The two things which make me prefer Bob's book are: 1) I feel he kept a more open mind than D. Self did, and 2) I feel Bob managed to stay well away from hard personal feelings and was thus more informative and educational. Just my feelings.
Which, of course, by no means eliminates D. Self as a competent designer, just more settled in his own ways which he proves to be perfectly functional. And that's fine, if you don't mind replicating the Lin topology no end. I have heard some excellent examples of D. Self-type amplifiers in my time, the last memorable two examples were both from Germany, a LAS Mega 1 power amp and Grundigs truly outstanding V5000 integrated amp, but that was done by 1984. LAS (Linear Audio Systems no longer exists, and Grundig is owned lock, stock and barrel by Philips, and is no longer a player by itself, it now makes rebadged Philips gear.
BTW, that LAS Mega 1 was my first encounter with a truly high current capability MONSTER, it could pump out in bursts as much as 80 Amperes on a nominal output of 100/160W into 8/4 Ohms. It was called "Mega 1" because its fequency response at 1W/4 Ohms hit 1 MHz. Very much VFB.
Thermal cancellation is cool , huh ?
I like to take the input stage and put it in the freezer (pitchfork villager style) ,
and see 5.3ma at VAS. Then , "blast the hell" out of the amp for an
hour and still see 5.3x ma. Really.
Must get ready for hot weather here - it's a comin'.
After that hour of abuse , AB output stage 16mv > 15.7mv bias. A happy
amp.
Real "live" testing of the goods is "one up" on simulation .
OS
Hi OS,
-Chris
That would be my feeling as well. Stability - what a concept! I am okay with a slight negative temperature co-efficient. Marantz made product like this in the 70's.
At the very least! What happened beats what might happen any day. The difference between the two is called confusion.
-Chris
ISBN 1461451876. It looks like a good book, based on the table of contents.
The interesting thing about so-called current feedback is how far it goes back, as it is really nothing more than series feedback, for example the ancient topology of tube power amps with feedback from an output transformer tap to an input tube's unbypassed or partially-bypassed cathode. When Nelson "invented" current-feedback op amps, the topology took advantage of complementary bipolar semiconductors which allow a symmetrical arrangement, and of course the use of very fast parts at that. The patent is old, but I don't think too many worried about using the approach even when it was in force.
Erik Edvardsen used a current-feedback topology in the power amp section of the old and famous NAD 3020 integrated amplifier. It is single-ended, with a single input transistor and temperature compensation via a diode. The feedback divider resistor is a.c.-coupled to common so there is reasonably low output offset. Within its limitations it's still a pretty decent low-power amplifier.
Thermals are one of the last frontiers, although many approaches have existed going way back, since scope vertical amplifiers needed maximum bandwidth and could not use straightforward negative feedback. The "discovery" of "memory distortion" mostly recapitulates the work of scope designers. In the defense of recent practitioners, the scope stuff was hard to find for a long time, as it was mostly proprietary. A classic circuit is a compensated differential pair, with emitter and collector resistors scaled so that each half swings in dissipation with the signal but in such a way that each side increases or decreases equally, one with more current and less voltage, and the other vice versa. A good coverage of the topic can be found in Feucht's book Handbook of Analog Circuit Design.
Damir, don`t bother, ignore him. DVV designed one headphone amp in his whole life and not one power amp, but he`s known as "I know it all" person. There is no one photo of the power amp he designed, so ...
... let him talk and wright 50 row posts about same things over and over again... there is no help... 
Last edited:
Yes , as well as scopes ... Moog and Pearlman (ARP) needed tight thermals
in their electronic instruments. 1V/oct scaling would be thrown to the wind
without the oscillators and filters being thermally stable.
Some of Moog's filters are "works of art" , both sonically and thermally.
