EDIT: I've just re-read and noticed your comment that the inputs of U17C are linked. That part of the circuit is correct then (assuming that the component values achieve the correct time constant). Can you confirm the identity of C31? Also, how old are the HEF parts?
EDIT1: I note that the signal defining the state of the inputs of U17C (via D23 and R81) is coming from the Q output of U13A, not VDD.
HEF parts are from 3 years ago. Capacitor C31 is a 220pF ceramic disk
Sorry @aless2056, I've been trying to get my head around the schematics but without any luck. Unfortunately it is one of those hand drafted 1970s/80s designs which uses lots of R/C time constants and looping discrete logic. The original designer would have had it all clearly in his head, but unless any design documentation has survived, it is very difficult to follow.
The problem also is that the circuit does 'work', but it is marginal. Without a scope to carefully compare circuit timings between the HEF and CD 4093 ICs, any solution is going to be down to experimentation. The timing differences between the devices will be subtle.
I would suggest playing around with passive component values, as I said, there are a number or R/C time constants. The simplest way is probably by temporarily adding 'padding' components on top of existing passives rather than substituting them. eg. try adding a resistor of approx. 5 times the value of a resistor to speed up an R/C time constant by 20% or patching on a capacitor of 20% of the value of the existing capacitor to slow it down.
As a starting point, I would suggest the R/C delays directly related to U17, so R89/C55 on the input of U17A, R86/C54 on the input of U17B, R81/C31 on the input of U17C. Maybe try adjusting each one by 10-20% either way as I described above. Also try adjusting them with a working HEF4093 fitted to see if you can make the circuit fail in the same way, this may be a pointer to where things are marginal. If these don't have any effect, then it may be necessary to widen the area of search to parts of the circuit further from IC17 but still related to its inputs and outputs.
I'm sorry I can't offer a more 'scientific' solution, but without fully understanding the function of the circuit and having it in front of me, with appropriate test equipment, it's all I can think of. It's never easy to track down a marginal problem in a circuit like this.
I also notice that there are several adjustable pots on the board. I would be reluctant to touch these unless you understand their function, but mark their starting position carefully so that you can return them. Also check that the supply voltages are in spec - I can't see if the 4000 series logic is powered from +5 or +15V but if it is the latter, different manufacturers parts might behave differently if the supply exceeds the 18V absolute maximum limit.