Because it is not so obvious. Because, most of the time, "linearizing" request adding stages, I.E. adding poles and the problem they add in closed loops.
Bill, bear in mind that what Richard was talking about will also depend on your digital source. If that's good, likely no great change will happen.
I didn't say it isn't being done. Hwever, when I compare a say input diff. stage being run at say <1 mA, collector resistors say 2K and emitter degeneration resistor values od say 10 Ohms or so, that's not inspiring. Whne the same input diff. stage is run at say 2 mA nd a total gain of 5, 6 or 7:1, that I think is a completely different affair.
More or less the same for the VAS stage, some run it at say 4mA and ohers run it at 20...30 mA; which one do you think will demonstrate a wider bandwidth? Otala and Lohstrioh limited the VAS gain at 11:1, at a current of 20 mA, with lead compnesation, that's how they got such a wide open loop bandwidth.
Not sure what you are referring to, but you are entitled to your OPINION. There is no technical basis for this opinion. Please be aware that I like to listen to 50's/60's performances on LP and don't really care about this stuff. Modern man seems to be a devolution artistically.
DVV, you have a valid point. We found that departure from OP AMP considerations to make faster slew rate, AND higher open loop bandwidth, at the cost of lower open loop gain, made better sounding amplifiers. I hold that to be valid, even today. Mostly the answer is now beyond TIM, but in removing other subtle distortions.
Last edited:
Scott, it is my OPINION too. About technical basis, well, don't it seem obvious, in a servoed system that, faster it is closer it is to the original it has to follow or compensate ? Think, as an example to antishake systems in cameras.
of course, it matter the system makes not too much errors (Distortion) it is a question of optimal balance between the two.
This is the worse "feedback chasing its tail" type of explanation I've seen lately.
- it is very obvious for anybody with a minimal EE education.
- linearization doesn't necessary require additional stages.
- there's nothing wrong in adding poles (if you know what you are doing)
- poles don't add problems to the closed loop (if you know what you are doing, it's called "compensation")
What's left in the quoted message is entirely correct.
Would you please define those "other subtle distortions"? And a measured example would be nice.
I think I very clearly stated that I have show a preference, meaning that in general I will prefer a wide bandwidth, low GNFB. This is over the last say 30 years or so. Again, I said "mostly".I'll skip posting a list. I might add only that I know for a fact there are others of a similar opinion.
As for your closing sentence, sadly, I have to agree with you. My favorite LP is an RCA release from the early 60ies, in mono, with Harry Belafonte singing gospek songs in a chuch accompanied by 64 male and female singers. Also some Decca Phase 4 recordings from the late 60ies.
Sooner or later, new evidence will surface which might clear this up. Right now, if 8 out of 10 amps I enjoy are of the widebandwidth and low GNFB type I can't help noticing such a fact. Much the same with swapping standard op amps with very fast ones - that usually brings about an improvement of the sound, not often a revelation, but a definite improvement.
The problem with IC's is not just speed, but limited Class A output, and no matched complementary jfet topologies available. I have found all IC's a compromise, but just like you, I have found the higher open loop bandwidth and lower feedback ones best.
For PMA, look seriously into PIM, for the next distortion mechanism. After that, who knows for sure? Ron Quan has done several papers for the AES on this distortion.
For PMA, look seriously into PIM, for the next distortion mechanism. After that, who knows for sure? Ron Quan has done several papers for the AES on this distortion.
Given the apparent absence of bias-controlled listening tests to support this perceived need after the same 30+ years, it looks like improved sound quality may have nothing to do with your preference.
Its no coincidence that the AES papers relating to improved listening test technology (ABX) came out after several years of seemingly incredible claims about the bandwidth required for acceptable sound quality. And its no coincidence that after that, the claims pretty well disappeared from their publications.
There are plenty of other non-technical explanations for personal preference in the absence of reliable listening tests, such as appeal to authority, fascination with larger numbers, style, peer pressure, etc.
If you are insistent that an IC Opamp should deliver more current in class A to be useful then that is an exceedingly easy problem to solve - see my 'Symphony' preamp. However, note that into 600 ohms, modern devices deliver -120 dBV didtortion performance for about $10 (AD797). An LM4562 is about $3. And, you get superb PSRR to boot.
It cannot be correct that only complementary JFET input stages work best. Those discrete designs, despite you earnest selection efforts, are no where near as linear as an integrated IC JFET long tail pair. And the JFET input stage is not as linear as a degenerated bipolar stage, which also happens to to tolerate RF ingress just as well.
- Status
- Not open for further replies.