Hi all,
What should be the correct location for Vbe transistor in a CFP output amplifier , should I connect it to the heatsink for the main output transistor or should I connect it to the drivers..Can't the driver the bias transistor and the output transistor all sit in on a common heatsink....
This is what is happening I have constructed a P3A with all the output transistor driver and bias on a common heatsink...When I bias the transistor as per recommended 75ma the amp keeps drawing more and more current as it heats up ...And the bias keeps increasing ... So I went ahead and seperate the bias transistor and drivers to a second smaller heatsink but it does the same thing ... So where should I put my bias transistor...
And my heatsink is a bit small so is that issue ..??
Regards
What should be the correct location for Vbe transistor in a CFP output amplifier , should I connect it to the heatsink for the main output transistor or should I connect it to the drivers..Can't the driver the bias transistor and the output transistor all sit in on a common heatsink....
This is what is happening I have constructed a P3A with all the output transistor driver and bias on a common heatsink...When I bias the transistor as per recommended 75ma the amp keeps drawing more and more current as it heats up ...And the bias keeps increasing ... So I went ahead and seperate the bias transistor and drivers to a second smaller heatsink but it does the same thing ... So where should I put my bias transistor...
And my heatsink is a bit small so is that issue ..??
Regards
Big Jim ,
Thanks for the reply , but why does my bias keep on increasing as the amp heats up , I have tried both variations .. VBE on main heatsink and VBE on driver heatsink... Is this normal ??
Thanks for the reply , but why does my bias keep on increasing as the amp heats up , I have tried both variations .. VBE on main heatsink and VBE on driver heatsink... Is this normal ??
You need a MUCH bigger heatsink.
If its too small, no amount of bias compensation can stop thermal runaway and device destruction.
If its too small, no amount of bias compensation can stop thermal runaway and device destruction.
If you bias a CFP at 75mA (which is very high for CFP and takes you well into gm-doubling territory) then there is an argument for putting the Vbe transistor so it senses the output temperature. At more normal CFP bias (around 10mA) it should sense the driver temperature, as the output will take very little current.
Exactly. The page where P3A CFP amp construction is described.
P3A is mentioned by the OP.
The comment I was referring to is next to the red paper pencil icon on that page.
P3A is a great amp btw and is playing as I type.
Note - Vbe spreader transistor (Q9) is thermally coupled with one of the drivers in P3A.
This is important for normal operation.
I have checked there is no oscillation , is stable even with square wave at >20khz with a load of 4E...
Attachments
i have tried putting the vbe on the driver transistor but still the bias keeps on increasing as the heatsink heats up ....
If you put a picture up of your amp we can see how big the heat sink is.
I have had CFP's running with no heat sink and the output transistors didn't heat up. But I do tend to use minimal bias current.
Why should the Vbe multiplier in a CFP OPS amp preferably be thermally connected to the CFP driver transistor instead of to any of the slave transistors (Q7, Q8)?
60-80W Power Amplifier
60-80W Power Amplifier
find the attached picture , i have tried using a bias of 10ma the bias drift definitely slows down a lot as compared to 75ma.. but isnt it the work of the VBE transistor to track the temperature rise in the heatsink and adjust the bias accordingly.... I also found that putting a fan to heatsink makes the bias becomes dead steady , no drift at all ... then I went ahead and tried something that i shouldn't try , i put my soldering iron very close to the VBE transistor and the bias drops instantly , so that means the VBE trasistor is working as it should then, why doesnt it track the heatsink and keep the bias stable .....?? this i have tried while mounting the VBE directly to the main heat-sink as well as with a separate heat-sink with the driver
And I have made the PCB such that i can try other CFP variations , just to see how things work ... thats why some points are not populated ...
Attachments
Last edited:
Hi sekhar,
What is the bandwidth of your digital oscilloscope? Oscillation can occur in the MHz or tens of MHz. You've got to be very careful making assumptions using a digital 'scope. Your oscillation could be there, but be invisible. Do you have an analogue 'scope you can try as a sanity test?
-Chris
What is the bandwidth of your digital oscilloscope? Oscillation can occur in the MHz or tens of MHz. You've got to be very careful making assumptions using a digital 'scope. Your oscillation could be there, but be invisible. Do you have an analogue 'scope you can try as a sanity test?
Because bias current control has limits beyond which it can't control the bias current. That's why small heat sinks can cause thermal run away. Same if you block the airflow or raise the bias current too high. Previous posters asking about the heat sink size were right to do so. Your heat sinks are too small for 75 mA of bias current. You proved that yourself.
-Chris
thanks i was thinking the same thing, that my heatsink is too small but it happens even with 10ma bias .... and my scope has 50mhz bandwidth ...
Hi sekhar,
Okay, you just have to be careful with your sampling frequency (time / Div). Digital scopes bring a whole bunch of complications to the table and they are not like analogue scopes at all. I keep an analogue 'scope on my bench, but my digital one is an Agilent. The only thing I use the analogue scope for these days are eye patterns and similar waveforms. The Agilent will do them, but analogue scopes are cleaner. So are the current Keysight digital 'scopes. Mine is quite old now.
-Chris
Okay, you just have to be careful with your sampling frequency (time / Div). Digital scopes bring a whole bunch of complications to the table and they are not like analogue scopes at all. I keep an analogue 'scope on my bench, but my digital one is an Agilent. The only thing I use the analogue scope for these days are eye patterns and similar waveforms. The Agilent will do them, but analogue scopes are cleaner. So are the current Keysight digital 'scopes. Mine is quite old now.
-Chris
Agilent that's nice , mine is a cheap chinasium one .... i am planning to get a analogue one , but in india everything so damn costly due to the import duties 🙁
Hi sekhar,
The best thing you can do is experiment with your 'scope. Try different signals and vary your sampling frequency in the 'scope. Try situations you know are wrong as far as frequency setting are.
If you have switchable probes (X1/X10), use epoxy and glue them in the X10 position. Also, buy some X100 high voltage probes that are available on Ebay.
-Chris
The best thing you can do is experiment with your 'scope. Try different signals and vary your sampling frequency in the 'scope. Try situations you know are wrong as far as frequency setting are.
If you have switchable probes (X1/X10), use epoxy and glue them in the X10 position. Also, buy some X100 high voltage probes that are available on Ebay.
-Chris
- Status
- Not open for further replies.