THIS output stage. For your consideration.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Disabled Account
Joined 2007
Hi,
Playing around with difference output stages, I tried this.

Comments? POS or worth trying?
 

Attachments

  • a1.gif
    a1.gif
    16.3 KB · Views: 919
You seem to be burning a lot of power in the drivers. Idle alone appears to be around 70 watts total in Q2,5,6,8,9,&12 for a standard AB output condition. Can’t tell were you want to idle the actual outputs at from the schematic given, that would make for additional heat as well. Better have some big heatsinks if you try it.
 
Disabled Account
Joined 2007
kaos said:
You seem to be burning a lot of power in the drivers. Idle alone appears to be around 70 watts total in Q2,5,6,8,9,&12 for a standard AB output condition.


Hi,
It was actually for a class B amp.
My idea was (probably silly) to have these driver transistors share about 10% (or more) of the load.
It works fine in the simulation, but it's nice to have those who are more educated give an opinion.

Am I wrong in thinking it has some voltage gain? Seems so in the sim.
 
Disabled Account
Joined 2007
These bias generator diodes (circled in pic) will track if mounted on the drivers heat sink, right? Should the drivers be mounted on a separate heatsink from these CFP outputs?
It's a little confusing, since I'm using driver transistors as outputs, but I'm referring to the actual drivers.
 

Attachments

  • a1.gif
    a1.gif
    13.7 KB · Views: 686
Hi,
look at the first CFP pair, Q2, Q3 & R2.
The Vbe of Q3 is about 600mV.
R2 sees all this voltage and tries to pass 300mA. That's your problem!
Change R2 and all the others from 2r0 to between 100r to 200r.

If you want to run the drivers hard, to keep up their speed, then maybe try as low as 30r.
 
Disabled Account
Joined 2007
AndrewT said:

If you want to run the drivers hard, to keep up their speed, then maybe try as low as 30r.


Hi Andrew,
Remember, I know not what I do. :)

It is my idea here to use these "drivers" as outputs, that they would share a substantial part of the output current. That's the reason for the low value collector resistors.
Simulation says..(that sounds Richard Dawson from Family feud "survey says..") that it will work. Simulation actually says it will work very well.
You know the simulator is my new best friend. ;)

I attach the full schematic to put things in context. This would be a subwoofer amp. Maximum 280 watts into 8 ohms.
 

Attachments

  • my_amp_classb_cfp_outputs.2.pdf
    38 KB · Views: 123
Disabled Account
Joined 2007
Going back to the simulator and changing the collector resistors to 4 ohms works a bit better. I am now seeing ~78mA through these drivers at max output.
The interesting thing is these drivers maintain a near constant current as the output power drops. The outputs (MLJ21193/94 will drop below these at low level.
This is what I see from the sim.
 
Hi,
maybe a bit more explanation is warranted.
Apply a signal to the base of the driver.
The driver Vbe varies with the signal.
The driver current varies with the Vbe.
The voltage across the 2r0 varies with the driver current.
So we have signal voltage at the driver base causes output voltage across the collector resistor.

Now the Vbe of the output transistor equals the voltage developed across that collector resistor. Change the voltage on the resistor and you change the voltage on the output base.
Increase the Vbe and you increase the collector current.

So, back to the beginning.
Vary the input voltage and the result is a current change in the output device.
But look at the collector resistor on the driver. It sees 600mV+-60mV (as a sensible maximum range). That's +-10% of variation from constant current. The output device will change Ic from near zero @ 540mVbe to near destruction @ 660mVbe.

Now look at the driver IC vs Vbe graph.
How much voltage change (delta Vbe) is required to change the driver Ic by +-10%?

Yes, tiny changes in driver current result in enormous changes in output current. Your sim should show that, if you apply a signal and monitor the IC of both driver and output.
 
Disabled Account
Joined 2007
AndrewT said:

Yes, tiny changes in driver current result in enormous changes in output current. Your sim should show that, if you apply a signal and monitor the IC of both driver and output.

I am trying to understand what it is you're saying. What I'm lost on is why is the sim showing better performance with the lower value resistor? The sim works by the rules that govern a real circuit.
I have something to show here. This is the CFP operating with the 4 ohm resistor and the 200 ohm. Output power is nearly identical, but THD is much lower with the 4 ohm collector resistor.
Also, current through the output is reduced, with the "driver" conducting the difference.
 

Attachments

  • a1.gif
    a1.gif
    18.6 KB · Views: 387
You have a choice.
learn how the circuit operates from first principles (pencil and paper), or use the sim to investigate the operation.

But, ignore the complexities of distortion etc, until you understand enough about it's operation to ask the right questions and then able to interpret the answers the sim gives to those questions.

You are forgetting to look at what q16 and r22 do TOGETHER.
They are effectively a CCS drawing near constant current through the driver, whereas the output is controlling all of the variation in output current.
 
Disabled Account
Joined 2007
AndrewT said:
You have a choice.
learn how the circuit operates from first principles (pencil and paper), or use the sim to investigate the operation.

But, ignore the complexities of distortion etc, until you understand enough about it's operation to ask the right questions and then able to interpret the answers the sim gives to those questions.


When I was younger, I could have learned this stuff easily, but I didn't have the time (or thought I didn't). Now I'm older, I have more time, but learning is a bit tougher. I have some motivation and the simulator. I'll learn by osmosis, gradually.

It's almost as if you are saying I shouldn't use the simulator until I understand every last detail, like I might hurt myself with it or something.
A straight yes or no will do, and the reason why. I understand where you're coming from, but it's not the way I do things.
Example:
Tell me I can't do it (not because it can't be done, but just that I can't do it) , and I'll do my darnedest to do it.
Tell me "no it won't work and here's why" and I'll try to figure out why, and eventually will find out why.

Long and short is that I need a specific thing to focus on, not bits and pieces.
I really appreciate your help and attention Andrew. Please don't take what I had to say above in the wrong way.:)
 
Dear MJL21193,
I know it's a while ago, but just stumbled over this thread and felt a comment coming on...
The first of Ur outlined topologies looks the best to me, providing you add some say ntc's for temperature compensation in order to prevent thermal run away.
The second topology compared to Bryston to me looks like having all possibilities of instability both thermal, current hugging and stability wise.:eek: On the second pair's of MJL's the resistor's should at lest move to the emitter side of the transistors in order to prevent some current hugging.:flame: Some base resistor on all MJL's would probably also help with linearisation.:cool:
Just some 'food for thought for U' with the best intentions and regards:)
a1greatdane
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.