DIY RAM EXPANSION

[Dave]

That's very....spacious

I only add that it is best to use the slowest GAL you can - at least a 20 or 25. 15 heads into marginal territory for some QLs, and 12, 10 or 7 often have issues with noise.

Good job

[tcat]

Dave,

Thank you for continued encouragment, I am afraid my GAL is 15ns, I was not able to get 20 or 25 in DIL. This might be possibly a source of trouble.

TCAT

[Dave]

Try it. It will work.

The problem comes if you have a noisy system, or one with lots of expansions putting all kinds of noise on the expansion bus. The faster GALs are sometimes quick enough to recognize spurious input as a valid signal. This means the device mostly works, but you might get the odd data corruption, etc. At 15 I'd say you're OK, as I am conservative and gave a good margin on that.

Many QubIDEs actually use a 15 and a 20, or a 20 and a 25, because the signals used by one of the GALs are a little more prone to noise than with the other.

If this is the only device in the system, and it's not on a backplane (aka an antenna designed to pick up every bit of noise in the street) it's very unlikely you'll have any problems at all.

An over-abundance of caution. That is all

I do have a question. Why did you choose that particular memory? I have found the 628512 is easy to get and very affordable in a few different packages. I didn't really look at other devices, so I am curious what motivated you to choose that?

[tcat]

Dave,

I somehow missed your question, I am replying now.
The memory part# 628512 is exactly the one I chose, but the inscription on the chip itself I got is AS6C4008. Mine is Alliance brand.

TCAT

[Dave]

Ok.

[tcat]

Dave,

Would you mind looking at my GAL equations I posted earlier in the thread, I really recoded logic what Brane suggested. Leaving out FCx now for simplicity. To get logical 1 or 0 I hard wired CLK and OE inputs to VCC and GND acccordingly. I actually only need 1, so 0 is left for convenience.

http://theqlforum.com/viewtopic.php?f=2&t=877&start=10

Also, do I have to care about power on state, what happens when voltage goes up, and QL performs some RAM checks, etc.

Many thanks
TCAT

[Dave]

I'm not the best person to look at your GAL equations. I have someone behind the scenes who does custom logic for me. They read this forum, and if there is any huge flaw with your work, they will do a lengthy post explaining nuances so subtle you will be enlightened about many subjects However, it looks good. Just keep an eye on which signals are active high and active low.

The QL takes a while to power up and start the memory check - as long as the expansion doesn't have so large an electrolytic on it that it takes longer to come up than the QL, there shouldn't be a problem. Looking at what you've done, I'd say you're going to be fine.

[tcat]

Hi All,

I had some time for soldering, please see how it went, but my first humble attempt ends in GREEN screen after QL powers up. The actual boot starts with usual colour random bits, folowed by GREEN screen, and takes considerably longer with the card inside. Having no osciloscope, all I can check is the supply voltage, seems correclty at some 5.33 V, measured on VCC pins.

I will be probing all tracks with my mutlimeter just to make sure, I have not left out anything.

Meanhile any suggestion of yours greately appreciated.

Many thanks in advance.
TCAT

[tofro]

The RAM test is divided into two parts, the first where it checks that it is possible to store ones, zeros and random pattern in every RAM location, and the second part a few seconds later when it checks that the stored data is still there. A white screen means that the first part of the test has failed and a green screen means that the second part has failed.

This would hint that refresh is not working as intended.

Tobias

[tofro]

The RAM test is divided into two parts, the first where it checks that it is possible to store ones, zeros and random pattern in every RAM location, and the second part a few seconds later when it checks that the stored data is still there. A white screen means that the first part of the test has failed and a green screen means that the second part has failed.

This would hint that refresh is not working as intended.
On second thought, that's probably nonsense with static RAM

Tobias