Did you get anywhere with this? Just looking at the board it looks like there is a fair bit of tarnish on the chip legs in places, especially where they are in sockets. This *will* stop things from working properly although I do note you've swapped out the socket and chip for the LM1889 with no improvement
I've been testing a set of spare boards I had and all of them had the same problem. Worse I could only find a single LM1889 that worked between them and from my cache of spares.
The single common factor is that they seem unable to run the crystal for the colour carrier signal, without it even if the rest of the chip is working you will see nothing. In theory (at least) you can feed the necessary 4.433MHz signal to the LAG pin of IC9 and it solves the issue.
I've had a modicum of success in "fixing" faulty LM1889s but only when sourcing them from ZX Spectrums. Putting them in a Dragon and leaving them for a bit eventually sees the clock spring to life and can then be transferred back to a Spectrum and continue working (for a while before the clock gives up again).
I'm working on a total replacement for the LM1889 but it is still early days and has a way to go!
Dragon 32 black and white output
Re: Dragon 32 black and white output
FWIW when you're looking for them, the "NTE846" seems to be identical to the LM1889N.
Re: Dragon 32 black and white output
Easier to find but I'd hoped for a nice cheap option... 
It does give me a viable price cap on the replacement device

It does give me a viable price cap on the replacement device