Please have a look at http://www.pha.inecnet.cz/macura/buffer_en.html , it is easier to see the image than to explain 
Oh, sorry for confusion. RG58 is used in case that audio buffer serves as improvement of output circuits of the CD player.
For headphone amp you do not need to use RG58. But R12 resistor placed directly at the headphone connector will do its job, you can also try to change its value or to omitt R12 at all, this may depend on headphones type. The readers who have built the circuit as headphone amp are saying that they use R12=50 Ohms.
For headphone amp you do not need to use RG58. But R12 resistor placed directly at the headphone connector will do its job, you can also try to change its value or to omitt R12 at all, this may depend on headphones type. The readers who have built the circuit as headphone amp are saying that they use R12=50 Ohms.
!BUF634 outside feedback loop
It really does seem naughty to use BUF634 outside a feedback loop. I have been running a bluish Corda headamp this way (an OPA637 driving stacked BUF634).
I recently experimented by placing the BUF634 inside the feedback loop of the OPA637.
Harmonic distortion levels at useful volume dropped by a factor of 10, to 0.03% or less. Before, harmonic distorion levels were a very strong function of output level: no more. Output levels are +-0.5 db from at least 7 Hz to almost 200kHz, and a square wave looks fine and shows no ringing I can see. As a side benefit, DC offset dropped from 25 mV to 0.3-0.4mV.
Norman
It really does seem naughty to use BUF634 outside a feedback loop. I have been running a bluish Corda headamp this way (an OPA637 driving stacked BUF634).
I recently experimented by placing the BUF634 inside the feedback loop of the OPA637.
Harmonic distortion levels at useful volume dropped by a factor of 10, to 0.03% or less. Before, harmonic distorion levels were a very strong function of output level: no more. Output levels are +-0.5 db from at least 7 Hz to almost 200kHz, and a square wave looks fine and shows no ringing I can see. As a side benefit, DC offset dropped from 25 mV to 0.3-0.4mV.
Norman
I agree with you, can't see any advantage of a separate buffer.
You already have a heavyli feedbacked opamp so what have you to win with a separate buffer? The BUF634 is so fast so it interferes very little and can be including in the feedback loop without any problems.
My offset voltage dropped to 80 uV at the output when I used the AD8610, even better!
You already have a heavyli feedbacked opamp so what have you to win with a separate buffer? The BUF634 is so fast so it interferes very little and can be including in the feedback loop without any problems.
My offset voltage dropped to 80 uV at the output when I used the AD8610, even better!
i think some people just have philsophical objections to putting too much stuff in the feedback loop... the general goal being to try to keep it as small and tight as possible (both electrically and physically). that said, if the thing doesn't sound any beter with it outside the loop, then there's no point (measured performance should always be better with it inside the loop).
I have tested my SMD amp more now and I have been taking the jumpers on and off. They are for meduim speed, high speed mode, 100 ohms out, and 100 ohms inside the feedback.
As you can see at the pictures I have used jumpers with "flärp"(*) and I have changed them on and off. The only thing that I really hear is when I'm putting a 100 ohms in series with the headphones Sennheiser HD545 (high impedance). I get a small level shift and the sonics may be changed. I'm not certain. 100 ohms inside the feedback loop creates no change in sound and also different speed of the buffer. I have only the original length of the cable. Maybe if I have a longer cable I would notice any difference.
(*) Flärp is this thing at the top of the jumpers, easier to get then off! Very neat.
An externally hosted image should be here but it was not working when we last tested it.
As you can see at the pictures I have used jumpers with "flärp"(*) and I have changed them on and off. The only thing that I really hear is when I'm putting a 100 ohms in series with the headphones Sennheiser HD545 (high impedance). I get a small level shift and the sonics may be changed. I'm not certain. 100 ohms inside the feedback loop creates no change in sound and also different speed of the buffer. I have only the original length of the cable. Maybe if I have a longer cable I would notice any difference.
(*) Flärp is this thing at the top of the jumpers, easier to get then off! Very neat.
Just some interesting reading at www.linkwitzlab.com about linearizing headphones...
Apparently, you need two notch filters with practically all headphones to get subjectively flat frequency response. One of these is linked to a very physically real problem: the cavity resonance between the headphone and your eardrum!
Apparently, you need two notch filters with practically all headphones to get subjectively flat frequency response. One of these is linked to a very physically real problem: the cavity resonance between the headphone and your eardrum!
Norman,
Do you incorporate the 200pF cap in the FB loop of the OPA637 as suggested in the application notes for the BUF634 when the latter is inside the FB loop?
This weekend I tried comparing the sound of the BUF634 when inside the FB loop vs when outside the loop. I did have the 200pF cap present when the BUF634 was in the FB loop, but removed it when the BUF634 was outside the loop. My driving op amp was the AD8610 with a 2 mA CCS. I do not believe I could hear any difference.
I also tried running the AD8610 by itself (all versions were driving about 1m of cable to a power amplifier; not for headphone use). It sounded rather anemic without a buffer, but had a more rounded, smoother sound than when the BUF634 was added. However, an output buffer was decidedly an improvement as this gave vastly superior dynamics and bottom end response.
In addition, I compared another buffer with the BUF634 and it allowed the AD8610 to have the same tonality as when it was run alone, but with the better dynamics that the BUF634 afforded. I'm presently trying to source more of this part, which is now difficult to come by.
Overall, the BUF634 seems to add to the sound of the AD8610 rather than remain neutral by adding a brighter high end. It seemed to accentuate the upper midrange (and even tape hiss). Has anyone else noticed this effect, running with or without the BUF634?
Regards, Robert
Do you incorporate the 200pF cap in the FB loop of the OPA637 as suggested in the application notes for the BUF634 when the latter is inside the FB loop?
