I might point out that the 'recommended' listening tests can't tell a 4558 from a 1000V/us IC, by the very nature of the test conditions demanded. That is why I don't use them. Esperado please note. '-)
In the alleged case above your post it s actually the VAS that
is working close to class C.
cool "Frakensitein's" Strawman - conflate several deliberate misinterpretations
actually the 1-2 v/us/V was from Quan's measurement data - the op amps that were too good to measure on his rig
I didn't propose Quun's tests as some fundamentally new arbiter audible circuit quality, didn't taunt people with "...just you see" when last year some pointed out he hadn't tested even the TL072, NE5534
do I have to find the posts? - the search will start with “find posts by username: John Curl”
and where did the 4558 enter the discussion – it doesn't manage 2 V/us/V for even 2 Vrms - there really aren't any defienders of its audio quality - wasn't it just proposed as an extremely poor op amp to comapre to
actually the 1-2 v/us/V was from Quan's measurement data - the op amps that were too good to measure on his rig
I didn't propose Quun's tests as some fundamentally new arbiter audible circuit quality, didn't taunt people with "...just you see" when last year some pointed out he hadn't tested even the TL072, NE5534
do I have to find the posts? - the search will start with “find posts by username: John Curl”
and where did the 4558 enter the discussion – it doesn't manage 2 V/us/V for even 2 Vrms - there really aren't any defienders of its audio quality - wasn't it just proposed as an extremely poor op amp to comapre to
Last edited:
Ron's test do test something UNIQUE, but they still do not have the sensitivity to measure differences in most of the 'better' IC's. The test residual is still too high, but it is just a beginning.
Dunno about 1000V/us IC but I'm confident the 'recommended' listening test can distinguish between Blowtorch & evil 4558 ... and more important allow a reliable judgement of which sounds better.
JC, I'm open to suggestions on how to improve my ABC Blind Listening Test. So far, it looks like it answers most of the criticisms that various posters have brought up.
In the meantime , are you releasing your Hirata test results on Blowtorch?
When are you going to let Quan test Blowtorch and show it is beyond blame? 😱
But is is sooo simple, John. Take any low THD current feedback OPA at 1000V/µs and enough drive capability for your charge, at any gain between 1 and 10, listen, compare with a strait wire and make yourself your mind.
Since how many years had-you stopped to try things and stay stuck on your 35 year's old convictions ?
Since how many years had-you stopped to try things and stay stuck on your 35 year's old convictions ?
Richard, I don't have YOUR headphone schematic handy, but if I remember, you used a pair of complementary jfets as your input stage. What were their numbers, please? Then a discussion can follow.
Hard to measure what's happen in an OPA/-) but your question have an interest. Some people prefer Class A because the crossing distortion.
We have to notice that this distortion happens where the slew rate of any signal is maximum.
When the servo is fast enough to correct-it > 100 times faster than recommended by JC😉, we can understand why cross distortion is no more a problem with very fast OPAs even at HF.
So, please, just try, and get back with your sincere reports.
I don't write those things to bore-you, but to share my experience, and, who knows, let-you discover a new track, passionating to explore.
Why keep raising the class AB thing? I've built a buffer using an LM4562 biased into class A and th distortion is down at 2-3 ppm as measured on one of the latest high end AP systems. So opamps and class A operation is a complete non issue. If you are prepared to go to the expense of building a discrete stage, you should not feel any reservations about adding 3 small signal transistors and a some passives to create a first class line stage. And, if you don't like BIP input opamps, swap for a JFET type, of which there are many fine examples.
Here are the test results
http://hifisonix.com/wordpress/wp-c...niversal-Small-Signal-Class-A-Buffer-V1.0.pdf
The design here is for a headphone buffer. Just scale down the output buffer currents for line level operation and run at c. 16dB stage gain.
