IBM 5150 Won't Power Up

After a very busy few months, I had some time this weekend for tinkering and decided to power up my IBM 5150.  Getting this computer running was a real scavenger hunt and it took months to track down a Model F keyboard (necessary because this was the only keyboard with requisite 5 pin DIN connector) and a video graphics adapter with a composite video output. But I finally got it working and was able to boot the dinosaur up on PC-DOS 3.3. So you can imagine my frustration when I flipped the monolithic red power switch to hear soft "click" rather than the airplane turbine whir that usually indicates a successful startup. I probably would have just tied the fifty-some pound computer to a rope and used it as a boat anchor if that click sound hadn't been so familiar. After spending most of my Sophomore year in an electronics lab, I will always know the sound of a resettable fuse detecting a short circuit.

The power supply directs its fare to the rest of the 5150's internals via four bunches of wires - two for the 5.25" disk drives (numbered p10 and p11), and two that connect to the system board (p8 and p9). I unplugged each of these in turn while testing the power and found that the fuse was triggered only when p8 was connected to the system board. The good news was that I knew roughly where the problem was, the bad news was that the problem was in the system board. Much more difficult to replace, if necessary, than a disk drive.

The power supply connections on the system board. +12V is third from the right.

Either way, I found a link to the 5150's technical manual and began flipping through it. The System Board section has ten pages of logic diagrams that detail the schematic layout of the board. P8 (labeled p1 on the board) provides 12 volts to the board while p9 (p2 on the board) supplies 5 volts. Sure enough, a quick check with a multimeter showed a short between the +12V pin and ground. To make sure it wasn't one of the cards in the extension slots that was causing the short, I removed the disk drive controller, memory expansion and video graphics adapter cards one at a time while powering up but none of these made a difference so the short was happening somewhere on the system board.

At something of a loss, I searched the Internet for talk of short circuits in old IBM's and didn't find much but  there was mention of tantalum capacitors and their tendency to fail after not being used for long periods of time. This seemed like as good a place to start as any so a quick look at the technical reference showed that the each of the power pins is decoupled by a 10uf capacitor. Decoupling capacitors connect the high and low lines to ground in order to reduce noise that might effect a sensitive integrated circuit. For an excellent explanation, watch this video.

After removing the system board, I tracked the shorted +12V pin until I found the decoupling capacitor. IBM used three-lead capacitors on their system boards. The middle lead is the anode and the outer leads form the cathode. The capacitors are configured this way so that they cannot be inserted backwards, leading to a thermal runaway (also known as an explosion).

You can see the power supply pins at the top of the board. The third pin from the right is the +12V pin. Follow it down and you can see where it connects to the center lead of the decoupling capacitor.

The faulty decoupling capacitor.

It took a little coaxing to get the capacitor off the board but after removing it a quick test with the multimeter revealed that the short had disappeared. Just to make sure I tested the capacitor as well and, sure enough, it was shorted. I made a quick trip to RadioShack and (for $1.79!) I got a 10uf tantalum capacitor to replace the faulty one. Making sure to place the anode (marked with a plus sign and a longer lead) in what was the center hole for the three-lead cap, I soldered the new one into place.

After removing the original capacitor...

... and adding the new one.

After quickly reassembling the rest of the components I flipped the big red switch and was greeted with the deafening sound of the power supply trying to suck the air out of my lungs! The 5150 runs like new and I can pleasantly wait for the next component to fail!

Atari 800 Keyboard

So with the video connection in place, I powered the Atari up again to see the default memo pad program that comes preloaded into ROM. I typed out my first memo, "Hello world!" and looked up to see "hoo" after the prompt. After testing all of the keys, the only ones I found to be working were 'h', 'a', 'o', '9' and 'break'.

Getting to the keyboard of the 800 is fairly straightforward but it's important to keep track of parts as it requires some disassembly. The first step is to remove the bottom part of the case. Flipping the computer over exposes the five screws that hold the bottom half of the plastic casing to the top. After removing these screws, slide your fingernails into the joint between the halves of the casing on the backside (TV cable side) of the box and pry the halves apart. It's important to start on the back because the joystick ports in the front protrude from the casing and make it difficult to remove the front first. Now the only thing attaching the bottom cover should be a pair of wires connecting the speaker to the main board. The speaker is permanently attached to the red kill switch on the bottom cover so disconnect the wires and put the lower cover and speaker aside.

Now to remove the main board. There are four screws, identical to those you just removed holding the board to the upper cover. The two on the right side (where the auxiliary ports and power supply are) go through the aluminium casing. The left two go through the main board. Remove all of them and slowly lift the front of the board. The last connection will be a ribbon of wires running from the keyboard assembly into the main board.

Carefully record their orientation and remove the wires. Now you can slowly remove the main board assembly and put it aside. All that should remain is the keyboard attached to the upper cover. Finally, to extract the keyboard, remove the four screws securing it to the case and gently pull it away.

