Re: SIMM 32 MB :-(

From: Magnus Kollberg <Magnus.Kollberg_at_emw.ericsson.se>
Date: Wed, 25 Feb 1998 08:55:41 +0100

Hi,

> you surely remember me asking if 32 MB EDO SIMM will work in AB40. Since I
> got six positive responses, I went and bought the SIMM. I installed it and
> turned on my computer. Doug's TK 5.07 found both SIMMs, performed mem test
> and announced 48 MB of FastRAM to the TOS.

Cool. What does Doug's mem test do?
 
> So I started TTRAMTST 1.2, just to be sure. Soon it found an error - it
> wrote $ff but read $fb from about 100 addresses (with 4 bytes offset at
> about $28b8xxx). I couldn't believe it so I tried Speichertest 1.0 and
> later also RAM.XFS and TreeCheck. Still there were the errors!

Yuck!
 
> Interesting is that Linux bootstrap finds 48 MB in bank 1! That's strange,
> since banks are 32 MB max.

Says who? :-)

I have seen figures for 128MB AB040s but I don't know of anyone testing it.
Or maybe someone have?
 
> So I tried to exchange the two SIMMs, so 16 MB was in bank 1 and 32 MB in
> bank 2. Results under TOS were the same. However Linux bootstrap found 16
> MB in bank 1 but only 28 MB in bank 2!!

Wacky.
 
> So I guess that:
>
> 1) my AB40 is broken and can't handle 32 MB SIMMs properly - instead it
> mirrors last 1/4 somewhere else, or the last 1/4 loses data
>
> or
>
> 2) the SIMM is broken and loses data
>
> What do you think? Can anybody come with an advice?

Try the SIM in a PC and test it. Should be the esiest way to find out where the
problem is.

Does the errors indicate that a address line is broken? Should be easy to see
if you look at which address are broken. It can't be a data line because then
you should have the errors on all addresses.
 
> BTW, I don't understand the SIMM - it's doublesided (16 chips), no
> manufacturer (just some strange logo with circle and square). The chips are
> V53C17405AK50. What's that - 50 ns? I doubt it.
>
> As for my AB40 - there are JP1, JP2, but they're free. Do you think a
> jumper could help me?

Doesn't sound like it.
 
> Anyway, I'm tired and sad it doesn't work.

Me too.

There is a way to test if the AB040 is are working.

Make a program which puts a "walking" 1 on the data bus. Just take a
random address or an address you like and write a 1 and then left shift it
one step at a time and read it back each time. If this works the data
bus is alright.

Now do the same for the address bus but be sure to write different data at
each adress to be sure that you does't read from the same address. Now you
will have tested all address lines and data lines and should be able to
know if the AB040 works or not and see which address line fails. If it does
it's probably a SIM fault (not unlikely). Then there could ofcourse be
a problem somewhere else, but this is something you could try.

If this does not work I guess you have to make a patch to "remove" the
faulty addresses. :-\

I do think the problem is with the SIM.....

//Magnus Kollberg
Received on on. feb. 25 1998 - 10:19:00 CET

This archive was generated by hypermail 2.3.0 : ti. nov. 03 2015 - 20:07:53 CET