Last modified: Sep 04, 2018
The previous section describes how I constructed the Rear I/O board. This board however has to
ultimately be integrated into the rear panel of the computer. The ports of the rear I/O board
have to be available from the outside of the machine and appear tasteful.
A further consideration is that I want the whole project to be modular. This has in previous
sections meant that I don't solder cables directly to motherboards, nor hack the end connectors
off pre-existing cables, instead opting to make conversion cables from new parts that can bridge
What modularity means for the rear I/O panel is that I put all the ports onto an inset board,
rather than fixing them directly to the bottom plate of the chassis. The rest of this section
shows how I've made that I/O board mount into the rear panel, and how it can easily be swapped
for an alternative board in the future. For example, if a USB mouse or keyboard adapter becomes
available at some point.
The I/O board has on it four ports, at least, this version of the board has four ports. These are
two PS/2 ports (for keyboard and mouse), and two Ethernet ports. Of the two ethernet ports only
one actually does anything, but the ethernet board I got happens to have two jacks on it.
Originally I'd hoped to make the second jack a simple passthrough, unfortunately this is not
possible due to the way ethernet works.
Somehow the these 4 ports have to be made available through the wooden rear panel of the chassis.
Iniitially I was considering doing precision cuts in the wood panel around the shape of the
ports. This however is really problematic for a host of reasons. The first is that it is very
difficult to make precision cuts in wood unless you're quite skilled and have high quality tools.
The second difficulty is that the I/O boards themselves, in particular the C=Key board, have
other dinguses, such as a jumper block, right side by side with the PS/2 port jack. This means
I would have had to cut the opening wider than the jack, wide enough to go around the jumpers
and the jumpers would be visible from the rear of the machine. Not to mention the fact that the
whole front edge of the C=Key's board which is about 2 1/2" wide is flush with the front of the
jack. One way to deal with that would have been to set the jack back 1/4 or a 1/2" from the back
of the machine, but it would not have been pretty.
The third problem, and in my opinion the clincher, is that the whole point of having the I/O
board be a separate inset board, upon which the parts are mounted, was to make the Chassis not be
directly married to those particular parts. It wouldn't make sense to go to that trouble only to
make the rear panel custom cut to the shape of those ports.
After I thought about it for a while, it dawned on me that the obvious solution to all of the
above stated problems is to use a thin metal plate. How that plate works will be the subject of
the rest of this section.
The point of the metal plate is that all the precision cutting will be done through it. And the
rear panel of wood therefore needs only a generic rectangular cut out. This cut out can now
support a variety of configurations of I/O board. You might wonder, well, what other
configurations are you really likely to have? Perhaps an alternative board could omit the
ethernet ports, and instead have an SD card slot and an IEC serial port. Or perhaps another board
could have the digimax output, and
8 bit stereo sampler input, or a USB port for connecting
external storage to an internal 1541 Ultimate II+.
The rear wooden panel then only requires a relatively simple rectangular cutout for the rear I/O
board. This cutout is made even easier by the fact that it extends over from the side of the
cutout that was previously made for mounting the AT power supply.
The wood, HobbyBoard, is a soft pine wood. It's very easy to work with, but it is unfortunately
not as strong in some places as you would want it to be. On the top plate, I reinforced the
underside using metal L-brackets to support the handle, for example. After the cutout is made for
the rear I/O board this left the rear panel with a weak spot.
The board is quite wide, and it thins out in the middle. The board isn't load bearing in any way,
but I felt more comfortable adding a bit of simple reinforcement. A straight bar across the gap
makes the board just a bit more robust.
When I first settled on the idea of using a metal plate for the rear I/O ports, I didn't know
what materials would be available to fit the bill. But a thorough search around the home–building
supplies store, Rona, my favourite go–to for tools and
materials here in Canada, revealed a whole section full of sheet metal. The sheets come in a
range of sizes and with a healthy range of thicknesses and types of metal.
I got a sheet of aluminum, 8" wide by about 30" high. Aluminum is lightweight, stiff but pliable.
