Thinking Inside the Box

[Superstrand]

I have a provisional patent that I am willing to donate to Diy. In cursory explanation certain amplifier classes excel at different nuances, my little idea is to contain in one chassis a class a, a/b, d, tripath(TM) and a class g amp, with a circuit comparing the input to the accuracy of the output stage and simultaneously selecting the most accurate, distortion free, and efficient amplifier matching the waveform of the input (it may change between amplification methods several hundred times per second). In addition, the user could override and select a particular amplification method if desired. I am working on the prototype, and would love any feedback (pun intended) from the group. Also, if anyone knows if this has already been done, then for God sakes let me know so I can just buy the amp (although that would be less fun).

 

[Ian Finch]

..... with a circuit comparing the input to the accuracy of the output stage and simultaneously selecting the most accurate, distortion free, and efficient amplifier matching the waveform of the input....

Error correction by amplifier selection? A power amplifier has finite thermal settling times before the prediction or assessment of which amplifier type will best perform can be precisely made - and you are talking about high precision here.

Unless all amplifier options are already driving similar loads with the same signal, it seems unlikely you would have any means of meeting valid comparison criteria. Simulation and prediction firmware might be more fruitful.

 

[Superstrand]

I figure it will be a challenge, it reminds me a tiny bit of how Mr. Carver controlled his switching power supplies, or the old NAD power envelopes, I think that with some super fast switching that this may be feasible, I will know soon, I have the guts laying all over the living room floor waiting for an enclosure. Sure there are challenges to be met here, but I don't think that it is insurmountable, the goal being an algorithm that compares a set of variables and selects the best amp for the job at say 100khz or more. Of course it may all sound like rubbish when I get it together, but in my design I can simply override the automatic switching and manually select the amplification method that I wish to use for a given work of music. In either event, I get all of my favorite amps into one box that I can manually control if my comparison algorithm turns out to be rubbish.

 

[Boscoe]

Sounds a bit weird. How do you know each one is performing unless you have it on? The performance will vary when it's attached to the speaker so you'll never know if one is better than the other and if you do the switching will introduce more problems than one amp on its own!

 

[Ian Finch]

You can't, but you can predict it if your algorithm is at least as good as say, SPICE. I'm guessing it would take a lot of expensive modelling and trials to do it. though.

 

[DF96]

If you can accurately determine which amp is best then you can accurately determine what you have to do to improve it e.g. via an auxiliary Class A amp.

To be honest, I think it is a silly idea. The problems of achieving fast seemless noiseless switching of audio power are much harder than merely amplifying an audio signal. So if you can solve the switching problem then you don't need a switch!!

 

[Superstrand]

Manual switch would be for taste or comparative listening only, in auto mode the device would automatically compare the input signal to the predicted output and do the selecting of the most accurate amplification method based on the predicted output accuracy, power, thd, etc., with reference to the actual output.

 

[Superstrand]

It will be a while before I post on this again, as I am busy building it. Cheers