Try out the standalone memory test from EXOS 2.4 beta! (twp8ukc.rom)

It is a big bug in the original EXOS test, it is use a CALL instructions to test the system segment. But if the onboard memory is bad then the return address corrupted, the system is frozen, no picture, no sound, no life sign... only see a random picture, because the Nick is running uninitialized and using the poweron random filled bytes from the video memory.

In this test version included my "BEEP test" routine, it is start testing the system segment, writing 00h,FFh,55h,AAh to each bytes and reading back, and say with beeping sounds the readed bits, low sound for 0 high sound for 1.
With this method you get confirmed the machine is least partialy working. And if you detected which bit is bad then know which memory IC is needed to replaced.
This test not use CALL instructions, only run in the ROM.

The BEEP test is started immediately if you put the EPROM to the motherboard ROM socket, using as test EPROM.
If you use on the normal place (the full EXOS 2.4 or the standalone version in the cartridge at the 04h segment), the BEEP test can be forced by pressing down the B key at power up.
And normaly before testing start the system segment (using CALL instructions), test the first byte with all 00-FFh values, if error detected then the BEEP test started.

One additional experience: I found few defective Z80 cpus in dead EPs which is cannot write memory correctly.

What size PROM does this EXOS 2.4 require?

It is a 64K (27C512). For using this need a connect new address line, A15, from the expansion port B9 pin.
(The BASIC is included, then no cartridge needed.)

The standalone memory test part, can be used in the cartridge. It is a 16K (27C128), or practicaly also add the BASIC ROM, then the total is a 32K (27C256).

If you made the new test EPROM, and if you get picture then try the Advanced test (pressing A at power up or pressing C+A when pressing Reset), it is more accurate then the original EXOS RAM test.
When all segments tested it is waiting for key press. This is for Refresh testing, you wait few minutes before pressing key, then it is will check again the previously writed pseudo random patterns.

I think it is about 99% the problem with the memory or the CPU.

Are these good enough?  I have not checked the spec of the 4164's that are already there,

It is good, 150ns.
150,200,300ns IC-s built in the EP machines. If you want to make 6 or 7.12Mhz turbo then the 150ns is recommended

Starting to get a little confused!

I powered it up with EXOS 2.0 and for the first couple of times it was OK
but I glimpsed a memory error (tv had box in the way)! 

After it has been on a few times, then it seems to stop working!

I put the EXOS 2.4 Beta 2 in and this has no effect, that is assuming I have
programmed them correctly 

I checked the voltages.

PSU output +13v, +10.5 under load.

+5A is running at +4.90
+5B is running at +5.07
+12 is running at  +11.72

Well this is testing part 1!  Need to try and find out if I have programmed the
EPROM correctly....

Russ

looks like eprom not programmed correctly as it did not
work in my working ep128.

Try and get it reprogrammed soon

I am using ATMEL 27C512R IC's and file EXOS24UK.ROM