a wonderfull sziklai

[Lumba Ogir]

Hi,
I would not make the voltage amplifying section complementary symmetrical.

 

[catalin]

C32 is too small.
It needs to be >= 22k/680r * sqrt(2) * 4u7F >= 215uF. Use 220uF.

C42 seems a bit big @ 47pF
R = 150 at output also seems big, only 8mA passing Q60 & 51.

Does the mirror load on the LTPs work when you have dual complementary VAS?

The 100 uF(still think that 47u is good) comes from low frequency ~5hz(not audible) => f=1/(2*3.14*R*C) ;= 0.16/(680*C);;;C=0.16/(f*R)=0.16/(680*47u)=5 hz.

C42 is calculated to minimise the noise band > pole at 154khz.(0.16/(22k*f)
Q60 and q51 working at 1mA ! not 8 . 2sc5171/1930 is working at 8-10ma. big input impedance for not oveloading vas => small thd . about the mirror ltp ...look at stochino's amp !

 

[catalin]

Catalin,

May I suggest you to not reference CCS's biasing resistors to ground?
Try to make the input stage + VAS as much as possible a constant current compound stage.
Keep the ground as clean as possible and use a separated PSU for output stage, and for inputs stages a good ccs+shunt feedbackless PSU

Regards,
Mihai

An capacitor + another R to ground will be better ? or do you want to say something else ?
Thanks !

 

[catalin]

Hi,
I would not make the voltage amplifying section complementary symmetrical.

Can you explain why?
Thank you

 

[homemodder]

Catalin, you are absolutely right about stochinos amp, but look more carefully and youll see why his vas looks the way it does. He was clever and found a way to define the vas current, try define it on your design.

In your case the mirror you hae there will help but i dont think it will overcome the problem completly

 

[catalin]

"output offset will be all over the place,"
Explain pls
Thank you

 

[Lumba Ogir]

catalin,
for a more favorable distortion pattern. Then I would use the resources to linearize the open-loop.

 

[homemodder]

Here is some info by members with much more theory knowledge than me and some diffrent ways to look at the problem. Theres many more threads I can refer to. The Jlh solution works but it has some flaws too, read the whole thread.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=16796

 

[catalin]

catalin,
for a more favorable distortion pattern. Then I would use the resources to linearize the open-loop.

I think that distortion is less(crossover,pozitive and negative rail have their way,etc,) with complementary design and a lot other advantages .

 

[catalin]

Here is some info by members with much more theory knowledge than me and some diffrent ways to look at the problem. Theres many more threads I can refer to. The Jlh solution works but it has some flaws too, read the whole thread.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=16796

It' not like his problems.
In that design the current vas don't change with the vcc.just one aspect

 

[Lumba Ogir]

catalin,
well in that case... keep it that way.

 

[catalin]

And the problem is solved by chosing with atention the resistor in the mirror(220),the resistor in input vas(2.2k) and the negative feedback from the emitter vas(75 ohms) .

 

[homemodder]

Catalin the designs in that thread especially the first 2 are exactly what you have, taking away the cascodes.

You going to play a lot around with values, keep in mind that real transistors have do not match so well compared to simmed ones.
If this type of biasing worked we would have already seen thousands of amps using it, look at the date of that thread 2003, there are threads even earlier than this.

 

[catalin]

catalin,
well in that case... keep it that way.

Lumba, even with the simetrical amp the open loop to 20khz is still 65db

 

[catalin]

For everyone ,I just simulate most defavorable case ...
One of the input transistor with a beta common-100.
The dc output is increasing with 30mV from the original (200uV) .

 

[Lumba Ogir]

catalin,
we should leave it hanging.

 

[Stee]

can i say one idea?

about crossover distortion...
put a mix of CFP and EF

 

[catalin]

Let's discuss a little abot Current feedback !

Let's discuss a little about current feedback amps .
I obtain with this circuit :
Slew rate 200V/uS . Phase margin > 70 degrees .Thd 0.0001 % .Open loop gain 137 db ; 90db at 20khz !
Bandwith 22mhz without input filter .
What disadvantages ant what advantages it brings ?
I think that the stability is one problem .
What do you think ?

 

[Stee]

very intresting Catalin

if Q88 is 2SA1302 maybe more stable
because in CFP OS --> hFE paramenter is foundamental
between complementary in common collectors

also

have just one point to drive Output Stage... see figure

can improve stability because front-end have a common load

 

[Stee]

see this circuit

on ebay

http://stores.ebay.it/Blekok-Ham-and-Audio-Parts-Kits-DIY