Things were quiet here on the bLog in April. I was busy working on computers but I experienced a string of repair failures. As discouraging as these can be I think it's important to write about them as a way to document lessons learned and to perhaps help others who may experience similar challenges.
The machines were a Compaq 386SX machine, a TI Travelmate 486 laptop, and an IBM PS/note Laptop.
The Compaq Deskpro 386S/20N was labelled as "not working, needs recap" but as my friend John Bumstead has said, when we are troubleshooting machines we need to disregard notes on the machine and do diagnosis ourselves. In this case however it was obvious by doing a visual inspection on the board that it did need a recap. As I removed the capacitors it became apparent that this board had suffered alot of damage from capacitor leakage. There were many, many lifted pads and sections of the board that were heavily corroded with traces coming up. I did some trace repair during the recap, but it was not enough to bring this board back. I'm not sure if there are Vias that are missing or if the board has middle layers but after spending many hours with the board with no signs of life, I'm setting this machine aside for now. This is such a cool little small form factor PC that I have put this on the wish list and I'll be looking for another parts machine, although I wouldn't be willing to pay to much knowing how bad the caps were on this particular one I had here. Sadly, the end may be near from these Deskpros, although I'm hopeful that ones stored in better conditions may have faired better than this one.
The next repair attempt was a TI Travelmate 4000 Laptop. This machine was reported as working but the Floppy Drive was not reading disks. The previous owner suggested replacing a belt inside of the floppy drive. I have done this on larger machines and so I thought this would be a repair I could handle. I even found a bLog article explaining how another owner had been able to do this. For me, I personally feel like a retro project is not complete till the machine can be rebuilt using its own drives, so I started in on this repair knowing that the machine was partially working and could be left as-is. Unfortunately, after getting all the way into the machine I discovered that I was not going to be able to take the drive apart far enough to get easy access to the belts. I was not able to thread a belt onto the machine from the floppy slot although I did try for quite some time. During the re-assembly process I also caused additional damage so sadly this machine will have to be split up for parts. Although I'm very disappointed, I don't regret starting the repair for the reasons I've already described. This is the only one of the three repairs that I consider a total loss.
The final repair attempt of these first three is an IBM PS/note portable computer. This machine was described as "working for a while, then not working" The note indicated that they believed that the power supply circuit needed recapping. This is also a repair that I feel quite comfortable doing, and did. Sadly, after the Power Supply circuit inside the machine was recapped the machine still wouldn't start. This leads me to think that perhaps the Power Supply Brick is also needing a recap, so this repair that is on hold, and not a failure (at least not yet).
After all these challenges I was able to turn things around a bit. I bought a couple of "project" computers from a trusted local seller on Craigslist in my area. He was selling two machines at low cost because he knew they had something wrong with them and needed repair. I've had a high success record for machines from this seller so I jumped on them as soon as he listed them.
The first machine is a DTK KEEN-2530 machine from around 1991. The reported symptom was that one of the four SIPP Memory Modules had a RAM chip broken off of it. Since this machine is a full 32-Bit machine, a 386DX-25, it requires four matching memory modules in order to Post. I decided to work on the memory issue before doing a through cleaning on the machine, but I did a light cleaning on the outside of the case and disconnected the power supply and tested it first before trying to fire up the machine for the first time.
I've never worked on a machine before with SIP or SIPP memory, so I had to do some research. According to Wikipedia and another source, this SIP is basically a 30 pin SIMM with pins soldered onto it. My first instinct was to try to repair the SIP with the chip broken off, but after a careful examination that did not look easy with broken traces and multiple layers. I tried searching on FleaBay for a part, but none were available at the time that I looked, so I started digging in on what I had to see if I had an equivalent. It turns out the machine had 1 MB SIPPs in the 9-chip parity arrangement. I was able to find a stick of very similar memory in my stock but the one I had was a SIM and not a SIP, and a 70ns part instead of the 80ns chips on the good SIPs. I found some solid wire that would fit into the SIP socket and started soldering, adding 30 pins to this SIM memory to turn it into a SIP. I was very gratified at the end of the soldering and leg trimming exercise to discover that this is a working system, my SIP was functional and the machine booted for the first time! The connection is a bit flaky when I move the system around, so I've hot glued the home-made SIP to its neighbor in hopes that this will hold it in place better. In the long term I may end up soldering the two end pins into the socket below to prevent movement. I think this could have been avoided if I'd have had some lead frames on hand in order to cut and solder on proper leads instead of the solid core wire. but I didn't have those parts on hand.
