Weird problem....gain leaks thru to output even with master disconnected

[daz]

Ok, so i'm looking at the DSP i use in my homebrew's loop that i recently added and i notice the global direct mix is set to off ! Last time i used this processor thats how i used it, for wet signal only into a second amp. So i'm totally puzzled as to how i am getting any dry signal and not just repearts. (i mainly just use delay) It hits m to pull the DSP out and the amp still has signal getting to the speaker. Now befor you start telling me it;s the switching jack, i didn;'t have any and used simple jacks with no switching because not only didn't i have any switching jacks at the moment, but i will always have the DSP in the loop and if not i'd just patch cord it. Besides, switching jacks get dirty and problematic.

Anyways, the signal getting thru is puzzling. If i pull the PI tube it goes quiet. But if i unsolder both the grids, 2 and 7 of the PI it still gets thru ! Splain that to me lucy ! The gain knob will turn the volume down but the master has no effect whether i leave it all wired as is or unsolder it's output that goes to the send. It would seem the signal that is getting to the output is before the master since the master has no effect and the gain knob does. But then where is it getting in? Oh, and the singnal that gets thru is small. Like bedroom volume wuith the gain on 10 and thin sounding. So obviously only the input stage must be getting thru. If i pull the PI it goes dead yet with the PI grids unsoldered it still gets thru? WTF?!!!

I don't have a schematic, but it's basically like a JCM800 2204 preamp with a regular gain stage in place of the cold bias stage, cathode follower and all. then master coming from the tone stack then out thru a cap into the typical marshall PI. Output is 6V6 that was once a el34 18 watt. NF loop and fixed bias. With the loop the master goes to the send jack and return to the pi input cap. This has got to be something i can understand being a non tech. But i DO understand routing and i built it so i'm not a total non tech, but i cannot find anything thats connected wrong. Any ideas?

[eschertron]

Inside the amp some wires are radio transmitters, and some are radio receivers. That's my hypothesis; it 'splains why the signal is heard even with the master disconnected.

Alt hypothesis: Signal on the B+.

[daz]

Inside the amp some wires are radio transmitters, and some are radio receivers. That's my hypothesis; it 'splains why the signal is heard even with the master disconnected.

Alt hypothesis: Signal on the B+.

I wondered about that but wasn't sure if that could happen, tho i seem to recall it being talked about. (B+) I don't think it could be wires because the signal seems too loud for that but i'll chopstick it a bit then if nothing i'll look at the B+. Thanks.

[daz]

I can't see anything that would look like signal going to B+. How would i test for that?

[eschertron]

As much as ripple rejection is lacking in the preamp nodes, so also the single-ended current through the nodes works its way backwards. The technique of matching out-of-phase stages on a single PSU node helps minimize both effects. Having said that, as much as the PI and power tubes reject ripple, they'd also reject the signal.

How much signal do you really get at the output? if you have 300vpp on the last preamp plate, how much reduction do you need to make the output quiet? 0.03vpp on the power tube grids is a reduction of 300/0.03 = 10000, or 80dB. Probably still audible at the output What do you have as grid leaks on the 6V6s? not enough loading there?

edit: If there's absolutely no hum at the output, I'd expect no signal getting in that way either.

[daz]

5.6k at the power tube grids, and no hum. It's a quiet amp as sloppy as it is.

As much as ripple rejection is lacking in the preamp nodes, so also the single-ended current through the nodes works its way backwards. The technique of matching out-of-phase stages on a single PSU node helps minimize both effects. Having said that, as much as the PI and power tubes reject ripple, they'd also reject the signal.

How much signal do you really get at the output? if you have 300vpp on the last preamp plate, how much reduction do you need to make the output quiet? 0.03vpp on the power tube grids is a reduction of 300/0.03 = 10000, or 80dB. Probably still audible at the output What do you have as grid leaks on the 6V6s? not enough loading there?

edit: If there's absolutely no hum at the output, I'd expect no signal getting in that way either.

[eschertron]

no hum. It's a quiet amp as sloppy as it is.

I'm sure it's a thing of beauty!

5.6k at the power tube grids

...and what are the grid leaks?

[daz]

heres a schematic of the EL34 amp i built and it's very similar. Ignore the bias circuit as thats a single circuit unlike the dual in this schematic. The PI to the power tubes and NF are pretty much the same and the preamp/tone stack CF all that is close with nothing different enough that it would throw off your troubleshooting ideas. No choke either, just a resistor.
EDIT: forgot to mention, and maybe this is the reason but you tell me....unlike the schematic there are only 2 filter nodes for the AX7's. One for the PI and one for V1 and V2.

[g1]

When you remove everything from the FX loop (including any patch cable), do you get bleed through?

[daz]

When you remove everything from the FX loop (including any patch cable), do you get bleed through?

Yes, I can unsolder the output lug of the master and no difference. Still bleeds thru. Like i said tho, not the full volume or anything. But with the gain knob on 10 it's about as loud as the master just past that point where volume kicks in, maybe 1.5-2 on the master.(with loop patched to work as normal) I can pull any of the 3 AX7's and that kills all signal. Yet I can disconnect the PI grids and sound still goes thru. So it sounds like what eschertron said, signal on the B+ because i see no other pathway for signal to get to the PA tubes. But i don't know how to check that.