Overschoot problem with amp

[Rudy]

Hello,

After building this amp and after some measering, i noticed a problem in my amp, witch is build on a breadbord, the problem is that everything looks ok to verry mutch ok except for the squarewaves, there is amassive amount of overschoot, not only at 10KHz, but also at 1KHz, witch is not normal at all to my knowing.

does somebody has any clew why and how to solve it ?
i can try about annything becouse the amp is still in the "investigatation" and "optimalisation" stage.

some info
first stage quisite current = 2mA
second stage quisite current = 8mA
VAS quisite current = 5mA
Output stage quisite current = 300mA
sinusoutput is ok from 5Hz ... 500KHz @ fullpower
supply is based on 500VA transformer with 4 *10.000uF and 2 * 470nF
distortion < 0.01% ( i can't measure any lower for now ) @ 1...20Khz



@ 1KHz ( 2V per devision )


@ 10KHz ( 2V per devision )


Rudy

[PMA]

Play with components on positions C15, R31 and similar and also C14, C18.

[djmiddelkoop]

This problem has to do with the relation between (to much)openloop amplification and the amount of feedback.

What you see is the feedback trying to correct ,read fighting, the amp while the amp amplifies to much (openloop).

If you lower the openloop amplification and at the same time make the amplifier more lineair before adding feedback, the sound will improve and the overshoot will mostly disappear.

The first and second stage complies, but the third (T9 and 10) doesn't.

Adding a resistor to either side of C24 to GND will make this stage more lineair and it will amplify less. Add a resonable value capacitor in series with this resistor to not influence the DC offset of the amp.

A good starting value would be 10k with 1uF.
If the results are not satisfying, lower the resistor value and at the same time increase the capacitor value.

succes,

Dick.

[Rudy]

thanks

@ PMA
this doesn't seems to change mutch for the squarewaves

@ Dick
The problem is indeed T9 and T10, but the solution you gave does not resolve the initial problem, and knowing that the VAS stage is only 5mA, i do not want to draw any current more then neccesary out off the vas ( this also adds distortion, and a lot more then i tought ).

There does not seems to be an easy sollution to this, as whenever i get a good sqaure out off the amp, the sinus distortion goes way up to 0.05%, witch is not acceptabel.

[MikeB]

Hi !

The reason for this overshooting is incorrect feedbackcompensation,
someone simply "forgot" to place a capacitor in paralell to r17,r18.
Something like 33pF should do it. You need to test/sim/measure
with different values. If this cap get's too big, the amp will oscillate.

Mike

[Rudy]

@ Mike

this does indeed helps a LOT

until witch frequentie should overschoot be invisibel ? 10Khz ? 20 KHz ?

thanks

ps. that "someone" is me

[Cortez]

Rudy !
Please make a "before-after" test with music and check the
results in your amp with this feedback correction 33pF cap and without it !
Wait for your results...
Thx !

[MikeB]

Originally posted by Rudy

until witch frequentie should overschoot be invisibel ? 10Khz ? 20 KHz ?

thanks

ps. that "someone" is me

Hmm, there never should be any overshooting, no matter what
"frequency" a squarewave has, the rise/falltime is allways the same !
For capacitive and highly inductive loads this is something else, but
for pure resistive loads there's no "reason" for overshooting.
Me personally believe that overshooting is a soundquality-killer.
(That's at least my experience)
The "squarewave" should have rounded corners. It's possible that
some other caps in your amp are too big, for example the 1nF in
second diffamp could need a resistor in series (~100ohm). (c14/c18)
And you could place them on collectorside of these bjts, caps at
emitters are risky, these could create local oscillation.
Then you could be able to reduce c21/c17.
The same is for first diffamppair, you could place a single RC between
collectors, like 220p+100ohm.

Are the outputdevices darlingtons ?

EDIT: I made the experience that finetuning feedbackcaps has a
very big impact on sounding of the amp !

Mike

[Rudy]

Mike

I am back to the simulations and seeing wat is the best solution to the problem, but for now no "good" solution yet.

About the output devices, they or darlingtong's with hfe>5000 and have a resistor of 0.22ohms already inside them, they also have the temperature compensating diodes onboard.

Rudy

[MikeB]

Hi !

Yes, it's a bit tricky, you need to play around with all these caps,
my experience is, that there's a ratio between this cap (the 33p)
and all others. Only if the amp is fast enough, you will suceed in
having a stable amp without overshooting and reasonable bandwidth.
I think, this is the reason why most good sounding amps are simple.
I make finetuning of feedback always with squarewave and acsweep.

I attached a shot of a 20khz-squarewave from my amp, but only
sim'd, i have no scope... (damn expensive these things !)

Mike