Taming Ringing...Square-Wave Response - Page 2 - diyAudio
Go Back   Home > Forums > Amplifiers > Solid State

Solid State Talk all about solid state amplification.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 18th April 2004, 08:02 AM   #11
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
...looks like the designer tried to design a fast amp.
...fast transistors, small compensation caps.... compensation of
VAS reduced above 2.8Mhz, which might correspond to another
pole of the following current gain stage....

... combination of low compensation and relatively strong feedback...

All in all:
- difficult to calm down.
- already critical alignement mixed up by changing several
transistors.

I do not agree 100% that "let it ring except you can hear it", would be good. Is the ringing frequency very high? If yes, then this ringing
may also cause heat in the output stage as the power transistors
need some more microseconds for current decrease than for
increase...

For the first approach my proposals would be:
- increase the 10pf (C7 ?) in the VAS in small steps and
at the same time decrease R18 (keep C7 x R18 unchanged).
- if you reach C7=47pF and R18= 1.2kOhms without major improvement then check what happens if you keep 47 pf and go on reducing R18.

Good Luck
Markus
  Reply With Quote
Old 18th April 2004, 08:15 AM   #12
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
...there it is... 290kHz....
At this frequency the ringing may already cause some undesirable
heat, especially if you operate at higher levels.

So just be carefully.
How did you toast the old transistors?
Did you overload, short circuit, or somehow else stress the amp?
If it simply burned just during normal operation, without obvious reason, then I suspect that already the original design had ringing
issues.

15 years back when several fast amps entered the market and were
promoted to be the perfect solution, I saw an ultra fast high end amp cooking itself into death by high frequency oscilation, just because the wrong speaker wires were used. The capacitance of the cable caused additional pole and the compensation went unstable.....
Well, in fact I am quite conservative in this. I prefer non critical
compensation with an phase margin of 90 or more, even if I have
to work with less feedback for this....

Bye
Markus
  Reply With Quote
Old 18th April 2004, 08:35 AM   #13
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Default or simply reduce feedback

another approach might be to decrease the feedback,
simply by increasing the closed loop gain.

I.e. change the feedback resistor (R23? cant read it).
Giving a big step here from 33k to 68k should also result in
less ringing......

OK, now we have flooded you with tons of proposals which might
turn out to be all rubbish.... hopefully not, but you never know until
real life examination is done.
  Reply With Quote
Old 18th April 2004, 10:18 AM   #14
diyAudio Member
 
EchoWars's Avatar
 
Join Date: Dec 2002
Location: Left of the Dial
No problem guys...

ChocoHolic..

I bought the amp dead. Was in beautiful condition, but one channel fried. I found shorted transistors in the dead channel, and leaky E-C junctions in the other. The drivers were also damaged, but I had originals of the same type to replace them with. Also a couple of pre-drivers were shorted. At this point, all transistors are original except the input pairs, and the final output devices.

I changed the Zobel cap to .1f, and it made a small improvement. Not near enough to call it fixed, or even close. Going to leave it in anyway.

However, adding a 100pf cap from the base of Q5 to the collector makes a real improvement, and adding a 12pf cap across the feedback resistor almost eliminates the problem completely. With both caps in place, there is only a small amount of overshoot with a 10KHz square wave, and it settles down very quickly, about 1 cycle.

Does this fix make sense? Tempted to say 'well, if it works, it works', but is there another way I should approach this?

How about taking that 5.6K completely out and replacing with a 0-ohm jumper, and putting in a 100pf cap in place of the original 10pf C7?
  Reply With Quote
Old 18th April 2004, 06:25 PM   #15
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
... yes I think, your observations make sense.

Increasing the cap at Q5 defines the dominant pole in the bode plot.
But unfortunately the drawback of this are:
- higher TIM distorsions (input stage has to deliver more and signal in order to drive the slow follwing stages at high frequencies....)
- feedback at high frequencies is getting poor
This is the reason why this way of compensation has its limitations.

I would expect that the time constant of the 5k6 and the 10pf
are related to some other pole of the amp. So I would try to keep
the time constant of these two components constant. But probably
the difference will be very small and using a jumper instead of 5k6
could work as well.

12pf across the feedback is also fine, as it helps to bring the
phase of the feedback signal towards "earlier" (simply said...).
I think it would be OK to experiment with values up to
47pF here.

Looks like you may find a good combination of the compensation
across Q5 and the feedback.

