Anyway I decided to use Schmitt-triggers and gave up with the noise. It is as it is.
By the way - those ground problems may be caused through default grounding of crt monitors and oscilloscopes as well (building a ground loop). So measuring with unconnected TV equipment may give different (better) results. 80 mV noise is quite much - about 10% of full white level (700mV) and well seen on white background.
this is my test zeddy most of the passives have been moved to the back side, and no ear or mic circuit exsists
the last time i posted i have way too much noise in the video signal, this is now muchly reduced
with just slight vertical banding showing on my LCD tv using the composite input.
speedtest reports exact number of frames which is always a good start.
Then i had a brainwave... since as i use onboard oscilator i dont use the OSC pin this can be repurposed into an
extra input... memory above 32K check.
activate special feature... M1NOT yes please
BTW do you have the barebone 32K RAM on your testing PCB?
So some design outlines,,
Composite 1v p-p video output with back porch, shorting link to select hi-z or 75ohm output. most monitors / composite inputs are 75ohm and will work fine with link removed, hi-z inputs will need link shorting ( it adds a 75ohm load to video circuit )
outline size no bigger than 40pin dip socket slightly higher than original ULA
on board xtal oscillator for accurate frequency control.
original osc pin converted to optional input used to enable M1not.
Tape input pre-amp included for improved tape loading.