Lou - N2MIY
Veteran Member
One of my teacart RX02s went on the fritz. It would not come ready and I could not tell why. To zero in on the problem, I swapped cards with another known good RX02 and narrowed the problem down to the M7744 controller module.
I put that M7744 in the rack mount RX02 I already had out on the bench. I hooked up the logic analyzer to the address lines on the microprogram roms to see where the microcode was getting hung up. (for those who don't know, the RX02 uses two 4-bit slice 2901 microprocessors) Fortunately, Chuck Dickman has posted the disassembled code in the RX02 Roms : http://www.chdickman.com/rx02/ look down under RX documentation.
So I watched the code start from location 0000 and followed it in the disassembled code listing. The problem was amazingly clear - a return from a JMS was not returning to the right address. The two MSB of the return address were stuck as zero. The RX02 technical manual explains explictly how JMP/S works on page 5-29 and 30. The stack for subroutine return addresses is in the 2909 microsequencers. My broke bits were in the most-signficant 2909 at E66 (the one on the left in this photo: http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17398 ).
I had no spare 2909 around, but I did have half a dozen 2911s. 2911 is a simplified lower pin count version of 2909. It's simplified by getting rid of the four usually unused OR inputs and the D and R inputs are tied together. My bad 2909 at E66 could not be substituted with a 2911 because it used the separate D and R inputs (see the print set, sheet 7). However, the other two 2909s at E59 and E47 were eligible candidates for 2911 substitution.
I carefully desoldered the E47 2909. I desoldered the bad 2909 at E66, put in a socket, and plugged in the one from E47. At E47, I soldered in a long tail wire wrap socket and made a 2911 adapter socket that engaged the back side tails. See here: http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17397 and here with the adapter plugged in: http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17396 . The finished product looks like the first picture earlier in this post.
The repair worked perfectly and this M7744 is back in business. If I ever find any real 2909s, I can cut off the wire wrap tails and plug it into the socket on the front of the board.
Lou (not bad for half a day's work.)
PS. The Grayhill dip switches (http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17400) on this board were awful. Don had mentioned this problem in an earlier thread. I needed to change the position of S1 to select RX01 mode and had to work this switch back and forth a bunch to get a low resistance connection. So, watch out for these Grayhill switches. I had a bad experience with one on a DL11-W in the past.
I put that M7744 in the rack mount RX02 I already had out on the bench. I hooked up the logic analyzer to the address lines on the microprogram roms to see where the microcode was getting hung up. (for those who don't know, the RX02 uses two 4-bit slice 2901 microprocessors) Fortunately, Chuck Dickman has posted the disassembled code in the RX02 Roms : http://www.chdickman.com/rx02/ look down under RX documentation.
So I watched the code start from location 0000 and followed it in the disassembled code listing. The problem was amazingly clear - a return from a JMS was not returning to the right address. The two MSB of the return address were stuck as zero. The RX02 technical manual explains explictly how JMP/S works on page 5-29 and 30. The stack for subroutine return addresses is in the 2909 microsequencers. My broke bits were in the most-signficant 2909 at E66 (the one on the left in this photo: http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17398 ).
I had no spare 2909 around, but I did have half a dozen 2911s. 2911 is a simplified lower pin count version of 2909. It's simplified by getting rid of the four usually unused OR inputs and the D and R inputs are tied together. My bad 2909 at E66 could not be substituted with a 2911 because it used the separate D and R inputs (see the print set, sheet 7). However, the other two 2909s at E59 and E47 were eligible candidates for 2911 substitution.
I carefully desoldered the E47 2909. I desoldered the bad 2909 at E66, put in a socket, and plugged in the one from E47. At E47, I soldered in a long tail wire wrap socket and made a 2911 adapter socket that engaged the back side tails. See here: http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17397 and here with the adapter plugged in: http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17396 . The finished product looks like the first picture earlier in this post.
The repair worked perfectly and this M7744 is back in business. If I ever find any real 2909s, I can cut off the wire wrap tails and plug it into the socket on the front of the board.
Lou (not bad for half a day's work.)
PS. The Grayhill dip switches (http://www.vintage-computer.com/vcforum/album.php?albumid=63&attachmentid=17400) on this board were awful. Don had mentioned this problem in an earlier thread. I needed to change the position of S1 to select RX01 mode and had to work this switch back and forth a bunch to get a low resistance connection. So, watch out for these Grayhill switches. I had a bad experience with one on a DL11-W in the past.
, 3 bad