Adcom GFA-585 behaves strangely

[BrianEno]

Dear all,

Have a Adcom GFA-585 that does something odd.
When I start it up it takes .460 Amps from AC.
But when I turn the volume up so that it exceeds this point it jumps suddenly to .706 Amps AC intake. When I turn the volume completely counterclockwise it still takes the .706 Amps from the AC. When I switch the Off and On it starts again with .460.
The bias is exactly on spec: 24mV between both testpoints and I measure only a couple of millivolts DC that do not increase when the amp takes more AC in.
I installed a Zobelnetwork with a 10 Ohm Moxresistor and a 220pF highgrade cap that did not solve it.
I also completely cleaned both inputboards from leaked electrolytic fluid from the Elna 'Long life' caps ans replaced all electrolityc caps, the biastrimmers, the DC servo chip and a couple of affectedr resistors and transistors.
It plays well now but behaves odd.

Any suggestions?
Thanks

[djoffe]

check for oscillation...do you have a scope to look for supersonic oscillation?

[BrianEno]

Originally Posted by djoffe

check for oscillation...do you have a scope to look for supersonic oscillation?

Hi Djoffe,

I'm in the progress of acquiring one.
Thought that the Zobel network would kill the oscillation...

[CBS240]

Not necessarily. Zobel filter is designed mostly to prevent RF garbage from entering FB loop but acts as an output filter to very high freq. It can be damaged by large RF signals. Check the resistor in the Zobel of the channel if you find it to be occilation, it could be open. With an RF occilation I smoked a 1/2W resistor once, burned it in half, and it was in series with a only 100nf cap!

[BrianEno]

Thanks CBS240
I installed a 5 Watt noninductive MOX resistors that are still in good condition.
I really have to buy a scope.

[CBS240]

yes. It is hard to tell what is really going on without one. I have heard that cheaper PC based solutions are getting much better. I paid ~$400 retail for a 30MHz analog scope, but it was brand new. Check Flea bay, but use caution.

[djoffe]

Your zobel may be part of the problem. The 10 Ohm is quite typical, but the 220 pF is MUCH lower than is typical. A typical cap value for the Zobel is in the range from around 0.022 uF to 0.1 uF. It would be worth a try with the larger cap...

[BrianEno]

Originally Posted by djoffe

Your zobel may be part of the problem. The 10 Ohm is quite typical, but the 220 pF is MUCH lower than is typical. A typical cap value for the Zobel is in the range from around 0.022 uF to 0.1 uF. It would be worth a try with the larger cap...

Sorry made a mistake there.
It's a 220nF cap and maybe that's even to big being 0.22uF.

I found out that there's already a Zobel situated on the inputboards with a
.047 cap and a 6.8 Ohm resistor....

I also discovered that when I disconnect the AC energymeter and connect my Variac wich has an Amperemeter that there is no jump at all...

I did however turn up the default value of 24 mV across the testingpoints for bias to 48 mV. The temperature of the coolingfins stays at 48 degrees Celcius and having a pretty efficient small speaker the amp will stay in Class A mode most of the time.
Read some articles from Nelson Pass regarding this matter and decided to crank up the bias. The overheatingsensors are set for 85 degrees Celcius so I reckon when I stay under 50 degrees there's no problem.
I also connected two big powertransistors of 5 Ohm and had the amp conduct 40 Volts across them for a couple of minutes, that's 320 Watts continiously, and the amp had no problem with that. After I let the amp cool down for a while the bias still measured the same before the experiment.
DC values are neglible low 1 or 2 mV.
I think the amp is pretty healthy now after being "mistreated" with electrolytic fluid from those Elna caps.
After I'm convinced that everything is okay I will test it on my main system and swap it temporary for my Threshold mono's to drive my fullrange Acoustats.
I'm anxious to see if the amber distortion leds will come on soon are late in that situation.