Again as Neikeel and I have suggested, a stable steady bias is the corner stone of the power section. If the amp can't maintain a steady unwavering bias signal with no power tubes installed, then that portion of the amp will need investigated as to why it will not hold a steady bias voltage. Changing tubes and trying other shotgun approaches will only add to your frustration level if there is a bias problem. If the bias circuit is stable over a suitable period of time, then that is the first checkmark. After proving an empty power tube bias, then add the load into the circuit, which is the power tubes. With no signal injected, again the power section needs to be stable and consistent. Only then can you be confident going forward. That oscillation that started after you saw the spark is probably something compromised by the meter tip touching two components, or a single component and ground simultaneously . It could have been a short from plate to filament. Prove your bias, then add your load, and you'll be farther along.