Amp oscillates with new speakers

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have changed the pair of speakers recently and the power amp maybe oscillating when it is connected with them. Sometimes, there is a tone of about 300 Hz which lasts for half of a second after the power amp is switched on. And it reproduces the 300Hz range very muddily and boomy. This is more obvious when the volume is high.
Could it be oscillation and can I fix the problem with a zobel network? What should be the value of R and C if my speakers are rated 6 ohm.
 
My equipment lineup:

CDP: SONY DVP S7000
PRE: DIY Ground Grid
Power: DZ phase 120
Speakers: Pioneer S-A4SPT-PM

My previous speakers were Hales Concept 2 and they never made this problem.
 

Attachments

  • dzphase120.jpg
    dzphase120.jpg
    42.5 KB · Views: 412
When I have answered you, I didn't know your amplifier's schematic, now, I have seen it. You should make two things,
if the problem is the amp stability (like it seems being):

1) add an antioscillation network.

2) upgrading the (simple) frequency compensation system in your amplifier.

the second way is simple because the frequency compensation system is simple in the amp. Infact, this system is based on two capacitors only, C1 and C2 in the schematic, in the theory, for giving greater stability to the amplifier it would have to be sufficient to increase the value of these capacitors.
I do NOT assume any responsibility and any liability for these modifications , even if I think that it's a good strategy (I have more than ten years in analog design).
First, in the theory, you should try to increase X2 this capacitor value (the c1 and c2 values is the same, I suppose) , if the amp doesn't oscillate, after you can descrease it. If the amp oscillates with X2 capacitor value, you must increase it again until the amp doesn't oscillate.

I would like to know the components values on the schematics, if it's possible (I doesn't interest to construct it because I suppose that it has the high not excellent), but I think that is a good idea.

For the first point , many people think that to add and inductance to the amp out isn't an audiophile thing, :) .
I think that some amp oscillate without it with some speakers and It's a good idea to apply an swicth for bypassing the inductance for the other speakers, because often is an bad idea to supercompensate the amp for a few speakers.

if you don't like the second point I will write about the first point.

ppa
 
on the output 0.7uH in parallel with 10R to the speakers -- I wind 12 turns of enamel wire on a 1/2 inch dowel -- ymmv.


also 300Hz oscillation -- sounds like something is fritzing up in the power supply for a moment -- although this frequency sounds high and that little fuzziness at turn-on only lasts for some hundreds of milliseconds -- with a spectrum analyzer you can see the power supply harmonics under load and how they change -- consider that one of your filter caps has a problem, bad solder connection in the power supply, bad decoupling caps etc. you don't need a megabuck HP analyzer, a freeware soundcard spec analyzer on the amp output will tell you how you are doing in eliminating the problem.

fwiw -- If you put a 330 ohm resistor in series with C3 it will lower the Q of the feedback compensation -- this will help oscillation in the tens of kHz arena. sometimes a cap with a high Q can be problematic here -- i know it's not the problem at hand, just something I found useful.
 
Just a possibility that the crossover in your new loudspeaker has a high Q or capacitive crossover circuit without series resistors on the section involved. Do you have capacitive LS leads ?

You could try a 0.22 ohms 3 to 5 Watt wirewound in series with your amplifier output terminals (not ground as well).

Cheers ............ Graham.
 
the ymmv's LC network (0,7 uH and 10 ohm) is good , this is the typical value in several amp, I would add also to the LC network a polyester or polypropylene 100nF capacitor in series to a 10 ohm 5W resistance between the output and the out ground to avoid local oscillation on the mosfet final stage.
Consider you that without the LC network the amp can oscillate,
even if the RC net it's added.



ppa
 
just out of curiosity...... what is the damping factor of the amp? since it uses global feedback i would assume it's at least 100..... you could also have a problem with the dc servo, since the dc servo is also part of the ac feedback loop. usually dc servos are in their own separate feedback loop, and it's rare that they cause problems, but if your dc servo happens to have some gain at 300 hz along with a phase inversion, you could have a low frequency instability problem. post the schematic for the dc servo.....
 
Dear All,

I have read the methods you guys provided and I consider that adding network to the output is more convinient to experiment.
But I don't think there are bad caps in the power supply because the problem exists in both channels simutaneousely in playing CDs. And it has no problem with the previous speakers so I think the interation between my new speakers and the amp introduces this problem.
I am posting all the component values and the dc servo schematics for your analysis before I take the final method. Thanks a lot!
 

Attachments

  • phase dcserv.jpg
    phase dcserv.jpg
    39.4 KB · Views: 176
for knowing if LC network will work well it's sufficient that you connect a resistance of a few ohm in series to the loudspeaker, if the sound is cleaned up then can you connect the LC network.
Obviously consider you that one resistance in series to the loudspeaker changes its frequency response.


ppa
 
What value are the resistors in the power supply filter for the input circuit? That kind of front end has a pretty bad power supply rejection, all designs I have seen have used regulated power supplies for the front end... Does the frequency change if you change the value of those resistors or filter capacitors for the input stage?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.