High-end preamp for my GC

carlosfm · 2004-06-08 10:37 pm

air,

Gain or 10 for a pre?

For headphones yes, but for a pre it's too much, I don't know how you tested...
Volume at the bare minimum?
I always said that the buffer is beneficial if you have long cables.
I have my pre on the equipment rack at a corner of the room and my power amp on a middle position between the speakers.
From pre to power goes a 5m interconnect.
But testing with the power amp near the pre and a short cable I don't notice any difference with or without a buffer.
I have a small pre for my tests.
When you use op-amps and buffers you have to take care of a proper PCB layout, it's not as simple as just op-amps.

I like much more to have long interconnects (with a good pre, of course) and small speaker cables than the reverse situation I had before.

air · 2004-06-09 5:09 pm

carlosfm:

Actually, I said wrong. I had gain set to ~7, and yes, it was way too high. Now I have changed to ~3, and the sound while using buf is much better. The sound with and without buf is almost the same now.

I understand that it might be good to have the buf there, in case of long interconnections, but I believe that in my case, it might be just as good without the buf (at the moment).

I havent made a proper PCB yet, so that might also cause some changes.

Thank you!

carlosfm · 2004-06-09 5:42 pm

Maby more dynamics with the buffer.

Dr.H · 2004-06-15 3:44 pm

The pre that keeps giving...

Replaced my ALPS blue pot last night with a DACT CT2 stepped attenuator and ... the pre just keeps improving! It is almost as though spaces in between instruments have become visible. Highs have taken on much more air, bass is a bit more resolved.

Also paralled the BUF634 (i.e. I am now using 2x BUF634 per channel), another step in the right direction, with improved clarity and resolution on an already good pre.

Will probably bias the OPA627 into Class A with a constant current source based on cascoded JFETS (2n3819 with 2sk170). 2n3819 a good choice with IDSS typically 10mA and low input capacitance of 2.2pF. Of course 2sk170 is needed to bring the IDSS down to around 2-4 mA, which is all the OPA627 needs.

In an IV stage, biasing the OPA627 with the above JFETS gave the OPA627 a more open expressive sound, highs and mids improving notably. Without the biasing, the OPA627 is detailed but dark.

Anyone else played around with optimising the pre?

Hope you can make sense of the descriptions...

Have fun!
Ryan

carlosfm · 2004-06-15 4:08 pm

Ryan, nice to see that you're happy and improving this pre.
I don't know if biasing to class A is going make such a difference here, the BUF634s are already doing part of that job.
Don't extend too much the feedback loop, it has to be as short as possible, so adding more and more buffers may not be beneficial, but I think that 2 buffers may be the sweet spot.

And every chip has to have a correct PSU bypassing.
Someday I'll make a new board with two buffers, for now I'm busy with other things...

Deam, those DACT stepped attenuators are expensive, aren't they?

This pre is a never ending story. :bawling:

Someday I'm gonna make a new board with two buffers and a pair of regulators for each chip, each op-amp, each buffer... :devilr:

Two trafos, one per channel?
Two trafos for each chip?

Dr.H · 2004-06-15 4:13 pm

Hi Carlosfm,

I agree, it is expensive, but I must tell you that is is very worthwhile. The ability to see into the performance is greatly enhanced and soundstage is incredible. For me, I am finally hearing highs with air and a kind of snow-falling delicacy. Really beautiful.
The device is still brand new, and apparentlky it gets better with break-in of 2-3 weeks.

On the BUF634, I have simply stacked them so no real change in feedback loop length.

I'll report on the biasing of OPA627...

Nuuk · 2004-06-15 4:53 pm

Most things in hi-fi are subjective but IMHO, an attenutaor is better than a pot and if you cannot hear the difference, your system is not up to the job!

I'm currently building my new buffered IGC's and will be biasing the opamps into class A but using BJT's as it's easier to implement.

carlosfm · 2004-06-15 5:05 pm

Dr.H said:
On the BUF634, I have simply stacked them so no real change in feedback loop length.

Don't stack them, Ryan...
I suppose you use the DIP8 version.
You should bypass each chip.
As you don't stack op-amps, you don't stack buffers.
I'm not saying it doesn't work, but it could be much better.
I use the TO220 version, with heatsinks.
Maby that's why I have a (very) tight bass and never thought of fiddling any more with this, like more buffers and so on...
The TO220 version is better, and has more current (250 against 200ma).

dickiegeorge · 2004-06-16 1:04 am

carlosfm said:
Two trafos for each chip?

Or two batteries for each chip :spin:

Dr.H · 2004-06-18 10:07 am

carlosfm · 2004-06-18 10:19 am

Ryan, people can do what they want, I know some people do that.
You can see thousands of pages on the net with TDA1543 dacs stacked, 4 and even 8 chips paralleled.
I tested it and didn't like.
One is better.

Try the TO220 BUF634, with heatsinks.
Serious business.

Dr.H · 2004-06-18 11:09 am

Hi Carlos,

Yeah, many opinions, what sounds good to me may be a sows ear to you. The TO-220 samples of BUF634 arrived today, so should be interesting to compare with the DIL versions, will report here.

So far the best mod to the pre has been the CT2 DACT pot. Still breaking in and sounds great.

Tell me, the deep bass that you found with JUST the LM3875, do you still have that now that you have included the pre? I don't, but don't get me wrong, the bass is still deep, just not as deep as with the LM3875 alone. I also found that on its own the LM3875 sounded a bit bright or tizzy in the tops. My setup is NIGC, details below.
I've read somewhere (probably in headwize.com) that the OPA627 cuts off low bass and high highs when compared with something like a AD8610 or AD8055. Just waiting for those to arrive.

