From perry Tue Apr 24 19:48:21 1990 Received: by cbmvax.cbm.commodore.com (5.57/UUCP-Project/Commodore Jan 13 1990) id AA28737; Tue, 24 Apr 90 19:48:18 EDT Date: Tue, 24 Apr 90 19:48:18 EDT From: perry (Perry Weinthal) Message-Id: <9004242348.AA28737@cbmvax.cbm.commodore.com> To: albright, bbaker, daveh, grr, gunther, perry, phil, rick, thomas Subject: A2500/30 problems. Cc: augi Status: R A2630 Fault Analysis, by Perry Weinthal, PA CBM-WC The A2500/30 is currently having DMA problems with the A2630 and A2091. The symptoms are only apparent while performing a DMA transfer while using 32 bit RAM on the A2630. A DMA transfer to either chip RAM or to 16 bit expansion RAM will be performed properly. ------------ Late Breaking News! ---------------------------------------- Well, guess what, all the junk below can be reduced to the following. When the jumper was placed on the A2630 (J302) to allow for operation WITHOUT the 68000, the system worked! This leads me to suspect bus loading or the equations of PAL U304 and the others dependent on it. However, I can not point out what is the actual problem. I have provided the following as a guide, I would like to help in the actual debugging of this problem. Dave or George please call me when you are ready to work on it. ------------------------------------------------------------------------- The DMA problem is cited by only one of the byte of the word transferred being incorrect, and with bits being dropped -- a $FF becomes $5E, but a $00 will not become a $03. This was apparent only when a $F was part of the test data, ie. $AAFF became $AAF0, $F05A became $005A. Connecting a logic analyzer up to the 68000 socket caused the problem to disappear leading me to suspect bus loading on the CPU side. However, after closer investigation this abnormality disappeared without any reasonable explanation. The next test was to take a working DMAC and try it in a non-functional A2091, this also yielded negative results. Other items tested, Gary and buster from the same working machine. The revision to 6.2 for the A2000 switched the tri-stated bus from a pull-high to a pull-low bus. This was done by moving the resistor paks RP900 and RP901 to the leads of the U603 and U604 on pins 1-10. The common of the resistor paks are connected to ground, The CPU data lines and _DIR line of both U603 and U604 are loaded by the RP. The value currently being used is 3.3K -- which is correct. When I moved the common of the resistor pak from ground to +5V, the screen would blink shades of gray when the 68030 tried to run, but the system would work just fine on the 68000 when the CPU selected was the 68000 from the CPU select menu. I was informed that the bus condition should make no difference to the A2630 ROMs, during a brief discussion with Johann George. If this is the case, then I can only infer that the results of the system being unable to boot-up would be caused by bus loading problems. Sample info: A2630 test code results ~30 units. A2630 hardware testing 3 units. A2630 w/o 68000 1 unit. - with ~40 passes before leaving Test code A2500/30 Burn-in, written by Linda Thomas Have a good Day. Perry Weinthal, x9380 From perry Tue Apr 24 20:12:56 1990 Received: by cbmvax.cbm.commodore.com (5.57/UUCP-Project/Commodore Jan 13 1990) id AA29184; Tue, 24 Apr 90 20:12:54 EDT Date: Tue, 24 Apr 90 20:12:54 EDT From: perry (Perry Weinthal) Message-Id: <9004250012.AA29184@cbmvax.cbm.commodore.com> To: albright, augi, bbaker, daveh, grr, gunther, perry, phil, rick, thomas Subject: A2500/30 more news "I LIED" Status: R Well, did that get your attention? I should of gone home and got the bad news tommorow. Well, here is the bad news. With the CPU bus set-up as tri-stated LOW, the DMA test failed 38 out of ~180 cycles. Atleast, we have a clue! I have set-up the CPU bus for tri-stated HIGH, and will see the positive results in the morning! ^ | + And I never met Murphy :}] Well, have a good nite. Perry From perry Wed Apr 25 10:12:58 1990 Received: by cbmvax.cbm.commodore.com (5.57/UUCP-Project/Commodore Jan 13 1990) id AA15516; Wed, 25 Apr 90 10:12:55 EDT Date: Wed, 25 Apr 90 10:12:55 EDT From: perry (Perry Weinthal) Message-Id: <9004251412.AA15516@cbmvax.cbm.commodore.com> To: albright, augi, bbaker, daveh, grr, gunther, perry, phil, rick, thomas Subject: A2500/30 GOOD NEWS, in the morning! Status: R Hello Gang, This system survived the night with over 3600 DMA passes and not even a single FAILURE! What did I do to get the wonderful results? Pulled the 68000 out set the jumper J302 and placed 3.3K pull-ups on the CPU side of U603 and U604, without connecting to the _DIR line on pin 1. Dave, I would like to help resolve the problem even further. I have three sets of PCB (A2000, A2091, A2630) that exhibit the problem, only one has been modified. Perry, x9380 (I told you, I never met Murphy)