Simple Symetrical Amplifier

[Jay]

Originally Posted by Lazy Cat

So Bryan if there is a beauty in simplicity, here's the simplest SSA version of them all just for you.

The simplest amp is a single ended single transistor amp, but the power is too low. How about simple SSA to give close to 50W. My preferred circuit is the combination between this simplest SSA with the basic SSA MOSFET you posted earlier. Something like this (if possible, but why not):

EDIT: another driver stage can be added if the above is not optimum.

[MiiB]

I concider the input and cascode as one device, one better device, I would put them back in...
and I would add a jfet as a current scouce in the vas stage to revent thermal current drift

[Lazy Cat]

Hi Jay

Yes, MiiB has the point here but if you still insist to omit cascodes than you shoud do it in a right manner.

Input pair without cascodes will of course work if you'll choose low noise, high Vce (65 V or more) transistors like BC546B/BC556B, 2SA970/2SC2240 or similar.

No input cascodes also mean higher power dissipation on input pair so no more than 2 mA collector bias current is recommendable (Rc 330 ohm).

Instead of 820 ohm resistor I suggest you to use simple Vbe multiplier completely installed on a PCB to act as adjustable zener only. It will provide you more stable output DC bias conditions as being independent to VAS thermal bias fluctuations.

Otherwise this version will work nicely also with lower rails potential like +/- 35 V and will easily give you 50 Wrms/8 ohms, since 28,28 Vp is exactly the figure you need for that kind of output power.

[MiiB]

Yes a vbe...not mounted on a heatsink....or a jfet currenscouce in series with the bias spread resistor..

[Jay]

Originally Posted by MiiB

I concider the input and cascode as one device, one better device, I would put them back in...

Better in term of measured performance, right? What is the best possible measured performance without adding the cascode?

TO-92 is fine for the cascode but in the schematic posted by LC lately it was BF470. Without it, if ever possible (acceptable measured performance/stability), it will sound better I believe.

Originally Posted by MiiB

and I would add a jfet as a current scouce in the vas stage to revent thermal current drift

What kind of current source? The one suggested by Salas? Can't it be made stable without adding the extra fet (coz it looks strange hehe)? How about using TO-3, big heatsink (or even use of fan), PTP wiring, or Vgs multiplier in the bias circuit?

BTW, how about your SSA MOSFET with 60V rail, are they safe? I have just lost 4 pairs laterals already

[Jay]

Originally Posted by Lazy Cat

Hi Jay

Yes, MiiB has the point here but if you still insist to omit cascodes than you shoud do it in a right manner.

Input pair without cascodes will of course work if you'll choose low noise, high Vce (65 V or more) transistors like BC546B/BC556B, 2SA970/2SC2240 or similar.

No input cascodes also mean higher power dissipation on input pair so no more than 2 mA collector bias current is recommendable (Rc 330 ohm).

Instead of 820 ohm resistor I suggest you to use simple Vbe multiplier completely installed on a PCB to act as adjustable zener only. It will provide you more stable output DC bias conditions as being independent to VAS thermal bias fluctuations.

Otherwise this version will work nicely also with lower rails potential like +/- 35 V and will easily give you 50 Wrms/8 ohms, since 28,28 Vp is exactly the figure you need for that kind of output power.

Thanks LC, I will try to figure it out.

For the worse noise rejection without the cascode, my plan was to use regulated supply for the front end including the driver (so only the laterals use classic supply). This way, the FET regulator may be compared with simple RC filters.

The higher voltage was not primary for power. I just think the mosfet will need them.

[MiiB]

The thing is that the drivers increase current when they and the input stage heat up...this can be avoided with the use of the jfet ccs ( as per suggested by salas)...then yuor laterals won't run away

[Jay]

Originally Posted by MiiB

The thing is that the drivers increase current when they and the input stage heat up...this can be avoided with the use of the jfet ccs ( as per suggested by salas)...then yuor laterals won't run away

I thought the FET solution was for stable offset as the lateral has negative thermal coefficient anyway. I believe you didn't mean thermal runaway.

How about small fan (I have 3x3cm) on top of the BJT drivers, with adjustable speed, or speed that change with temperature sensing as found in computer power supplies

[MiiB]

Your right about the fets....but what about the VAS/drivers...?. When they heat up the current increases and thus the sprread increases

[Jay]

Originally Posted by Lazy Cat

Otherwise this version will work nicely also with lower rails potential like +/- 35 V and will easily give you 50 Wrms/8 ohms, since 28,28 Vp is exactly the figure you need for that kind of output power.

Are you sure that it can give 50W@8R with 35V supply? What a good circuit! But which schematic? BTW, your simplest SSA is without the input cascode. Ruwe added one TO-92 cascode transistor to that. I can accept both circuits, but currently I'm looking for power (I'm building a big 3-way speaker).

What is the maximum power (and supply rail) can be had from both circuits? I believe the cascode will bring more benefit (or is it a must) when higher supply rail is used in such a simple circuit. I will use at least 46V rail (preferably more) and 2 output pairs. I expect at least 70W to replace my Goldmund clone (Mimesis 3) and I want to start with the simplest circuit, i.e. Ruwe version (because I believe the simplest SSA wont make it there).

I guess Ruwe circuit is not optimal but very potential soundwise. But if this circuit cannot give comparable power to Mimesis 3, I will start with my previous idea, or add another driver, i.e. Wahab's schematic without the input cascode.