andy_c said:
I bumped into this thing while reading the Wikipedia "audiophiles" entry. I'm still speechless after reading it myself.
Is this where "feedback goes 'round and 'round" came from? The veritable sonic Droste, indeed. Martin IIRC did some interesting stuff when he mounted miniature accelerometers on tone arms.
andy_c said:
I bumped into this thing while reading the Wikipedia "audiophiles" entry. I'm still speechless after reading it myself
That’s pretty amusing. I was reading through my 1959 HiFi Annual and Mullard application handbooks on the weekend for ideas for my current valve amp project:
http://www.diyaudio.com/forums/showthread.php?postid=1423788#post1423788
In the former, THD, IMD and their reduction in a test amplifier via both the use of greater global negative feedback and the linearization of individual stages within the loop is discussed at length. At the time, <1% THD and <1% IMD was a generally accepted performance goal for any amplifier intended for high quality reproduction, but figures close to 0.1% were achieved with the more advanced feedback techniques. An audible improvement in sound quality was reported in an A-B comparative test.
This isn’t a one-off article, but common of the era, and the amplifier circuits in the Mullard handbook generally follow the low distortion, high global nfb design techniques developed towards the end of the 50’s - specifically developed to address the sonic shortcomings of amplifiers that were quantifiable and real.
Back then there was a degree of scientific procedure applied to amplifier design, and a zero nfb single ended triode amplifier with 5% THD and 10% IMD (which is one of the alleged paths to “HiFi Heaven” today, according to many esteemed journals
) would have been considered, WRT HiFi reproduction, the same as the cat’s whisker WRT radio.
G.Kleinschmidt said:...Back then there was a degree of scientific procedure applied to amplifier design, and a zero nfb single ended triode amplifier with 5% THD and 10% IMD (which is one of the alleged paths to “HiFi Heaven” today, according to many esteemed journals
Also, the mainstream audio press has really gone downhill since the '70s. Back then, people like Richard Heyser, Bob, and Dr. Leach were doing articles or reviews for Audio.
The first time I was exposed to discussion of feedback and its possible pitfalls as a lay person was by an article in Audio by Dr. Leach. Even though it made use of some of Otala's results that have since been refuted, the caliber of that article was worlds ahead of anything Stereophile has ever published.
john curl said:
The specific thing I'm referring to is his assertion that a wide open-loop bandwidth is a necessary condition for minimizing TIM distortion. I'm sure in our previous conversation in the "What's your reasoning..." thread years back that you've agreed that his solution with wide open-loop bandwidth was sufficient, but not necessary, for minimizing TIM. I also remember that you did not concede this point with respect to PIM. I understand your argument for that, and have even provided a rough analysis of that here. That page is a result of our previous conversation and my conceding your point in that regard.
I just want to say: Four of us, Matti Otala, Walt Jung, Marshal Leach, and I spent about 10 years researching TIM, IIM, etc BEFORE Bob Cordell wrote anything about it. We rebutted him very strongly, back in 1980, BUT he keeps inferring that he has overruled us in some way. No way, that I can see.
john curl said:
I don't wish to make this a "John vs. Bob" personal issue. I just wanted to mention the argument that attempts to connect open-loop bandwidth with TIM. Marshall got rid of the VAS collector resistor in his amp in the late '70s and explained to his class at the time (with me in attendance) why he did so. That was his realization that it was the gain-bandwidth product, not the open-loop bandwidth, that was important for TIM. I don't know who convinced him of this, but someone did. Or maybe he convinced himself. At any rate, he is an open-minded guy.
I AGREE with you that TIM can be OK with low open loop bandwith. Matti Otala made an error in not accepting that in the '70's. This was more 'politics' than science, BUT everything else seems to be in order, AND PIM is directly correlated to open loop bandwidth, and is still so, no matter what Bob has measured.
john curl said:
This remains a bone of contention I think. The analysis I did on my linked page assumes that the PIM (AKA AM-to-PM conversion) of the amp without feedback is zero.
But there's a fly in the ointment here. Think about an output stage near clipping - especially one with many output devices in parallel. Its input capacitance will increase a lot as the output voltage gets close to the rail. So the bandwidth, and therefore the phase shift, from input to output will vary with signal level. As the output signal gets close to the rail, the phase lag of the output stage/driver combo will increase. So this means the open-loop amplifier has PIM, even if it has no feedback. The feedback doesn't "know" whether distortion is AM-to-AM or AM-to-PM or come combination thereof. It just reduces it.
So in one case, the feedback causes PIM that didn't exist to begin with (for an open-loop amp with no PIM by itself). On the other hand, the feedback will decrease the inherent PIM in the open-loop amp. So which wins? I think Bob's measurements reveal that.
andy_c said:
The problem with such articles is that they try to draw a technical conclusions from a commercial reality.
If I had to chose a commercial amp, I would prefer a tube or no/low feedback design, just to be sure the output bias is high enough and it sounds OK.
As a hobbyist I use low feedback, high feedback, global error correction and so on always with high bias...
Regards,
Adam