I have a situation where it would be easier for me to mount my output transistors for EACH channel on separate heatsinks (1 output, 1 heatsink).
In other words, I have 4 output transistors, 2 per channel, but I need to mount each transistor on it's own heatsink if that is not going to be a problem... i.e.... could there be thermal tracking problems or some other reason I should not do this?
In other words, I have 4 output transistors, 2 per channel, but I need to mount each transistor on it's own heatsink if that is not going to be a problem... i.e.... could there be thermal tracking problems or some other reason I should not do this?
Common Heatsink For Each Channel
If I understand you correctly, then yes. What you are proposing is what many manufacturers do. The Left and Right channels in their stereo amplifiers have their own dedicated heatsink. For example, the left channel heatsink is located on the left side of the amp and the right channel heatsink is on the left.
Having thermal tracking is important on each channel, because each channel has it's own bias circuit and the output devices need to all be ideally at the same temperature so that they conduct identically. The transfer characteristics of output devices change with temperature. All the NPN and PNP output devices on a given channel would be on a common heatsink. They do not need to be on the same heatsink as an other channel.
If on the other hand you are saying that you want to mount each individual transistor on a separate heatsink thank NO. Not a good idea.
I hope that helped
Xsipower
If I understand you correctly, then yes. What you are proposing is what many manufacturers do. The Left and Right channels in their stereo amplifiers have their own dedicated heatsink. For example, the left channel heatsink is located on the left side of the amp and the right channel heatsink is on the left.
Having thermal tracking is important on each channel, because each channel has it's own bias circuit and the output devices need to all be ideally at the same temperature so that they conduct identically. The transfer characteristics of output devices change with temperature. All the NPN and PNP output devices on a given channel would be on a common heatsink. They do not need to be on the same heatsink as an other channel.
If on the other hand you are saying that you want to mount each individual transistor on a separate heatsink thank NO. Not a good idea.
I hope that helped
Xsipower
Thanks for the quick reply and yes, each transistor would have it's own heatsink. I failed to mention also that this is closely based on EPE Magazines 20W Class A design. I redesigned the driver boards and a few other small mods, but being Class A, I was afraid that thermal tracking between the outputs may be an issue.
I wouldn't be able to bolt them together. They are actually from Dell workstations CPU heatsinks. They are very large (not heatpipe), and I think the sq/in/deg would satisfy the heat requirements. The look really cool when arranged, but just one output per sink is all I can do. I'll keep looking around, I really need about a 12x6 vertical fin sink (2 of them). Thanks for the reply!
That's about the size Aiwa used in some of their receivers. Maybe not quite 6". Any repair shops close? We used to have tons of old heat sinks.
Hi,
It can and has been done (separate device heatsinks) in class aB.
In your case the CFP output transistors have nothing to do with
the bias arrangements (bias is Q10, Q11 and Q13) and there
is nothing stopping you using 4 separate output heatsinks.
There are no output device thermal tracking issues at all.
Q11 and Q13 have their own separate heatsinks and
Q10 the bias sensor is attached to one of them, Q11.
rgds, sreten.
It can and has been done (separate device heatsinks) in class aB.
In your case the CFP output transistors have nothing to do with
the bias arrangements (bias is Q10, Q11 and Q13) and there
is nothing stopping you using 4 separate output heatsinks.
There are no output device thermal tracking issues at all.
Q11 and Q13 have their own separate heatsinks and
Q10 the bias sensor is attached to one of them, Q11.
rgds, sreten.
Last edited:
That's good to know! I may re-think this and try the separate sinks. I redesigned the driver board so that all three (drivers and bias sense transistors) all share the same heatsink. I designed a board to accommodate the outputs, collector resistors, and filter assy separate from the driver board(s). Hope it turns out.
Hi,
FWIW its the type of amplifier that will work with any
substantial heatsinking even if its not quite enough.
Direct mounting the outputs won't change how
hot the heatsinks get, it will reduce the the die
temperature of the transistors to get that hot.
Die temperatures determine devices dying .....
The amplifier pretty much is a shameless ripoff
of D. Selfs ideas, not that there is anything much
wrong with that other than a lack of originality.
Sensibly though they ignored his utterly pointless
Class A bias controller, for something simple.
As described it does 20W class A into 8R
and clips at 25W into 8R. Into 4R it will
do 10W class A and clip about 45W.
rgds, sreten.
If it runs too hot, just reduce the bias current.
FWIW its the type of amplifier that will work with any
substantial heatsinking even if its not quite enough.
Direct mounting the outputs won't change how
hot the heatsinks get, it will reduce the the die
temperature of the transistors to get that hot.
Die temperatures determine devices dying .....
The amplifier pretty much is a shameless ripoff
of D. Selfs ideas, not that there is anything much
wrong with that other than a lack of originality.
Sensibly though they ignored his utterly pointless
Class A bias controller, for something simple.
As described it does 20W class A into 8R
and clips at 25W into 8R. Into 4R it will
do 10W class A and clip about 45W.
rgds, sreten.
If it runs too hot, just reduce the bias current.
Last edited:
I looked at Self's amp and almost opted for it, but this one seemed a bit simpler in design and "elegant" if you will. The specs looked very impressive. I still don't get the output arrangement though, B+ & B- to emitter's (respectively) and collector to collector for output... seems a bit strange, but I've seen many with this configuration.
I wonder what would happen if you just swapped the NPN & PNP outputs and put the B+/- to the collectors instead????
I wonder what would happen if you just swapped the NPN & PNP outputs and put the B+/- to the collectors instead????
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.