1940's Amp High Volume Oscillation

[bobloblaws]

Hey guys, it's been awhile. but I have this amp back on my bench to try and figure out what is causing the squealing, or at least try to get rid of it somehow. No need to review the entirety of this thread, here's the Coles notes version. Contrary to what was discussed earlier, the guitar pickup is not a factor in the high pitched oscillation. In fact I can repro the problem without even a cable plugged into the input jack, all I need to do is have the (mic channel) volume and the tone knob turned up to a certain level and unshort the 6SJ7 grid by pushing the hot tab on the input jack away from the shorting tab. As I alluded to before, the length of input cable is a factor i.e. the shorter the input cable the less I have to turn up the volume for the oscillation to start happening. Reducing the total gain of the first two preamp stages is a factor. For example, removing either or both of the cathode bypass caps changes the position that the volume control needs to be at for the amp to start oscillating. Turning the tone control down also changes the position that the volume control needs to be at for it to start oscillating. I was comparing the first stage (6SJ7) to a Valco 8418 circuit which uses the same preamp tube in the first stage. They are quite a bit different, which is to be expected since my amp was not designed for electric guitar. But my amp has this oscillation problem regardless of what is being plugged in as I just mentioned (even with nothing plugged in). I'm wondering if anybody has any ideas of what I might try in terms of trouble shooting or even trial and error modifications to try and solve this. On the 6SJ7 my plate voltage is 137V, screen is 15V, cathode is 0.71V. Note that my schematic shows that someone modified the circuit at some point. It looks like originally the 6SJ7 cathode was connected directly to ground, now it has the same cathode configuration as the 6N7. I've experimented with swapping out older components for new ones (and changing the 6SJ7 tube) and also adding a grid stopper on the input but the oscillation persists.

Valco Harmony 8418.pdf

[Enzo]

Well, that is what the shorting tab is for on the input jack.

Ages ago I was interested in getting the guitar away from the amp. But you also found just having a long cord int eh jack could kill oscillation. Cord has capacitance, and the longer the cord the more it has, so long enough cord can dampen higher freq oscillation that flows through the input stage.

I note the inpit stage 6SJ7 has been converted from grid leak bias to cathode bias.

[bobloblaws]

Well, that is what the shorting tab is for on the input jack.

Yes, I mentioned this to illustrate that even the length of wire from the jack to the grid (which happens to be about 3 or 4 inches shielded) makes a difference. It basically constitutes the shortest cable run that I've been dealing with. I could just have easily plugged a dummy connector into the jack to lift the tab.

I note the inpit stage 6SJ7 has been converted from grid leak bias to cathode bias.

I'm not familiar with grid leak bias. Do you think I should try restoring it to spec? Apart from the connection that is whited out from cathode to ground, what else is hidden by the white out on the grid end? Diagram 1-25 on this page shows a cap in parallel with the grid leak resistor.

http://electriciantraining.tpub.com/...Biasing-46.htm

[Enzo]

Grid leak is not used much these days, but was more common in the octal era. In grid leak, the cathode is grounded and the grid has a LARGE resistance to ground..10 meg being common, or at least 5 meg. I think you are running the tube hotter once converted to cathode bias. There MUST be an input cap so any DC building up sees only the 10 meg.

[bobloblaws]

Grid leak is not used much these days, but was more common in the octal era. In grid leak, the cathode is grounded and the grid has a LARGE resistance to ground..10 meg being common, or at least 5 meg. I think you are running the tube hotter once converted to cathode bias. There MUST be an input cap so any DC building up sees only the 10 meg.

Interesting. I hadn't noticed that the value of the input resistor had been changed with whiteout as well.

In the diagram that I linked to the cap is polarized and the negative end is connected to the grid. Looking closely at my doctored schematic there appears to be a negative symbol still showing above the input line on the grid side. So I'm assuming there was also a polarized cap with the same orientation in my amp. The difference is that in the linked diagram the resistor is connected to the negative side of the cap and in my schematic the resistor is connected to the positive side of the cap. Would either way make any difference? I don't quite follow the part where built up DC "sees" only the 10 meg resistor. And if the purpose is only to block DC, why a polarized cap?

Also, if you'll indulge me for another moment, when you talk about "DC building up", what does that mean? I'm going to guess it has to do with the capacitance of the input cable that you mentioned. I also don't understand why the cap can (apparently) be eliminated if the stage is changed to cathode bias.

Finally, if I go ahead and try this, what value cap would be appropriate? Is the .005 cap (C4) in the Valco 8418 schematic there for the same reason (despite the stage being cathode biased)?

Thanks, Enzo.

[HaroldBrooks]

As you describe the oszillation as medium speed tremolo effect, it would be a kind of motorboating. This is often caused by a positive feedback between stages via unsufficiently decoupled/filtered supply voltages, in other words too small filter cap C-values. 4µ and 8µ are already extremely low and electrolytics lose capacitance over time. You should try new filter caps and if necessary increase the values moderately.

Sorry if this was already said, but make sure the 10k resistor between the filter cap stages are good as well.

I caused the oscillation thing by using too low a resistor between stages, in a misguided attempt to boost the later B+ values for the preamp and PI tube.

It started an oscillation just like yours, that also responded to the volume pot (lower cancelled the oscillation). I went back up in resistance, and the problem went away entirely.

Good luck with the fix for the squeal, of that I know little.

[bobloblaws]

Sorry if this was already said, but make sure the 10k resistor between the filter cap stages are good as well.

I caused the oscillation thing by using too low a resistor between stages, in a misguided attempt to boost the later B+ values for the preamp and PI tube.

It started an oscillation just like yours, that also responded to the volume pot (lower cancelled the oscillation). I went back up in resistance, and the problem went away entirely.

Good luck with the fix for the squeal, of that I know little.

Yes, I did try new resistors in place of both 10K's since they looked ancient but there was no change. (I did not try changing them both at the same time however).Considering your experience, I wonder if bumping up the value of those resistors would do it.

[TimmyP1955]

Perhaps I missed something in the conversation (short attention span): Shouldn't there be a filter cap from the junction of R7 and R9 to ground? As it sets, not only is the preamp on the same node as the amp, but there's 390 Ohms between that node and C1B. AND: The screen node is connected to C1B. I'm thinking that the 6V6's plate and screen connections are reversed, as the B+ is lower than the screen supply.

[bobloblaws]

Perhaps I missed something in the conversation (short attention span): Shouldn't there be a filter cap from the junction of R7 and R9 to ground? As it sets, not only is the preamp on the same node as the amp, but there's 390 Ohms between that node and C1B. AND: The screen node is connected to C1B. I'm thinking that the 6V6's plate and screen connections are reversed, as the B+ is lower than the screen supply.

I think you are looking at the schematic for the Valco 8418 that I also uploaded for comparison. My amp is the circuit in "schematic.jpg" from message #31 in this thread as well as way back in the original post.

[bobloblaws]

Grid leak is not used much these days, but was more common in the octal era. In grid leak, the cathode is grounded and the grid has a LARGE resistance to ground..10 meg being common, or at least 5 meg. I think you are running the tube hotter once converted to cathode bias. There MUST be an input cap so any DC building up sees only the 10 meg.

Enzo, can you advise on an appropriate cap value and type for trying out grid leak bias?

[Enzo]

Input cap is just a series cap at the input jack. 0.1uf comes to mind as it passes everything audio.