acthom@ihlts.ATT.COM (thomas) (06/09/88)
OK you hackers on the net, I have a stickler of a problem. I bought a 1764 - 256K ram expander for my c64. I read the articles on the net about expanding it to 512K, and, after waiting for the warranty to expire, I popped the cartridge open. Things looked easy. There, hunkered down in the front left was a great, square controller chip; in the middle was a row of traces just waiting for sockets. So I cleaned out the holes, soldered in sockets, and plugged in 256K memory chips. Oddly, the 512K trace was already cut. Having modified the RAMtest program according the the GEOworld article (a couple of pokes in the basic portion to access the upper banks), I gave the now 512K expander a try. The test whizzed thru the lower 256K of memory and picked up an error in the middle of the new 256K. And rerunning the test picked up what looked like a hard error at the start of the new 256K bank. Ah, a bad chip, I thought; plugged in a spare, and it did the same thing! Two bad chips? For a while it looked like the expander liked TI 256K chips and nothing else (but I could only find 7 of the TI chips, so I'll never really know, the 8th other chip masked any other problems). After weeks of probing and borrowing every 150ns and 120ns chip I could find, I brought the little bugger into my local CBM repair depot, where the fellow in charge said "You did WHAT!" Fortunately he was impressed with my soldering job and the sockets, so he decided to take a look at it; to no avail. The memory chips are good, they work in a 1750. Another 1764 controller chip does the same thing. So he asked me to get back on the net, and ask around if anyone else has seen this problem: If you have modified your 1764, and it has the square chip (not the old rectangular one), and you have had a problem of the new memory working somewhat, and then going to a consistent failure mode ... how did you fix it? (The current theory is that there are two versions of the controller chip, and the ones that they put in the 1764 are the ones that failed the 1750s, but could handle 256K.) Does anyone out there have any ideas? Thanks, Al Thomas.
hedley@cbmvax.UUCP (Hedley Davis) (06/10/88)
In article <1068@ihlts.ATT.COM> acthom@ihlts.ATT.COM (thomas) writes: > >The memory chips are good, they work in a 1750. Another 1764 controller >chip does the same thing. > > > (The current theory is that there are two versions of the controller > chip, and the ones that they put in the 1764 are the ones that > failed the 1750s, but could handle 256K.) > >Does anyone out there have any ideas? > Thanks, Al Thomas. Sounds like something to do with your PCB. 1) Remove all ram from board, socket lower 256 K. Try old chips to verify you didn't bugger something in this step. Then once that works, try new chips in lower 256 K. If failure, then your rams are fubar. If good, then something is bogus about the second 256 K, and it probably has to do with the PCB patterns. You have an open or a short or a cold solder joint, or something silly like that. Note that chip inside of the little square package, and the longer 64 pin package is actually the same, only the package has been changed. As for your comment that we are sorting controller chips, sorry bosco, but we don't. All chips shipped pass the same tests. One tester program. One other comment, about that jumper. The jumper is misleadingly labeled. What it does is two things. #1) cause a bit in the status register to read one or zero. #2) Change the address multiplexing for the DRAMS for 64Kx1 or 256Kx1 ram chips. The labeling is due to historical reasons regarding revised plans in the engineering & marketing issues surrounding the product line. ( Harumph ! :-) ). Hedley