Question regarding oscillation in a Plexi style amp

[psychepool]

Hello,
I previously posted a thread about this.

https://ampgarage.com/forum/viewtopic.php?f=6&t=38953

I built this amp now, but there's an issue and would like to ask for advice.

It is an amplifier that allows switching between 5F6A, JTM45, and SLP modes.
After make it, all mode switches function correctly, and the sound is great.
However, there is one problem: oscillation occurs when turning up the Presence pot in SLP mode.

Here is the current status:

1. During construction, when I tested only the power amp section by connecting the signal to the PI input, it was completely silent.

2. There is no oscillation related to the Presence pot setting in any mode other than SLP mode.

3. Oscillation occurs when the "Bright Channel" mode switch (which toggles between 2.7k/680n cathode, 2.2n coupling cap, and 4.7n bright cap) is set to SLP mode.
There is no oscillation when this switch is set to 5F6A/JTM mode (820R/220uF common cathode, 22n coupling cap, 100pF bright cap).

4. There is an NFB switch that toggles between 22k, 56k, and 100k resistors (connected to the 8-ohm tap); the oscillation worsens as the resistance value increases.
With the 22k setting, oscillation starts around the 2 o'clock position on the Presence pot; with the 100k setting, it starts around the 10 o'clock position.

5. Naturally, the amount of the oscillation varies depending on the Volume pot setting. It begins to set in slightly when the SLP Bright Volume is around the 9 to 10 o'clock position.

6. Engaging the "Fat" switch (a 680n bypass capacitor connected to the 820R cathode resistor on V2 pin 3) significantly reduces oscillation. The point where oscillation begins shifted from the "10 o'clock" position on the Presence control to around "2 o'clock."

7. I tried various grounding configurations, but there was no change.

8. Both B+ voltage and bias levels are within normal ranges.

9. Shielded cables are used for all signal lines (Input jack to grid stopper lines on V1 pins 2 & 7; channel coupling capacitors to volume pot pin 3; mix resistor output to V2 pin 2; V2 pin 8 to EQ coupling capacitor and slope resistor; Treble pot pin 2 to PI coupling capacitor).

I have described the situation to the best of my recollection.

The build is based on the circuit below.

Bassman & reverb.gif

A reverb circuit is included, as shown in the schematic above.
The reverb circuit components are installed and B+ power is supplied, but the reverb potentiometer and reverb input/output lines are not connected.
(The 100k mix resistor was removed from the PI input, and Treble pot pin 2 is connected directly to the PI input.)
Just to be sure, I removed the reverb tube, but there was no significant change.

Here is the current build status.
For reference, it uses Hammond Deluxe Reverb transformers and a pair of 6V6 tubes. The bias is set to approximately 19–20mA.

1000039521.jpg

The layout is unusual—making it difficult to avoid noise—but I would still like to try and resolve the issue.
Please let me know if there are any measures I can take.

[pdf64]

A cramped layout like that, eg putting the speaker circuitry within a couple of inches of the input circuitry, kinda invites such trouble. I suppose you're committed to it now. At the very least, screen off the ouput stuff from the input stuff.

[Stevem]

I have run into this when installing a mod kit into a Tiny Terror.

You have a possibility of two separate problems going on or both of them combined.

1) your in out jack is too close to your output jack.

2) and or your input jack is too close to a coupling cap.

If you remove the input jack such that you can get it even 3 inches out of the chassis I bet your issue gets resolved.

On the previously mentioned Tiny Terror I had no choice but to make a grounded shield out of copper sheet that I insulated on both the inside and outside due to how tight the chassis and layout was.

Years ago there was a aftermarket Kit of a Fender model ( deluxe I think it was) and once the volume was cranked open a certain amount it woukd go off into oscillating and be flat out useless.

The simple cure was to pivot a coupling cap that was near the input jack around from being mounted with one end pointing at the jack ( perpendicular) to being mounted horizontal.

Another build tip that can help keep these issues from popping up is to keep wires that feed a down steam grid of a tube as short as possible since a grid with gain acts like an antenna .

To this end it’s best to make a plate wire longer and place the coupling cap right the hell on the grid terminal of the down stream gain stage.

Also coupling caps of the non disc type or silver mica type have an end that acts like a shield when you have a grid hooked up to its other end.

With coupling caps that have a line on one end then that should be the down stream end going to the next grid.

[Roe]

oscillating NFB/presence circuit is common. Fixes include:
-shielded headcab
- moving purple wire from presence pot to NFB resistor (27k-100k)
- changing the impedance of the nfb loop, replacing 5k pot with 50R-220R (w. 270R-4.7k resistor)

[psychepool]

Thank you for the replies. I’ve successfully find the root cause. As you suggested, the issue stemmed from the input and output sections being too close to each other.
Since the top control plate is detachable (as shown in the attached photo above), I connected it to the main chassis with clips and then lifted the plate up into the air.
Even lifting it slightly significantly reduced the oscillation, and raising it by about 3cm made it disappear completely.

The reason I hadn't suspected this issue was, as I mentioned earlier, that the circuit had previously worked perfectly without any oscillation when configured as a Deluxe Reverb.
To be precise, it wasn't a standard Deluxe Reverb circuit but one based on Channel 1 of the Mesa Boogie Rectoverb. Consequently, the circuit was nearly identical to that of a vintage Marshall's Phase Inverter (PI) and included a Presence pot; yet, back then, there was absolutely no oscillation associated with the Presence control.

I had often heard that proximity between input and output sections could cause problems, but since I’ve always built amps using small chassis—and never encountered significant issues despite close spacing—I hadn't considered it for main issue.
However, it seems the SLP circuit's specific high-frequency characteristics made it susceptible to this particular problem.

