I'm restoring some pre-TTL DEC stuff (R and W series logic) and I need
a handful of front panel bulbs. Unlike the later bulbs with a plastic
base and strong wires to solder to the PCB or to plug into socket pins,
these are like a kernel of corn, with two fine wires coming out of the
glass envelope at a slight angle with no supports of any kind.
Some of the bulbs have burned-out filiaments. Many of the ones I need
to replace have broken wires externally. I am going to attempt to
solder new wires to the stub, but I don't expect the attempt to be an
overwhelming success. I might try a harder solder so that when I put
them back in the frontpanel PCB, they won't give way when I put enough
heat on the other end to install them.
About all I know about these bulbs is that they are fed a nominal 12VDC
from the W-series driver boards. The front panel itself is literally
just
a PCB and a bunch of bulbs; no active circuits (unlike,
say, the front
panel of a PDP-8/L). Testing should be easy - feed 12VDC at a few mA
to each set of fingers and check the bulbs, one by one.
If DEC was underfeeding these bulbs to extend their life, I would
expect
that the bulb should be rated at 14V-16V. I measured
slightly over
12VDC
in circuit, but well within a 5% tolerance.
I have checked the online manuals I could find, but no mention is made
of the nature of the bulbs for 1966/1967-era DEC equipment. Any ideas?
I relamped the PDP-9 I am resoring for the Rhode Island Computer Museum
a little over a year ago, and what follows is my description of that
process. Given the time frame of the PDP-9, I suspect you're looking at
a similar situation. I bought my 1764 lamps off the shelf from Mouser
(
www.mouser.com):
The PDP-9 lamp board on which I wasted so much time and solder today is,
I'm sorry to say, a nasty bit of work - not DEC's finest hour.
Consider, it's 60 (not 56 as I previously reported) type 1762 T-1 3/4
wire-terminated lamps soldered into a board with 60 driver transistors,
an electrolytic cap and a handful of connectors. It's buried deep
inside the front panel, accessible, and only with difficulty, by
climbing into the main system rack. Type 1762 lamps are very
interesting. They're roughly 3/16" in diameter by 3/8" long, domed at
the top (where they show through a masonite light blocker panel to the
laminated plastic front panel of the -9), and roughly conical at the
bottom. The wire leads protrude TO THE SIDES about 1/16" up from the
bottom, and the leads are bent down so that they can be soldered into
the board. I've seen oddball lamps before, but these are about as outr?
as little incandescent lamps get. The wires are hard-drawn (ie: very
brittle) copper, and they appear to poke through the glass without much
of a seal. Over time the copper has corroded at the glass interface
(Lots of green and blue ambergris. 'Looks good on an old roof, but is
very disconcerting to see in electronic circuitry.), and the lamps break
loose if you even think about looking at them. Also, enough tungsten
has boiled off the filaments over the years that the top domes of most
of the lamps are greatly darkened.
I didn't bother to use any of the old lamps (although I salvaged what I
could for posterity), but replaced all 60 of 'em with type 1764 lamps.
Type 1762 lamps are damned-near impossible to find anymore (and given
their construction I'm not surprised one bit); type 1764s are identical
electrically (28V, 40ma, 340mA max cold surge current, CF-2 filament
structure, 4000 hrs), and are the same size mechanically, but the wire
leads come out the bottom through a more-conventional, and far less
fragile, frit seal. Also, the leads appear to be tinned Kovar. They're
stiffer but more malleable than the copper leads, they appear to "wet"
where they seal with the glass, and they're magnetic. They also solder
quite well.
I dunno where DEC got those 1762s from. Maybe Ken got a good deal on
'em or something, and any insight from old DECies would be helpful here,
too. But they sure did suck.
In any event, all 60 lamps were removed. All 120 contacts (staked
eyelets, by the way, pressed into a single-sided non-plated-through-hole
tinned glass-epoxy board roughly 3" wide by 18" long) were cleaned - in
large part to remove the drek left on the board from previous
re-lampings. (I found evidence for at least one, and probably two,
re-lampings, as well as the replacement of at least one of the driver
transistors. One of the jobs was done by a tech whose training appears
to have been at Hormel. He or she didn't know the difference between
careful soldering and slaughtering hogs!) All 60 new lamps were
installed and aligned (And I carefully made sure to get all their
polarities right! <grin>), and the soldering job was inspected both
visually under a magnifier and with an Ohm meter. The visual inspection
found only one cold solder joint, and the meter found only one short
(from a "sailor joint": gobs of solder), both of which were corrected.
The driver transistors had previously been verified by junction
checking. We should be set to go.
I would recommend should a re-lamping ever again be required that LEDs
rather than incandescent lamps be used. The availability of suitable
lamps is questionable in the future (I just happened to find a stock of
300 at Mouser Electronics, now down to 250), and the ability of the lamp
circuit board to take too many more re-lampings is questionable. A
standard 20mA T-1 yellow LED (watch that polarity!) in series with a 680
Ohm 1/2W resistor to handle the roughly 15V drive voltage, "air-bridge"
constructed to fit into the hole in the masonite light blocker of the
front panel, would be a suitable permanent replacement for one of these
lamps.