I'm not suggesting that AD603 or AD605 be used in preamps - these are the only two chips in the family that I have experience with. I have no idea about competitors - I'd guess that the architecture is patented but its not new so its possible the patents have expired or are close to expiring.
I obtained these chips because I was looking for a simple and cheap solution for building a DAC. They fulfill that function very well indeed giving tremendous bang for the buck. What do you mean by 'extreme end of audio'? If you're referring to the chip in that paper having 500MHz of bandwidth, I wasn't suggesting that particular chip be used, just giving an example of the X-amp architecture.
I must admit that I have tried to prove that there are better opamps for audio than NE5532 and LM4562. I tried many of them and concluded that Doug is right. There are not "better". Some sound "different" but that is not the same as "better". If one prefers some subtle, or more than subtle, coloration than one can try video opamps with excessive slew rates, or high speed current feedback opamps, but for audio above opams are optimum. Low distortion, low noise and low price. But it is not forbidden to use opamp as some kind of tone control or "aural exciter". Serve yourself!
But I noticed that I prefer the sound of single opamps. They sound marginally better than the same opamp in dual package.
But I noticed that I prefer the sound of single opamps. They sound marginally better than the same opamp in dual package.
For me J-fet opamps are no-no. Their sound is much warmer, bottom end is more prominent but the "speed" is good. Overall, the sound is more valve-like. Presentation is more laid back and softer. You can listen all day with j-fets, never experience fatigue but also never experience excitement. The loss of precision is problem with j-fets. If circuit has only one j-fet inside it can be tolerated, but if sound goes through several overall coloration is over the top. Using low noise j-fet like OPA 2134 does not help a lot. Bipolars sound thinner but somehow more exciting. From time to time you can experience agresiveness with bipolars, but sound is always more involving and resolution of fine detail is always better.
Abraxalito, I was curious about your suggestion...
I think Doug Self's preamp. designs would be considered by specification alone, to be at the extreme end of audio preamplifiers. Apparently, this latest version also pushes control performance further beyond the reach of current types of electronic devices, so now seems a good time to redouble efforts at improving electronic types, maybe as you were suggesting.
What makes Doug’s preamp circuits so hard to better, both subjectively and in the lab, is decision to opt for Baxandall active gain control. This is the core of the circuit. I am absolutely positive that Doug’s preamp would sound great even with low-down 4558. But I would never build his preamp because there is so many things that I never use: balance regulator, tone controls, tape outputs... It would be interesting to try single j-fet opamp at some strategic position in the circuit to tune it to personal taste. (Use opamp sockets!)
For minimalist, active gain control followed by an inverting buffer, to have output signal in phase, would suffice. Just three dual opamps and a pot. LM 4562 would fit nicely because it’s inverting input is much more consistent with noninverting than other audio opamps (read Doug’s Small Signal Audio book).
Please note that LM 4562 needs 2-3 months of constant use to break in. The change is not great but only then the sound stabilizes completely. After that, the sound gets a pinch of sharpness but retains a certain calm. Overall, the charm of this part lies not in some euphonic colorations, but in the absence of any! Once you tune your ears to such distortionless sound, music listening becomes intoxicating experience.
For minimalist, active gain control followed by an inverting buffer, to have output signal in phase, would suffice. Just three dual opamps and a pot. LM 4562 would fit nicely because it’s inverting input is much more consistent with noninverting than other audio opamps (read Doug’s Small Signal Audio book).
Please note that LM 4562 needs 2-3 months of constant use to break in. The change is not great but only then the sound stabilizes completely. After that, the sound gets a pinch of sharpness but retains a certain calm. Overall, the charm of this part lies not in some euphonic colorations, but in the absence of any! Once you tune your ears to such distortionless sound, music listening becomes intoxicating experience.
Hi Guys
DS is certainly a "character". You can love him and hate him at the same time!
His engineering skill is without question. He uses the latest test equipment to verify the design performance and his goal is always to extinguish noise and distortion. I don't think even subjectivist listeners want too much of either.
He is arrogant at times but has earned some right to be so. On the other hand, when he says "I have come up with a better method than this but this is not the place to disclose it" he is being over-the-top. He has his preferences, likes, dislikes, just like everyone here. He is just a silly monkey like everyone else.
His PA book places emphasis on the Lin circuit - but that happens to be what 99% of all amps are, so he is justified to try to show how to make it perform well. He does tend to flippantly dismiss very good alternatives, but part if this is because he is English and the English have a rich history of penny-pinching audio designs. He has had to design amps to a budget while trying to attain best performance. I pity those who bought the early editions of this book as they have to buy the same stuff over and over again to acquire the extra bits added in subsequent editions. Cordell's book has more breadth and a truly useful iterative development showing small circuit changes and how they improve performance.
