Not sure the message ever gets through, one thing to look out for on bi-polars is the big base charge hit when entering/leaving quasi-sat, but yes gm upon C doesn't look so good when C varies 300%.
Scott
Isn't it a good idea to make sure that the active BJT (s) is always in the active region and thereby never enters or leaves the quasi sat region?
S
Yes, but that's device dependent and not usually specified. You just have to be careful and make your own measurements.
Here's my own measurement of a MPSA42, whose quasi-saturation took D. Self by surprise (APAD 6ed p.186). Ibe is about 90uA. That's a seriously quasi saturated transistor!
Attachments
Last edited:
Yes, this, IMO, is the basis for various design attempts. Sometimes the total thd is larger in some of these attempts but seems more cohesive and similar from low to high in sound character. --- in contrast with lowest distortion at low freq and higher distortion at higher freqs characteristic sound. Or, so it goes. Wide ol bw AND very low distortion is better.
-RNM
_____________________
"Ultimately, Physics only describes the part of reality that is susceptible to mathematics."
https://www.newscientist.com/articl...pid=ILC%7CNSNS%7C2016-GLOBAL-webpush-CULTURE1
Last edited:
Hi Mark,
But note that the BJT emitter followers have virtually no load on them, as they feed MOSFET gates. Nevertheless, the early effect of the predrivers can present a very slight load to the VAS.
Again, DC gain does not matter much anyway.
Note that MOSFETs have near-infinite input resistance, but finite input impedance as a result of their gate capacitances.
Not sure which MOSFETs John uses, but, depending on their size, their gate-drain capacitance may present a load at ac that can have some influence at higher frequencies. Gate-source capacitance will also have an influence, since driving the heavy load of the output BJTs may reduce the gain of the source followers below unity, reducing the bootstrapping effect of Cgs.
Cheers,
Bob
This is a reasonable question. Most amplifiers have falling loop gain within the audio band, so a given amount of static distortion will rise with frequency due to declining feedback. As a result, one can observe that the shape of the THD spectrum will change, so that higher order harmonics are more emphasized IN A RELATIVE SENSE.
Suppose you have 2 amplifiers each with 0.001% 5th harmonic at 10 kHz (2 kHz fundamental). One has flat open loop gain and the other has declining open-loop gain. They both have the same loop gain at 10kHz. Would you choose the one with higher 3rd harmonic at 5 kHz just so you could have a flatter distortion spectrum?
Remember, for a given amount of stability for a given circuit design, wider open-loop bandwidth sounds like a good thing, but it means LESS loop gain at lower frequencies.
Moreover, it often is not simple-minded harmonics and harmonic distortion that create the most corruption of sound quality. Rather, it is more often intermodulation distortion among the many frequencies that exist simultaneously together in music. Even a simple-minded view with two or three sinusoids results in intermod products at lower frequencies than the tone frequencies themselves. More loop gain at these lower frequencies will better reduce the amplitude of these lower-frequency intermod products.
If you hear "spit" on a cymbal with an amplifier with poor HF linearity, that "spit" is from intermodulation products at lower, often more audible, frequencies.
Remember, we don't listen to individual sinusoids.
Cheers,
Bob
It seems unlikely that he made a serious design error but we must concede it certainly is a possibility. He posted the full and complete schematic in his Blowtorch thread, you could look it up. As I recall the output stage was a double not a triple (but don't trust my foggy memory), suggesting that he felt the current gain of the MOSFET source follower was adequate. Final loudspeaker driver was five or six parallel Sanken 50 MHz BJT emitter followers running at ~100 mA each, if memory serves. Best to trace back to the original source and set aside my imperfect remembrances.
Yes, this is important. A good example is the common 2-transistor current mirror, where one transistor is diode-connected. It has 0V collector-base. While this is adequate with many devices in many situations, it can detract from the linearity of the mirror, since the other transistor usually has much larger Vcb in most regions of operation. Some transistors begin to gradually enter the quasi-saturation region at Vcb=0V. A 3-transistor current mirror with a "helper" transistor (EF buffer supplying the base currents of the mirror transistors) puts 1 Vbe of Vcb on that "diode connected" transistor.
Cheers,
Bob
I have already responded to this criticism in the thread where you first posted it, but you have neither replied in that thread nor included my response in the quote.
So I have included my response, only added "audibly" to clarify the quote when taken out of context from the first thread.
That thread was about an amplifier with a few PPM distortion at 20 kHz.
Well yes, exactly, that's precisely what the PPM distortion amplifier in the OP was all about.
My point was that it is simply unnecessary to push the OPCF down any further.
So very low corner frequencies may be undesirable, not because they make the amplifier "slower", as often erroneously claimed, but if they add extra complexity for minimal benefit.
Best wishes
David
Last edited:
Bear in mind that lowest distortion at low frequencies in no way implies higher distortion at high frequencies.
Cheers,
Bob
Hi David,
Indeed, amplifiers whose loop gain goes higher, unhindered, at low frequencies, are often simpler. High gain at low frequencies is often a natural result of those circuit techniques that increase stage linearity.
Cheers,
Bob
Yes indeed. If that is a natural consequence then let it happen.
I do not advocate flat Return Ratio across the audio spectrum as an objective, hence my dismay that JCX's quote of my comments, out of context, could mislead.
Self has a couple of examples of how to raise the frequency corner to make the amplifier look "faster" to the naive.
They are deliberately dumb, and funny, makes the point pretty well.
Best wishes
David
I never said nor implied it. Just saying what/why the design/ers re constant feedback vs freq is about. compared to designs with low dist at low freq and higher dist at high freq's characteristic sound of non-constant fb amount.
THx-RNMarsh
Last edited:
the mosfet follower c may not be as large is thought in this topology etal. [Bootstrapped?]
The high peak drive current required for all the BJT OPS devices is easily handled with mosfet driver.
THx-RNMarsh
Last edited:
That was my question, but if distortion is low enough what would you choose?
Intermodulation distortion produced at high frequencies will not be reduced with more loop gain at low frequencies, will it?
I think it does. It is simply something that is not in the input signal, so feedback will reduce it, the more so, the more loop gain is available at that low frequency.
Reading Bob's very good post above I believe we should more emphasize the falling of distortion at lower frequencies, rather than thinking the high frequency distortion rises, which it doesn't.
Jan
Thinking about it some more. Suppose I have an amp with the following, not uncommon, distortion profile:
10kHz -80dB
1kHz -100dB
100Hz -120dB
What the flat distortion proponents are basically saying is that if you raise the 100Hz distortion a 100 times, and raise the 1kHz distortion 10 times, your amp will sound better.
I really have a hard time to believe it and/or think of any reason why this would be so. Does anyone know of something that supports this other than personal opinion?
Jan
10kHz -80dB
1kHz -100dB
100Hz -120dB
What the flat distortion proponents are basically saying is that if you raise the 100Hz distortion a 100 times, and raise the 1kHz distortion 10 times, your amp will sound better.
I really have a hard time to believe it and/or think of any reason why this would be so. Does anyone know of something that supports this other than personal opinion?
Jan