From the OPA627 datasheet.
"Power supply connections should be bypassed with good
high frequency capacitors positioned close to the op amp
pins. In most cases 0.1mF ceramic capacitors are adequate.
The OPA627/637 is capable of high output current (in
excess of 45mA). Applications with low impedance loads or
capacitive loads with fast transient signals demand large
currents from the power supplies. Larger bypass capacitors
such as 1mF solid tantalum capacitors may improve dynamic
performance in these applications."
We are talking here of op-amps used for audio applications, music ("fast transient signals").😉
Just forget the tantalums and use 22~100uf electrolythics bypassed with 0.1uf ceramic or poly, ok?😀
"Power supply connections should be bypassed with good
high frequency capacitors positioned close to the op amp
pins. In most cases 0.1mF ceramic capacitors are adequate.
The OPA627/637 is capable of high output current (in
excess of 45mA). Applications with low impedance loads or
capacitive loads with fast transient signals demand large
currents from the power supplies. Larger bypass capacitors
such as 1mF solid tantalum capacitors may improve dynamic
performance in these applications."
We are talking here of op-amps used for audio applications, music ("fast transient signals").😉
Just forget the tantalums and use 22~100uf electrolythics bypassed with 0.1uf ceramic or poly, ok?😀
Carlos, try to get out established prejudice and think with me. By ordinary conception is preamp volume control before output amp. If this amp have SNR for example 100 dB ( pot of volume in position " up " and Rg for example 100 ohm ). If by pot slide volume on - 40 dB, SNR will be only 60 dB, respectively worse, 'cos you must count also thermal noise of pot. Second disadvantage of this solution is, that if is on input of this amp any passive RC low pass filter, his frequency will be changed ( depend on volume position ). Both these disadvantages are trying to solve by volume control at the output , but this solution is simply only at the first look. SNR is clear - with falling volume noise fall too and SNR is constant. The same clear is input filter, Rg is constant too, so frequency is the same. But in this case is main problem choice of regulator. Pot or rotary switch you can't to use, 'cos this attenuator must have, in this case, very low summary resistance, which can't to do with pot and do it with switch is realy difficult. I guess, that in Pass's are by relay switched switch, connected as pot, with relatively high value of resistance and mainly with nonconstant output impedance, which is wrong. More elegant is serial combinations of T or Pii attenuators, with defined the same input and output impedance. If you make it on 50 ohms, you can connect it straight beside your preamp with buffer and with only 7- bit logic you get 128 dB of attenuation. Better configuration isn't possible to get. Concrete solution have a few " little " problems, but about it maybe in next time. 😉 .
What nobody but Nuuk seems to have used, or reported on results, is a simpler buffer using just one chip instead of two like the one Carlosfm is so happy about, inspired on a Pavel Macura application.
In fact the confusion about the e-mail from a different Carlos to Pavel Macura was due to a question I mailed to him on possible improvements or lack of them when using just one chip buffer with an IGC instead of two.
If the buffer is the same box as the IGC you don't really need drive force to go through the cable, as it will be short o non-existent (e.g.: if on same board). Biasing into class-A, with FET or resistor, might then be more meaningful, as there wouldn't be a BUF634 to do that.
Trying different chips on that function would also be interesting. Using good regulation as well as proper bypassing, as quite rightly Carlosfm defends, are also valid here.
Soon I will be reporting my results here, but maybe there others that tried that path.
Carlos
In fact the confusion about the e-mail from a different Carlos to Pavel Macura was due to a question I mailed to him on possible improvements or lack of them when using just one chip buffer with an IGC instead of two.
If the buffer is the same box as the IGC you don't really need drive force to go through the cable, as it will be short o non-existent (e.g.: if on same board). Biasing into class-A, with FET or resistor, might then be more meaningful, as there wouldn't be a BUF634 to do that.
Trying different chips on that function would also be interesting. Using good regulation as well as proper bypassing, as quite rightly Carlosfm defends, are also valid here.
Soon I will be reporting my results here, but maybe there others that tried that path.
Carlos
carlmart said:
Who said I didn't try, Carlos?
I have several GCs, you know.
My power amp is inverting LM3886, has an input buffer (OPA627) and I use exactly the same circuit on another GC I have.
Just change feedback resistor from 220k to 330k, a volume pot and that's it, as simple as that.
I always told everyone that the reason I made this pre is because I NEED it.
One reason is that I wanted to have a standalone pre.
The other reason is that I have my system (including the pre) on one corner of the room, and to connect the left speaker to the amp I needed a 6m cable, and for the right speaker, 3 meters.
I've put the power amp on a central position.
So, I have a 5m interconnect between pre and power, and a 2 meter pair of speaker cable, this is much better.
