Another quasi-complementary design

[darkfenriz]

Hi quasi again!
I have one general question to you:
I've been always a bit suspicious about driving mosfets like this, it seems to faster swith-on then swith-off (it looks like a bjt-current-source for 'on' vs. 220resistor for 'off'). Have you considered additional emitter follower stage?

cheers

[Workhorse]

Originally posted by richie00boy
D5 needs an additional normal diode in series with it to prevent output rail spikes due to reactive load from reaching up to the upper MOSFET gate and turning it on spuriously.

Hi buddy,

IF i am not wrong , when the spike is present and as it goes to the upper mosfet , in every case the zener will clamp it to 12V if +ve with reference to VGS or to 0.6V if -ve with reference to VGS.
Acc to me there is no need for extra diode in series.


PLease correct me if i am wrong.

regards,
kanwar

[sajti]

Hi quasi!

Nice design, I'm interesting to build it! My only problem is the loss on the positive swing. As I count, there is 4.7V on R4 and R9. The Vgs of the IRFs, is about 6-7V with the peak current. This means that You can get only 39-40V peak on the posisive half of the output signal. This looks serious power loss for me

sajti

[quasi]

Originally posted by darkfenriz
Hi quasi again!
I have one general question to you:
I've been always a bit suspicious about driving mosfets like this, it seems to faster swith-on then swith-off (it looks like a bjt-current-source for 'on' vs. 220resistor for 'off'). Have you considered additional emitter follower stage?

cheers

Hey darkfenriz;

Yes in my other design in the thread "Power Amp Under Development" I run a 2nd driver stage that is controlled by the second constant current source.

With a very similar cct to this one though I have not noticed any problems with the output drive. The second stage constant current source of around 36mA (18 mA per side) is a sufficiently low "impedance" to turn the FET on and 220 ohms should discharge the gate quickly too. There are probably differences between the turn on "slew" and the "turn off" slew but in practice on a similar cct even the square wave response doesn't seem to show anything bad.

Cheers

[quasi]

Originally posted by sajti
Hi quasi!

Nice design, I'm interesting to build it! My only problem is the loss on the positive swing. As I count, there is 4.7V on R4 and R9. The Vgs of the IRFs, is about 6-7V with the peak current. This means that You can get only 39-40V peak on the posisive half of the output signal. This looks serious power loss for me

sajti

Yes you are right of course. The 4.7v is necessary to provide enough voltage for the second constant current source (T5) to work properly.

If you cut the track between C8 and R17, you can run the top left hand side of the amp with an additional 10v. This will provide sufficient voltage to drive the positive output stage to close the 50v rail.

Here is the revised cct.

Cheers

[darkfenriz]

thank you for reply!!
if I could ask you for further sharing of your knowledge and experience:

ftp://ftp.elektroda.net/download/Audio/Projekty/Wzmacniacz%20HiFi%20250W.pdf

here is another (very similar topo) amp I found (sorry for it's in Polish) and resistors responsible for swithing off are 1.2k. For me it is asking for troubles, what do you think?

[sajti]

Originally posted by darkfenriz
thank you for reply!!
if I could ask you for further sharing of your knowledge and experience:

ftp://ftp.elektroda.net/download/Audio/Projekty/Wzmacniacz%20HiFi%20250W.pdf

here is another (very similar topo) amp I found (sorry for it's in Polish) and resistors responsible for swithing off are 1.2k. For me it is asking for troubles, what do you think?

God! The amp looks quite good, but the overcomplicated bias compensation circuit???????


sajti

[richie00boy]

quasi,

Thanks for the compliment You do not need to include a normal diode in the lower zener diode because there is absolutely no risk of current reaching up through it.

Originally posted by Workhorse
Hi buddy,

IF i am not wrong , when the spike is present and as it goes to the upper mosfet , in every case the zener will clamp it to 12V if +ve with reference to VGS or to 0.6V if -ve with reference to VGS.
Acc to me there is no need for extra diode in series.


PLease correct me if i am wrong.

regards,
kanwar

Hi Kanwar,

No, you must include the diode. As you say the voltage will be clamped at 12V, but surely it's far better to stop the spurious transients even being a problem. They will cause distortion.

[quasi]

Originally posted by darkfenriz
thank you for reply!!
if I could ask you for further sharing of your knowledge and experience:

ftp://ftp.elektroda.net/download/Audio/Projekty/Wzmacniacz%20HiFi%20250W.pdf

here is another (very similar topo) amp I found (sorry for it's in Polish) and resistors responsible for swithing off are 1.2k. For me it is asking for troubles, what do you think?

I agree that 1.2K seems high for the gate to source resistor. 3 x IRF640's have less input capacitance than 2 x IRFP250's but I think it is still too high.

By using the 1.2K this means that the drive current is only 3.3mA and this is too low I believe for good FET control.

And the temperature compensation cct is very impressive, but I would never use it. It may do a better job than my one transistor (wired as a diode) but both will settle down to a consistant bias current. In any case +/- 10% in bias current is not important provided you still have plenty.

Note though that in my cct if the sensing transistor is too far from the FETS .......kaboom! is a real possibility.

Cheers

[quasi]

Originally posted by richie00boy
quasi,

You do not need to include a normal diode in the lower zener diode because there is absolutely no risk of current reaching up through it.



Hi Kanwar,

No, you must include the diode. As you say the voltage will be clamped at 12V, but surely it's far better to stop the spurious transients even being a problem. They will cause distortion.

I'f I use the diode I will put on both zeners, only so that if they turn on under over-current conditions they do so symmetrically.

The FETs have built in reverse diodes. Would these clamp back EMF sufficiently to reduce the problem?


Cheers