U009@CCIW.BITNET (11/30/89)
Further to the previous postings regarding my problems, George Dinolt (whom I cannot reply to, sorry George) wrote with similar experiences. He stated: > After a lot of experimenting I determined that, on my system at least, > the problems had to do with the position of the cable from the Memory > Management unit to the Datafree board. ... > ... test this yourself by just letting the Datafree memory board hang > loose outside your system (You can leave the plastic cover off. You > can even leave the metal shielding off for testing purposes. My ST > sits at the edge of the desk so the memory board literaly hung from > the two cables in my testing. Dave Clemans (also unreachable) wrote: > I would rather doubt a MMU chip problem. It is MUCH more likely > that you are seeing ringing and other noise problems. A good way > to check for this is to run a decent memory tester program when > the board is set up for 4MB, and see where it gets errors, and what > the patterns of error bits are. If you get randomness, it is almost > definitely ringing or other noise problems on the dram control lines. Well there were never any memory errors with the memory testers I used and the system was working OK under Spectre, Multifinder with lots loaded, but the two comments together rang (!) the bell. Now, the Datafree board is supposed to be able to fit above the top shield of the motherboard (according to their documentation), but I found there was insufficient room between it and the keyboard, so I left it off. Maybe newer systems are different. George's suggestions about operation when the board and cables dangled out the front of the disasembled ST, in particular when the ribbon from the MMU was routed away from the motherboard was a great help. I made a shield from aluminum foil covered on both sides with "MacTac" sticky plastic (for insulation). I made it about 1/2 inch larger than the memory board dimensions but left a strip of foil about 1/4 inch wide uncovered along the front (long) side. I folded the exposed strip back, punched a couple of holes thru it an the insulated part where the screws holding the motherboard were. I then fastened it down with the original screws and a couple of small washers. Dave's comments about ringing and noise suggested I should take a look at shielding and/or terminating the ribbon cables. I figured I could do both simultaneously by increasing the capacitance of the ribbon's conductors as follows: I also completely wrapped the MMU cable with a long narrow strip of aluminum foil, then covered it with tape to insulate it. I wrapped a bare piece of wire with the foil at the MMU end and soldered the wire to the bare groundplane on the mother board. The added capacity on the wires would slow the rise time of the pulses and reduce the ringing, if any. After re-assembling the machine, the problems seem to be gone. I copied small files til the floppy was full several times and no crashes. I was going to supply power to the board thru a separate pair of wires too, but so far, it has been unnecessary. Make sure the foil doesn't touch anything except the ground plane at the front of the motherboard. Heat dissipation doesn't seem a problem because the shield is covering the old memory, which is disabled anyway. Thanks again George and Dave for the leads. I now seem to be fully operational with 4 MB of memory. Regards, Stu Beal, VE3MWM, (U009@CCIW.BITNET), National Water Research Institute, Burlington, Ontario, Canada.