You will find on this thread my repply to someone where I recommend for an integrated amp just the OPA627, no need to use the buffer.
I know the OPA627 very well, alone and not alone.😀
carlmart said:
I was surprized when I've read your post because I had no idea of what you were talking about...
I'm at home now and I can clarify this:
It was Pavel Dudek (Upupa Epops) that sent me a mail and I repplied him, BTW very nice guy.😉
Pavel Macura (PMA) confused me with that question, and because I had no access to Upupa Epops' mail I've got really confused and not sure which Pavel It was.😕
Then you came in and I really got more confused.😕
All is clear now.
Carlos (carlmart)😀 ,
Look at my boards, here (open the pic at post #3):
http://www.diyaudio.com/forums/showthread.php?s=&threadid=35482
Look at my boards, here (open the pic at post #3):
http://www.diyaudio.com/forums/showthread.php?s=&threadid=35482
Capacitors at OA PSU pins
PSRR decreases with increasing frequency (and is different for V+ and V-). For the HF mess, PSU pins must be considered as further inputs of the opamp. That's why 100nF capacitors must be placed directly at the OA PSU pins and the return ground path must be as short ans as thick as possible.
PSRR decreases with increasing frequency (and is different for V+ and V-). For the HF mess, PSU pins must be considered as further inputs of the opamp. That's why 100nF capacitors must be placed directly at the OA PSU pins and the return ground path must be as short ans as thick as possible.
carlosfm said:
This is not a question, but reply. Somebody asked hereabove why 100nF caps must be placed directly at OA pins. Before I happened to reply, you posted several more posts 😉 So I have just better edited that post 😉
carlosfm said:
Sorry to have missed those comments.
I do remember the reasons why you used the two chip buffer for. They make quite sense.
From the photos it seems that your buffer is on the same board as the IGC, right?
What other chips did you try on it, besides OPA627? Socketed?
Carlos
Just curious:
Has anybody ever tried a simple source follower with for example IRF610.
Seems that´d be the most simple, purest solution.!?
What do you mean, that you get with this ? Only little bit bigger distortion on lower impedance 😉 .
High input- low output impedance (certainly lower than DIP-package opamps) and maybe "purer" sound than a rather "complex" opamp.
A source follower biased into class A should give minimal distortion. No?
Good UGS opamp have lover output impedance than any source follower 😉 and the same is about distortion 😎. But you have thruth - in some cases can follower sounded better.
Thanks for clearing that up, Pavel.
As said just curious if anybody has tried and still am what a buffer does anyway to the main chipamp so gonna try that soon.
In what order might be the output impedance of a source follower biased with 50mA or so?
As said just curious if anybody has tried and still am what a buffer does anyway to the main chipamp so gonna try that soon.
In what order might be the output impedance of a source follower biased with 50mA or so?
carlmart said:
Yes.
Edit: And the regulators for them are on the board too.
If I want to make an integrated amp, just add a pot before the board.
Socketed.carlmart said:
With good machined sockets.
The way I bypass there's no harm to use a socket, and it's very practical.
When I make it on veroboard I put 100nf multi-layer ceramics inside😱 the socket, between + and ground and - and ground, and the electrolythics are so close that they touch the socket, it's difficult to take out the op-amp.😀
My favourite op-amps, in order of preference:
- OPA627/637
- LM6171/2
- OPA2132
- OPA2604
- OPA228/2228/4228😀
I tested the AD797 on my phonopre and liked it, but put back the OPA637 and it's better.😀
Not much experience with the AD797, but it seams good.
I have here looking at me some AD8620 op-amps, but haven't tried them yet, the same with some others.
Ah, I don't like the AD825 at all.
joensd said:
You can make a headphone amp with this and blow your ears with a festival of sound and dynamics.😀
😎
I´ve etched a PCB for OPA627/BUF634 headphone amp.
The power supply is nowhere near perfect as it is a (big) wallwart with 35V regulated and on the PCB it´s split by another BUF634 to give approximately +-17V.
-I´ve found that combination to sound very detailed, no noise and very analytical.
But I´ve also found that layout is very critical as popping in a OPA604 for example the sound gets real harsh although it is a "recommended audio opamp".
-In comparison a mini-Zen biased with a resistor or little 6V bulb gives me at least the same brilliant sound reproduction but is nowhere near as sensitive to layout than opamp circuitry is.
(for example : I´ve decoupled the opamps on the PCB with 100nF,4,7uF, 47uF near the IC´s)
The last two points make me wanna rather use the simpler circuit as I don´t wanna etch PCB´s weekly and time certainly is a limiting factor.
Then again I can well imagine once you get it right the opamp might be far superior to the simple "one-gain(buffer)-stage" topology.
- Status
- Not open for further replies.