Simple MOSFET amp problem

[luixssj3]

Hi guys
I build this classical MOSFET amp because I have some scrap:


This is the PCB




Little photo :



Iknow is not the best design, is not the best sounding amp etc.. I build it for fun... so don't kick me

I have some trouble with it, the Vbe multiplier is oversensitive, if I set the bias (100mA per MOSFET) this slowly decrease as themperature increase and if I rebias at 100mA when the amp restart at cool the bias is over 100mA and I risk some thermal runaway....

Now I set the bias at 70mA ad is OK for the start, no runaway but after the heatsink is hot the bias get 50mA or less...

In your experience is happened thing like this?
Is there any solution?

Sorry for my bad english and thanx in advance...

bye

[lineup]

you could try
to let the temp sensing transistor
have a bit less direct contact with heatsink

maybe put in some millimeter of some material
between this little transistor and the heatsink

this will make the temp adjustment a bit slower
and less sensitive to variations


from looking at your schematic
another way
exchange the MJE340 in your schematic
for a transistor
that better match the temp charateristics of those MOSFETs
http://img205.imageshack.us/img205/3...wmosfetvl4.gif
Often used is BD139
and there are a few other you can try to experiment with

also you might try putting those drivers
MJE15030 + MJE15031 onto a separate smaller heatsink
without temperature contact with main heatsink

[luixssj3]

Thanx for your reply, I'll try to put some space from Vbe mult and heatsink...

I'll let you know...

[lineup]

Originally posted by luixssj3
Thanx for your reply, I'll try to put some space from Vbe mult and heatsink...

I'll let you know...

Yes, that is a good first try.
Sometimes it can be enough just loosen the screw a little bit.

I know old amplifiers, where the temp sensing resistor
was never in actually contact with heatsink, but very close to it.
Often used was NTC thermistor = Negative Temp Coefficient
is a resistor that gets lower resistance as temp get higher.

But also most transistors, like MJ340, BD139
has got a very linear negative coefficient.
Where Vbe ( volt from Base to Emitter pin ) sinks like 2 mV / per 1 degree C increase.
The hotter a transistor gets, the lower VBE gets.

In audio amplifiers
This gives a better control than using only resistors and/or thermistor for temp sensor circuit.
Still this is one detail that often require some trial and error.
To arrange this sensor in a way that gives perfect temp compensation,
is not that easy .. especially when you design an amplfier from scratch.

Now most experienced audio designers have this knowledge in their head
they know just about what will work.

This is why is no big problem if follow build instructions when you build
for example an audio amplifier by Professor Leach.
See reference link below!


Another good way, I did not mention:
Bigger Heatsink
When having these problems, it can often indicate,
you need a bigger HEATSINK for you appplication.
Big heatsink will not become as hot,
and so the overall effect will be just the same as
making Temp Sensor less senitive!


lineup post became half an article, again ... why cant I keep it short?
Appendix A.
Audio Related Things, by Professor Leach
http://users.ece.gatech.edu/~mleach/audiothings.html

[luixssj3]

Thanks for all the info!!!

I will trix with Vbe multiplier, I see some schematic have a diode on the emitter or a little resistance, I'll try it...

Distancing the Vbe transistor have some positive effect...

The heathsink is 20x10x5 centimeter, I think is OK, the themperature afther an hour is about 40°C you can still put your hand on it, maybe I'll try to put some fan...

[AndrewT]

Hi,
vertical FETs, is this what the amp was designed for?

Move the drivers off the main heatsink and keep them cool.

I prefer a small reduction in bias as the amp warms up and hopefully even less bias when it starts to overheat. A form of self protection.
The extra bias when cold is no problem and it helps get the devices up to temperature.

BTW. 200W into 4r from +-45Vrails is a bit optimistic. even 100W into 8r will be difficult (impossible?) to justify.

[jackinnj]

did you measure the Vce on the Vbe multiplier ?

I think that the 10K value is a bit excessive

[Nelson Pass]

I'm surprised that none of you have mentioned using a Mosfet
as the bias transistor in a Vgs multiplier - it's a lot less tweaky.

[darkfenriz]

Originally posted by Nelson Pass
I'm surprised that none of you have mentioned using a Mosfet
as the bias transistor in a Vgs multiplier - it's a lot less tweaky.

With bipolar drivers on heatsink?
Are you sure??

[GRollins]

From what I saw, the outputs were MOSFETs, not bipolars. Did I miss something?

Grey