The keyboard has three components. The key assembly is purely mechanical and has no electronic components. Problems with the keys should be fairly obvious. To detach it, remove the dozen or so small screws and it should come off easily. One of the screws is positioned under the wires that connect the keyboard to the main board. On my Atari these wires were stuck to the back of the keyboard with double sided tape. If this is the case with your box as well, simply use a razor blade to peel the tape from the back of the keyboard to access the last screw.

If all the keys depress and there is no visible malfunction you can put this part aside for now. You're left with a a printed circuit board onto which is taped a sheet of Mylar.

The underside of the Mylar is covered in electrical connections that have been painted on with conductive paint. When a key is pressed a contact point on the Mylar is pushed into a contact point on the PCB and a unique connection is made. Often, the conductive paint can crack or chip which leads to broken connections and faulty keys. In order to check the Mylar, it must be removed very carefully. First, the electrical tape must be removed, care being taken not to tear the delicate Mylar. Next the Mylar can be slowly peeled off the PCB. It's likely that the keyboard was never serviced so the Mylar will stick to the PCB, especially around the screw holes. Be patient - the last thing you want to do is tear the Mylar.

With the Mylar fully removed, you can inspect for cracks. If you don't see anything, it's time to take out the multi-meter. There are eight strips at the bottom of the Mylar where the conductive ink makes contact with the PCB. Each of these is at the head of a trail of nodes on the Mylar. To test a strip, simply place one terminal of the multi-meter at the strip and the other at the furthest connected node - if there's a fork in the trail, test all of the ends. In this manner you should be able to quickly test all of the connections on the Mylar. If you do find a break, it's a quick fix. Just pick up some conductive paint at your local electronics store (RadioShack sells this) and carefully paint over the break.

If you were unable to find a break, the adventure continues. Another point where the connections tend to break is the contact point between the Mylar and the PCB.

The conductive strips on the Mylar need to be pressed firmly into the terminals on the PCB. There is a ridge on the button assembly that is supposed to do keep the two layers firmly connected but this can fail over time.

In fact, this is what ended up being wrong with my keyboard. The easy fix here is to add a strip of electrical tape on the top side of the Mylar, right above the connections. This makes the whole assembly a little thicker and makes ensures that the Mylar is sandwiched more tightly to the PCB.

With the new tape job in place I reassembled the keyboard for a quick test and voila!

It worked!

Atari 800 TV Adapter

My latest acquisition is an Atari 800 box that I won on ebay for about $100. It came with an Atari 1050 Disk Drive, joystick, trackball, number pad and about seven and a half million games on 5.25" floppies. I pulled it out of the box, plugged the RCA cable into the composite video jack (yellow of the familiar red-white-yellow threesome) of my TV and powered it up to see... nothing. Apparently when the Atari came out, RCA connectors had not yet been adopted for composite video so the signal run through the cable is a broadcast TV (VHF) signal such as you might receive from an antenna.

The quick fix that worked for me was this little "phono plug to 'F' jack" adapter that allows you to plug the broadcast TV signal into the jack that is expecting it: the cable/antenna jack in the back of the TV. I picked it up at RadioShack for something like $5 but you can probably find it on eBay for next to nothing.

After fitting the Atari cable into the adapter and then plugging the other side into the TV I was able to get the Atari up and running. This link goes into a little more detail if you're curious.
One nice thing about the 800 and a few of the later models is that they have a monitor output in addition to the TV cable. The monitor jack is a 5 pin DIN connector which is apparently quite easy to convert to s-video so this might be a project for the future.

printf("Hello world");

This being my inaugural blog post, I'll take a second to lay out what I'm hoping to accomplish with this blog. I've always enjoyed tinkering with electronics and have found the internet to be an invaluable resource when my desk is covered in scavenged parts that don't seem to want to assemble themselves into a cohesive structure. Since this situation is ubiquitous throughout DIY-land I thought I'd contribute what I can to the collective body of knowledge. So hopefully when you, my reader, encounter this blog in a moment of crisis, I'll be able to save you some grief the way countless anonymous tinkerers have helped me with their blogs.

That said, my interest over the past couple of years has settled on computers. My formal education is in physics and (more specifically) astrophysics so most of what I post will be stuff I've picked up on the fly either on my own, working in labs or talking with people who actually know what they're doing. The bulk of these posts (I'm predicting, although my attention tends to wander) will be on vintage computers. I've acquired over the past few months an Apple ][e, an IBM 5150 and most recently an Atari 800, in various states of functionality. These machines are fascinating because despite being computers in every sense of the word, they are still simple enough for someone with no formal training to deconstruct and understand. If you're still reading this, you're clearly keen on getting your hands dirty so rather than waxing philosophical I'll shut up and start preparing material for this blog.

Happy soldering!