Leaky coupling caps, local feedback in preamp

[nsubulysses]

I was messing around with a build recently and noticed something. When I would clip in my capacitance substitution box across a coupling cap in the 3rd gain stage there were certain settings that caused the noise floor to drop SIGNIFICANTLY, and other settings where it would jump back up to normal. I came to realize it's because there are a few caps in my 1950s (probably) Eico cap substitution box that are leaky. I didn't measure to confirm it, but I had a leaky cap in the garbage that I removed from a 1940s Masco amp earlier in the day so I grabbed it and clipped it in and sure enough, the noise floor dropped a ton.

is this some sort of feedback scenario or what is going on? Is the hiss and hum being nulled because it's bleeding back into the power supply with the leaky cap and canceling out with another gain stage? Some CMRR or something?

It actually sounds awesome and the noise floor is super low with a leaky .01 cap clipped in over the .022
No free lunch and what not so I found the stipulation, when you turn down your guitar volume the signal becomes very blattey and choppy as if there is a noise gate on the amp. sounds like a distortion pedal with dying battery. With guitar volume on probably 5 and above, amp sounds normal

I was able to mess around with resistors 2.2M to 10M and get similar noise reduction but less blattey guitar sound with volume rolled down. Less affected settings (10M, 6.8M) have highest noise floor (but still lower than normal) and most normal sound. ie most true sound to no resistor clipped in. Going down in value the noise floor lowers but the guitar gets more blattey. Also if I go to 2.2M or lower the amp turns into a signal generator.

Now I'm wondering if anyone can shed light on this scenario, or guide me towards resources that might talk about noise rejection with local feedback, or things like that.

I circled the cap in this crude schematic

[Mike Sulzer]

If the capacitor is there to block dc and you let some through with a leaky capacitor, I guess the stage is not biased right anymore.

[Old Tele man]

A leaking capacitor allows current flow in what would normally ONLY be a changing DC-voltage path.

So, that leaking 0.01uF coupling capacitor is effectively connecting the plate voltage (Vp) of the driving tube to 'connect' to the 1MΩ control grid resistor, thus forming a DC-voltage divider (Vp-thru 100KΩ to 1MΩ to ground) circuit that is over-powering the existing cathode-current developed bias voltage.