A little of both actually - I got this case from tone007, and he and I did some measurements before I bought it, just to make sure it would actually fit.

Indeed it is a flat 128, and it just barely fits!  I think there's only about 3/4 of an inch to spare at the keyboard/joyport (back) end.   

Indeed, a made-to-fit plate is a lot less work.  In the previous tower, I got lazy and used a 5 1/4" knockout panel to block the cutout.   

damn. well, I tried. 

I guess you could say that...  Ladies and gentlemen, THAT is what happens when a hardware project suffers the physical equivalent of software bloat!   

One thing I wish I had were really good pictures of the very first one from the 1998 expo.   I didn't have a good camera available at the time, nor had it occurred to me to take pictures during or after the build process.  I eventually borrowed a few images from James Frazier's photos page from that event a few years ago (the only such page that appears to exist).

Robert, seeing as how you seem to be able to make it to just about every major event, I don't suppose you have good pictures of the tower from that year also?   

To Dr. V:

The direct connection did look nice, but in the end I am sorry I did it, as it did cosmetic damage to my SuperCPU.   

Indeed, it is quite easy to purchase the female connector to make such a cable from.  You can get one for a reasonable price from Digikey, for example.  Right angle connectors seem to be quite rare now.

To get a male card edge, however, you'll probably have to get one the hard way.  I recycled a dead cartridge to get the connector.  Other possibilities include buying and cutting down an "8 bit baby" board, or perhaps just etching one the old fashioned way.

The power supply is indeed a standard ATX power supply, and isn't actually modified too much.  I replaced one of the power connectors with one specific to the 128 and 128-DCR, and opened the case to get access to the 110VAC mains plug and to tap into the 12VDC rail (I wanted to keep the splices hidden).

The HD faceplate is sort of custom - it is actually built from a standard 5 1/4" PC knockout panel.  I drilled holes to match the CMD HD LED board lights and switches, epoxied threaded stand-offs to the back side, and attached the LED board.  The label is actually a spare CMD HD-20 label, cut down and re-arranged into a smaller footprint.  Eventually I epoxied that front panel to the CD-ROM casing.  Before I cut that label up, I made a very high resolution scan of it, and created a cleaned-up idealized version just in case it ends up being needed some time in the future.

I used the same method to build the Control Panel, and based the look of it on the CMD HD label, if that isn't obvious.   

The mention of Maurice refers to when he and I discussed making a single composite front panel for the machine, but this was during a time when I wasn't putting things into individual 5 1/4" drive bays with individual faceplates, so I no longer need such a thing.

To Robert:

No Comment^TM   

Thanks for the compliment 

Actually, I used a flat 128 mainboard this time more as a matter of availability and cost than anything else.  I couldn't locate a 128-DCR that was within my budget, known to work, and actually available for sale.  After some discussion with folks on IRC, it was agreed that my mainboard was the reason I was having issues, and that a flat 128 should work fine.  The one I have now was graciously donated by Al Anger.

Having the 1571 integrated into the 128-DCR mainboard does make for fewer boards inside the machine, but does have one caveat:  you have to extend the 1571's wiring to reach the connectors regardless of which mainboard you use, but with the 128-DCR you also have to extend the read/write head cables as well.  This can add noise and cause signal loss, and if your drive is excessively worn, it may play up a bit.

Using a flat 128 plus a separate 1571 board makes it a little more cluttered, but the read/write head cables will reach fine without modification and you can use the 1571 board to hide other cables.  Having the separate 1571 is not without its tradeoffs, though:  it can get in the way of the longer 5 1/4" devices.

As for choosing the case itself, you could probably pick most anything that suits your fancy.  That said, there are several things I learned while building these things:

1. Case size:  You need something that is at least standard ATX form factor.  It can be a server case if you like, but you don't necessarily need anything that big unless you have a ton of gadgets to add.     Anything smaller than standard ATX probably won't work without heavy modifications.  Also make sure there are plenty of slots on back so that you have places to mount connectors and such.

2. Board mounting space:  The 128-DCR board certainly is a big sucker, at 15 1/8" wide by 12 1/2" deep.  The flat 128 mainboard, by comparison, is 16 1/4" wide by 10 1/8" deep.  You'll need to add at least an inch to the two edges of the mainboard where the connectors are to give you room to plug stuff in (right angle connectors helps here).  If you intend to stack boards, leave at least 1" between them, and if you have a SuperCPU 128 MMU Adapter installed, you'll need at least 1 1/2" of clearance because of the micro clips.