It is much stronger in pieces that are small. It is susceptible to scratches, but it can also be
sanded or buffed depending on the look you want it to have. It's not a surprise that this
metal is a favorite of Apple for the bodies of MacBooks, iMacs, iPhones and iPads.
I thought I was going to need to use the dremel to cut off a 1 3/4" high strip. But it turned out
to be much easier than that. I drew on the inside of the sheet with pencil as a guide for where
the cut should be. Then, using a ruler as a straight edge, and a good exacto knife, I repeatedly
scored the metal along the guide line.
Each score actually removes long thin slivers of aluminum, you do have to be careful because you
don't want to get those slivers stuck under your skin. Definitely avoid brushing away the bits
with your bare hands.
After a few passes pressing heavily into the metal to pull out those slivers, the sheet can be
bent. Put the sheet along a sharp table edge, and push it down firmly, slowly and evenly. The
metal will naturally bend along the score line. After bending it nearly 90°, slowly bend it back
up and that is enough for fatigue to break the metal along the score line. It's really easy.
After it was broken off, I used a metal file to remove the burrs and dull the edge. There is
actually an unexpected side effect to bending the metal as described above. It isn't a perfect
fold right on the score line, it does give a slight curve to the metal at the edge, curving in
the direction of the initial bend. The bend back up has much less resistance because the sheet
is in the process of breaking off, so the curve is left in the original direction. At first I
thought I would have to straighten it out, but it ends up being really quite useful. The slight
curve causes the sheet to tightly hug the rear wooden panel, which you'll see again below.
The original sheet is 8" across, but for our rear panel we only need it 7 1/4" wide. So I did the
same scoring, bending and breaking trick to remove a final 3/4" piece. This also resulted in a
slight curve which is perfect, as it matches the direction of the curve along the long edge.
Placing the piece of cut metal over the rear panel's I/O cutout shows how nicely it fits and how
well it works. You can see that the curve on the top edge and right edge cause the metal to pull
up against the rear panel. It really worked out perfectly. The metal plate is about a 1/2" wider
and higher than the I/O cutout from the wood. So the metal plate overlaps it, covering up any
imperfections in the wood cut. Gives it a really nice appearance in my opinion.
When mounting the various boards (C=Key, MicroMys, Ethernet Mount) to the I/O board, the jacks
themselves were all aligned to the edge of the board. The problem I encountered here is with that
pair of ethernet jacks in the wall mounting plastic case. I considered removing the board from
the surrounding plastic case, but the holes in the PCB through which to put a screw are oversized.
That means it would need a washer and would not be too stable and would tend to shift. Also the
jacks themselves have big soldered leads on the underside of the PCB, so the PCB wouldn't sit
flush with the board, the back side would sit lower than the jack side. All around inconvenient.
Because of this, I left the PCB inside the bottom half of its plastic case. The plastic case is
designed to hold the PCB perfectly flat and stably, and is also designed to be mounted to a
surface. However, the plastic case has a larger footprint than the PCB, and so with the jacks
aligned to the front of the board, about a 1/4" of the plastic bottom case juts out past the
edge of the I/O board. That outer case bit has to be removed.
A dremel is a very handy tool for this sort of job. Even basic dremels come with a variety of
swappable bits. One is a tiny drum sander. You loosen a screw at the end of the drum and slide
a cylindrical sanding piece onto the rubber drum. When you tighten the screw the rubber is
squished, such that it deforms and presses against the inside of the sanding piece.
The plastic case actually starts to melt a bit with the heat caused by the sanding. But it very
easily sloughs off. With a bit of care you can get the case edge to be flat and aligned with both
the I/O Board front edge and the two ethernet jacks. I covered the C=Key and Micromys boards with
a bit of plastic, taped down, to prevent any bits from getting into them during the sanding.
The dremel really is the lifesaver here, doing this sanding by hand would be almost impossible.
It would take 10 times as long and put lots of unnecessary stress on how the plastic case is
mounted to the board.
