Adventures in Vintage Microcomputing
The IPX has two serial ports on the motherboard.These are set up via the admintool.
I have PortA set to ADM20 @ 1200 baud – The kaypro can talk to it.
I have PortB set to VT100em @ 19200 baud – This works with the IBM Terminal set for VT100 emulation.
/opt/wp/wpbin/wp runs a text version of wordperfect on the terminal.
It’s a bit anachonistic, but here’s a Kaypro accessing a BBS via the IPX:
I think it’s still better than hanging a RaspberryPi off the serial port.
I had some issues with the DEC RRD42-DA (NEC CDR-1410A) CD-ROM drive. It had very leaky electrolytic capacitors.
A recap followed, and a happy ending achieved.
The Sony power supply gave me some grief not long after i got the machine. When i pulled the board out, it appeared to be recapped by a previous owner, but the PCB was still covered with a lot of residue from the previous capacitors. I gave it a good clean, dried it out, tested it, and with the machine booting, i claimed victory.
But then, sometime later, the machine refused to start up. There was nothing on the screen. The supply voltages were fine, but I wasn’t sure about the Power On Reset (POR) (grey wire) which seemed to always be low.
I took the unit out and put a disk on it for minimum load. The POR line stayed low, but I found a maintenance manual which said that it was open-collector and once I added a pullup resistor it worked fine.
I gave the supply a clean again – the solder resist is coming off where the capacitors leaked, so I scraped it away and tinned the tracks for protection. I put it back in and the POR signal started to do its thing, but still there was no boot.
The service manual suggested putting a monitor on serial port A and this showed diagnostics running continuously.
Depending on how much was plugged in (eg hard disk drive, floppy) the diagnostics would restart. Once the drive spun up, this would sometimes come good. A closer look at the POR showed that it was “going off” sporadically – seemingly worse with load. The supply lines seemed fine.
I took the supply out again and loaded it up with some more disks. Sure enough, the POR line misbehaved.
I couldn’t find schematics, so I was flying somewhat blind. I worked back from the POR line, which eventually led to what I think is an LM358 dual op amp (very hard to read), the output of which was indirectly driving the POR line. The amp is configured as an integrator. Despite this, it was clearly not riding out some bumps in whatever it was sensing – which seems to be the 5V line – although not directly because the 5V was clean.
The power supply is switched mode. What was coming across the transformer was very complicated, and I couldn’t make sense of it. It looked like the signal coming across the isolation transformer was the issue, with some significant lumps and bumps at 100Hz, so I moved back to the high voltage side – which I hate.
It has a bridge rectifier with a large filter cap, which looked like it was the only original electrolytic original unit. The bridge rectifier seemed fine. There is a series surge limiting resistor that’s shorted by a relay controlled from the low voltage side by an optical isolator. Another optical isolator provides the feedback path for regulation.
The DC voltage on the filter cap was about 240V. Usually I’d expect something a bit higher than that – around the 320V to 340V mark – the peak of the 240V waveform. There also seemed to be a very large AC component. I didn’t want to put the scope on, but I thought the cap might not be at its best.
I pulled it out – it measured a few nF. It wouldn’t hold charge from a battery.
The supply still works as long as the load is light – no hard disk drive; a remarkable effort from the switcher given the fault. It seemed to be quite effective at tracking the rectified AC although this probably made life difficult for the switching transistor.
Once the replacement cap was fitted, the DC level and 100Hz ripple on the rectifier were much healthier, the POR good, and the system was happy.
The HP tape drive that came with the IPX was unreliable. I bought a Sony SDT-S7000 DAT-2 drive.
The ID has to be set to 4 for device /dev/rmt/0.
The Switches have to be:
From the Sony Unix Configuration Guide v2.07:
To write files to a tape: tar cvf /dev/rmt/0 <files>
To list files on tape: tar tvf /dev/rmt/0
To get files from tape: tar xvf /dev/rmt/0
My first workplace used a combination of PCs and Macintosh computers (mainly Macintosh Plus). I would do technical stuff on my PC and write up manuals on a Macintosh Plus. When i left that workplace, i left the Macintosh world behind and had no particular reason to return to it until i developed an interest in vintage computers.
There are two things that made the macs standout to me in the mid to late eighties. The first was the GUI and the ability to be able to swap between programs and copy and paste between them. The second was the simple and low cost networking. These were really impressive capabilities.
I found the Macintosh ecosystem very addictive, and at one time i had over a dozen machines, including:
As i pushed my hobby back further, space and time both became issues, so i had to take some difficult decisions. I chose to focus on the compact Macs of the eighties, which are easily stored and quickly set up. I shuffled on most of the nineties machines and a couple of the duplicate eighties machines. Hopefully they’re entertaining their new owners!
