JTM45 rectifier question

[martin manning]

I think there is variation between PT's, Rectifier tubes, Chokes.

All that stuff is very straightforward. What you are doing with the whole-amp choke in the power supply is reducing the ripple voltage at the OT CT. The ghosting (really a parasitic oscillation) is evidently some interaction with that.

[Reeltarded]

Edit: the high pitch noise came from a pedal with dying battery LOL

OOof! When evaluating any amp, guitar-wire-amp!

[Tony Bones]

I think there is variation between PT's, Rectifier tubes, Chokes.

All that stuff is very straightforward. What you are doing with the whole-amp choke in the power supply is reducing the ripple voltage at the OT CT. The ghosting (really a parasitic oscillation) is evidently some interaction with that.

Interesting. My guess has been that ghosting was intermodulation at the screens.

I hope that SPeter gets around to trying the amp with the CT connected before the 3H choke. That will provide additional filtering to the screens, but leave the plates with just the first cap.

[SPeter]

Edit: the high pitch noise came from a pedal with dying battery LOL

OOof! When evaluating any amp, guitar-wire-amp!

I know I was excited and left it in there accidentally from last night's fooling around.

[SPeter]

I hope that SPeter gets around to trying the amp with the CT connected before the 3H choke. That will provide additional filtering to the screens, but leave the plates with just the first cap.

You mean this right?

modified layout.JPG

[SPeter]

the F&T 16+16uf caps only handle 450v but the 32+32uf caps handle 500v. I'd use 32+32uf on the mains personally

Good point, I do not want to increase filtering though, it may turn the amp into a JCM 900 or something LOL is there any 16/16 or 20/20 caps out there which can handle 500V?

[sluckey]

JJ has a 16/16 @ 500v

[Tony Bones]

You mean this right?

Exactly. I'm interested in knowing if both connections eliminate the ghosting, and I'm also interested in your opinion of the difference in tone and feel of the amp.

[martin manning]

Yes this should restore the normal level of ripple at the CT, but provide very clean DC to the screens. The usual setup has a 20H choke feeding the screen node, so much more than the typical 4H, and a couple of iterations have already been tried here adding more capacitance and more inductance IIRC.

[SPeter]

Tested the mod, ghost notes are back full force!!! Yuck! So OT CT front of the 3H choke is not good. What does this mean? What else can I try before I go back to the no ghost setup?

[Tony Bones]

I'm good. Thank you for taking the time!

I had been guessing that the ghost notes were the result of intermodulation between the ripple frequency (100 or 120 Hz) and the note being played. And, since the screens are way more sensitive than the plates to ripple, I wanted to see if improved filtering to the screens but not the plates would be enough to fix the problem.

But Martin was right (always a safe bet. ) The interaction causing the ghosts is at the plates. Martin suggested it is actually parasitic oscillation, not intermodulation. I need to think about that.

Again, thanks for taking the time ...

[Roe]

I rarely have problem with ghosting at the plates. I have build many plexis and experimented with low filtering on the 100w amps, with both el34s and kt66s. the 34s ghost more than 66s, perhaps because they draw more screen current. Marshall went from 32uf to 50uf mains filtering. For me, 33uf works but it does tend to ghost if you really crank the amp. 50uf kills most of the ghosting. In either case, 16uf on the screens seems sufficient with a 3h,110 ohms 250ma choke

[martin manning]

Ok, so we're learning some things here. How about going to a CRC filter for the reservoir? I think a 32u-250Ω-32u might do it, and if that's not enough, the resistor could be increased to 500Ω at the expense of a bit more plate voltage. That resistor should be a 10W, BTW, and you could parallel a couple of 1k 5W to try that out. You have two two-section caps available, so a 32-32 with the resistor across the positive terminals serves as the reservoir and the other one (16-16) will serve the screens and the PI. The resistor will shield the rectifier from the second reservoir cap, but I would move the standby switch after the reservoir. The resistor also shields the plates from the first cap, and so this solution might minimize the sound and feel impact.

[Roe]

With a CLC filter you get more sag and lower voltage, making the tube rectifier dispensable in many cases (particularly in an ac30)

[martin manning]

Roe, did you mean to say CRC? A choke’s DCR is lower than the 250-500 ohms I’m proposing in the CRC, so it will have reduced loss of static B+ and less sag. My idea here is to lower the B+ ripple enough to kill the ghosting without using two chokes, and keeping the existing 2x multi-section caps.