Please have a look at the excellent models (5532 and others) that DIYA member @Nazar_lv kindly provided:
https://www.diyaudio.com/community/threads/accurate-spice-macromodels-for-some-op-amps.337687/
He is an expert and has published a whitepaper on the topic "WP-1004 Designing an Accurate SPICE Macromodel for Ultra-Low Power, Rail-to-Rail CMOS Op Amps"
And precisely this point, which I agree with 100%, is the exciting one - and everything should revolve around it. Understanding and penetrating the past in order to learn from it. The learning vehicle is then also the finale, a new design, a new and modern current dumping amplifier.
🙂
I also agree with you here, with one decisive exception: If there were two independent circuit parts, i.e. two amplifiers that were also fully functional as standalone units, then this (simple and very quick) test would be appropriate.
With the Quad 405, however, this is not possible because the 47 Ohm resistor is not simply a feedforward branch, but an integral part of the last stage of the VAS. I had hoped to have worked this out in my long monologues. Obviously, I failed on this (all-explaining) point.
I'll try to put it in simple, clear terms:
With the 47Ohm resistor pulled out, one permanently change the last stage, “the class A amplifier” (portion), it no longer works as it should, its output resistance is no longer relatively low impedance, but high impedance, its idle current setting is also destroyed, its idle gain "Leerlaufverstärkung" drops ..!
At the very beginning of this discussion thread, jxdking already indirectly described this stage correctly, but in my opinion not concisely enough due to the perspective taken.
The idea of testing the quad in this way and thus proving the superior power of the CD is absurd, because this procedure is comparable to chopping off one leg of a sprinter's two legs in order to now feel confirmed that the sprinter with two legs simply runs the 100m faster ..!
HBt.
Attachments
I immediately recognize an inconsistency, namely the following:
it leads to the assumptions that
a) a quad amplifier does not need to pass quality control because its technology is overwhelming - that means quad does not need QM or QC
b) for all other competitors, the two points listed would be problematic, namely the long-term stability and the permanent inclusion of the self-heating of the semiconductors in the circuit design - in short "thermal tracking"
Neither of these points has ever really been a problem for any of our competitors.
I have two very old amplifiers (Revox A78, since 1978 and Dual CV121, since 1976) in operation to this day without any problems, no recalibration has (been) necessary.
Mr. Self shows DIY how to get the best out of a simple basic circuit by taking absolute care. No more and no less. His contributions certainly play a key role today. It was very good that Douglas got involved in the mix - it's very cleansing.
I agree with you.
😎
greetings,
HBt.
Let's take a look at the gap:
2PI * f * t = sin^-1 [u(t) / u_p]
t = ( arc sin[ 0.7V / Upeak] ) / (omega)
What we don't want is a dead zone, but at the same time we don't want a linear first order interpolation, i.e. no connecting straight line... we want a jump, an immediate increase to 0.7V (the classic potential separation in the electric field of the BJT's junction = diffusion voltage), i.e. in an infinitely short time the operational amplifier (or in the Quad 405 the SE-SRPP class A part) must raise its low-impedance (and this miracle of low-impedance is also only possible with nfb) output to +/-0.7V. An infinite DC gain can make this possible. The converse is the unconditional requirement for a large SR value.
Unfortunately, @EdGr was completely right with his objection.
HBt.
The problem with each jump is its components, which consist of sine and cosine shaped time courses of individual voltages. An infinite summation of individual high-frequency oscillations /swing and a DC component.
Each jump also stimulates the system - this is well known from control engineering and serves for simple system identification, namely h(t) ..!
So we actually want to avoid the need to jump.
HBt.
Each jump also stimulates the system - this is well known from control engineering and serves for simple system identification, namely h(t) ..!
So we actually want to avoid the need to jump.
HBt.
Now we want a jump because we need it so that no dead zone is created, but at the same time we want to avoid jumps at all costs. What now? The solution is a VBE bias voltage and this should swim with the signal as much as possible, follow it, usually it does this very well. However, this circuit trick cannot prevent the two complementary BJTs from being switched on and off to form a push-pull output stage. The switching distortions and the existing non-linearities remain.
Only now can the miracle of feedback and feedforward appear.
The Quad 405 seems to solve the jump problem in an ingenious way. In reality, it cannot be solved; solving it means avoiding any switching. We can filter the artifacts or modulation products or integrate them into the noise, but that's all. Of course we can mask them, but they are still there. Only SE Class A solves this one problem.
Without any difficulty we can use the methods of correlation (cross and auto) to read from the noise, the fog, all the necessary secrets in plain text ..!
😉
Only now can the miracle of feedback and feedforward appear.
The Quad 405 seems to solve the jump problem in an ingenious way. In reality, it cannot be solved; solving it means avoiding any switching. We can filter the artifacts or modulation products or integrate them into the noise, but that's all. Of course we can mask them, but they are still there. Only SE Class A solves this one problem.
Without any difficulty we can use the methods of correlation (cross and auto) to read from the noise, the fog, all the necessary secrets in plain text ..!
😉