buffer output stage to the A 47 - maybe even the Szekeres, but I suppose buffer chips would be a lot easier to implement.
Can anyone tell me whether there is a noticeable difference in sound quality between single channel and dual channel chips?
Is there any benefit to be gained from using double the amount of circuits - ie 2 left and 2 right, and then joining them up at the output?
Sorry about all the questions, but I really do want to know.

 

[Earwax]

Quote:

Originally Posted by hihopes I am not an absolute green-horn. I have built a chu moy (twice - it was harder than I expected to translate it onto stripboard the first time.) But i am still pretty much a beginner. I think for a start I will try a few ideas with chips. I like the A47 idea, but have only come across a circuit diagram so far. Is there a full article on it that someone can point me to? I am also interested in biasing op-amps into class A using a j-fet cascode. Is it possible to do this in the A 47, or are the op-amps already doing this to each other? Perhaps I can add a buffer output stage to the A 47 - maybe even the Szekeres, but I suppose buffer chips would be a lot easier to implement. Can anyone tell me whether there is a noticeable difference in sound quality between single channel and dual channel chips? Is there any benefit to be gained from using double the amount of circuits - ie 2 left and 2 right, and then joining them up at the output? Sorry about all the questions, but I really do want to know.

The original A47 article is at headwize http://headwize.com/projects/showfil...heared_prj.htm and the first thread about it http://headwize.com/ubb/showpage5.ph...d=661&fdays=20

I wish I knew the answers to your technical questions, but I don't...