As you can see in these photos, the result is really nice. Those jacks are all lined up perfectly
along the edge of that I/O board. And we can place the I/O board where it's going to go inside
the main chassis and put the metal plate against it to see how thin the plate is and how well it
lines up with the board and ports.
You can see in the photo on the right that even though the jumper block and the rest of the PCB
of the C=Key board is aligned with the PS/2 jack, it doesn't matter at all now that our final
back panel is a very thin piece of sheet aluminum.
With the metal plate in position, I didn't so much as measure the positions of the holes for the
ports, as I traced them from the back. After tracing the holes, I did do measurements to make
sure they correlated. And I eyeballed them to see how they'd come together.
Next, I clamped the metal plate down to the workbench, and got to work dremeling out the first
hole. Once again, you really need to protect yourself. You should be wearing protective glasses.
Even my cheap 40$ dremel 100 series spins at 35,000 RPM. That's 583 revolutions per second. You
don't want a shard of aluminum going in your eye at those speeds. Not to mention the fact that
there are 3" to 4" sparks flying off the surface as you cut. You should also wear ear plugs
because it only takes a few seconds of the wrong frequency and loudness to permenantly destroy
your ability to hear certain audio ranges. And lastly, you should wear a good quality face mask,
because nothing sucks more than breathing metal dust into your lungs.
After working around the 4 sides and then popping the little square of metal out of the panel,
it's necessary to use metal file to get rid of the burrs and dull the edges. You can also get
into the corners and straighten them up. The resulting hole looks pretty good. It's pretty
straight and square.
Comparing it to the actual I/O board though, unfortunately the whole was a bit tall. It's a
great fit at the top, but it goes a bit further below the jack than I would have liked. That's
partially the result of eyeballing the bottom of the jack, because that was the only side I
wasn't able to trace. But also partially my inexperience using a dremel and knowing where to
place the bit relative to the pencil line.
It wasn't so aggregious that I wanted to start over from scratch with a whole new metal plate.
And all you can do is learn from your mistakes. I remeasured the remaining holes and was careful
to set the dremel bit on the inside of the penciled guide lines.
The final three holes didn't have that problem. I'm pretty pleased with how they turned out.
These photos show how the 4 holes line up. Pretty good. Not CNC machined good, but pretty good
considering I did this by hand and this is the first time I've ever done anything like this. I'm
quite pleased actually. Whatever gaps there are between the whole and the port behind the hole
will be less noticeable when the machine is sealed up and dark inside.
You can see in the photo on the left how the metal plate extends beyond the edges of the I/O
board by a some fractions of an inch on either side. That's the portion of the metal plate that
overlaps the cutout in the rear wooden panel.
Fixing the metal plate to the I/O board was actually quite straightforward. I used some masking
tap to hold the plate where I wanted it. And draw along the top of the tape to show where the
wood goes to. And measured out three spots for screws, equidistantly spread out along the length
of the I/O board.
It's necessary to pre-drill the screw holes, especially through the metal plate. This was not
particularly difficult. The I/O board itself is not as thick as the 3/4" HobbyBoard, it's only
1/2" thick. The screws are quite short, only about a 1/4" deep with flat heads. The result is
kind of industrial looking, but I think it's sharp. The rear of the AT power supply which is
right beside this panel also has screws exposed, so it's in good company.
After the metal plate is on, we can finally see what it looks like sitting in the chassis. The
first photo is without the surrounding wooden rear panel.
The second two photos show a couple of different angles of the metal panel in place with the
wooden rear panel behind it. I really like this. The metal panel was a great idea. It completely
obscures the simplistic wood cut out, and allows those ports to come right flush with the back
side of the machine.
Now you can really see how subtle the curve at the edge of the metal plate is, imperceptibly
curved, but it does help to remove any gap between the metal plate and the wood panel behind it.
Now the I/O board has its outer metal cover attached to it, and this whole board, as we'll see
soon, can slide in and out of the back of the chassis. Once it's been pushed into place within
the chassis however, we no longer want it to slide. The I/O board has to be mounted to the
inside bottom plate of the chassis.
