The 5534 circuit is freely available though there are some which show erroneous PNP i/ps.
I was hoping for some idea how you replaced the input stage with FETs using the compensation pins on the hybrids you sold to Wilson Audio.
What performance advantages did you establish all those years ago for this mod?
Check the THS4601, you may find a better trade. Also JFET input, but higher open loop gain, lower open loop bandwidth (you can't escape the GBW trade, pick your poison, but still well over 1kHz, ). Also lower SR to 100V/uS.
THS4631 is indeed hard to to manage, stability wise. Requires outstanding layout and decoupling. Also, doesn't take very well capacitive loads. But then it offers objective performance not available in any possible discrete implementation.
BTW, if one of the resident audio experts could specify the requirements (other than "sounds good" or any other similar lyrics) for a "high end" gain stage, then I am sure there are dozens of op amps today, meeting and exceeding those requirements.
It was printed in the Siliconix databook.
3Pins to short, to de-active input stage. 4 JFET: Two as a differential, cascoded by two others.
If you bias the output stage I to class A and buffer it, all of these issues will go away. Try it and use some good devices - get your tech to wire up a few and AB test them.
Impossible, as each time you have to consider the whole system.
Source impedance (including capacitance), load impedance (including capacitance), signals characteristics (presence of HF or fast transients etc.), required gain, inverting/non inverting. Each configuration will result in a different choice.
You often do not have a choice.
Last edited:
SR > 100V/us
noise < 1nV/SQR(Hz)
FET input
all class a
output stage bias > 20 mA
Which OpAmp would I choose for that ?
(The JC-1 line stage designed by John 40 years ago would meet
these specs, BTW)
Not good enough, that's not a spec, it's a wish list. You don't specify how the device is build internally, but numbers about what you want to get out of it, under certain input conditions. Otherwise, you may as well specify the sex of the electrons and holes.
What exactly means "all class A"? And why output stage bias > 20 mA and not > 5mA or > 50mA? And noise <1nV/rtHz for a JFET amp, are you planning to use such for a MC input, or as a line amp as discussed here?
I see this going nowhere, though.
Last edited:
I never got "rid of" the A21, in fact my friend bought one to (replaced his Krell which he damaged) drive his 3.6 , the A21 was not able to drive the Big Bembeh's 1ohm load,(many have failed , i can forward a list of the ones that can ) intrusive protection circuit in my situation...
I sense an agenda ......!
Why ...?
Last edited:
Yeah something like that
Hmmm... tell me how do you do this, or in which kind of audio chain (the circuit) can you insert a preamp without changing the sound?
Yes, I have heard this argument very often. But I sensed something wrong with the "logic" but could not really explain.
I guess ESP nulls the OP-amp sound by applying it inversely proportional again in playback after making the recording...
Straight wire no sound , like MIT cable ...
Oh I remember now how Christophe did this. Very good idea. No listening, only comparing signal. But I believe there will be difference sound-wise
But what else can be done. Cool...certainly pro mixing boards contained 100's of op amps - but I suspect many signals went through many fewer op amps in series depending on the amount of chained processing boxes required for that feed - maybe a dozen for some feeds?
with ADC at the mic, everything else in DAW I guess digital audio must really minimize "op amp sound"
with ADC at the mic, everything else in DAW I guess digital audio must really minimize "op amp sound"
Last edited:
Some of my personal tips:
Before to test an OPA, several things are required. Proper decoupling right at the power pins is VERY important, and the decoupling cap technology can have a sonic impact.
If we use several OPAs, it is important their respective PSUs to be well isolated from each other.
It can be a good idea, to isolate the problem, when evaluating, to power an OPA with batteries.
Because the output pin is also the feedback source it is a requisite to protect-it with a serial resistance from any load influence, specially RFI/EMI.
Try to avoid as much as possible parasitic cap at the - input pin when you do the lay-out.
It is always a good practice to ensure there is overshoot on square waves with a adjustable cap in // with the feedback resistance.
With current feedback OPAs, use the recommended value for the feedback resistance.
If you want to use a current feedback as a low pass filter (paralleling a cap to the feedback resistance), you can do-it with a miniature ceramic coil in serial with the - input to avoid the bandwitch peak (an instability) at the closed loop transition frequency.