The EQ in Doug's Precision preamp had all the tone control filtering as a side chain mixed into the main audio path by a add/subtract stage that corrected signal phase after the Baxandall volume circuit. The EQ could be switched out completely and used modest boost/cut to correct for room/speaker irregularities. The use of first-order filters keeps all the phase and delay issues benign and is far more "musical". he even admitted the last part so he is not without human emotion and a desire to enjoy music he listens to.
The Baxandall EQ is easier for most to understand and is highly traditional. The controls can be set to a 'flat' position and the inverting gain stage is still there to correct phase for the Bax volume circuit.Noise contribution by this stage is negligible especially if you start paralleling opamps. Doug challenged his readers to come up with a lower noise pre than the PP above. It seemed obvious to just parallel opamps in it and reduce R values to get lower noise.
Low noise is pretty important if you want to hear just the music at low listening levels. I always find the shape of power amp THD curves versus power to be very disappointing, as the lower end rises due entirely to noise. In my own listening room, a 1W per channel listening level is way too loud. The ideal amp and pre would have as low noise as possible and low THD at those levels. Of course, the room noise sets the minimum SPL and the maximum dynamic range for music.
The 4562 is pretty phenomenal but its current noise is higher than for 5532. At the lower impedances of the new DS pre one might be able to use the 4562. Note that there is a quad version of this called the LME49740, so the chip cost can be held reasonable.
I have no snobbery with respect to technology to attain good sound - subjectively satisfying sound. Tube stuff can sound okay with certain music but for modern pop and hard rock there is too much smearing of mids. Same goes for most of the simple circuits Pass promotes. They are good for simple music - which is what I notice about reviews of such pieces. Solid-state is cleanest and its deficiencies are easily eradicated - usually by close matching of symmetric components, or other labours. ICs work for most people most of the time even though a well-designed discrete circuit can provide better performance in many cases. Discrete circuits also allow the hobbyist to feel more like he is "doing" something.
The first part of most design/build article series will seem a bit theoretical. This sets a ground work for what is to come and also handily pays the author an extra stipend.
Music is supposed to be enjoyable as is building gear to listen to music.
Have fun
DS is certainly a "character". You can love him and hate him at the same time!
His engineering skill is without question. He uses the latest test equipment to verify the design performance and his goal is always to extinguish noise and distortion. I don't think even subjectivist listeners want too much of either.
He is arrogant at times but has earned some right to be so. On the other hand, when he says "I have come up with a better method than this but this is not the place to disclose it" he is being over-the-top. He has his preferences, likes, dislikes, just like everyone here. He is just a silly monkey like everyone else.
His PA book places emphasis on the Lin circuit - but that happens to be what 99% of all amps are, so he is justified to try to show how to make it perform well. He does tend to flippantly dismiss very good alternatives, but part if this is because he is English and the English have a rich history of penny-pinching audio designs. He has had to design amps to a budget while trying to attain best performance. I pity those who bought the early editions of this book as they have to buy the same stuff over and over again to acquire the extra bits added in subsequent editions. Cordell's book has more breadth and a truly useful iterative development showing small circuit changes and how they improve performance.
The EQ in Doug's Precision preamp had all the tone control filtering as a side chain mixed into the main audio path by a add/subtract stage that corrected signal phase after the Baxandall volume circuit. The EQ could be switched out completely and used modest boost/cut to correct for room/speaker irregularities. The use of first-order filters keeps all the phase and delay issues benign and is far more "musical". he even admitted the last part so he is not without human emotion and a desire to enjoy music he listens to.
The Baxandall EQ is easier for most to understand and is highly traditional. The controls can be set to a 'flat' position and the inverting gain stage is still there to correct phase for the Bax volume circuit.Noise contribution by this stage is negligible especially if you start paralleling opamps. Doug challenged his readers to come up with a lower noise pre than the PP above. It seemed obvious to just parallel opamps in it and reduce R values to get lower noise.
Low noise is pretty important if you want to hear just the music at low listening levels. I always find the shape of power amp THD curves versus power to be very disappointing, as the lower end rises due entirely to noise. In my own listening room, a 1W per channel listening level is way too loud. The ideal amp and pre would have as low noise as possible and low THD at those levels. Of course, the room noise sets the minimum SPL and the maximum dynamic range for music.
The 4562 is pretty phenomenal but its current noise is higher than for 5532. At the lower impedances of the new DS pre one might be able to use the 4562. Note that there is a quad version of this called the LME49740, so the chip cost can be held reasonable.