Not shown in any photos, I shortened the depth of the I/O board by about an inch after I knew how
big the components on it were going to be. Then I picked some arbitrary spots behind the
components for where to put bolt holes. The only real consideration here is that once these bolt
holes are made in the bottom chassis, they will become the standard locations for these holes in
all the subsequent alternative I/O boards I may choose to produce in the future.
I initially drew the spots on the I/O board where I wanted the holes and then drilled through
just the I/O board. The bit used for these holes is big enough for the bolt to pass through
snuggly but without needing to be screwed through the wood.
With the I/O board having its holes already in place I drew through those holes onto the inside
of the bottom plate of the chassis. Then I took some quick measurements to make sure the distances
correlated, before drilling the holes through the bottom chassis.
Again, the holes in the bottom chassis were drilled large enough for the bolts to be snuggly
pushed through by hand. The bolts are not intended to grip the wood itself for their hold. That
would not be workable as the expectation is that these bolts will be removed and reapplied as
many times as the I/O board will be replaced. If they were merely screwed into the wood, the
wood would eventually wear out.
Instead I used threaded wood inserts. These have four sharp spikes that dig into the wood on the
underside of the chassis, pointing up. And contain a threaded nut that inserts into the hole. I
actually had to make the hole in the bottom of the chassis just a hair bigger than the bolt to
accommodate these inserts.
On the inside, I used two stacked washers to give a larger surface area for the bolt head to
pull down on without damaging the wooden I/O board. These bolts simply pass through the I/O
board, pass through the bottom chassis holes and screw down into the wood nut inserts on the
other side. The tighter you tighten these down, the more securely the bottom mounted wood inserts
dig into the underside of the chassis.
The I/O board is 1/2" thick, and the HobbyBoard bottom plate of the chassis is 3/4" thick. So, I
got 1 1/4" bolts. In theory they would reach flush with the bottom of chassis, inside the wood
nut inserts. However, the two stacked washers add an extra 1/8" or so of distance. So you can
really crank them down without risking having the bolts stick out past the bottom of the
One last thing to note about these bottom mount metal bits. There are actually 10 metal bits on
the bottom chassis. The 4 screws on each side holding the bottom plate to the side plates. Plus
now these two I/O board mounting inserts. This is not going to be a problem for any surface we
put the machine down onto. Before the project is fully completed, the bottom plate will be
painted, and 4 rubber feet will be addeded. So having a few metal bits on the bottom surface of
the chassis isn't an issue, they'll never touch the surface that the end product will sit upon.
With all that inner work complete, it's finally time to screw down the rear panel, and
permenantly attach it to the back of the chassis.
Three screws seemed appropriate. These are the same screws used to hold the top plate to the
sides, and the bottom plate to the sides. It requires predrilling the holes as it did for those
in the top and bottom plates. When deciding where to put the screws some care was needed to make
sure the screws going in the back would not collide on the inside with the screws coming up from
There isn't a whole lot to say about the process of screwing the back plate to the chassis. Except
that it becomes obvious that when this plate was cut with the circular saw the blade curved off
at the end making the left side slightly shorter than the right. Because of all the cutouts for
the I/O board and the power supply, most of this variation in height didn't matter. Only the
left most strip of wood that is screwed down the side shows up with a small gap at the bottom.
Structurally it makes absolutely no difference. Wood filler to the rescue. Once this gap is
stuffed with some wood filler, then sanded and painted, it will become completely invisible.
With the rear panel securely attached, we get to test out how the I/O board realistically slides
in an out of the chassis.
Looks like it works pretty well. Including with a ribbon cable attached to the C=Key adapter. I'd
say this whole rear panel is looking pretty nice. Here are a few more glory shots, closeups of
what the rear I/O panel looks like, with a cable installed, and with the drawer partially pulled
out. Very sexy.
This is a living document. The table of contents is a work in progress,
and the content is only partially written. It will continue to be written as
the project continues to be worked on.