Re: Cirtech GS 8 plusRAM woes...

["Michael J. Mahon" <mjmahon@aol.com>]

amotyka wrote:
> On Apr 7, 8:15 pm, Michael J. Mahon <mjma...@aol.com> wrote:
>
> > "ict@ccess" <gids...@sasktel.net> wrote:
> >
> > > On Apr 5, 10:47 am, amotyka <anthony.mot...@gmail.com> wrote:
> > >
> > > > Hi guys,
> > > > I have recently come into posession of a Cirtech GS 8 plusRAM board,
> > > > having given up on ever receiving the Sirius ram card I ordered a year
> > > > ago... Cirtech was a British company, which is probably why I can't
> > > > find much info about it on the net, only photo's, no manuals, and I
> > > > believe this was the first 8mb board available for the iigs. The
> > > > trouble is, ever since installing it I have been plagued by random
> > > > crashes. Replacing the card with my old 4mb AE GS-RAM III makes the
> > > > system (almost) impossible to crash. I have read old posts about 8mb
> > > > cards causing trouble with DMA and rom 3 machines, but I'm using a rom
> > > > 1 and the cards installed are reactive micro's Microdrive (not the
> > > > turbo version so no DMA), the apple scsi card (not the high-speed
> > > > version so again no DMA) with cd drive attached, uthernet card (no DMA
> > > > to my knowledge) and a Transwarp clocked at 16mhz. The card checks out
> > > > fine with mega memory tester.
> >
> > > > I'm going to run some more tests e.g. with the transwarp at a lower
> > > > clock and reducing the ram on the board, but would be interested to
> > > > hear if anyone else has encountered similar problems with 8mb boards,
> > > > or if anyone can give me some more info on this particular card. I
> > > > read elsewhere that this card also has a rom disk feature, and in fact
> > > > has two rows of header pins which I presume would be used to program
> > > > it.
> >
> > > > Thanks,
> >
> > > > Anthony
> >
> > > I have a Sirius 8 mb card and I was getting quite a few crashes as
> > > well.  I did the power supply mod in which you change the 18 guage
> > > wires to 12 guage.  This helped quite a bit and cut my crashes down to
> > > 1 a day.  I would liked to have done one more mod.  The ground trace
> > > is supposedly not heavy enough since it goes all the way to the
> > > opposite side of the board where the RAM slot is.  A ground wire could
> > > be soldered at the power connector and go to the ground terminal of
> > > the RAM slot.  I was lazy and ok with 1 crash per day, so I didn't
> > > test this one out.   But I would recommend trying a larger power
> > > supply you can get fromwww.reactivemicro.comand also do the ground
> > > wire to RAM slot mod.
> > > I use a ZIPGS rated at 12.5 Mhz in slot 3, ramfactor in slot 5 and
> > > CFFA in slot 7.
> >
> > > Rob
> >
> > That suggests to me that putting several 0.1uF capacitors between +5v and
> > ground, either right at the slot or, preferably, on the memory card itself.
> >
> > It shows the symptoms of being insufficiently decoupled.
> >
> > Ground bounce could also be an issue, but it would take a scope to tell for
> > sure.
> >
> > -michael  -  NadaNet 3.1:  http://home.comcast.net/~mjmahon
>
>
> Hi Michael,
>
> My card actually has around 40 capacitors soldered in between the ic
> sockets, so I don't think decoupling is a problem.
> I'm beginning to suspect that maybe the card can't keep up with the
> transwarp running so fast. My guess is the GAL(s) on the ram card are
> only supposed to go so fast, especially since this was the first 8mb
> board produced, its creators may never have expected the clock to go
> above 3mhz.
> I'm going to try running at at stock iigs speed, which will bring back
> painful memories, and see if it works. Then I may have to face a
> difficult choice between speed or memory...

I understand, but I thought that accelerator cards did not, in fact,
run the memory any faster than "regulation" speed.  They have caches
that run faster, but their interface with the rest of the system is
at standard speed.

Computer busses cannot, in general, be sped up by aftermarket
gear.  They have a timing design and that's that.  That applies
to the external memory bus of the IIgs as well.

Of course, as you increase the accelerator speed, certain timing
margins may change slightly in the wrong direction, and there may
be small changes in power dissipation in the memory due to a changing
proportion of reads and writes.  These effects would usually have no
practical effect on a machine with good timing and voltage margins,
but if your machine is "close to the line", they could certainly
affect stability.

It might be worth checking all the decoupling capacitors on the
board carefully.  I once had a 0.1uF cap decoupling the 6502 on
my ][+ develop a subtle break in one of its "legs", with the
result that the 6502 became quite unstable.  Resoldering the cap
fixed the problem completely.

You could have a similar problem with one or more DRAM chips,
resulting in stability issues only when certain address ranges
are used.

-michael

NadaNet 3.1 for Apple II parallel computing!
Home page:  http://home.comcast.net/~mjmahon/

"The wastebasket is our most important design
tool--and it's seriously underused."