You mean the 4028 at IC36? (not IC38)

What's your logic probe telling you about activity on the input enable signals to those two 4724 chips, and all the way back from the 4028 and the 4556 at IC32 which selects the 4028 chips? Is there activity or are they stuck high or low?

Are any of the outputs working properly on the 4724 chips at IC9 and IC10 on the interface board?

Your anomaly here is that an odd ( 3 from 8 ) number of outputs aren't working properly off the 4724's which takes away an obvious simple logic issue.

A logic probe can be your friend. I think it's also worth swapping the select line wires to the decoders on the CN12 connector at the CPU board, just incase it reveals anything of interest.

My only Zaccaria is in pieces at the moment and I can't remember if there's any sort of RAM test on power up.
The two RAM chips "6514" and "2114" are pin compatible. The 6514 is just a low power version of the 2114. If you suspect RAM and they're socketed, try temporarily swapping them around - of course you'll lose any battery backed settings.

Both the Philips and Intel 5101 datasheets specify that the 5101 is pin compatible with the 2101.
5101 is a very low power CMOS version of the 2101 NMOS RAM chip.

Very interesting and well done!
I guess Zaccaria aren't doing any RAM tests on powerup.
Thanks for posting the resolution.