I can still create a pretty solid network with:
I have had three Macintosh SE/30. The first one (pictured above) was a purchase from a well known auction site in about 2018. These days everyone seems to want an SE/30 but considering it was in working condition, had 32M of memory, and came with a network card, keyboard, and mouse the cost was pretty good. Given that it was a working unit, i powered it up immediately. It booted up fine on the hard disk, but i did notice the audio was very low which was a hint that the logic board capacitors needed work.
On first use, i broke it. I was very much a beginner with old macs. I put in a disk and when it ejected the eject drive gear broke. There’s a lesson.
Of course, the gear probably would have broken anyway; they go brittle. I was able to buy a replacement online.
I now clean and lubricate 3.5″ drives on arrival. IPA is generally sufficient for cleaning. Silicon grease or bike chain lube seem to be ok lubricants.
I then discovered that Macs are hard to open. I had to by the required torx bit, and then gradually lever the case open (i later got a mac cracker).
I noticed that the insulating card on the back of the analogue board had been replaced with piece of plain old cardboard. It was probably fine, but I replaced it with the appropriate material. It wasn’t quite as large as the analogue board, but it covered all the active areas. Later, i swapped in the correct card.
Premium Apple Service Centre repair!
The SE/30 has a reputation for leaky logic card capacitors. i replaced them with tantalums. I also replaced the PRAM battery.
The hard disk had System 7.5 installed. With a Daynaport Ethernet card and plenty of memory available, I wanted to get as much network capability as possible. There is a very good networking guide here:
http://www.applefool.com/se30/
I won’t describe the hoops i went through to make the OS installation disks back in 2018, because nobody would do it the same way now. Greaseweazle is now available and is a much easier way to write disks. A FloppyEmu is an alternative method, which emulates a floppy disk drive, and avoids the need for floppy disks at all.
At least the SE/30 can read PC format 1.44MB disks. I installed System 7.5.3 and then System 7.5.5 from images downloaded from the web.
To get the networking running required the drivers and some additional files for the Daynaport network card and Open Transport. Transfer of unarchived files from anything other than a mac running a classic operating system is a frustrating experience.
The old Macintosh OS used a novel file system that had two forks stored under the one filename. Once they come through a PC (or virtually any other computer) one of the forks has disappeared so that the resulting file is often unusable once transferred.
There are some specific formats that survive the process – Stuffit and binary files, although sometimes these need to be opened from the Stuffit application – the application associations are lost when the resource fork disappears.
Those archives, once transferred to the SE/30 can then be unpacked in their native environment which retains the resource fork.
Once open transport and either a browser or FTP client are setup, files can be downloaded without any need for floppies or serial transfers. Serial transfers were required for files larger than 1.4MB.
There was a version of Stuffit on the hard drive already, and i was able to transfer Mackermit across via a floppy disk.
Then I made up a serial cable to connect from the SE/30 to a PC. I used kermit95 to transfer the files I needed across to the SE/30. It’s important to configure Kermit to use the binary protocol.
Using kermit I transferred:
The Daynaport card has a significant limitation in that it only seems to support half-duplex operation. This means that the connecting device needs to be able to be configured for 10Mbit half-duplex.
At the time i didn’t have a managed switch (i now have several), so I added a second card to a PC, configured it as required, and then bridged the networks. This required a crossover cable to the mac.
This allowed primitive browsing (forget https) including downloading of files. I found that although it was difficult to browse to a file, it was possible to discover it on another computer and then type in the (nearly always) lengthy url into Internet Explorer 3.01 to complete the download.
FTP as a client or server is also quite viable.
With the SE/30 setup, it was much easier to set up other Macintoshes.
Within a couple of weeks of purchase, the system completely died – black screen. I thought this could have been the analog card or the power supply. When I took the cover off and turned the machine on, I was intrigued to see a beautiful blue corona on the flyback transformer. This is the component that creates the high voltage for the cathode ray tube.
The tracks are at 5 o’clock on the image above. It did cross my mind that, as the fault was visible, it may be fixable from the outside.
All the macs seem to have some high voltage putty around the anode lead. Perhaps this putty had failed, or perhaps a crack had opened up.
There is a lot of fear out there on the internet about disconnecting the lead on the CRT. I used a grounded screwdriver to discharge the CRT (a non-event) and then some long-nosed pliers to allow the high-tension lead to be disconnected.
I bought some high voltage dope (like an insulating varnish) and after carefully cleaning I applied three coats. The cleaning included scratching out the two carbonised, self-sustaining, tracks that had been burnt into the outer shell of the flyback transformer.
