my new 4.5kW class-d amp (crown BCA topology)

[Gyula]

So I suggest you to listen it with an average loudspeaker, at a few watts of power, because there's a remarkable problem with the BCA's transient response. You can see it in my early attached "bca.zip" simulator. And you also can refer to my simulator in your papers and thesis.

Gyula

 

[mzzj]

So I suggest you to listen it with an average loudspeaker, at a few watts of power, because there's a remarkable problem with the BCA's transient response. You can see it in my early attached "bca.zip" simulator. And you also can refer to my simulator in your papers and thesis.

Gyula

I am not sure what i am supposed to see in your simulator response? Is lowest blue trace showing output voltage with some sort of asymmetry in rising edge of sine or what?

Thanks for the offer but this is purely hobby project and i am not going to write any sort of papers from this

 

[Tekko]

That is a ugly design But if it works

 

[mzzj]

That is a ugly design But if it works

Ugly, where? I would call it professional
speaking of ugly smps:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=68662

 

[Tekko]

That smps no longer exists, it quit its duty after two weeks of developement.

It is ugly but actually p2p wiring is the optimal solution for high freq stuff such as class d amps, as long as wires are kept short and ones leading amps is kept as short as possible together with as thick as possible. If possible the components legs should be the only wires connecting the components.

Using cir boards may look good but does induce nasty capacitances if you dont pay attention and is a very good pcb designer.

How does the thing sound ?

 

[mzzj]

How does the thing sound ?

well, ..Powerfull

"reproduces respectable amounts of bass without any hesitation"

 

[Gyula]

The problem of the BCA lies in the subtraction of the two coils' current. If the circulating current isn't set to a value larger than half of the maximum output current, the coils are taking turns at getting exhausted during the half periods of the signal. And the output is being distorted. In this distortion period, the rising of the output current is halved. This is resulting in a drowning sound, mostly under the high amplitude bass. You can see it with the simulator if you raise the input signal with button 'i', set the coils' loss to maximum with button 'v', optionally switch on a 40 kHz LP IIR filtering for output with button 'e', and then reduce the coils' loss with button 'c'. The other functions are present in the cpp file.

Gyula

 

[mzzj]

The problem of the BCA lies in the subtraction of the two coils' current. If the circulating current isn't set to a value larger than half of the maximum output current, the coils are taking turns at getting exhausted during the half periods of the signal.

Gyula

OK, why didnt you tell me this in firstplace Thanks, clarified few oddities to me greatly.

I noticed this behaviour on scope screen, at minimal bias nonlinearity appears near zero crossing and when i turn up "bias" nonlinearity sifths to higher amplitudes. noticed also on spice but i wasnt sure if spice is doing its own or if its real behaviour.

Orginally i didnt have any control on circulating current and circuit was running with 30amps "bias". Needless to say that losses on output stage were a bit high. Couple of weeks ago i actually had a topic here asking why output stage is taking kiloamperes of current....

thanks again, saved me from scrathing my head a lot

 

[Gyula]

I was just out of curiosity if you find it out. I also thought a lot on this problem. I think, there should be a repose duty-factor controller, which varies the bias current efficiently. Or other way to maintain non-zero coil currents all the run. This is a very interesting control problem, mainly with paralleled output stages.
But I think, UcD is more economic.
Now, I should to learn for my exams rapidly..

Good Luck!

Gyula