Hi Patrick -
There are several ways to tame down a final amplifier that is RF unstable. I've had mucho experience with the 4-1000A in both class B and plate modulated class C service.
1) The first way is neutralization. You've already done this, but be sure that it is working properly. I usually "cold" neutralize it. With just fils on, inject a signal into the grid as usual and hook the scope probe to the 50 ohm C2 plate loading cap. You should see a definite null in pass-thru signal as the optimum neutralization point is passed. You have a variable way to tune the feedback, right?
No adjustment but it may be time to consider it.
2) Another VERY effective method is to add a loading cap/resistor from grid to ground right at the tube pin. Usually a .001 uF (heavy duty brown 1"X1" type) in series with a non-inductive resistor like 50 watt Glowbar works well. The resistor value can be anywhere from 600 ohms to 6K. The lower the value, the harder to drive, but the more stable the amplifier. All of my big rigs have a grid loading cap/resistor just as good practice, starting with about 3K or so. A very stable amp may not need it below 40M, but some will.
I did this before with a 3K globar NI resistor of 50-60 and IIRC the stage was stable. I removed it because I needed the existing front panel switch to change input coils on the MB-40 tank. I don't recall if there was some tuning issue or not with the resistor.
3) Be sure to sprinkle a lot of .001 1KV BYPASS disc caps onto ALL incoming leads. This means keying lines, AC, grid voltage input, screen voltage in, etc. Use a C-L-C input for the grid D.C. and screen D.C. close to the tube pins. (500V > .001, 2.5mH, .001 > tube pin)
This has to be revisited. LC filters are presently at the rear panel but not on all connections. The pins of the tube except grid drive are all very heavily bypassed right at the socket.
4) Be absolutely sure the RF INPUT leads and parts cannot see any RF OUTPUT leads and parts. Watch the paths and routing of other DC leads too. Keep RF leads away from ALL other stuff. Use shielding/barriers. Study the underside and top of chassis and look for possible coupling. A common problem can be the antenna relay that might have the input and output leads close together on a common relay. Two separate relays with a shield between them is a better idea. I always use a barrier strip on the back of the rig with .001 caps bypassing every incoming lead before entering the critical inside.. Feed-thru caps work well too .
The input is well separated from the output and the RF jacks are some 25" apart, although the input and output coaxial cables are through the same hole in the overall cabinet. There is no TR switch on it and a separater antenna is used for receive. The grid tank circuit compartment is under the chassis in a box with just the grid lead, neutralizing cap lead, and the coax cables for 50 Ohm drive and grid bias poking through. Bias and 50 Ohm input RF drive go from the back panel to that box through shielded cables. No output, HV, or other thing comes through the chassis from the plate area except the neutralizing 'capacitor' rod.
I should find pictures because there is probably room for improvement.
There's more, but try these first and report back. I am quite confident that the .001/ 1K-2K non-inductive resistor(s) from grid to ground will do the trick. Be sure to figure how much power this power resistor will dissipate and rob from input drive watts. Good thing that it is under the chassis so that the blower air will help it cool a lot.
Let us know.
T
Ok so I should try the globar resistor and capacitor right at the socket? That can be done. This picture is old, the globar resistow was removed.