Cheers Markus
  Reply With Quote
Old 18th April 2004, 07:11 PM   #16
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
....is there a reason why you don't want to set the 5k6 down to
560 ohms in combination with 100pF....?
Why would you prefer a jumper and 100pF?

For my taste the 100pF at this stage is already quite
high. The pole frequency will be given approx.
by R9 and C7 DIVIDED through the gain of the VAS!
R9 defines the driving resistance (if we neglect the output impedance of Q2). The gain of the VAS influences the resulting pole because
C7 has to be charged not only with the small amplitude of the
input voltage of the VAS input... No, it is seeing the full voltage
swing of the VAS output. More precisely the factor would be the gain+1, as the input moves to the other direction than the output...
....the same behaviour like a linear miller capacity.....
If we now caculate 100pF and 5k6 make 100kHz, dived by the gain of the VAS about 100 (?? no detailed idea just a number).
Tehn this means already above 1kHz the compensated gain of the VAS may decrease..... At 20kHz the voltage gain of the compensated VAS may be down to about 5..10 (again just a number)!
You can try to measure the signal at the input of the VAS and the
output of the VAS at 20kHz.
You will see at high frequencies the differential stage will have to deliver quite high levels. So a short fast signal at the input of the amp can drive the INPUT stage to clipping. TIM distorsion means transient intermodulation distorsion...... Even without hard clipping of the differential the distorsion of the input stage will increase with
the required signal level....

You see in frequency compensation, you cannot win.
You will always loose in one or the other way....


Don´t worry, these poor looking values are not a special
drawback of your amp. They are normal in most amps.

Many OP amps have the dominant pole even much lower than
1kHz.....

Bye Markus
  Reply With Quote
Old 19th April 2004, 02:28 AM   #17
diyAudio Member
 
EchoWars's Avatar
 
Join Date: Dec 2002
Location: Left of the Dial
Quote:
....is there a reason why you don't want to set the 5k6 down to 560 ohms in combination with 100pF....?
Why would you prefer a jumper and 100pF?
I'm just going by what has made an impriovement in my bench experiments.

Seems that if I use 100pf and 560 ohms...providing the same pole frequency...that I haven't gained anything. I also agree that a 100pf cap is a little large. The alternative would seem to be a compromise...perhaps a 47pf cap instead. Adjusting the resistor value would mean putting in a 1.2K resistor instead of the 5.6K, but I do not have the caps here to attempt this.

I'll order several different sized caps and see what they do.
  Reply With Quote
Old 19th April 2004, 02:42 AM   #18
fab is offline fab  Canada
diyAudio Member
 
fab's Avatar
 
Join Date: Mar 2004
Location: Canada
Quote:
Originally posted by EchoWars
No problem guys...

... At this point, all transistors are original except the input pairs, and the final output devices.

Just a thought:
I see that you have done a lot of trials w/o getting the performance of the working channel but have you looked at the input pairs transistors... This is not neglectable. I noticed that the scjematics includes a compensation cap at Q1. This cap value may not be appropriate for the new transistor?
  Reply With Quote
Old 19th April 2004, 02:57 AM   #19
diyAudio Member
 
EchoWars's Avatar
 
Join Date: Dec 2002
Location: Left of the Dial
The Zetex transistors I used as replacements are rated at 100MHz ft just like the original 2SA726 ones. I have also added another compensation cap to Q1, and it changed nothing. The key here seems to be compensation on Q5 and/or a compensation cap across the feedback resistor.
  Reply With Quote
Old 19th April 2004, 08:12 PM   #20
diyAudio Member
 
Join Date: Dec 2003
Location: Munich
Hi Echo!
100pF and 560 Ohms or 10pF and 5k6 is not setting the pole.
This time constant is setting the frequency above whitch the compensation network will stop it's influence (about 2.8MHz).
The pole is given by the driving resistance (mainly R9) and C7 and the
low frequency gain of the VAS (resulting in some single kHz).

But in fact, I would expect that you do not observe a major influence
if you use 560 Ohms or a jumper....
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Sine wave - Square & Triangle wave generator using Transistors / OP-Amps lineup Solid State 20 9th October 2006 12:15 AM
Microphone turns triangle wave into square wave? Circlotron Analogue Source 18 25th January 2003 06:44 AM
CD square wave pre-ringing a filter problem? Circlotron Digital Source 77 8th November 2002 02:48 AM


New To Site? Need Help?

All times are GMT. The time now is 05:13 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2