Will report here on the benefit, if any of biasing OPA627 in teh pre in Class A using cascoded JFETS's set for around 3mA.

Ciao
Ryan
NIGC, LM3875, 300VA per channel, MUR860, 1000uFx2 per LM3875, unregulated PS, 25VAC secondaries, rectified to 31V DC.

Pre is OPA627 with BUF634 ala Pavel Macura; Separate 50VA toroids per channel, regulated via LM317 to +-18V, All opamps bypassed to gnd with 100nF on pins

carlosfm · 2004-06-18 11:40 am

Dr.H said:
Hi Carlos,

Yeah, many opinions, what sounds good to me may be a sows ear to you. The TO-220 samples of BUF634 arrived today, so should be interesting to compare with the DIL versions, will report here.

So far the best mod to the pre has been the CT2 DACT pot. Still breaking in and sounds great.

Tell me, the deep bass that you found with JUST the LM3875, do you still have that now that you have included the pre? I don't, but don't get me wrong, the bass is still deep, just not as deep as with the LM3875 alone. I also found that on its own the LM3875 sounded a bit bright or tizzy in the tops. My setup is NIGC, details below.
I've read somewhere (probably in headwize.com) that the OPA627 cuts off low bass and high highs when compared with something like a AD8610 or AD8055. Just waiting for those to arrive.

Will report here on the benefit, if any of biasing OPA627 in teh pre in Class A using cascoded JFETS's set for around 3mA.

Ciao
Ryan
NIGC, LM3875, 300VA per channel, MUR860, 1000uFx2 per LM3875, unregulated PS, 25VAC secondaries, rectified to 31V DC.

Pre is OPA627 with BUF634 ala Pavel Macura; Separate 50VA toroids per channel, regulated via LM317 to +-18V, All opamps bypassed to gnd with 100nF on pins

Ryan, what I have now is:
- My pre.

- LM3886 BIGC (with OPA627), with regulated PSU. This is a power amp, and stands between the speakers, with a 5 meter interconnect coming from the pre, on my system rack.

- Epos 11 speakers.

- My sub, with 2//OPA549s and regulated PSU.

The OPA627 MUST have capacitance on the PSU pins, not just 100nf. From 22 to 100uf is what I recommend.
If you don't have the space now, put them under the PCB.
Lack of bass? No way!
Lack of highs? No way!
Just do it as I say, you'll be surprized. :angel:

Note: plenty of work, tests, several implementations, etc. to get to this actual system configuration and now I'm on audio nirvana.

Upupa Epops · 2004-06-18 11:55 am

Have somebody here preamp based on opam + buffer in different combinations of both ?

Dr.H · 2004-06-18 12:11 pm

Hi again Carlos,

On the pre, the OPA627 and BUF634 have about 5 x 47uF electrolytics within 5cm of their V- and V+ pins, with the 100nF soldered directly to the V- and V+ pins.

Do you believe that the 5cm distance would make a difference?

I also have 100nF from V- to V+ on the op-amps and of course the 200pF cap from the output og the 627 to input. I will see whether this cap is really needed.

Always good to tap into anothers experience, thanks for the feedback.

Upupa Epops · 2004-06-18 12:29 pm

Best preamp not only for GC

By my own experience, best results give preamp, which is build as small discrete power amp, working in all stages in class A. In this case you can much more " overdimensed " output stage and all compensation can be tuned for concrete amplify. Try anybody someting like this ?

carlosfm · 2004-06-18 12:38 pm

Dr.H said:
Hi again Carlos,
Do you believe that the 5cm distance would make a difference?

Yes it does! :bawling:

Dr.H said:
I also have 100nF from V- to V+ on the op-amps and of course the 200pF cap from the output og the 627 to input. I will see whether this cap is really needed.

It's needed, without it the treble is a little on the bright side.
Much better with the cap.

carlosfm · 2004-06-18 2:01 pm

Re: Best preamp not only for GC

Upupa Epops said:
By my own experience, best results give preamp, which is build as small discrete power amp, working in all stages in class A. In this case you can much more " overdimensed " output stage and all compensation can be tuned for concrete amplify. Try anybody someting like this ?

Something like the Bride of Zen pre?
It maybe good, never tried it, but that volume pot on the output really makes me scratch my head.
Makes me wanna put an op-amp buffer after it. :clown:

Nuuk · 2004-06-18 4:51 pm

The OPA627 MUST have capacitance on the PSU pins, not just 100nf.

Does that apply to all opamps or just the OPA627? And why?

carlosfm · 2004-06-18 5:11 pm

Hi Nuuk!
I explained this a zillion times didn't I?

It applies to almost all the OPAs, they like capacitance on the PSU pins.
Without that you can't believe the difference, it won't be pretty.
With more or less effect or more or less noticeable depending on the circuit the result is bass detached from the rest of the music, slow, dominant, mucks up midband and treble.
Suffice to say, on a normal circuit, some people prefer the NE5532/4.

Some op-amps are not so picky, but they like it too, and I can guarantee you that on a circuit I make my way I can test any op-amp (OPAs or not) without any problem and know that I'll have a reliable conclusion.

Did you ever opened an Arcam 7SE cdp?
The only PSU caps for the analog stage are on the right side of the board.
I repeat: the ONLY.
Some 5 cm to the left there's one op-amp.

Some 15 cm to the left there's another op-amp. :bawling:

Now put there a good, modern, fast op-amp and it will not deliver. :bawling:

But the solution is easy, of course.

High-end preamp for my GC

Banned

Member

Banned

Member

Banned

Member

Member

Banned

Member

Member

Banned

Member

Banned

R.I.P.

Member

R.I.P.

Banned

Banned

Member

Banned