The finishing touches on this build included desoldering a dead Dallas RTC and installing a socket for that, adding a 1.44MB Floppy Disk Drive, and testing the early Seagate IDE HDD in the machine. The machine cleaned up quite nicely and is a good time capsule 386 machine.
.jpeg)
The next repair candidate is an HP Vectra QS-16S from 1989. This machine was reported to have a failed capacitor which released the "Magic Smoke". After this failure, the owner powered the machine down quickly. I find that this seller is very accurate in his descriptions of failures so I knew these symptoms would have been reported accurately. What I didn't know what what kind of failure I was dealing with, my initial assumption was that it was a capacitor failure inside the Power Supply. This made the purchase a bit more of a risk because HP machines of this era have a proprietary power supply, so if I couldn't find the fault and correct it then the repair may have to go on hold waiting for a difficult to find part.
So, as usual I started the repair by spending some time learning about the machine. The HP Computer Museum is a great resource for these machines and did not disappoint with information about the HP Vectra QS-16S. I also did a bit of light cleaning before starting to take the machine apart from inspection and repair. As mentioned above, because I suspected the power supply, I didn't even try to test it, but immediately opened it up to look for blown RIFAs or other capacitors with obvious bulging or leaking, but there were none inside the Power Supply. On putting it back together and testing I was a bit surprised to see all the voltages responding as expected, so I needed to dig deeper into the machine.
This machine has the processor and RAM on a daughter card. I removed everything from the case and did some cleaning. As I inspected the main board I noticed a burned surface mount capacitor (C1)! It was near the keyboard connector on the main board and I can only assume it's part of the power supply section of the machine. The cap was so burned that I was worried about not being able to figure out the value of it, but after a close inspection every other capacitor on the board had the same shape, size, and value, 22uF, so I was able to find a replacement cap from my stock and install that on the board. I did some more cleaning as I reassembled the machine as this one had gotten quite dusty inside. Once I got the daughter card and VGA card re-installed, I was thrilled to see the machine boot for the first time!
Once I started reattaching drives and devices too the machine, I entered another no-start condition. After isolating the fault two one of the two MFM hard drives (identical ST-251s) I was able to remove the second drive get the machine booting again. There was a burnt tantalum on the ST-251 board that was causing the machine to not be able to start. I set the drive aside and came back to it later, but it didn't end up back in this machine.
The next hurdle was the Floppy Disk Drive. As I've already mentioned in this article, it is important to me to have computers that can start up on their own. The 1.2MB 5 1/4" Drive in this machine was cleaned and tested, but it is exhibiting some odd symptoms. The drive spindle rotates and the heads are moving freely, but it can't read or format disks. In the short term the easiest path forward (and probably the most reliable path forward in the long term) is to add a 3 1/2" High Density 1.44MB Floppy Drive to the machine. This would have been one of the factory options, so I purchased a replacement drive and bezel kit at Free Geek Twin Cities and installed that. Since this HP Computer does not have its BIOS setup program in ROM, I also made a couple of floppy disks from images on the Internet in order to set the BIOS settings and deal with Error messages on boot up. The machine now starts and runs without error. It's a great looking machine...
.jpeg)
The picture above shows the HP Vectra up and running with my copy of IBM OS/2 Version 1.2, but my interest in and history with OS/2 is a story for another time.
If you've made it this far thanks for reading and let me know if you have had any interesting repair successes or failures you'd like to share either in the comments section, or you can email me at the address shown in the Bio section of my bLog.
