Why are decoupling caps between plate and grid lower values like 0.1uF-0.33uF and the output decoupling cap higher values like 1.5-2.2uF? I understand the math behind the output decoupler where-by a lower value causes the cutoff knee to be at a higher frequency there-by rolling off the low end. I am curious as to the choice of values for caps in tube circuits and why these values are different.

Update:

After thinking about it, here is my guess: on the line level output, there is a impedance to ground that causes the final output cap to act like a filter, rolling off the low end, the lower value the cap. based on a formula, you can work out what the ideal cap value is here usually 1.5uF for a 20Hz knee roll-off. but in the middle of a circuit, i am guessing the impedance are such that you can use a much lower value cap before you get this roll off. i see 0.1uF and 0.22uF often used for these. presumably you can use larger values (within reason) in the same way - just don't go lower because you'll get low end roll off... something like that, right?

Update:

After thinking about it, here is my guess: on the line level output, there is a impedance to ground that causes the final output cap to act like a filter, rolling off the low end, the lower value the cap. based on a formula, you can work out what the ideal cap value is here usually 1.5uF for a 20Hz knee roll-off. but in the middle of a circuit, i am guessing the impedance are such that you can use a much lower value cap before you get this roll off. i see 0.1uF and 0.22uF often used for these. presumably you can use larger values (within reason) in the same way - just don't go lower because you'll get low end roll off... something like that, right?