I have no snobbery with respect to technology to attain good sound - subjectively satisfying sound. Tube stuff can sound okay with certain music but for modern pop and hard rock there is too much smearing of mids. Same goes for most of the simple circuits Pass promotes. They are good for simple music - which is what I notice about reviews of such pieces. Solid-state is cleanest and its deficiencies are easily eradicated - usually by close matching of symmetric components, or other labours. ICs work for most people most of the time even though a well-designed discrete circuit can provide better performance in many cases. Discrete circuits also allow the hobbyist to feel more like he is "doing" something.
The first part of most design/build article series will seem a bit theoretical. This sets a ground work for what is to come and also handily pays the author an extra stipend.
Music is supposed to be enjoyable as is building gear to listen to music.
Have fun
Hi Guys
I would like to add that the Bax volume circuit is a very good idea!
Most playback systems have way more gain than is required and this requires adding attenuation somewhere. The gain-att-gain topology has inherent noise penalties and can impose a headroom bottle-neck for some signal sources. I much prefer to have variable gain, so that even if noise rises with gain, the noise is not always there diminishing low-level performance.
It would be interesting to see what performance could be had using discrete opamps as DS illustrates in the "Small Signal Audio Design" in the Bax volume circuit.
Have fun
I would like to add that the Bax volume circuit is a very good idea!
Most playback systems have way more gain than is required and this requires adding attenuation somewhere. The gain-att-gain topology has inherent noise penalties and can impose a headroom bottle-neck for some signal sources. I much prefer to have variable gain, so that even if noise rises with gain, the noise is not always there diminishing low-level performance.
It would be interesting to see what performance could be had using discrete opamps as DS illustrates in the "Small Signal Audio Design" in the Bax volume circuit.
Have fun
I pity those who bought the early editions of this book as they have to buy the same stuff over and over again to acquire the extra bits added in subsequent editions.
I bought editions #1, #3 and #5. I do not complain. Each time, I gave or sold the former edition and each time I very much enjoy the newer one, the extra bits being largely worth to get it.
I bought editions #1, #3 and #5. I do not complain. Each time, I gave or sold the former edition and each time I very much enjoy the newer one, the extra bits being largely worth to get it.
I request's you to please explain, What will be the situation here? Both, when (1k resistor shunts the pot as well as when two pairs of 560 ohms shunting the pot).
What's wrong with using, 560 ohms pair with 10K pot, if its ok in every way?
Please, do consider, that i want to use a linear pot.
Thanks.
Last edited:
The noise factor of the preamp will be slightly worse (slightly as in on paper). A real benefit is the loading on the opamps is reduced substantially.
Dougs older "1996 precision preamp" was more complex than this one and used the higher impedance values. It was ultra quiet too. You won't hear any difference noise wise.
The noise performance would be significantly worse. Exactly how much worse I cannot say without doing the experiment.
The whole essence of the preamplifier design is that it shows how to use 1K pots without resorting to the complications of discrete circuitry to drive the low impedances. To build it with 10K pots is to really miss the point.
See the last part of the reply in kevinkr's post... "the situation will be different, but the taper resulting from such a connection would not be the most useful I could think of."
So it's OK apart from the "taper" or law of the pot. What that means is that rotating the pot a few degrees will not give the same change in output voltage on the wiper as a pot with no resistors connected to the wiper. You would find all the "action" happens very close to each end of the rotation.
Hi Noddy55
Off-topic, but if you want to understand and minipulate potentiometer law (taper) with external resistors, try this article, which covers most issues. Potentiometers (Beginners' Guide to Pots)
Off-topic, but if you want to understand and minipulate potentiometer law (taper) with external resistors, try this article, which covers most issues. Potentiometers (Beginners' Guide to Pots)
Well as I said it's basically Self's discrete opamp circuit as published in his book. Although most of the component values I ended up with are different I think I would still be breaching copyright by posting the schematic as the topology is the same. I recommend getting a copy of the book - it's excellent.
The circuit is basically a Lin toplogy - degenerated LTP input with current mirror, simple common emitter VAS (I tried both darlington and cascode enhancements and found they made the sound worse), and single-ended emitter follower output stage with constant current source (again I found complementary push-pull sounded worse). I suppose the differences in sound quality I heard with the different configurations came down to the distribution of harmonics, but I have no way of measuring distortion that low. I could see it in Spice though.
Would Mr Self object if I posted my simulation schematic here given the component values are different? IIRC he had the circuit free to view on his old website so maybe he won't mind. If he doesn't I will post it.