Quasicomp mosfet PA-amp idea

[edl]

Hi Folks,

Yesterday I posted two circut based on Quasi's output topology in this thread:
Power amp under development
I don't want to steal that thread, so I opened this one.

I updated the referred circuts, I changed the previous stages with an OPA and a cascode-stage.

I going to bulid it at the weekend, but before I start it, I would hear some feedbacks. Is it operable at all? Unfortanetly I cannot simulate it.

On the margin, please, don't suggest me "high-end type", audiophile modifications in the circut. It's coming to be a public-adress amplifier, and I would like to keep things simple and cheap.

Thanks in advance!

[darkfenriz]

Seems workable and relieble if you spend some time on proper frequency compensation. As far as I know this op-amp is internally compensated and you;ve added external compensation too for two-pole compensation. Also you've added another miller pole and global phase lead. I'm lost here and can only tell you, that this may or may not be one potential problem.
10nF parallel with 100R at sziklai is intended to be phase lead compensation of the pair?

best regards

[Eva]

It looks good while keeping the part count under strict control

However, have you evaluated the rail losses?

In the upper rail you will have approx 1.3V from the source resistors, plus two diode drops from the VAS and the gate driver, plus 5 to 6 volts from the gate. This makes a total of 7.7 to 8.7V, very high for a PA amplifier.

In the lower rail, you have also the same losses as in the positive rail, plus 2 to 3 additional volts due to the emitter resistor of the quasi-complementary driver, making a total of 9.7 to 10.7V, and thus, producing asymmetric clipping (in my PA experience that makes much easier to blow voice coils).

As it can be seen, rail losses are a pity in MOSFET output stages, and the problem is not trivial when only N-channel devices are employed. You can bootstrap the upper rail with just three diodes, a resistor and a capacitor, but the lower rail is a chore..

[edl]

Dear Darkferniz,

Thanks for your beneficial opinion.

-You're absolutely right about inner comepensation of that device, the manufacturer already intergated a 15pF cap so as to compensate. Other amps I've seen that using NE5534 in this stage are using 33pF external compensation, so I going to try this value first.

-Yes, I'm long to use miller capacitors between the base and the collector of the VAS transistors. I chose 68pF, I think it's a good compromise to take into account the amplifiers stability and the high-frequency transfer.
In addition I added a 22pF paralel with the 22k feedback resistor, forming an RC-circut with about 330kHz frequency. This calling to achieve extremely stability.

-I was thinking about that 10nF cap paralel with the 100ohm resistor in the hexfet's driver circut. I wanted to use it to make that stage more simmertic, so that that capacitor simulates gate capacitance.
I found that this composes more drawback than benefit, so I going to erase that cap from the circut.

All the best, thanks again,
Edl

[edl]

Dear Eva,

Thank for your feedback chiefly for your accurate calculations.

Yes, I took rail losses into account, but I wouldn't use higher rails for the driver, because it assumes more complex supplying.
But that 8-10V loss is too much, so I am dismal throught this fact... I'm really doubtful now.

Best regards,

[Eva]

I made a mistake, in the lower rail there are no VAS losses due to the quasi-complementary driver, so their losses are similar to those of the positive rail. However, the circuit shows an interesting behaviour near clipping because the transfer function suddenly flips the phase by 180 degrees when the driver saturates. In practice, the circuit will probably oscillate more or less wildly until the output is no longer asked to approach the rail (I have experienced it in my N-channel project).

[edl]

Hi Eva,

And what do you think, how can I get rid of that undesirable oscillation?

Thanks,