16K RAM Pack Schematic / Chip Layout

[PokeMon]

Okay, now it boots.

The RAM pack is not seen yet however. I have been using a working RAM pack for initial experiments to eliminate extraneous unknowns. POKEing 0's and 1's to 8000h PEEKs back whatever was in there before, not the value POKEd.

So when you try to read (with PEEK) from S0 and $8000 - do you get same values ?
And try $0001 and $8001 - will be the same ?

That would point that the /ROMCS hack doesn't work.
Just a thought.

So now RAM pack can be connected without boot failures.
Can you check the wiring in my last schematic, especially the diode to /ROMCS and it's direction ?
You told you experimented with it ...


By the way, the RAM pack can not be found by the system, it is just accessible from program in address range $8000-$BFFF (if ROM shadow copy is disabled).

[TMAOne]

So when you try to read (with PEEK) from S0 and $8000 - do you get same values ?
And try $0001 and $8001 - will be the same ?

Yes, what I'm seeing starting the loop at $8000, is an echo of what is at $0000. After a POKE, the value remains unchanged.

I tried playing with the diode again but it doesn't seem to make any difference either way around. I will check all my connections again.

[PokeMon]

Yes it seems that disabling ROM in section $8000-$BFFF doesn't work.
But I use same technique in my ZX81XRAM which works very well.
I only use a 74HCT4075 (3 input OR) and disable ROM on A15=high.
I use an additional resistor but this is not needed, just to avoid to much stress for the ULA when pushing up and ULA is pulling down.

Clipboard0000.gif (12.29 KiB) Viewed 5541 times

What type of diode do you use ?
Can you measure the /ROMCS voltage with your scope when A15 goes high or the minimum and maximum values ?
Are you sure you connected to the right pin ?
Pin 23 of connector (underside) or pin 20 of the ROM ?

[gozzo]

Another way to disable the ROM in the 32768-49152(and also up to 65535), connect an NPN transistor, collector to +5v, base to A15 from zx81, emitter to zx81 ROMCS ... seen this done somewhere....

[TMAOne]

Success!

My jury-rigging had caught up to me. I had a cold solder joint at the diode.

The board is now merrily testing 16K. Write 0, read back 0, write 255, read back 255, working so far. (This is with a good pack.) So the circuit works, and you're a genius (again) PokeMon.

[PokeMon]

Congratulations.
Now you have to program a test utility for the ram packs.
I would do in both directions, maybe rotate a zero bit between other 7 bits set and after rotate a set bit between all other reset bits.
Rotating or shifting could be done best in assembler.

There are several algorhythms in internet for properly RAM tests - up to you how deep you will do.

If you have some time you could implement a test utility with the so called march algorhythm.

Take a look here
http://www.ece.uc.edu/~wjone/Memory.pdf

Or here
http://larc.ee.nthu.edu.tw/~cww/n/625/6 ... MT0703.pdf

Good luck !

[TMAOne]

Thank-you for the links. Interesting material. I guess there's quite the science to diagnosing the spookier classes of errors.

The tiny program I have already is, (I see now), a simple "march algorithm", intended to spot "stuck at" faults. It is good enough to figure out if a 4116 has gone to chip heaven. Likely, the first byte tested will tell me what I want to know. I've had a fleeting thought of making it smarter, ie. look at the differences that get read back and display something like "REPLACE D3 CHIP", but I don't think I'll bother. Or, maybe I will use it as a goal for learning Z80 assembler, which I've never gotten around to doing.

Unfortunately the packs I currently have which need repair are not in a condition (yet) to make use of this circuit. (One has logic issues, two still have voltage issues.) I don't have room on my bench to keep this "spider" set up the way it is. But now that it's been set up once, next time I'm in that position I know I can set it up again with two tiny pieces of cellophane tape, and six temporary solder joints! That's much better than unsoldering and prying out chips at random until I happen to hit the one that's causing the trouble.

Thank-you PokeMon. It's been a fun ride.

I don't know if sirmorris still has thoughts of making one or more of these up as interface boards to sit between a stock ZX81 and a RAM pack under test. It would be a slick tool for hardware-capable collectors who want to preserve the traditional RAM pack, wobble and all. But as many folks, (including myself), have pointed out, there are better ways to feed RAM to a Zeddy.

Ian

[macjohnmcc]

Would someone be so kind as to post of a clear photo of the chip side of the board for the 16KB RAM pack with the memory chips on it? I desoldered all of the memory chips and in the process forgot where pin 1 is for the chips.

I feel like a dope.

[zx81jens]

Hi there!

I know that feeling (right and left are often switched )

I found that with google: http://www.zdnet.com/photos/sinclair-zx ... 51/6196518

Greetz
Jens

[1024MAK]

I can post a picture of the later single board design (with memory chips mounted on both sides) if that is the type you have.
Another way to tell which way round a chip is fitted, is to compare the pin-out diagram in the data sheet with some easy to trace tracks on the PCB.
So for example, it is normally not hard to work out which PCB tracks are the 0V and +5V supply lines. You can use a meter on the resistance or continuity range and check to the edge connector pins as well.

Mark