When you have the choice, prefer an inverting solution. If not prefer a current feedback OPA.
Before to test an OPA, several things are required. Proper decoupling right at the power pins is VERY important, and the decoupling cap technology can have a sonic impact.
If we use several OPAs, it is important their respective PSUs to be well isolated from each other.
It can be a good idea, to isolate the problem, when evaluating, to power an OPA with batteries.
Because the output pin is also the feedback source it is a requisite to protect-it with a serial resistance from any load influence, specially RFI/EMI.
Try to avoid as much as possible parasitic cap at the - input pin when you do the lay-out.
It is always a good practice to ensure there is overshoot on square waves with a adjustable cap in // with the feedback resistance.
With current feedback OPAs, use the recommended value for the feedback resistance.
If you want to use a current feedback as a low pass filter (paralleling a cap to the feedback resistance), you can do-it with a miniature ceramic coil in serial with the - input to avoid the bandwitch peak (an instability) at the closed loop transition frequency.
When you have the choice, prefer an inverting solution. If not prefer a current feedback OPA.
A good place at home can be in a line input of your best preamp. Feeding both an other line input and the input of your test circuit. If you need to evaluate a circuit with gain, build an attenuator before the gain stage to be tested. Of course, you take care of correct impedance for both side of your attenuator.
You can too feed your test circuit with a good signal and switch between input and output to feed your actual preamp.
You just have to change the source in your preamp to test.
You can feed your amp directly from our test circuit if you want to avoid a preamp in the path: add an attenuator at the output of your test circuit to set accurate listening level.
Of course, better is your system, more the differences will be obvious.
With a mixing desk, it is so easy
About NE5534, i have found a definition witch reflect exactly what i feel:
"I intensely dislike the way it makes all recordings sound vague, distant and boring, lifeless, including those that are supposed to sound not like that."
"Don't overlook anything" as this app. note explains
Thanks Scott
http://www.analog.com/static/import...018494695982855668424783486554001060AN348.pdf
George
Thanks Scott
http://www.analog.com/static/import...018494695982855668424783486554001060AN348.pdf
George
Wayne and PMA, thanks for your inputs on the A21. I have one on my test bench right now, AND I am going to give it to my business partner, because it will NOT drive my WATT's properly, which are 4 ohms and below in drive Z. With 4 ohm and below, the limited number of output devices causes a small, but measurable amount of higher order distortion, which puts me off for MY set-up. My colleague is more relaxed, even though he also has a Vendetta equipped CTC Blowtorch that we made together.
Gk7, you got the essence right, but the numbers slightly wrong.
The Levinson JC-1 was a bipolar based pre-preamp with a rated input noise of 0.4nV/rt Hz, much less than 1nV/rt Hz
The Levinson JC-2 had an input noise of about 1.5nV/rt Hz in its more conventional phono input stage, and a fairly high slew rate, mostly limited by the necessity to drive fairly large RIAA caps with only a 60 ohm load. The line stage, however, was the complementary differential jfet input design with >20ma output idle current with 100V/us slew rate. It was not as quiet THEN, but it could be as quiet as 1nV/rt Hz today with lower noise complementary jfets available This later evolved into the JC-3 power amp with 100V/us slew rate. This was all done between 1972 and 1974, yes about 40 years ago.
Overdesign at its best, I would say. Now EVERYBODY does it. '-)
The Levinson JC-1 was a bipolar based pre-preamp with a rated input noise of 0.4nV/rt Hz, much less than 1nV/rt Hz
The Levinson JC-2 had an input noise of about 1.5nV/rt Hz in its more conventional phono input stage, and a fairly high slew rate, mostly limited by the necessity to drive fairly large RIAA caps with only a 60 ohm load. The line stage, however, was the complementary differential jfet input design with >20ma output idle current with 100V/us slew rate. It was not as quiet THEN, but it could be as quiet as 1nV/rt Hz today with lower noise complementary jfets available This later evolved into the JC-3 power amp with 100V/us slew rate. This was all done between 1972 and 1974, yes about 40 years ago.
Overdesign at its best, I would say. Now EVERYBODY does it. '-)
- Status
- Not open for further replies.