BTW, jcx proposed a neat solution for the opamp class A bias. Bootstrap the output by removing the output current source load and simply tie a resistor from the opamp output to the emitter of the buffer transistor. Select for about 1mA
Here are the test results
http://hifisonix.com/wordpress/wp-c...niversal-Small-Signal-Class-A-Buffer-V1.0.pdf
The design here is for a headphone buffer. Just scale down the output buffer currents for line level operation and run at c. 16dB stage gain.
BTW, jcx proposed a neat solution for the opamp class A bias. Bootstrap the output by removing the output current source load and simply tie a resistor from the opamp output to the emitter of the buffer transistor. Select for about 1mA
Last edited:
Congratulations for your "error correction" article, i just discovered, Bonsai.
I had yet experimented such a thing, with similar methods, and similar results. That is exactly the kind of things i consider the right track on modern high end audio, because it is realistic and industrial, because it push quality to the top with no magic or virgins inside. Just good understanding of the systems we are dealing with.
Of course, i'm on your side too about class AB.
In order to move on, let me state that the 2N5457-2N5460 'complementary' jfet pair is from 1969 or so. This is known because of the sequential designation of the 2N parts. For example, the 2N4403 became available in about 1968, and the 2N5680 was available in 1969. I know, because I used them in that time frame.
These are known as LONG GATE complementary parts, and have rather low Gm of 1000 umho or somewhat more.
So called SHORT GATE complementary parts became available in about 1972 and have a typical Gm of 5000 umho to 10,000. They can be fairly quiet, but not MC pre-preamp quiet. These parts have been continuously available for at least 40 years.
The third kind of complementary jfet is called the MESHED GATE, and this design originated in Japan. These have a Gm of 10,000-50,000 and are usually the best parts that are available for audio design. Unfortunately, most types have been discontinued and available parts are relatively expensive, today.
There is an another set of parts that are optimized for RF use, like the 2N3819, 2N4416, etc, that are often useful, but usually have NO complement, just N channel.
This is all there is to it.
These are known as LONG GATE complementary parts, and have rather low Gm of 1000 umho or somewhat more.
So called SHORT GATE complementary parts became available in about 1972 and have a typical Gm of 5000 umho to 10,000. They can be fairly quiet, but not MC pre-preamp quiet. These parts have been continuously available for at least 40 years.
The third kind of complementary jfet is called the MESHED GATE, and this design originated in Japan. These have a Gm of 10,000-50,000 and are usually the best parts that are available for audio design. Unfortunately, most types have been discontinued and available parts are relatively expensive, today.
There is an another set of parts that are optimized for RF use, like the 2N3819, 2N4416, etc, that are often useful, but usually have NO complement, just N channel.
This is all there is to it.
jFEts for line level stages -
Oh those. They are the 2N5457 and 2N5460. Run at 1mA approx. Good for line levels. -RNM
Oh those. They are the 2N5457 and 2N5460. Run at 1mA approx. Good for line levels. -RNM
John Linsley Hood buffer, maybe?
This is just simulation with diferent BJTs used, JLH used BC212/182.
dado
Attachments
No. At my house it is called the RNM Buffer/Amp. circuit. 🙂 My own derivation is in Linear Audio #3. Its an outgrowth of my earlier designs that were published later in The Audio Ameature. That is when I first started talking to john Curl about it. In those early days, I used all bjt in complimentary push-pull circuits and only jFEt in differential (2N5564). have you seen them? Later, Borbely did the same topolgy in all jFETs and also published in TAA. You might do a sim of the amplifier version as used for a headphone driver. The one I made has been measured to be at or lower than .0005% THD+N. I have MicroCap 10 for sim work, Audio-Precison, HP339A and now a ShibaSoku AD725D to test audio circuits with. I never sim'ed it.... just designed and built it and listened to my headphones with it. Later, I measured it and then did some other work on headphone auto-EQ that I folded into the Linear Audio article on a nice little, cheap headphone amp. Thx-RNMarsh
Last edited:
Thank you Esperado! I still need to work on the clipping behaviour - ill do AFEc in a practical design once I've got that sorted out. A job for 2013 after my new preamp is finished.
- Status
- Not open for further replies.