I know this is an ancient thread but after tuning the mica cap to achieve lowest reflected power I find that the output of my amp decreases by about 60w.
Anytime you see this, you know the the input transformers have not been wound with a ratio that matches the 50 ohm input. Depending on the number of turns, it could be higher or lower. When you set the input to match whatever the impedance the two input transformers show when combined (half of whatever one of the two input transformers represents), the amplifier makes more output than it does when tuned at a "perfect" 50 ohm match.
That's because maximum power transfer occurs when the load impedance, macthes the source impedance. In this case, the source is your radios transmitter (50 ohms) and the load is the input section of the amplifier (something other than 50 ohms). Because maximum power transfer is not occurring at minimum input VSWR, we know the two input transformers are not each providing a 100 ohm load, to each side of the input splitter. Which, is required for the input of that splitter to show a 50 ohm load to your radio.
Adding to this problem is that these RF transformers only allow certain fixed ratios for impedance matching and sometimes you find yourself needing a match that is in between two fixed values. You have two choices on how to handle that problem. Take the easy way out and just add a padder cap. Deciding on the value, based on how much of a compromise you want to make between a good 50 ohm input match, or more drive power actually making its way to the input of the transistors, by matching the different input impedance they show at the input splitter.
The second choice is to install additional parts to form a circuit that can take whatever impedance is actually at the input splitter and match it back to the 50 ohm radio. There are a wide range of circuits that can do this, including the "L", "T" and "PI" networks.
...runna pile-up on 6m SSB