VSSA Lateral MosFet Amplifier

[pergo]

have your try a LED string for bias, instead of 1k resistor?
More fashion

[sonnya]

Sorry but it is depending on the unlinear millercaps in the bjt's .

I Think it is a NO GO as some AMP maybe Will be stable and others Dont.

How does distortion change when this AMP exit class A.

[Jay]

Originally Posted by sonnya

How does distortion change when this AMP exit class A.

I thought much worse, but no, only a little worse. So class-A might not be a good idea. And if I'm not mistaken, just around 30W. Too low.

[Bigun]

Originally Posted by pergo

have your try a LED string for bias, instead of 1k resistor?
More fashion

It's a valid idea in my books, possibly allows elimination of the capacitor. But you do get more control over the bias current through the output when using a variable resistor.

[pergo]

yeah,
i do not simulate yet the circuit, i don't know the real voltage drop on 1k resistor.

[Bigun]

Originally Posted by sonnya

Sorry but it is depending on the unlinear millercaps in the bjt's .

we don't care !

[dadod]

Originally Posted by Bigun

It's a valid idea in my books, possibly allows elimination of the capacitor. But you do get more control over the bias current through the output when using a variable resistor.

With the bias resistor of 1 kohm amp is in A class(more then 4A of bias). For B class it should be between 160 and 200 ohm depends of used bjts.
dado

[Bigun]

Originally Posted by dadod

With the bias resistor of 1 kohm amp is in A class(more then 4A of bias). For B class it should be between 160 and 200 ohm depends of used bjts.
dado

Actually, in my sims, only needs 120 Ohm for around 120mA of bias, which is about as low as you want to go without temperature compensation. But even at this low current, the distortion is still excellent.

[Lazy Cat]

Few more words about VSSA, tips for a potential builder to follow.

Input BC550C/560C pair has to be thermally coupled simply by gluing flats of their cases together or by greasing the flats and than firmly tied by plastic cable tie.

Each of VAS gain transistors should have small heatsink to dissipate its 0,5W thermal power. It would be also fine to have only one heatsink for both of them, in this case mandatorily connected to GND potential. 2SA1381/2SC3503 must be electrically isolated from the heatsink by mica insulators if they're not already in isolating cases.

10uF poliester film capacitors and 2,2mF/10V elcos should be quality performers here since they're in direct AC feedback to GND path.

Output transistors of course have to be mounted to the main heatsink (0,5°K/W), I suppose the best version would be that they're completely positioned underneath the PCB serving as spacers-holders for PCB.

There are three adjustments for VSSA to set it to normal working conditions.

Two 15k input DC current injection resistors should be trimmed to get 150mV on each 10 ohm VAS emitter resistors and at the same time to get 0mV DC offset at the output.

1k multiturn trimmer serves to adjust proper output mosfet's bias, something around 150mA is pretty optimal value here. After one hour when the amp is warmed up, readjust trimmer to set exact 150mA bias current.

At the end short explanation why so low distortions we get in so simple amp as it is.
Well beside high slew-rate there's also large bandwidth at 80dB of OLG, which is very high concerning this is CFB design, so 55dB of feedback correcting the output is fast and accurate to uV levels.

Another very important feature of VSSA is correct subtraction between input and output signal due to constant quiescent current provided simply by resistor, not by using some complex constant current source. How is this possible? Well 15k resistors are located between two almost constant DC points with no AC voltages present there, so constant DC and constant resistance gives constant current simply by using plain resistor. Even better performance results would be achieved by regulating and filtering rails DC to a mV level.

[sonnya]

2sc3503 and 2sa1381 is full body plastic