-----Original Message----- From: cctalk-bounces at classiccmp.org [mailto:cctalk-bounces at classiccmp.org] On Behalf Of Charles Sent: vrijdag 6 januari 2006 1:25 To: cctech at classiccmp.org Subject: Re: RL02 write faults, found it, now how to fix it? I spent an interesting day in the shack today. After blowing off the dust, I found that my old 7D01 logic analyzer won't trigger on the word recognizer but Ch. 0 would trigger, so I started hooking it up to the RL02 logic board and R/W boards using the drive error as a post-data trigger and looking backwards from there. Eventually I learned that Write Gate and Write Data are being correctly asserted, and the data bursts continue for 500+ uS (one sector) beginning a couple of uS after the Sector pulse. But approximately 4 to 6 uS after the Write Gate opens, and before the data could even begin (10+ uS), the DL5 DRIVE ERR signal asserts! I worked around the loop from there and found that the *first* thing to go wrong was the Write Data Error Latch ('LS279, E6 on logic board) which was setting its output pin 13 (DL5 WR DATA ERR H). This fatal error signal then latches DL5 DRIVE ERR H via the other "glue" logic and prevents any further action. But the R/W board that generates Write Data Error Pulse was *not* telling the latch to set! There is a 6K8 resistor to +5 and a 0.0056 uf cap to ground as a delay (set only) in the error circuit (a time constant of 38 uS) so to reach a logic 1 (2.0v) would take considerably longer than 6 uS anyway. Further exploration with the logic analyzer on "high" speed (20 ns/tick) shows that on the 'LS279 pin 15 (the latch input, WRITE DATA ERROR PLS L) there is a small (~180-200 ns) low pulse with a 20 ns glitch in the middle. It is NOT evident to the logic analyzer when examining the signal's source on the R/W board, (Write Data Error Detector E3, 7404 pin 8). But it's enough to set the WDE latch all right! Right about then, sparks started flying from the brush area of the spindle motor and the drive faulted on its own and spun down. Holy crap! =:^O I pulled the circuit breaker fast and gently spun the motor and could hear scraping noises. I figured the brushes were shot or something, so I took the two nuts off the through bolts. Of course the end housing wouldn't come off, way too tight a fit, and I didn't want to pull the blower motor. With a bright light I could see a piece of something blue sticking up, so I pulled it out with a pair of fine needlenose pliers. It was the tip of one of my logic analyzer leads that had sneaked into the vent holes unknown to me, and been cut off by the armature turning! I put the nuts back on (of course I dropped the second one inside the motor and had to fish that out too), crossed all my fingers and flipped the breaker. It spun right up and has been working fine since. Whew. Never a dull moment in Murphy's Laws. Anyhow, there is a 330 pf deglitching cap near the input to the 'LS279 but obviously it's not adequate. I need to put a fast scope on the line and see what the edges actually look like. Some resistive termination may be needed. This is on a very short ribbon cable to J6, about 6", inside the drive, and the chassis ground wires to each board are connected. Or maybe the Berg connector on one end is defective in the ground lead. More as I discover it. -Charles