diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Solid State (http://www.diyaudio.com/forums/solid-state/)
-   -   My new mini Class A amp. (http://www.diyaudio.com/forums/solid-state/99252-my-new-mini-class-amp.html)

GK 1st April 2007 01:35 PM

My new mini Class A amp.
 
Hereís a unique little amp Iíve designed for my study with 20kHz THD rather under 0.001% at full power, using common ďgarden varietyĒ transistors. In my study, I only require low power, and I like to have something to warm my feet on in winter, so the amplifier is rated at 12W rms into 4 ohms. Dissipation per channel is about 50W and quiescent current stability is rock solid.

I call the circuit unique because Iíve come up with a complementary VAS stage which allows the twin long tail pair (NPN/PNP) input stage topology to be used, whilst retaining current mirror loading and push-pull drive to each VAS stage. This significantly improves the linearity and gain of the long tail input pairs over typical circuits that use a low value resistors for the collector load for each LTP. I acheived this by using a differential stage for each VAS and picking-off the biasing point from the non-inverting side of the current mirror. This method works quite well, without compromising PSR.
In order to make the complementary VAS stages as linear as possible, I used both emitter follower buffering, emitter-degeneration and cascoding. Also, a triple emitter follower output stage is used, to keep the loading of the VAS very light.

The quiescent current of the VAS stage is inherently temperature compensating (note the extra diode connected transistor in each current mirror), and is set to 10mA with RV1 (to compensate for the Vf variations of typical red leds). I didnít use a Vbe multiplier to bias the output stage, as they donít perform very good when the emitter-emitter voltage of the output transistors is large, as it is in typical class A designs. Iq is set to 1.5A by RV2, giving 1V emitter-emitter, with a ballast resistance of 0.333 ohms in each leg. Temperature compensation is provided by the six transistors connected as diodes. Four of these transistors are TO-126 types, which mount easily onto the heatsink. The pre-driver transistors and their respective temperature compensation transistors are TO-92 types, not mounted on the heatsink, but thermally bonded to each other in pairs.

Separate, unregulated supply rails are provided for the output stage and the low-power circuits. The ~ +/-15V rails for the output stage are provided by a conventional PSU using a twin 12V secondary windings and 18800uF per rail (2 by 4*4700uF). The ~ +/-20V rails are provided also provided with a conventional PSU, but with a transformer having twin 15V secondary windings.

Iím currently laying out the final PCBís. If anyone is interested in the design, Iíll post the final design files here in the future along with photos of the finished product.

Cheers,
Glen

http://users.picknowl.com.au/~glenk/CLASSA.gif

fab 2nd April 2007 01:29 AM

Hi Glen

Have you tried a more simple topology like this one here http://www.diyaudio.com/forums/showt...59#post1120959 for symetrical diff input current mirror loading and push-pull drive to each VAS stage?
And if so, how does it compare to your developped circuit?

12W rms is not much. Are you sure your amp can handle dynamic music transient even at "low" level setting ? Do you think that your circuit could be also used in class AB (higher supply but same actual bias current)? This is what I have done for my amp (15 Wrms class A/ 75 Wrms class B).

GK 2nd April 2007 02:37 AM

Quote:

Originally posted by fab
Hi Glen

Have you tried a more simple topology like this one here http://www.diyaudio.com/forums/showt...59#post1120959 for symetrical diff input current mirror loading and push-pull drive to each VAS stage?
And if so, how does it compare to your developped circuit?

G'day fab.

I haven't seen it done that way before - there is always more than one way to skin a cat ;)
My circuit may be more complex, but it doesn't compromise the operation of the current mirror.


Quote:

12W rms is not much. Are you sure your amp can handle dynamic music transient even at "low" level setting ? Do you think that your circuit could be also used in class AB (higher supply but same actual bias current)? This is what I have done for my amp (15 Wrms class A/ 75 Wrms class B).

I've got an old but good pair of speakers from an old PYE record player in storage. I used to drive these with a pair of National LM380's as a teenager, and they could be heard a few houses down the street.
Coupled to my baby 12W class A amp, they'll be plenty loud enough for my study.

Cheers,
Glen

G33/33 2nd April 2007 04:13 AM

Re: My new mini Class A amp.
 
Quote:

Originally posted by G.Kleinschmidt

Iím currently laying out the final PCBís. If anyone is interested in the design, Iíll post the final design files here in the future along with photos of the finished product.

Cheers,
Glen



Hi Glen,

I'm interested !! pls keep us informed. I also interested with your PCB layout, there is no way I can make my own layout for such complex design:smash:

thanks alot


G33/33

GK 2nd April 2007 08:52 AM

BBZZZZZTTTT.....

The schematic I posted above is flawed. The current current steering ability of the VAS stages won't work reliably. I was mislead by a spice simulation and my physical prototype unit just worked by a fluke of transistor matching.

I have fixed the topology with a rather simple modification though, and am updating the schematic at the moment.

I have deleted the old schematic and I suggest that anyone who has downloaded it should do also, as the circuit won't work as drawn.


Cheers,
Glen

GK 2nd April 2007 10:45 AM

Okay boys and girls.......

The link (as above) is now attached to the updated schematic, with the revised VAS stages. The LTP load's are now two pairs of CCS's and the VAS drive is fully differential. Anyone who saved the previous schematic, please delete it and swap it with this one.

http://users.picknowl.com.au/~glenk/CLASSA.gif

Comments/Criticisms?


Cheers,
Glen

fab 3rd April 2007 02:44 AM

Only C8 and C9 are required for stability? no cap across feedbcak resistor?
Have you looked at square wave into RC parallel load yet?

The traditonal "zobel" only consists of C20? no series resistor?

I am not sure to see the "current mirror" in the LTP...

fab 3rd April 2007 02:48 AM

Re: My new mini Class A amp.
 
Quote:

Originally posted by G.Kleinschmidt
[B]... I didnít use a Vbe multiplier to bias the output stage, as they donít perform very good when the emitter-emitter voltage of the output transistors is large, as it is in typical class A designs. .]
Can you explain why Vbe mult is no better than a pot?

GK 3rd April 2007 03:10 AM

Quote:

Originally posted by fab
Only C8 and C9 are required for stability? no cap across feedbcak resistor?

Yes they are.


Quote:

Have you looked at square wave into RC parallel load yet?

When the final boards are finished, that's what I'll be doing to find the optimum value for C8 and C9.


Quote:

The traditonal "zobel" only consists of C20? no series resistor?

This is one derivation of the circuit that doesn't need a resistor in series with C20.


Quote:

I am not sure to see the "current mirror" in the LTP...

Thats probably because the circuit has since been modified and each current mirror has been replaced with a pair of current sources - like I said above.


Cheers,
Glen

GK 3rd April 2007 03:28 AM

Re: Re: My new mini Class A amp.
 
Quote:

Originally posted by fab


Can you explain why Vbe mult is no better than a pot?


I did not say that a Vbe multiplier is no better than a pot. I said that a Vbe multiplier isn't very good at biasing a class A stage when the quiescent emitter-emitter voltage across the output stage ballast resistors is large. In this design it is 1V, to give a quiescent current of 1.5A. A Vbe multiplier is substandard in this application because it provides overcompensation. If you donít believe me, get out a pen and paper and do the math.
Also, I am not biasing my output stage only with a trimpot, but with a trimpot in series with six diode-connected transistors.

Cheers,
Glen


All times are GMT. The time now is 10:05 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright ©1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2