Test tone gets through with volume at zero

[Paul P]

If the grounds are wired like the drawing, then ther is a bus running from preamp to first filter, then along that bus, the PI joins in, then the reverb. That means to me that for at least part of its trip back to the filter, the preamp ground shares copper with both the PI and the reverb. That is not the same as running a ground wire from the preamp to the filter, running one from the PI to the filter, and tunning one from the reverb to the filter. Three separate wires would be a real star.

That's why I said it was a sort of star. I prefer calling it a snowflake ground,
but maybe a tree would be better since there's really only one trunk. I'm
under the impression that sharing a wire is ok if everything is flowing in
the same direction, towards the main ground. I may be wrong.

Twisting B+ and ground won't achieve anything. Might look nice. We twist AC power wires, like the heaters, to prevent radiating AC into the circuits. Ground and B+ shouldn't be radiating anything, plus the point of twisting is the two wires would carry the same thing at opposite polarity and hopefully cancel. B+ and ground do not carry the same thing and there would be no cancellation even if it was a radiating situation. WOn't hurt anything, but shouldn't be involved in the problem.

I did this based on the content of an article provided by dai h. found at :

http://ampgarage.com/forum/viewtopic.php?t=220

The idea is that a loop can collect interference so if you minimize the loop
area you get less. The wires could just have been close and parallel but
twisting is neater.

I thought it made sense and since I was building from everyone's opinion I
figured what the heck.

My next post has a lot of information I collected today.

Paul P

[Paul P]

I have a lot of information which I've tried to summarize as much
as possible but this will be a long post. Unfortunately I haven't
found anything conclusive, just some strange behaviour that I can't
figure out.

I checked all my grounds with the scope and as far as I can tell
they're all identical with no signal, only this hazy sort of line
at normal frequencies. I did manage to get the following single
waveform on ground with the scope set to 1us horiz / 5mv vertical :



The frequency would be in the Mhz ? Is this from the scope itself ?

Running a wire straight to chassis or 1st filter ground from every
ground terminal in the amp, including the input jacks, made no
difference to the output.

Here is what B+ looks like. At the first filter cap(s) I get
the following waveform which my DMM says is 2.9vac RMS :



Next, the screen node right after the choke, is weird. At .1v/div
the trace is all over the place moving up and down and sideways
continuously :



(All following pictures will be with the scope set to 5mv/div)

At the currently unused PI node I get the following which still moves
up and down somewhat :



At the preamp/reverb recover/tremolo gainstage node I get the unwanted
signal, which is fairly stable :



I discovered that the strength of the unwanted signal on B+ varies
inversely with frequency. At many narrow frequency bands (I only
measured between 100hz and 4khz or so) the waveform is also doubled.




[Edit : I did not re-adjust the input level to 250mvac each time so the
differences may just have been due to this. That doesn't account for
the doubling of the wave though.]

The signal at the output, however, went the other way and decreased
in size with lower frequency. There are also narrow frequency bands
where the ripple will stay still on the larger wave (which is hum,
I'm presuming) but at most frequencies the ripple moves along the
larger wave.

I made a signal grounding probe by attaching a 0.1uf cap to the end of
a chopstick with a bit of lead at the tip for the probe with a wire at
the other end to 1st filter ground.

Using the probe at the power tube coupling caps (there are two of them
in parallel, one for each tube) caused the signal to disappear from the
output.

I went around the amp and discovered grounding the output of the normal
channel (the one with the signal input) had no effect on the output nor
did doing anything in the reverb circuit, which is also on the normal
channel.

Grounding the signal path in most of the vibrato channel preamp and
vibrato circuit with its extra gainstage caused about a 50% reduction
in the strength of the signal on the output.

Applying the grounding probe to any B+ had no effect.

Since the vibrato channel was having some effect I disconnected B+ from
the gainstage that follows the vibrato circuit. This removed the signal
from all of the vibrato channel (preamp, tremolo circuit, gainstage).
This also had the effect of reducing the waveform on the output by a
substantial amount, maybe half. Now, all of a sudden, grounding just about
any point in the normal channel, including the reverb circuit, caused the
signal to disappear almost completely from the output (very slight ripple
remained).

Here are two maps of what I've just described.



I really don't know what to make of this, other than the 20uf preamp node
decoupling cap not being able to remove the ripple coming from the normal
channel preamp's first triode's plate. This ripple then being sent on to the
other circuits. It does seem like each channel is contributing half of the
unwanted signal to the output. But then I don't understand why grounding
the signal within the normal channel anywhere (other than the input) did
not reduce the output by half when the vibrato channel was alive. Or why
the signal disappeared completely from the vibrato channel when I only
disabled the gainstage and not the other circuits of this channel.

Paul P

,

[Paul P]

I made some changes today that had a positive effect. First I moved the
tremolo gain stage B+ to the unused PI node. This makes sense since in a
push-pull amp the PI would be providing a gain stage in the same spot.
I then split the preamp node B+ in two with one node going to each channel.
So each channel now has it's own 20uf decoupling cap.

I also cleaned up the connections at the chassis ground a bit just to make
sure and updated the power supply dropping resistors to get the voltages
back to where they're supposed to be.

It appears that things are now much better. B+ at the preamp nodes is
much cleaner and no signal is getting from the normal channel into the
vibrato channel.

I've also discovered that my volume pot doesn't go completely to zero but
lets through a very small signal (a few mv). In fact if I turn the volume
up just a hair I can get the signal almost to zero before it rises for real.

The only thing I still wonder about is the apparent instability of B+ through
my choke. Is this normal ? I would have thought that the waveform would be
just a cleaner version of what is present at the first filter node. The
ripple is much reduced but it also varies quite a bit in intensity. Maybe
the screens have something to do with this ?

I'm done working for today but I'll post a follow-up with more details once
I have them.

Paul P

[Paul P]

I'm done working on this issue. I don't believe it's possible to completely
illiminate a really strong input signal from appearing on the output with
the volume off since the signal will appear at very low level on the preamp
node decoupling cap from where it can get into the preamp's second triode
on the other side of the volume pot. So there's no way to totally ground
the signal. The situation is made worse by further gain stages which will then
amplify this little signal.

The main problem I had was that since my vibrato channel preamp was also
on the same node it was getting infected by the ripple and then amplifying
it through several stages. By putting my vibrato on a separate node I now
have no signal coming across between the channels.

So this is what the output now looks like with a 250 mvac 1 khz input signal
(1mv/div) :

Paul P

,

[Enzo]

Then if the signal is riding the B+ node past a volume control, the stages on either side of said control should be on different nodes. That is what decoupling is all about - decoupling.

[Paul P]

I've been thinking the same, if problem there is. Fender put up to 6 triodes
on the same node/decoupling cap, on both sides of volume controls, and I
can't see how there wouldn't be some signal getting spread around. Is a
coupling cap a dead short (ac) to ground or is there some sort of slope/resistance
involved ? If the latter, a signal, however small, is bound to exist at the top
of the cap.

Paul P