3. Positioning issues:  There is one spot that I had minor troubles with in two of my towers, and that's the 5 1/4" drive bays.  Usually, these have a rolled edge that sticks out away from the drives, and this tends to interfere with the mainboard.  In my case, the positioning of the mainboard is such that the edge of the bays rests on the top of one of the mainboard's logic IC's.  It isn't in any danger of shorting anything out, so I just left it that way.  Also note that in a C128-DCR, there is a huge capacitor near the front edge, which may interfere with particularly long 5 1/4" devices.

4. Case width:  Wider is definitely better.  Most cases are around 7 inches wide at the bottom, but one thing I ran into with my case is that the bottom is narrower than I had planned for.  After mounting the mainboard, there wasn't enough room to set my SuperCPU and RAMLink upright in the bottom, so I turned them on their sides.  It was a good thing I did, as they are more secure and easier to access in this position anyway.  The negative here is that there isn't enough room to install my REU.  I need to build another short Expansion Port extender cable so I can position it out of the way.  If I could just lay my hands on a CMD 1750XL .... 

5.  Power:  As you know, the C128 needs 5VDC and 9VAC like the C64 does (the C128-DCR needs those plus 12VDC).  You'll need to find a way to power the C128's 9VAC rails (I used a 12V relay and a transformer fed from the 110VAC mains input), and if you use your RAMLink, you'll need to supply an always-on 9VDC to it (I used a wall-wart, also powered by that same 110VAC mains feed).  Practically all other C64/128 accessories can be powered from the 5VDC and/or 12VDC rails.  This is an ATX power supply you'll be dealing with, so you'll need to build a power on/off circuit also (very simple). 

6.  The back of the case will have a large rectangular cutout where an ATX I/O plate is supposed to snap in.  Of course a C128 doesn't have such a thing, so you'll need to have your Dremel (or a hacksaw and a file) so you can fabricate something suitable to cover it.  While this isn't pictured on my tower page, I cut a piece of steel from an old CD-ROM case, fashioned large tabs on one edge to fit into slots already present in the case, mounted my external IEC/Serial port on it, and bolted the other edge to the back of the case.

7.  Heat sinks where they're needed AND FANS!!!  'Nuff said?   

Yes it is possible, as the output from the port is just TTL-level signals like you might find on the IEC serial port.

That said, it was only intended to run one monitor at a time, so you might want to buffer the port first, perhaps with a 74*365 (hex non-inverting buffer) or a 74*210 (octal non-inverting buffer).  You could probably use a bunch of good transistors if that's all you have.

Whatever you use will require power, but you can draw that from the cassette port.

Yes, I did actually.  In stock drives, the mechanism is powered from a header located on the FD2000 controller board, and when I removed the controller from its case, I noticed that the regulator started getting really hot - enough so that when there was too much mechanism activity (particularly head movement), the controller would weird out.      It turned out that the drive's case normally serves as a heat sink in stock units, same as in the RAMLink.

The simplest solution was to make a heat sink similar to the one shown in the photos.  It worked well enough, but I've just been waiting for a chance to improve on it.

I replaced the regulator purely for cosmetic reasons (the pins were broken and re-soldered), but the new one has a slightly larger body anyway, so with the improved heat sink, it'll stay cooler and hopefully last a good long time.

I thought I'd post here to let those who don't visit my site regularly know that I've been working on a new C128 tower for the past couple of weeks, and it is just about finished.  Still a few minor things to do.   Note: this page has very high resolution photos, they might give your browser a stroke. ;-)

Link:  http://starbase.globalpc.net/~ezekowitz/vanessa/hobbies/c128tower.html

I've put the website back online. You can find it at the same address as before, and the content is as it was when I took it down (except that I removed some stuff related to the GNU Public License and just made everything I've released there public domain).  I'll update some of the old stuff soon.  Enjoy!

I took the website down because I felt the content had gotten outdated.  I still have the site's files, and I've been thinking about putting it back up, but I really can't think of anything to replace the old content with.  I still have the files it was comprised of, so if there's something you wanted, just let me know.