Simple feed forward error correction amp.

[Circlomanen]

Here is a simple feed forward error correction amp I built this morning.

YouTube

An ugly fire hazard rats nest wiring prototype playing through a cheap but nice sounding car audio coax driver.

 

[Circlomanen]

Nice simulated performance. It does sound muchbetter then it should considering the crude rats nest construction. I built this as a prototype to check the potential of my idea.

 

[Circlomanen]

An atmospheric picture of my masterpiece at work.

It does sound quite nice all the way down to 5 volts from the power supply.

 

[padamiecki]

Hi!
very nice!
BTW, did you throw away the circlotrons?

 

[thorstenlarsen]

Hi Circlomanen! How does the feedforward errorcorrection come in to play? T.

 

[Oneminde]

The Benchmark AHB2 is using a combo of NFB and Feed Forward, have you tried this and possibly measurements of IMD and THD in both scenarios, aka NFB alone and NFB + Feed Forward.

Nice work

 

[Circlomanen]

BTW, did you throw away the circlotrons?

No. Not really, but I have not built any circlotron since 2006. I don´t like the added complexity of having to use two floating power supplies per channel.

How does the feedforward errorcorrection come in to play? T.

Through the bifilar coil. It will mirror whatever happens on one side of the Mosfet and phase-invert it to the other side of the Mosfet, effectively canceling lots of distortion. This is the simplest way of building a Feed Forward Error Correction amp I can think of. And it works wonders for the sound quality. A rare combination of a high level of objective measured performance and perceived subjective sound quality due to low amounts of NFB and a simple design with few parts.

 

[N101N]

The scope for signal amplitude in the input stage is rather modest. At what current is it running? I am extremely suspicious of the magical power of that coil.

 

[Mark Tillotson]

The DC voltages marked on the output stage suggest a lot of resistance in the bifilar transformer, or is the standing current high? It looks like you should get good output swing this way, but I fear the output impedance is high and not very constant across the band?

 

[Circlomanen]

I am building a two channel version with a slight decrease of "ratsnestiness" and I simplified it slightly by removing the input p-channel Fet.

 

[Circlomanen]

Each winding has 1,4 ohm DCR and I am running 1,9 ampere through them in my new build.

The output impedance is low and seems consistent through a wide bandwidth.
The Feed Forward Error Correction/Nelson Pass SUSY-ish function is quite powerful and lowers distortion and output impedance substantially.

 

[padamiecki]

Did you "lose" something regarding sq after applying gnfb?

 

[Oneminde]

Q: Regarding transformer/inductor as error correction, were can I learn more about that ?