It did end up with some bubbles after operation, so perhaps it hadn’t cured for long enough. Nevertheless, it did the job and continues to do so.
In 2018 i bought a job lot of Macintosh computers in varying condition. Amongst the lot were two SE/30 computers which had crushed the previous owner’s spirit. They nearly did the same to me.
One machine had a black screen. The other had a stripey “simasimac” display, ie the startup process was not getting to the point where the video ram was initialised.
The previous owner had done some work on the second machine, replacing the surface mount caps with through-hole caps made to fit.
I recapped both boards with surface mount tantalums and thoroughly cleaned the boards. I don’t do a lot of surface mount work, so some of the caps are not as straight as they could be!
There was till no joy with either machine. Some posts on the web suggested that the corrosive juices of the capacitors could penetrate ICs – particularly the row with UE8 which is adjacent to 5 leaky electrolytics. I replaced the entire row on both machines.
After this, The black screen machine progressed to having a classic “simasimac” screen like the other one.
This machine had a broken ROM socket. When i held the ROM firmly in position i was happy to see that the machine would start up. I crafted a kludge to hold the ROM in place, but it really needed a new socket.
The audio on this machine was quiet and crackly. Initially i thought that a transistor had been eaten by the capacitor innards, but i was surprised to find that there was no -12V because a power connector pin had been eaten away. Those capacitors are nasty.
This machine also had a curiosity of a previous repair to the yolk connector. One pin was clearly not making good contact and had overheated. The pin had been bypassed.
The recap and clean had no effect on the other machine: still simasimac. I swapped every component that i could easily swap – no joy.
I did notice that messing with the RAM could cause some changes in behaviour – a curious collection of different chimes at different times – often two chimes. On the scope i could see that the boot behaviour varied. It looked like the RAM test might run for a short time or a long time.
Sometimes it changed just by touching signals with the scope probe.
I spent many hours, coming back several times over about three months, trying to work out what was going on. I Eventually found that it was very sensitive to the probe being on some of the CAS lines. I found that the resistor pack next to the RAM sockets and right next to a leaking capacitor had failed – the internal resistors were not isolated from each other. I replaced the pack and the machine fired up.
TCP/IP is required for applications such as FTP, Telnet, and browsers.
It was first implemented on Macs by MacTCP which was available for System 6 and was included as a part of System 7. In the absence of Ethernet, MacTCP uses AppleTalk as the “bearer”. TCP/IP packets are embedded in AppleTalk packets.
My Ethernet carrying SE/30 and a few other machines with Ethernet that i had at one time eg LC III and 660AV simply connect to TCP/IP services via Ethernet. I set up all of these machines to use the Open Transport TCP/IP implementation under System 7.5.5 or later.
My main interest is in the older machines, including the Mac Plus, SE, and SE/30. These machines (except for the one SE/30) only have LocalTalk connections.
The LocalTalk Bridge on the SE/30 (or various other machines with Ethernet) will happily move the AppleTalk packets to the Ethernet, where they are embedded in TCP/IP packets. This means that the original MacTCP packets are embedded in an AppleTalk packet, which is then encapsulated in a TCP/IP packet. They are effectively on a separate virtual TCP/IP network.
To overcome this, a router is required. After a little searching, I found that IPNetRouter is a viable solution. I also found that although it still requires registration, there is a free registration for non-commercial use. You have to go back in time to pick up a promo version:
http://web.archive.org/web/20080328045722/http://www.sustworks.com/site/news_promo.html
The versions cited in the article are for PPC. For 68000 use 1.54 with the same codes.
The router can run on any machine on the Ethernet network. The LocalTalk machines still need to be bridged to Ethernet, but not necessarily on the same machine as the router.
As per the instruction, i have set up a private network, 192.168.0.x behind the router. The MacTCP clients don’t seem to support DHCP, so the IP addresses have been allocated manually, with the router address being 192.168.0.1. The IP address on the Ethernet side is allocated by the network DHCP server.
This is a screenshot from an LC III that was running IPNetRouter but i’ve also run it on the SE/30 and a 660AV.
Example of a localtalk client:
This allows all System 7.5.5 machines to browse the internet using Netscape. The colour classic, running System 7.1 only has 6MB of memory, which is insufficient to support Netscape. It does work if swapping is turned on. I have not succeeded in getting the 6.0.8 machines to run a browser.
All machines support telnet and FTP clients.
I was able to log into an internet BBS and also into my SGI workstation.
I came back to this a year later and had a bit of trouble getting it going again. Fortunately, I had the images above. It seems that the configuration needs to be saved and then reloaded. Perhaps a default file is loaded on startup.
