Eliot Sound p3a oscillation

[ray_jay]

Hi guys,

I've have 4 x elliot sound p3a's ( 2 x 2channel boards )

http://sound.westhost.com/project3a.htm

Tested all 4 channels into 8 ohm dummy loads, 10KHz square waves all look ok.

These drive 2 x woofers per channel from an active crossover so I want to joint the two inputs of each board together.

One board now tests as before, so all ok.

The other board has an oscillation only on the -ve cycle of the square wave but ONLY if the input are joined together, if I test each channel separately all is ok as it was before.

If I separate the inputs with 2 x 100k resistors the oscillation disappears. ( but now I have a lower output )

Ok , the questions.

a) is the board that behaves just good luck ?, is joining the two inputs together a valid thing to do ?

b) If it is valid, could someone make a suggestion as to where to start looking for the problem, I will start tracing the board with the scope to see where the oscillation starts but I've just about had enough for tonight <groan>.

I understand basic transistor principles and can do basic audio calculations and use a scope ( sortof ! ) but this is all a bit beyond me. If it oscillated on both cycles I would expect wiring and layout, but just on the negative cycle ?. I'm using Rods boards and the

Attached is 10 kHz square wave , approx 4V p-p , oscillation is around 1uS cycle. It oscillates on the neg section of a sine wave too. sorry for the lousy photo but you will get the general idea.

Any thought or pointers would be appreciated,

thanks,

Ray

[ray_jay]

And 8v p-p, the oscillation appears differently here, doesn't seem to reach a maximum amplitude as it did in the 4v p-p scenario

cheers,
Ray

[jaycee]

Originally posted by ray_jay
a) is the board that behaves just good luck ?, is joining the two inputs together a valid thing to do ?

It is not recommended. You can probably get away with lower resistors though eg 10K.

Originally posted by ray_jay
b) If it is valid, could someone make a suggestion as to where to start looking for the problem, I will start tracing the board with the scope to see where the oscillation starts but I've just about had enough for tonight <groan>.

You may wish to try augmenting the 100nF rail capacitance with a 100uF capacitance also. I have personally found this cures any chance of an LC oscillator being set up on the power supply lines. Others (eg Professor Leach) may recommend removing the 100nF's and replacing solely with 100uF. It's a case of suck it and see, I guess.

[ray_jay]

Thanks jaycee,

I'll try the extra 100uF caps and see what happens.
Is it possible for just one supply rail to oscillate ?? I guess so but is it "normal" ?

cheers,
Ray

[ray_jay]

Fixed ( well the ringing is gone <g> ).

Added 1 x 220uF in to the -ve rail, all ok, went down as low as 1uF and that seemd to work as well, only one cap, in one rail, didn't seem to matter if it was the +ve or -ve, any clues to why guys ?

I'll just go ahead and put 100uF on both rails I think,

cheers,

Ray

[Upupa Epops]

Ray, try to connect .1 micro/ 100 V between both rails.

[ray_jay]

Hi Upupa,

thank you for the suggestion,

removed the 1uF I had on the _ve rail and put a MKP-X2 .1uF cap between + and - supply rails on the board, this solved the problem too although there still seemed a *little* ripple on the tops of a 20KHz square wave. 8V 10K square wave looked ok, I may check all these again later.

Now I'm confused, should I use 100uF on each rail *and* the .1 between rails ?, this would also give some small local reserve to this board as well because the PS lines to this board are 200mm long.

Why does the 0.1 between rails solve the problem?, I did look at the rails with a scope and they were ringing like crazy, but only with the -ve half of the squaure wave, with the +ve section there was only normal PS ripple. As I said this is all a bit beyond me, some LC resonance I assume but I'm willing to learn <smile>,

thanks,

Ray

[Upupa Epops]

Ray, MJL's are very fast devices and is easy to get them into oscilation, mainly if you have ( as you say ) long rails. For suppresion of oscilations is good in this case to get into base of output transistors low value serial resistor ( 2R2 - 10 R ).

[edl]

Dear Ray!

Solder 100uF elkos and 0,1uF folcaps on the rails.

[ray_jay]

Thanks for the help guys,it is appreciated.

I'll put 100uF between the +-rails and 0V ( already have 0.1 caps edl ) , measure square wave and then try some of the other scenarios. Upupa, it's not really practical to put the resistors into the base of the outputs so if some extra caps solve the stability issues I'll stop there. I think I will use this as an opportunity to have a bit of a play with 2 channels and see what I find,

thanks again,

regards,
Ray