Hello All,
As the Subject says;
I protoyped my amp and it works , sort off !!!
When i started the quiescent current was way to high and i popped several power transistors.
Now, the quiescent current is fine, at least , when i hold the heatsink. When i don not , it raises.
The power transistors are mounted on the heatsink with Mica sheets.
Any one an idea !
As the Subject says;
I protoyped my amp and it works , sort off !!!
When i started the quiescent current was way to high and i popped several power transistors.
Now, the quiescent current is fine, at least , when i hold the heatsink. When i don not , it raises.
The power transistors are mounted on the heatsink with Mica sheets.
Any one an idea !
Mr Evil said:
the oscillations I have inadvertently created are in the high tens of kHz -- it is really helpful to have a scope with FFT for this --
to cure the problem experienced by blu_line I would recommend 100nF caps on each of the power supply rails as nearby to the output transistors as possible. Polypropylene is (purportedly) best of all, but polyester (mylar) is much less expensive.) I hope you have a ground plane since a bypass cap with a long lead to the ground is about as useful as teats on a bull.
The classic '100nF in series with 10 ohms' RC network from the output of the circuit to ground is almost mandatory. Nearly all amplifiers will oscillate without these components
If you are designing your own circuit and oscillation persists you should have to change frequency compensation and improve layout
If you are designing your own circuit and oscillation persists you should have to change frequency compensation and improve layout
ANd that heat sink is a giant antenna if it isn't grounded. Make sure there is no resistive connection to anything else and then ground it. Either ground it with a wire, or at least with a cap.
A stability network shunting the output at high frequncies really is a must.
A scope on your output would show the oscillation instantly.
A stability network shunting the output at high frequncies really is a must.
A scope on your output would show the oscillation instantly.
schematic
For those interested , hereby my amp schematic.
On board are 4 x 10.000 uF and 4 x 100nF as well as 2 x 100 nF close to the current sources and mirrors (not drawn).
These last 2 helped to stop oscillation in for the most part.
I also removed the emitter degeneration resistors of 0.1 ohm in the emitters of the power transistors. This also helped a bit.
Did not try the snubber network yet !
grtz
Simon
For those interested , hereby my amp schematic.
On board are 4 x 10.000 uF and 4 x 100nF as well as 2 x 100 nF close to the current sources and mirrors (not drawn).
These last 2 helped to stop oscillation in for the most part.
I also removed the emitter degeneration resistors of 0.1 ohm in the emitters of the power transistors. This also helped a bit.
Did not try the snubber network yet !
grtz
Simon
I would not remove the emitter resistors if i had parallel output devices.
Haven't done any such thing myself but several people calim that the sziklai output stages are prone to oscillations. There has been suggestions on how to stabilize them, can't remember where though.
I like the diamond like output stage.
I am a bit confused with the gain stage since most of the circuit seems to be emitter followers but ia am kind of thick when it comes to these kinds of things.
If the emitter resistors on the power transistors caused the oscillations you could try to put them back and incease the current slightly to bias the output stage a bit more. I guess one of the problems with the Sziklai pair is at switchoff and if you have a low bias that could cause problems?
Haven't done any such thing myself but several people calim that the sziklai output stages are prone to oscillations. There has been suggestions on how to stabilize them, can't remember where though.
I like the diamond like output stage.
I am a bit confused with the gain stage since most of the circuit seems to be emitter followers but ia am kind of thick when it comes to these kinds of things.
If the emitter resistors on the power transistors caused the oscillations you could try to put them back and incease the current slightly to bias the output stage a bit more. I guess one of the problems with the Sziklai pair is at switchoff and if you have a low bias that could cause problems?
Check this pdf
http://www.eecg.utoronto.ca/~kphang/ece1371/cfa.pdf
and
http://www.elecdesign.com/Globals/PlanetEE/Content/2800.html
I had another link explaining diamond buffers and etc.
But can not find the link , sorry
You can check per-anders website also.
They are MPC71 types
grtz
Simon
Hi Blu_line !
You have no feedbackcompensation at all, without, an amp with
globalfeedback will always oscillate !!!
An amp must drop below unitygain at frequencies where phaseshift
reaches 180°, this is only possible with some caps forcing this
behaviour. Typically this freq is around ~10mhz.
Sim your amp with 20khz, with timestep at ~10ns, this should show
up the oscillations.
Mike
You have no feedbackcompensation at all, without, an amp with
globalfeedback will always oscillate !!!
An amp must drop below unitygain at frequencies where phaseshift
reaches 180°, this is only possible with some caps forcing this
behaviour. Typically this freq is around ~10mhz.
Sim your amp with 20khz, with timestep at ~10ns, this should show
up the oscillations.
Mike
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.