Ql problem

[tofro]

Tobias can you give me a link where I can see how I can check the memory with the scope?
Thanks Tassos

Well, not a link, I'm afraid.

Dave has already given an outline of what you need to do - With a simple scope you will, however, not be able to find single memory cell failures (but in about 99% of the cases, the whole chip will fail anyway).
If it's not a complete memory chip that fails, Minerva will come to the rescue.

What you want to do is
1. Visual inspection: Thoroughly inspect upper and under side of pcb for any obvious shorts between adjacent data or address lines

Now comes the tedious part: There's a lot of lines to check....

2a. Collect a number of additional hands. I know that most of us have only 2, but you could rather use like 3 or for for this job....
2b. Using your scope, make sure all the chips have a working 5V supply
3. For each chip, set a trigger on the raising edge of /W and check that you see pulses on the Q pin (that's a read cycle) This makes sure the data output is able to drive the data lines
3a. Set a trigger on the falling edge of /W and check that you see pulses on the D pin (that's a write cycle)
4. for each chip, make sure you can see pulses on the address lines, RAS, CAS, W, D and Q, just as Dave has written.

The pulses you see should go 0...0.8V for low, ~4.0 to 5V for H, anything in-between shows a problem.

You might need to press the RESET button every now and then, as you want the QL to stay "alive" during this process.

Hope this helps
Tobias

[cybervet]

Ok the scope arrived yesterday , it's a DSO nano simple scope, so Iam ready to test the memory.
I just wanted to ask what it means "3. For each chip, set a trigger on the raising edge of /W and check that you see pulses on the Q pin ".
I understand that I have to put my probe on the Q pin but I don't understand the "set a trigger on the raising edge"
Any help is greatly appreciated.
Tassos

[tofro]

Tassos,
"normal" scopes allow you to tell them when to start recording. That is the "trigger". Don't know whether yours is able to do that.
If your scope doesn't have a trigger input, just try and watch both Q and W side by side. Once /W goes high, that means someone wants to read the addressed cell and the RAM should set Q according to the content of RAM a few µseconds later. Note you won't necessarily see a "1" on Q every time you see /W going high (that means the addressed cell contains a 0 bit). But you should verify that the RAM is able to rise Q at least once.

Hope this helps,
Tobias

PS: Nice scope by the way - I've got an old Tektronix one about 100 times the size of it that needs a carriage to be moved around

[tofro]

Tassos,
checked the DSO manual. This guy seems to be single-channel only. In this case, I would just like to watch /W and Q and see they're changing levels and not stuck to high or low.

You still can do all the measurements, it's just a bit more work.

Regards,
Tobias

[Dave]

In this case, Minerva is the best way to locate the bad location. During boot, it finds the bad memory and allows the machine to start with reduced memory. You can then test the faulty location to work out which chip is bad.

I do have a Minerva ROM and some spare RAM chips, if it turns out to be more than a solder issue.

[cybervet]

I used the minerva and each time I start up the computer a different address is displayed on the screen, sometimes 2 different one after another and then stops at a black screen with the minerva logo down right
Any ideas?
Tassos

[tofro]

Tassos,

can you give some examples of the output you get from Minerva?

Maybe all of the failures can be traced down to one single faulty chip.

But maybe not - Then we need to look somewhere else.

Minerva does the RAM check from high to low addresses. If it finds faulty memory, it will restart the memory test from there, excluding the RAM from the faulty address to the end of RAM completely. And if the "good" RAM it has found in the end is not enough to run the OS, it will just sit there doing nothing.

Regards,
Tobias

[cybervet]

Usually I am getting :
FFFFFFFF
7F7F7F7F
00030000
then others follow I will try to write them down also.
Tassos

[cybervet]

After another start I am getting
00506636
00D0E636
00030000
and then
D2D2D2D3
D2D2D2D2
00010814
and then
D2D2D2D3
D2D2D2D2
00020360
and then MINERVA------> reset
FFFFFFFF
7F7F7F7F
00030000
and then MINERVA-----> reset
7BC0173A
FBC097BA
00030000
and then MINERVA---->reset
115BFFFF
115B7FFF
00030000
and then MINERVA-----> reset
MINERVA

[tofro]

Tassos,
the values are expected to mean
"Value written"
"Value read back"
"Memory location"

From the examples you're giving, I can't see a directly obvious pattern, which doesn't really point to a specific RAM chip. (Maybe someone else does?)

I would rather assume you're suffering from noise or spikes on the data bus or a weak power supply.

Some other questions:

Does one of the RAM chips feel particularily more warm than the others after a few minutes?

A very long time ago I used to find faulty RAM with the following method (It's a bit blunt, but sometimes works):

Find yourself a spare RAM chip that you know works.
One after the other, simply "piggy-pack" (i.e. just set it on top of the others, maybe with slightly inward bent pins to improve contact, no soldering) the known good RAM on top of one of the inbuilt chips (make sure pins don't overlap, chip orientation is correct, and also make sure the QL si switched off when placing the chip, also, don't leave the QL on for more than maybe 30s with the chip in place). Once you find a location where the RAM test runs through, you've got the faulty chip.

But I would rather assume the problem outside the RAM area.

Regards,
Tobias