Oscillation under load

[Craig405]

Hi,

I posted a question on this a few months back, it still remains an unsolved problem for me though.
I built a Quad 405-2 amplifier, with the well known mods done to it ( OPA134, MJ21194's, more capacitance)

Also an HT of 55v (5v above the original design).
It works very well without a load doing 80V pk-pk, 100w into 8 ohms.
But as the frequency goes up (roughly at 10khz and above) or as the load is reduced from 8 to 4 ohms the oscillation gets worse until the waveform becomes almost jagged. With oscillation in the order of several Mhz on positive cycles, crossover distortion becomes visible too.

I actually gave up on it before but have been away at university for a while and now want to get it working.

Somthing is wrong but i cant find out what, i have snubbersied, ferrite beaded, base stoppered but still cannot solve the problem.

Cheers
Craig

[Craig405]

ok i just got rid of the oscillation problem by putting 680nf in // with the speaker now it doesnt oscillate at all even under full power into 3 ohms.

I still get jagged waveforms on high frequencies.

I am sure that the capacitor solution is a terrible one but it works well enough and doesnt attenuate the signal within the 20khz range at all so im happy :P.

What might be causing the jagged sine waves?

Cheers
Craig

[anatech]

Hi Craig,
A few things can cause that. PCB layout is one of them. Experiment with the compensation caps. If you run a frequency sweep, you may see some peaking.

Does the current increase as you sweep the frequency up?

-Chris

[clem_o]

Hi Craig,

Agree with Chris - could be the layout of the amp is causing instabilities. Could you post what the "jagged" waveform looks like? Doesn't have to be a true picture, even just a hand-drawn figure might help.

Also, try 'scoping the supply waveforms going into the OPA134, see if you can spot some niggles when the oscillation happens...

Cheers!

Clem

[Craig405]

As i remember from last time i tried to repair this, the oscillation i get appears all over the PSU grounds and on the output at about 5mV pk to pk but gets worse at higher frequencies.

I am rebuilding the PSU and rewiring the lot today to eliminate any possibilities from this area.
I will try and do a sweep of the frequency response and get current readings from the supply as soon as its back in one piece.

The only problem i forsee in the sweeping is that i only have one ancient function generator and i think i need 2. Ill try and get some results today or tomorrow.

Also here i a picture of the waveform at higher frequencies with the oscillation also present.


Thanks
Craig

[phase_accurate]

Could you supply a scan/picture of the PCB layout ?

Regards

Charles

[jan.didden]

Craig,

Did you have the usual 0.1uF + 10 ohms across the output? If not, try that, or even keep the 0.68uF but put a 5 or 10 ohms in series. It probably still gets rid of the oscillation without loading the amp in the audio range too much.
Did you put in an output series inductor?

Jan Didden

[consort_ee_um]

Hi Craig,
I see you have been struggling with this amp for months now. This link is a brilliant bit of work on modding the 405 which may give you some info on how to solve the problem.

http://www.dc-daylight.ltd.uk/Valve...on/405_Qw_6.pdf

I would echo Janneman's thoughts on the zobel network but you might need to make the resistor lower than the 5 to 10 ohms he suggests. The standard 0.1uf and 10 ohms will not do much.

The waveform you have posted is classic slew rate limiting. I note from the above link that the 405 does not have good figures for its slew rate spec.. Perhaps you are expecting too much from the design. It would be worth noting the slope of the straight bit of the distorted sine waveform (in V/uS) and comparing that with the spec.
The presence of numerous inductors and capacitors for compensation shows that the design is touchy and any deviations from non-standard may be troublesome oscillation wise!

[consort_ee_um]

sorry, got links wrong:

http://www.dc-daylight.ltd.uk/Valve-...n/405_Qw_6.pdf

http://www.dc-daylight.ltd.uk/Valve-...-405-mods.html

[Craig405]

Thanks for replies everyone.
I have rewired the PSU and one channel alone with very short thick wire so thats fine now.

Here are current readings taken from the positive rail at various input frequencies;

10v pk-pk output into no load;
100hz = 89.7mA
1kHz = 89.8mA
17.5K = 89.4mA (first obvious signs of triangulation appear)
22.5K = 87.9 mA
60k = 86.3mA

with 5 0hm load, no input measured 86.5mA idle;
100hz = 90mA ( wave already looks bad)
10Khz = 89.9mA
20K = 89.2mA
25K = 88mA
30K = 87.3mA
60K = 86.5mA

As for the op amp supply it looks horrible even though it is transistor regulated, the input signal appears there with lots of oscillation all over it.

Jan yes i have the 0.1uf 10 ohm network in place, and have tried a few values of output inductor// resistor too but i still get the same trouble.

consort_ee_um thanks, i pretty much built my amp using tips from those sites they are very good.

I cant find my PCB layouts, i hadnt backed them up and probably deleted them a while ago (never thought id need them again..doh)

Thanks
Craig