Hi everyone,
Just a quick question that i havent seen addressed elsewhere.
Consider a class B output stage for a design where simplicity is very important and and quality isnt. Every design I have seen uses a separete emitter resistor for the top and bottom half. Is this really necessary?
Wouldnt just using 1 for either the top -or- bottom half be sufficient? The temperature stabilization issue concens the total emitter resistance so that shouldnt be a problem.
Any comments?
thanks for the discussion...
Doug
Just a quick question that i havent seen addressed elsewhere.
Consider a class B output stage for a design where simplicity is very important and and quality isnt. Every design I have seen uses a separete emitter resistor for the top and bottom half. Is this really necessary?
Wouldnt just using 1 for either the top -or- bottom half be sufficient? The temperature stabilization issue concens the total emitter resistance so that shouldnt be a problem.
Any comments?
thanks for the discussion...
Doug
If you only put an emitter resistor in the top half then the feedback point can come from directly the emitter of the bottom half. That shouldnt be a problem, should it?
It does make the gains of top/bottom asymmetrical, but we are considering a design with emphasis on simplicity and not on quality.
Doug
It does make the gains of top/bottom asymmetrical, but we are considering a design with emphasis on simplicity and not on quality.
Doug
Regarding the use of a single emitter resistor, as long as your bias generator and drivers do not have any connection to the output point, it will work fine regarding the bias itself. In general, the bias generator will satisfy this, and for the drivers you need the modified EF version (where the emitters of the drivers connect to each other through a resistor, and do not connect to the output).
However, regarding performance, only one side will see degeneration WRT the output NFB pick-off point so you will get elevated second harmonic - which might even be received positively by the listeners
I have seen this done on a MOSFET amp but with the intention of matching the transconductance of the N-ch and P-ch FET better - the resistor was in the source line of the N-ch MOSFET.
However, regarding performance, only one side will see degeneration WRT the output NFB pick-off point so you will get elevated second harmonic - which might even be received positively by the listeners
I have seen this done on a MOSFET amp but with the intention of matching the transconductance of the N-ch and P-ch FET better - the resistor was in the source line of the N-ch MOSFET.
Thanks very much for the info.
My intended application is a high quality amplifier but I am not so worried about the increase in 2nd harmonic distortion because the amp has a H-bridge output stage. I think the bridge output stage will cancel at least some of the 2nd harmonic.
I think you are right about the MOSFET output stage. I could certainly see using unequal resistors there in an effort to equalize transconductance of the P and N channel devices.
Doug
My intended application is a high quality amplifier but I am not so worried about the increase in 2nd harmonic distortion because the amp has a H-bridge output stage. I think the bridge output stage will cancel at least some of the 2nd harmonic.
I think you are right about the MOSFET output stage. I could certainly see using unequal resistors there in an effort to equalize transconductance of the P and N channel devices.
Doug
deleveld said:
It's not a perfect way to do it but you can tailor transconductance to something more symetrical for a load and output power it is most likely to be used in.
Regarding H-bridge, yes this will do. However, since youa re going to have a larger voltage drop from one power rail to output than from the other power rail (Iout x Re), you may want to make your ground reference truly floating and offset it a bit to get maximum output voltage.
Wavebourn, you make an interesting point.
I think one should try to maximise output swing before clipping occurs. I agree thats not a very objective statement. If fact it sounds like an ethical statement and I am not sure why it seems natural to describe it in those terms.
From an objective viewpoint you could consider that symmetrical clipping increases the probability that a random signal wont be clipped. Because you never know whether the signal peak will be +ve or -ve. An amp can only relaibly process signals smaller than its lowest clipping level.
I think that with a single emitter resistor then the clipping difference could very well be on the order of a few volts. If the difference in clipping levels is on the order of fractions of a volt, then the point of diminished returns has been reached. But when the clipping difference can be a few volts I think that some attention should, if possible, be paid to fixing the problem.
Doug
I think one should try to maximise output swing before clipping occurs. I agree thats not a very objective statement. If fact it sounds like an ethical statement and I am not sure why it seems natural to describe it in those terms.
From an objective viewpoint you could consider that symmetrical clipping increases the probability that a random signal wont be clipped. Because you never know whether the signal peak will be +ve or -ve. An amp can only relaibly process signals smaller than its lowest clipping level.
I think that with a single emitter resistor then the clipping difference could very well be on the order of a few volts. If the difference in clipping levels is on the order of fractions of a volt, then the point of diminished returns has been reached. But when the clipping difference can be a few volts I think that some attention should, if possible, be paid to fixing the problem.
Doug
Nordic said:
Isn't it odd, that harmonics are one of the downsides to perfect symmetry?
OR
Aren't odd harmonics on of the downsides to perfect symmetry?
Which statement is the one you wanted to make...
As far as we go, I agree, assymmetry is built into our genome.... is that why we prefer some forms of distortion?
deleveld said:
In the order of a few volts? One tenth ohm against 4 ohm load?
Let's check... Suppose, output voltage is 10V, it means 2.5A current, and 0.25V drop on 0.1 Ohm resistor, it means 0.625 W of power.
Now, you add 0.25V more of voltage drop claiming that output power before clipping will be higher? No, it will be less. Unaudible less, but anyway less.
So, what is the reason?
deleveld said:
I put it in the top, but either way would have done. The resistor was 1 ohm, the output transistors were TIP31/TIP32 (that's what I had in my junk box). Driver transistor connections weren't an issue because there weren't any and the output devices were biased and temperature compensated with two 1N4001 diodes connected in series.
Pout was a massive 1.5 watts rms into 8 ohms.
Probably not what you were expecting!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.