On Tue, Jan 8, 2013 at 7:46 AM, Al Kossow <aek at bitsavers.org> wrote: That's been my experience thus far... It looks like Christian Corti's worked out how to do this, but I don't know if he's made the archives public: http://computermuseum.informatik.uni-stuttgart.de/dev/ibm_5110/technik/en/i… The 5120 ROS is identical to the 5110 as far as I know. I don't know how applicable the above is to the 5100, I assume it's a bit different (and I don't have one to try it out on). It'd be nice to know what chips correspond to what portions of the ROS tests -- I have two bad Command/Language ROS boards from the two 5120s I have and they fail diagnostics in different places. I'm sure I could make one working board from the two if I knew which chips were bad on each one. It'd also be fun to find some software for this, but I haven't tracked any down. - Josh

Josh Dersch wrote: Yes, there might be older and newer versions of the code, and possibly the versions of executable and non-executable ROS may need to match, but generally they should be interchangeable between the 5110 and 5120. (The "computer" portion of the 5120 is considered a 5110 Model 3.) There are substantial differences between the 5100 and 5110/5120 software such that I don't think any of the ROS code, executable or non-executable, is interchangeable between them. I can read any of the ROS modules out-of-circuit, and can read the non-executable ROS in-circuit using a logic analyzer, but my 5100 has bad ROS modules so I haven't bothered reading those. CHM has a working 5100, but my request to read the ROS modules was not well-received. I can understand that they are worried that I might damage them, but they'll likely fail within the next 10 years anyhow. The Maintenance Information Manual has enough information to identify the failing module. I don't think there's much point to identifying it to the chip rather than just the failing module, since it's basically implausible to repair the modules. The simplest repair is to build a replacement module using a modern flash memory chip and a CPLD. That's how I plan to fix my 5100 if I can get a dump from a good machine. I've also considered building RAM modules suitable for replacement or expansion. Eric

Josh Dersch wrote: It's somewhat difficult to remove them without damaging them. The bigger problem is that the chip you replace it with can't really be expected to last much longer. At this point they're all around 30 years old. I'm not certain what the failure mode is, but they seem to be past their best-by date.