Apparently, I was incorrect that PMA was rewarded by Parasound. However, his associate was. Sorry about that.
There is a paper in the AES preprints that talks seriously about two stage compensation. I suspect that it might have a complex transient response, because of this. The circuit I designed was for continuous tone operation, so it didn't matter.
There is a paper in the AES preprints that talks seriously about two stage compensation. I suspect that it might have a complex transient response, because of this. The circuit I designed was for continuous tone operation, so it didn't matter.
scott wurcer said:
I have a question for Scott: I wonder how the first graph looks like for Ic greater than 1mA. Is hfe still constant between 1mA and 20mA?
In particular I'm interested in hfe linearity when operating the MAT02s at around 10 - 12 mA.
Thanks for any info you might be able to provide
The 1978 paper presented at the AES (LA) that goes into 2 pole equalization is:
'Design and Construction of High Slew Rate Amplifiers' by Takahashi et al, from Sansui.
As I feared, the transient response is suspect, with 2 pole compensation, BUT that might still be OK. It might have gone to the JAES, but I don't have that reference at this time.
'Design and Construction of High Slew Rate Amplifiers' by Takahashi et al, from Sansui.
As I feared, the transient response is suspect, with 2 pole compensation, BUT that might still be OK. It might have gone to the JAES, but I don't have that reference at this time.
Sidiy, the MAXIMUM rated current for the MAT02 is 20ma. However the max F(t) is at 10 ma, a departure from many transistors that I have noted. Usually, peak beta tracks closely with F(t) in my general experience. I would be careful, if I were you.
AndrewT said:
Hi Andrew,
1. By masking the overshoot using a LP filter (certainly not my style).
2. By using TMC instead, if applicable. However, TMC, opposed to TPC, only reduces distortion from the output stage.
Increasing the gain bandwidth, and reducing the amount of negative feedback allows SPC once more. I don't know how to get around it.
Dimitri, how about feedforward compensation?
Dimitri, how about feedforward compensation?
engineering is always about tradeoffs - saying you will not accept any overshoot closes off the possible gains elsewhere in the amplifier performance
pushing more than single pole gain into the outer loop will extract some cost in overshoot/frequency response flatness at high requencies - but for typical SS amps this occurs way beyond "audible" frequencies
10-20% overshoot that can be compensated by ~200KHz lpf would seem to be a possibly good trade off for >20 dB more gain over the entire audio spectrum - leading to as much as 100x reduction in input pair distortion from 3rd order gm nonlinearity
an additional "cost" is considering overload recovery of higher order gain systems - "nonlinear compensation" in the form of diode clamping or more complex schemes enable higher loop gains with reasonable overload/clipping response:
http://www.diyaudio.com/forums/showthread.php?postid=512806#post512806
for tailoring loop gain for "Bode's maximum gain" B J Lurie's book and Mitchell's papers are useful:
http://www.luriecontrol.com/index.htm
http://ukacc.group.shef.ac.uk/Control_Conferences/Control2004/Papers/063.pdf
pushing more than single pole gain into the outer loop will extract some cost in overshoot/frequency response flatness at high requencies - but for typical SS amps this occurs way beyond "audible" frequencies
10-20% overshoot that can be compensated by ~200KHz lpf would seem to be a possibly good trade off for >20 dB more gain over the entire audio spectrum - leading to as much as 100x reduction in input pair distortion from 3rd order gm nonlinearity
an additional "cost" is considering overload recovery of higher order gain systems - "nonlinear compensation" in the form of diode clamping or more complex schemes enable higher loop gains with reasonable overload/clipping response:
http://www.diyaudio.com/forums/showthread.php?postid=512806#post512806
for tailoring loop gain for "Bode's maximum gain" B J Lurie's book and Mitchell's papers are useful:
http://www.luriecontrol.com/index.htm
http://ukacc.group.shef.ac.uk/Control_Conferences/Control2004/Papers/063.pdf
PMA, talking 'sense' here will not get you anywhere. We are talking about 2 pole eq. SO that we can use LOTS of feedback. There are certain side effects, however.
This is just one more tool, not panacea. I have some anecdotal experience, obtained by reverse engineering of Japanese feedforward amps.
john curl said:
John
Every circuit has its pros and cons.
dimitri said:
Dimitri
Could you please explain your thougths?
dimitri said:
Indication of lower stability margin, especially with complex load. The input LPF just masks this. Input is not the only way how to excite possible oscillations.
PMA said:
Incorrect, unless you have a new definition of "stability margin".
An amp without overshooting may have a lower stability margin than one with overshooting.
Some claim they can hear an overshooting amp even LPF masked, but this is already a different story.
- Status
- Not open for further replies.