This weekend I tried comparing the sound of the BUF634 when inside the FB loop vs when outside the loop. I did have the 200pF cap present when the BUF634 was in the FB loop, but removed it when the BUF634 was outside the loop. My driving op amp was the AD8610 with a 2 mA CCS. I do not believe I could hear any difference.
I also tried running the AD8610 by itself (all versions were driving about 1m of cable to a power amplifier; not for headphone use). It sounded rather anemic without a buffer, but had a more rounded, smoother sound than when the BUF634 was added. However, an output buffer was decidedly an improvement as this gave vastly superior dynamics and bottom end response.
In addition, I compared another buffer with the BUF634 and it allowed the AD8610 to have the same tonality as when it was run alone, but with the better dynamics that the BUF634 afforded. I'm presently trying to source more of this part, which is now difficult to come by.
Overall, the BUF634 seems to add to the sound of the AD8610 rather than remain neutral by adding a brighter high end. It seemed to accentuate the upper midrange (and even tape hiss). Has anyone else noticed this effect, running with or without the BUF634?
Regards, Robert
Note, I have used OPA627, not 637! I have had no trouble without the capacitor but it's more safe to have this capacitor to prevent oscillations (long cables and low output impedance).
The AD8610 can't very well drive headphones alone at high levels so it's no surprise that a buffer is good. A buffer does also wonders in RIAA amps when the RIAA network is of a low impedance kind (in order to get low noise).
Capslock, I surely need an equalizer! I have a dip in my hearing around 6 kHz, very narrow also!
I have listened even more to AD8610 and BUF634, heaven is near.
I also like the extremely low output offset voltage and the fast startup without any glitches or bumps. The power down is also very controlled.
The AD8610 can't very well drive headphones alone at high levels so it's no surprise that a buffer is good. A buffer does also wonders in RIAA amps when the RIAA network is of a low impedance kind (in order to get low noise).
Capslock, I surely need an equalizer! I have a dip in my hearing around 6 kHz, very narrow also!
I have listened even more to AD8610 and BUF634, heaven is near.
I also like the extremely low output offset voltage and the fast startup without any glitches or bumps. The power down is also very controlled.
BUF634, continued
"how to stack buf634? Should I put output resistors to individual buffers?
The convention on the Headwize site is simply to solder one buffer on top of the other. Small individual output resistors would probably be a good idea, but I did not use one. In fact, as a headphone driver, I have only a shared 7.5 ohm output resistor for two stacked buf634. Works fine with high impedance headphones (HD600).
Stacking buffers when driving headphones clearly reduces distortion levels significantly. See the findings of "KurtW" on the relevant headphone DIY sites. [This also holds for NE5532, etc. driving low impedances.] I doubt it would make any difference with a high impedance load.
"Norman, Do you incorporate the 200pF cap in the FB loop of the OPA637 as suggested in the application notes for the BUF634 when the latter is inside the FB loop?"
Robert: I stared at that cap in the diagram for a long time, and in the end did not add it. Got 10 pF in there already (not that it does much). No evidence of any problems without the 200 pF.
"This weekend I tried comparing the sound of the BUF634 when inside the FB loop vs when outside the loop. I did have the 200pF cap present when the BUF634 was in the FB loop, but removed it when the BUF634 was outside the loop. My driving op amp was the AD8610 with a 2 mA CCS. I do not believe I could hear any difference."
I'm driving the OPA637 with a. 4.5 mA CCS: seems to sound better than 1 mA; didn't try 2 mA.
Norman
"how to stack buf634? Should I put output resistors to individual buffers?
The convention on the Headwize site is simply to solder one buffer on top of the other. Small individual output resistors would probably be a good idea, but I did not use one. In fact, as a headphone driver, I have only a shared 7.5 ohm output resistor for two stacked buf634. Works fine with high impedance headphones (HD600).
Stacking buffers when driving headphones clearly reduces distortion levels significantly. See the findings of "KurtW" on the relevant headphone DIY sites. [This also holds for NE5532, etc. driving low impedances.] I doubt it would make any difference with a high impedance load.
"Norman, Do you incorporate the 200pF cap in the FB loop of the OPA637 as suggested in the application notes for the BUF634 when the latter is inside the FB loop?"
Robert: I stared at that cap in the diagram for a long time, and in the end did not add it. Got 10 pF in there already (not that it does much). No evidence of any problems without the 200 pF.
"This weekend I tried comparing the sound of the BUF634 when inside the FB loop vs when outside the loop. I did have the 200pF cap present when the BUF634 was in the FB loop, but removed it when the BUF634 was outside the loop. My driving op amp was the AD8610 with a 2 mA CCS. I do not believe I could hear any difference."
I'm driving the OPA637 with a. 4.5 mA CCS: seems to sound better than 1 mA; didn't try 2 mA.
Norman
in tangent site, it said not all buffer can be stack directly.
So I assume adding individual output resistors can seperate them.
but I don't understand why in some design puting resistors before buffers.
also the output resistors inside and outside feedback, should them be designed to match headphone and how?
So I assume adding individual output resistors can seperate them.
but I don't understand why in some design puting resistors before buffers.
also the output resistors inside and outside feedback, should them be designed to match headphone and how?
This is an optional thing. Resistor inside the loop creates a very good current limitation AND a limitation of the max output level when using different headphones with different impedances plus I get a (rather) low output impedance. You could also feed the headphones through a resistor which some people prefer but I think it's better especially with long cables, to drive the phones with low impedance.yejun said:
Nornal value of this resistor is 100-150 ohms. This is "lagom"(*).
This is a burning subject I suppose?
(*) "Lagom" has no word in english! This means not too much, not too little
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.