OS
Considering the cost constraints and the parts that were available, I agree about Bob Moog's achievements. Getting unijunction transistors to work well enough as sawtooth oscillators with the current source appropriately dependent on control voltage was a stunt. It wasn't perfect, and in the UCLA Electronic Music Studio when I was studio engineer, briefly circa 1969, no one had figured out how to cool the place by calling the right number and knowing what to ask maintenance to turn on, so temperature drift was a problem. I think there was a thermostat but many times the physical plant would just disable whole sections of the HVAC for that section of the building where the little hole-in-the-wall of the studio was. James Cooper, who replaced me when I quit to go full-time for Astronomy, quickly made inquiries and knew whom to call.
Another Moog trick I liked was using a bipolar in an intentionally overdriven stage to polarity-transition between inverting and noninverting, which he used to turn sawtooth waveforms into triangular ones, prior to the sine wave shaper network.
Those synthesizers were made possible in many ways by the advent of plastic-packaged bipolars, the 2N3391 for example, which shared the last four digits of my UCLA Astronomy phone number for many years.
Ummm..... well that is partially not true.... there are definite criteria to be called a CFA and many of those mentioned do not qualify. But, it does go back aways and it is a shame more people here do not understand them..... Like I said most all audio companies change to CFA topology and operation for their best work and top of the line models. Go over to the (SS) and find out.
THx-RNMarsh
(SS)?
Ah, Solid-State.
One clarification to the statement about series feedback: the feedback point is a much lower impedance, as an inverting input, compared to the noninverting input. This is satisfied by the emitter of a bipolar versus its base. It is less-well-satisfied by a vacuum tube but still operant.
Based on things that people call current-mode, say in the wikipedia article, or for example the first schematic in the first entry in the "Slewmaster - CFA vs. VFA rumble" thread, the belief might be acquired that the arrangements have to be symmetrical, like the Nelson patent. But this is not an essential characteristic. Looking at most CFOAs one could also suppose that the needed inversion(s) must be supplied by various species of current mirrors---but some designers put a little gain there as well, despite this complicating stability. ST made an integrated amplifier that did this, and it was the devil to compensate for anything other than low loop gain (that is, high closed-loop gain). But it was possible and the part delivered some very decent performance.
Last edited:
Also, in an extreme simplification, the simplest complementary feedback pair beloved of Self is a sort of current feedback topology: input on the base, feedback to the input device emitter, the complementary transistor supplying a bunch of current gain. Load the output collector with a current sink and buffer with a single e-follower, and you've got some pretty respectable performance.
So would be a Baxandall (super) pair , by that metric.
BTW , member Vzaichenko has the non-symmetric CFA with just a
"singleton " Jfet at input.
Link - http://www.diyaudio.com/forums/solid-state/266418-compact-vfa.html#post4215557
Amazing for a 5 device input stage - built and tested.
OS
Actually---not quite. The Baxandall superpair (historically, invented about ten years before the Baxandall-Shallow article, by Frank S. Boxall) recycles the base current of the main device by pumping most of it back into the emitter. At Tektronix it was known as the super-alpha stage and believed to have been invented about the time of the Baxandall article by Larson, according to John Addis.
Note that the Hawksford cascode is another form of control-electrode recycling, in this case also predated by one from Aldridge published in 1962. Both Boxall and Aldridge were telephony engineers, which explains why so many missed their publications. I ran across them in an old TI publication, "Communications Handbook Part II" from 1965.
Walt Jung has a good contribution somewhere in here that establishes the chronology, including some things with JFETs and JFET-bipolar combinations by Csanky and Jaeger.
That is nice. Yes, a CFP input with feedback to the single JFET source, and an Aldridge (aka Hawksford) cascode as a feedback integrator transconductance stage. Note that the effectiveness of the base current recycling depends on the emitter resistor R26 value being large compared to the emitter impedance at Q6, as the current pumped back from Q5 splits proportionally at that point.
Not sure what the green LEDs in the drain of the JFET are doing.
Brad
That's the odd part though---the impedance at the SK117 drain should be fairly high, especially with the source feedback, making the effectiveness as a voltage drop with a negative tempco small. Perhaps the designer commented somewhere.
- Status
- Not open for further replies.