The LM1875 for best

[kokoriantz]

This is derived from class aa buffer of Technics, see bellow.
The left amp is the master, the slave tries to decrease the master's current to null as power steering. by un balancing the bridge with 200 instead of 100 ohms the slave provides only halve of the output current.

In this version two slave power amps are under the command of an opamp.
It requires precise adjustment to get the opamp steer the output with very little current. This amp can have less than 0.0003% distortion.

[kokoriantz]

I will start with this model as it doesn't need fine adjustments to function correctly. The central 100k resistors bias the amps 210ma, to be adjusted for linear dissipation. It can deliver 70 good watts, but with distortion of about 0.02%.
To correct the non linearity, I will not use composite but Hawksford error corrector. Mine, shown bellow called the Reactor, Is different type but works on the same principle. I need to adapt for differential feedback, and will act only on one side of the bridge .

[kokoriantz]

This is a trial differential reactor. As you notice the nonbiased bridge gets correction from a single side.

I need to math now the values for perfect correction.

[kokoriantz]

This is first trial with Hec , To stabilize I added temporarily C1. The saturation is an issue to review.
If you like this circuit, give a star.

[kokoriantz]

This is a different way of bridge. The right hand side is a unity inverter. The error circuit when adjusted to required gain by R26 goes to minimum and adds or subtracts the distortion or phase as shown bellow. The error opamp needs to be jfet as it functions also as dc servo. I used here OPA627 but AD8610/20 can replace. The AD8008, the ultra low distortion/noise 450Mhz CFD opamp is far superior, can be used only with single supply.
As I see this thread doesn't attract much interest.
Hayk

[kokoriantz]

Simpler paralleled but no bias, yet.

[mjona]

Originally Posted by kokoriantz

This is a different way of bridge. The right hand side is a unity inverter. The error circuit when adjusted to required gain by R26 goes to minimum and adds or subtracts the distortion or phase as shown bellow. The error opamp needs to be jfet as it functions also as dc servo. I used here OPA627 but AD8610/20 can replace. The AD8008, the ultra low distortion/noise 450Mhz CFD opamp is far superior, can be used only with single supply.
As I see this thread doesn't attract much interest.
Hayk

You have posted so many variations of a theme it is hard to know if these are simulations or circuits where you have done some breadboard testing.

While the LM1875 is relatively cheap the other op.amps are not, which is probably enough to put people off.

This device has a minimum closed loop gain of 10 which is for an 8R impedance. What level of driven impedances are you looking at that would change that situation.

The composite amplifier mentioned by adason that you dismissed was authored by a group that included Scott Wurcer who is responsible for some high spec IC designs.

Here is a link https://worldradiohistory.com/Archiv...cs-1992-11.pdf from page 36 onwards.

[kokoriantz]

Hi Mjona. I am glade that you give interest for this subject. I am continuing what FauxFrench has started on LM1875 in parallel configuration and used in a composite amplifier..
FF and chermann did lot of tests and measurments that I am relying upon, I proposed also several solutions and composites. For now I am away from home and I don't think I'll be back soon, so all these circuits are research to get uncompromised best from LM1875 and it is far to be reached. For now, thanks to this project, I created two new circuits, the Hec differential and floating interactive current amplifier. There are still lot of problems to resolve, here is an example.
I asked Sangram on post 16 to listen a music comparatively to LM3886 which can qualify to my standard as entry level HIFI. It remained a dead silence response, so I presume that it is at least as bad if not worst. To continue, I need first to resolve the open loop phase shift problem.
PS: AD8610 costs 1$ from Ali.

[mjona]

Sangram might have discovered as I did that LM1875 modules can cut in and out at higher power levels. I ran these singles from +/- 22 Volt regulated supplies I had on hand for my JLH existing 1996 power amplifier - which did not misbehave in this way.

I had some correspondence with Sakura Systems' proprietor in 1999 but did not ask him which chip he was using. The thought had been that the chip modules might be as good as the Class A and not run so hot and consume so much power.

I believe one high end manufacturer has made composite power amplifiers from parallel LM modules whether 3875 or 3886.

I have acquired a few commercial amplifiers in my time. One of these is a Technics Class
AA design - model SU-VX720. I have a service manual for this.

[mjona]

As a matter of interest I have tried stacking LM833N chips 3 high as an experiment. There may be something in odd numbers - things did not sound so good 4 high.