Class i and siblings

[ontoaba]

D2S? pioneer d2s?

link attachment:
Bob Cordell Interview: Negative Feedback
Square Law Class A Amps

[Edmond Stuart]

The links are correct. A shame I forgot to mention Ian Hegglun, as he is the one who elaborated on the D2S concept. See:
"Square law rules in Audio Power", EW+WW, Sept. 1995, pp. 751- 756.

Cheers,
E.

[Edmond Stuart]

Originally Posted by homemodder

Edmond
...........
Nice results, could you direct me to Ken Peters paper, I must have missed it.
...........

Hi Homemodder,

I don't know of any paper, but here you can see the schematic.
His design is a bit different. First, no local feedback around the OPS. Second, the bias control approaches the geometric mean (or hyperbolic) relationship between Ip and In in a different way. Instead of turning off the sense tranny at higher currents, in his design they are turned on by the logarithm of Ip respectively In (by means of Schottky diodes). In this way, in principle, the hyperbolic relationship is more accurate approximated. However, its difficult to match Vd of Schottky diodes with Vbe of transistors. Please look here for a more stable version.

Cheers,
E.

[Edmond Stuart]

By "First, no local feedback around the OPS" I mean no feedback WRT the audio signal (no differential feedback loop). WRT to the bias control, of course there is local feedback around the OPS (being a common mode control loop, CMCL).

[jan.didden]

Originally Posted by Edmond Stuart

The links are correct. A shame I forgot to mention Ian Hegglun, as he is the one who elaborated on the D2S concept. See:
"Square law rules in Audio Power", EW+WW, Sept. 1995, pp. 751- 756.

Cheers,
E.

Edmond,

Ian Hegglun elaborates about square law class A designs in his Linear Audio Vol 1 article: 'Square-law class-A: a family of high-efficiency class-A power amplifiers'.

This article treats issues beyond the design and the transfer function itself like efficiency calculations for square-law and curved-law, comparing favourably with for instance the Pass ZV5 and the well-known JLH class-A amp. Also treats issues like heatsink and power transformer requirements and the impact of using different output MOSFET devices.

jan didden

[Edmond Stuart]

Originally Posted by janneman

Edmond,
[snip]
This article treats issues beyond the design and the transfer function itself like efficiency calculations for square-law and curved-law, comparing favourably with for instance the Pass ZV5 and the well-known JLH class-A amp.
[snip]
jan didden

Hi Jan,

Did Ian Hegglun also compare D2S to AB2? If so, I'm eager to learn more (and buy a copy of Vol.1 ).

Cheers,
E.

[ontoaba]

There is still huge crossover distortion from that square law application (hegglun sch) even for 200mA bias. The most effective is Candy's style that added drop voltages to input plus remaining drop voltages to input again.

Is there any explanation why square law not perform so good in linear audio vol1?

About Kenpeter, he using current based negative feedback similar with this: http://www.diyaudio.com/forums/attac...-se-classa.png
http://www.diyaudio.com/forums/attac...w-tripleef.png
but with schottky nonlinearity.

[jan.didden]

Originally Posted by Edmond Stuart

Hi Jan,

Did Ian Hegglun also compare D2S to AB2? If so, I'm eager to learn more (and buy a copy of Vol.1 ).

Cheers,
E.

Ian's starting point is a class A amp but with much better efficiency. It's a real class A amp; if you look at the (measured) wingspread diagrams you'll see no ripples or discontinuities at the zero crossing. The distortion as such is not the lowest ever, but due to the transfer it is very benign. This is open loop, and he has a section on adding gain and feedback to make the distortion lower but that is not a prime consideration.
There an extensive section on the use of various types of MOSFETs and the impact on efficiency and performance.

In Appendix A he gives extensive info on efficiency comparisons for music signals between his square-law class-A, conventional class-A and class-B, with square-law class-A falling somewhere between the other two.
Of course, long-term efficiency depends on the type of music you play; for classical music linear class-A has about 3%, square-law class-A 8% and class-B 25%, while for sine waves the figures are 50%, 67% and 78%.
Another interesting tabel gives the ratio of power supply transformer VA to amp output power; for conventional class-A this is 6 (!) and for square-law class-A it is 3.2. So a 100W conventional class-A needs a 600VA xformer, while Ian's 100W version needs only a 320VA one.

Then Appendix B goes into the distortion analysis for the various topologies. This includes damping factor/ Zout as well.

Total article runs 36 pages.

Other articles you might be interested in are Ed Simon's article on resistor non-linearity and Scott Wurcer's first part of his low noise mic preamp. And several more - abstracts can be read on the website.

jan

[ontoaba]

Originally Posted by janneman

Ian's starting point is a class A amp but with much better efficiency. It's a real class A amp.
....
In Appendix A he gives extensive info on efficiency comparisons for music signals between his square-law class-A, conventional class-A and class-B, with square-law class-A falling somewhere between the other two.
....

jan

Hi, Jan,
I could make AB2+XM or my original XM to not doing switch at all for all freq up to 2ohm load working at smooth current curves with biasing the storage, the result must be much better efficiency compared to that square law, because dynamic biasing depend on freq and load, but it is not necessary I think. Sound quality is one most important and 0.005% THD is enough to get its clarity.

Is there any listening comparison between square-law and original classA?

[ontoaba]

This is not so good performance. Just 8ohm and 4 ohm.
I am using Pawel's asc thanks for his drawing: Square Law Class A Amps