Well you may have made major steps forward with Kal's replacement boards.
You yourself stated that your boards had seen a little too much heat
and the tracers were in less than ideal state.
From what I understand, (and I understand hammers better than circuitry)
Parasitic oscillations can be introduced a number of ways and one of those ways is thru poor signal path.
Overheated boards with rough tracers could certainly be a factor in inducing an amp toward oscillation.
=================================
Parasitic oscillation:
https://en.wikipedia.org/wiki/Parasitic_oscillation
Mitigation
Several measures are used to prevent parasitic oscillation. Amplifier circuits are laid out so that input and output wiring are not adjacent, preventing capacitive or
inductive coupling.
A metal shield may be placed over sensitive portions of the circuit.
Bypass capacitors may be put at power supply connections, to provide a low-impedance path for
AC signals and prevent interstage coupling through the power supply.
Where printed circuit boards are used, high- and low-power stages are separated and ground return traces are arranged so that heavy currents don't flow in mutually shared portions of the ground trace. In some cases the problem may only be solved by introduction of another feedback
neutralization network, calculated and adjusted to eliminate the negative feedback within the passband of the amplifying device.
========================================
Willy, what the above kinda points toward is that tracers/circuits, over and above being correctly engineered and laid out, should also be kept in good serviceable condition in order to stave off unwanted oscillations.
It may well have been the poor condition of board and tracers that allowed the signals to emanate and mix via inductive coupling (just my misguided opinion).
Also, since oscillation can happen at any/many frequencies,
this coupled with the G-9700 extreme flat frequency response well past our/human hearing,
The (extreme) fast high frequency transistors could be oscillating in a bandwidth well above our hearing perception
and thus we would never know if the amp was oscillating especially on a G97 with a frequency response well above 100 mhz
without putting it on a scope.
I would trust that the boards sent by Kal are top notch refurbished and tested
(I too bought a replacement set and they are top notch)
and that your days of oscillation may be behind you.
==================================================
It is good to hear that after trial and error you are making good headway with the '9700.
As you know, I have owned one since new.
In the 38 years I have owned my G9700,
I have learned that you are all in at about warp factor 5 and pushing to 6
is were clipping is on it's way.
(Personally, I run the loudness circuit in and the bass backed off.)
To ballpark/find clipping of an amp:
If you have a inductive amp meter ( the kind with the big c clamp that you hang over the power input/cord)
You can use it to help you figure out where your G is going to start to clip in relation to your volume settings.
If the G is rated (the tag on her backside) at maximum input of 600 watts,
then you could easily use your inductive c clamp amp meter (over the 110v power cord) to watch the amperage build toward 5 amps as you crank the Sansui.
->5 amps at 120 volts equals Sansui maximum input of 600 watts.
I would bet that you will notice that you are approaching 5 amps at a little over half volume (bass control/position dependent).
Attempting to draw any more than 5 amps from the wall will cause this particular amp to clip.
As you know, clipping an amp is the equivalent of running direct current thru the transistors and into your speakers.
When you clip the input sign wave, the top and bottom of every wave is flat topped.
Down the line at the speakers, the flat top portion of the wave
attempts to hold the speaker cone out
or "freeze" the cone in an extended position for and extended period of time.
Speaker cones need to consistently move and for that they need to see a constantly changing wave form,
not one that freezes at the top and bottom of the wave.
When -For a brief microsecond, the cone is held out (or in) at maximum travel,
this is seen back at the amp as a dead short.
Speaker over heats, amp over heats components overheat, boards, tracers etc all see a temporary flat topping or dead short as heat.
-------------------------------
Luckily, Sansui built amps with forgiving circuitry that for the most apart was able to
quickly revive when pushed into clipping.
But many amps do not settle back down or revive once you have pushed them into clipping.
And I have owned amps that would only revive if powered down.
My Sansui Amps were all able to quickly regain there composure just by backing off the volume button.
I know the G9700 is new to you and so many of us are used to amps were we can crank to 8, 9 or 10
and they may sound like sheet but we push them anyway.
The G9700 will give you well over a true 200 watts per channel extremely cleanly with a very high slew rate.
And this is what we love about the G-9700, It's clean (almost unsurpassed),
it's
extremely Faaaaast and it's inherently beautifully honest to beautifully honest recordings.
But it is also about all in by ~6 and will easily clip for some time before we trigger the protection circuits.
Hope I haven't overstepped my place and all will be taken into context.
Much Sansui Love,
moe
.
