Good sleuthing, Grainger! But I'm afraid I was not very clear - I meant to avoid anything physically larger, which would present a larger surface for capacitive pickup of noise.
It s probably OK to increase the capacitance of the 2200uF cap, but I can't provide any certainty - these two caps set the poles of the DC feedback loop (the bias servo) and feedback loops are potentially unstable if the poles are not working together correctly.
One other point - most electrolytics, including the polymer types such as the Os-Cons,have a good deal of leakage. Larger capacitance leads to larger leakage. The leakage must be substantially less than the EF86 quiescent current or the servo will effectively be shorted out. The other feature necessary in this cap is low ESR at high frequencies.
You may find these specs in a data sheet, if not then you can't predict whether it will work, you'll just have to try it and see. Check the EF86 plate voltage first, if it's too low then the cap leakage is probably the problem. Then listen, especially for distortion or incorrect equalization - i.e. changes in the high frequency response level. If it sounds good then it is good. Watch for very slow oscillation, which would indicate an unstable servo loop - not likely, just something to watch for.
