17 May, 1999
Just some pictures of the construction of Model 11 serial number 2. The electronics are my standardized Model 01 CPU/IO card, designed around a MicroMint PICSTIC. I make 'em as two boards, as I usually throw the Nixie driver board away when I embed a Model 01 in other projects.
Serial 2 is very small for a Nixie clock with 1" tubes. It uses a compact National display, so the six digits and NE-2 colons fit in six inches of width (1" round requires more than 8"). The clock is assembled in a stack, to fit the small (6.75w x 3.75h x 5d) wooden cabinet. The cabinet once housed a Weston D.C. Voltmeter in 1923.
Here the Nixie sockets being wired (a tedious 60 conductors of ribbon cable); a close-up of the socket, with driver board behind (which is actually the prototype PCB with wrong, huge capacitors).
Oops, here's yet another clock awaiting assembly, it's sockets (for round tubes) waiting metal work. Pretend you didn't see this. You can however see the empty cabinet for #2 in the upper left.
The basic electronics now working, tuning the firmware for the specific front panel controls (each is different slightly). And a better front view and rear view. You can see the "AM/PM" and one "colon" neon lamp lit, they're just tacked into place for testing.
The front panels are more work than any other subassembly, and sometimes more than the rest of the project. This one took two tries. Here is this one's first attempt, right out of the etching tank; right out of the tin-plate bath; the background pattern getting black enamel filled, and color-codes applied. The holes for the Nixies gets milled out next, not shown here. The panel you see here was not used, it had too many flaws; I milled it flat and started over again (not shown, too depressing). The final is near-perfect; I got tenth-mil resolution this time, tuning my processes; 2-point type is quite readable (sic), and you can see slight jaggies from the 600 dpi photomask (with a 5X magnifier) in the metal. Not bad!
Final assembly for #2 starts with mounting the electronics stack to the front panel (shown here in "lash up", power transformer wires just tacked on for testing). It is epoxied, to avoid through-bolts (unsightly on a small panel) and soldering (before, would complicate patterning and etching; after would destroy surface or enamel). Here's a decent side view showing the density of junk it takes to drive Nixies (for scale, the vertical aluminum spacers are one inch high). (This clock had to fit in a tiny volume, complete, hence the high-density). The front panel is face-down, with the six Nixies in their green sockets pointing down. The PCB at the top is CPU and low-voltage power supply; the other (2nd from top) PCB is the HV power supply (low-heat regulated 170V), Nixie decoder and NE-2 colon and AM/PM indicator drivers. The clot is caused by the ribbon cables necessary to be able to lay the thing out on the table to assemble and wire. In this particular model the Nixie assembly is only an inch from the PC board; usually it's many inches away. No two Model 11's are the same. Production to me is what is known in industry as "brassboard", the step after "breadboard"; a one-off, real-life device make to spec.
Here is the electronics assembly, in final trim, ready to install in the box. The three wires go to the AC line input connector/filter mounted on the back of the wooden cabinet. The screws attaching the rubber feet on the rear of the assembled unit also hold the entire electronics assembly in the cabinet. They screw into the ends of the standoffs shown in the previous pictures.
The completed unit with the cover closed (at 80 degree ambient, the temperature rise deep in the cabinet is only 20 degrees F with the cover closed after 12 hours operation); and the front view, cover open, as it stands on the table.