In any case, I’m delighted to have identified the issue thanks to your help.
Now I need to implement shielding, but given the cramped interior, I have to choose the right method.
Since the components are packed tightly together, I need to determine exactly which areas require shielding to ensure the shield doesn't make contact with any parts.

I don't have a photo of the area in its current state.
Please take a look at the photo below for now.

01.jpg

This is a photo from when the amp was configured as a Deluxe Reverb, before I rebuilt it as an SLP.
The rear plate is detached, so you can see the general layout.

Viewed from the side, the structure looks like this. I’m not sure exactly when this photo was taken, but I believe it dates back to when I built a different circuit previously.
This chassis was constructed about 15 years ago and has been reused several times.

03.jpg

I’d like to ask for advice on the most reliable way to handle shielding.
Is it necessary to wrap the input jack entirely, or is it sufficient to shield only the area facing the output jack?
If I shield the output jack side, should I shield just the immediate vicinity of the jack, or should I extend the shielding to cover the OT wiring as well?

[pdf64]

I suggest to form and fit a thin metal screening plate between all the output stuff and all the input stuff.
So, from the floor to the top, from the side of the chassis as far along as is feasible, eg past the V2 socket.
It'll need to be angled coming off the floor, then bent vertical.
Thereby creating separate isolated EM compartments, to eliminate as much as is feasible interaction between output and input circuits.
Twist all OT wires that carry opposing currents, so all secondary wores twisted together, all primary wires twisted together. That should help to reduce their radiated magnetic field, because effectively screening magnetic fields requires thick ferrous metal, for a thin plate an expensive metal such as mumetal would be needed.

[psychepool]

I ran a quick test. I looked around for a suitable object I had at home and spotted a string action gauge.

Here is a photo of the area around the input/output jacks:

01.jpg

As you can see, I don't need to cover it completely; simply blocking half of the output jack from the top makes the oscillation fully disappear.

02.jpg

Even with the top left open, blocking just the side stops the oscillation completely.

03.jpg

My initial plan was to use a piece of paper or plastic sheet lined with copper tape for shielding, but now that the specific area requiring coverage is clear, I think a more rigid material would be better for the final finish.
Since the internal space is so cramped, using materials that easily deform or shift could lead to a short from vibration or impact; instead, I plan to bend a sheet of aluminum (about 1mm thick) and bolt it securely in place. If the aluminum shielding proves insufficient, I can always add copper tape on top.

Thanks for the crucial help. I'll post the finished result later.

[pdf64]

Good work!
I suggest to do a proper job, not just the bare minimum necessary to get it stable now.
Because it would be a shame to do the work but then plug in a different speaker or change a valve and it starts with the oscillation again.
You can't get an amp too stable

[Stevem]

Here’s two photos of one of my most loved amps by me and the many who have played and recorded with it.

Look at these two photos of my Gibson ga77 made in 1954.

The 4 input jacks are mounted only 2.5” away from the OT which is mounted on the outside rear wall of the chassis, so it’s shielded from the input jacks which by the way are mounted and grounded to the the chassis and each jack has its own shielded cable to each preamp input grid.

Also notice the tanish white 6 inch long nylon or maybe plastic tube on the rear floor of the chassis.

This tube has the plate leads running thru it from the OT to the plate of each output tube.
These plate leads themselves are over 7” long!

Normally 7” long plate leads would be a very big oscillation issue, but each lead is sheilded separately and the B+ primary lead is run thru that same tube.

[psychepool]

I ordered a custom shield panel to fit the location; while waiting for it to arrive, I used the string action gauge shown in the image above—securing it to the chassis with copper tape so that it acted as a shorting ground. This improved things significantly, but there was still some oscillation when the Treble and Presence controls sets at full.

shield_01.jpg

A few days later, the ordered parts arrived. I covered the OT wires with copper tape, drilled a hole in the chassis, and securely fastened the shield with a screw...

shield_02.jpg

shield_03.jpg

shield_05.jpg

All oscillation vanished completely!
Aside from the oscillation, other lingering, unidentified noises also disappeared entirely!
SLP-style circuits—especially those with a 4.7n bright cap—are inherently noisy amps; while there is still the expected level of noise as you turn up the volume, the amp now produces a truly clean sound.

After playing around with the finished amp for a while, the 5F6A mode doesn't seem particularly worthwhile; I think it would be a better idea to focus on replicating the JTM45 and various vintage SLP models instead. I plan to add a few more features based on this idea and then finish the project by fabricating a new panel.

Thank you so much for your help!

Here’s two photos of one of my most loved amps by me and the many who have played and recorded with it.

Look at these two photos of my Gibson ga77 made in 1954.

The 4 input jacks are mounted only 2.5” away from the OT which is mounted on the outside rear wall of the chassis, so it’s shielded from the input jacks which by the way are mounted and grounded to the the chassis and each jack has its own shielded cable to each preamp input grid.

Also notice the tanish white 6 inch long nylon or maybe plastic tube on the rear floor of the chassis.

This tube has the plate leads running thru it from the OT to the plate of each output tube.
These plate leads themselves are over 7” long!

Normally 7” long plate leads would be a very big oscillation issue, but each lead is sheilded separately and the B+ primary lead is run thru that same tube.

Thanks for sharing that amazing amp. The oldest amp I’ve seen in person is a '70s Silverface Princeton Reverb; since I’ve never actually laid eyes on genuine vintage amps like these, I’m always struck with a sense of wonder whenever I see photos of them.

For this build, I went a step further and covered the OT wires with copper tape in addition to using the shield panel. It’s definitely making a difference this time. In previous builds, perhaps because the layout fundamentals were already sound, adding this kind of shielding didn't seem to make any noticeable difference.