1. Power Supply Chassis: Provides B+ and B- voltages for the tube plates.
2. Blower Motor/Ducting: A squirrel cage blower and duct work provide continuous cooling to all of the vacuum tube modules. Air is drawn in through filters that surround the power supply chassis.
3. Magnetic Drum: 4k-word of mem is available on a recirculating magnetic drum driven by an xx HP motor connected by v-belt.
4. Digital Display Unit: Provides the high voltage for the CRT, along with filament isolation and typical HV controls (astigmatism, focus, etc.). The Horizontal and Vertical control modules insert from the front.
5. Sequencer Unit: Houses the relays responsible for a controlled startup of each section.
6. Card Assembly 1: Contains the primary “logic” tube modules; top of unit houses the diode logic PCB.
7.Card Assembly 2: Contains the bulk of the tube modules, especially those related to reading and writing to the mag drum.
In addition, the base contains a massive Sola transformer that provides regulated 120V power to the filament transformers throughout the machine. A smaller (but still sizable) isolation transformer provides 120V to the Flexowriter. The control panel mounts to the top of the chassis and interfaces primarily with the Sequencer Unit.
Friday, December 23rd, 2016: “Want to go to Canada next week?,” I asked my wife. The response was met with some trepidation….”can I think about it?” 72 hours later we were on the road, staring down the barrel of a 3000+ mile journey. You see, the freight was arranged, but there are some delicate, irreplaceable parts to this machine. My concern was primarily the drum. A unit housing 74 magnetic heads that ride less than 0.001” off the surface. Any damage to that would be, well, not good. After posing the question to the ClassicCmp mailing list after failing to dig up any maintenance or setup manuals (was there a drum lock? Spoiler alert…there’s not), the collective opinion was that the drum should probably be hand carried back to the states. Flying, driving, and a combination thereof was considered, but it was Xmas break and a road trip sounded like fun. I’ll leave out the parts about ND closing all roads in their state the night we approached, the blizzard we somehow got through in a hatchback, the grouse that challenged the car head on at highway speeds in Montana (and lost), and other interesting stories…
Initial inspection: The comp had been donated by a civil engineering company in Vancouver to a known recycling agency. Originally, it had the external high speed reader/punch but its whereabouts were unknown. Quite possibly just the machine was saved as a non-functional artifact (I believe it was displayed at some point) and the reader/punch considered unnecessary bulk. But the comp itself was complete, sans power cord. All cards were present and connections looked good (no rust, corrosion or signs of abuse). The company had stenciled their name across the front of the machine in very unflattering paint, but I knew this going in. The Flexowriter was complete with the (often lost) paper tape supply holder on the back. Its chassis looked clean with no broken or missing keys or levers.
The drum and motor assembly was easy enough to remove and all module cards were marked, photographed and carefully wrapped for transport. The Flexo, control panel and CRT were loaded up to take back and the rest got palletized. Transporting the fragile parts home went better than expected, though I think we both now need a getaway from this getaway!
Jan 9: A Little History
While waiting on the machine, I started work on the Flexowriter, control panel and tube modules. Immediately apparently was a mix of late 1950’s soldered components and cards and late 60’s/early 70’s tubes and drum cert. This supports the theory of having lived two lives. I was curious who the original buyer or renter of the machine was and inquired with Christian Corti in Stuttgart. He has access to a list of the original POOL (user group) subscribers, but #393 failed to make the list. #394 was owned by Spink Engineering in Sacramento (company since sold), and #389 by Holley Carburetor. No news there.
I went up to the local university to go through their archives. Per Datamation Magazine, in August of 1965, Control Data bought out Librascope, which was General Precision’s computer ‘operation’. They acquired existing rental and service contracts, computer inventory and sales and service for the LGP-30, the 21 and the RPC-4000. An engineer shared a story at Ed Thelen’s LGP-30 page about helping refurbish these machines during this timeframe; those machines were then sold outright in the ballpark of $10k each. Small businesses that could make use of a computer were willing to rent or pay, even on an obsolete machine, to at least get their foot in the door. My #393 has all the signs of being part of that refurb, including the Canadian Control Data nameplate on the control panel where the General Precision emblem would have been, and has other Canadian marked parts, like the belt that drives the drum.
Jan 17: The Machine Arrives
The call came in around 1PM…a semi was headed my way with a 700lb delivery. “A loading dock?…no, I don’t have one of those.” After a couple of quick phone calls and some quick thinking by the trucker, we made use of the dock at his first place of delivery. A mere two miles from where I work. I rushed over with my truck, managed to get the temperamental tailgate to drop and backed up to a low spot adjacent to the building (perfect elevation!). Within 30 seconds, a fork truck popped out of the building with a shrink-wrapped 4’x4’ pallet and loaded it squarely in the back of the truck. Paperwork was exchanged and I was off.
I took my sweet time driving on our mid-January, ice covered roads; no sense in risking damage now. Once home, I figured I’d pull the weight out of the machine, one chassis at a time, which I did. And yet, still could only get one end in the air an inch or two. Even backed up to my sloped unloading spot in front of my shop…this would be too big of a drop. To give you an idea of the weight of this chassis, imagine lifting a 50’s Steelcase tanker desk. Now imagine that desk full of power transformers. Short of owning a forklift, I moved the truck into the yard, grabbed the utility tractor and lined the flat of the bucket up with the tailgate’s edge. Now I could ease the LGP-30 into the (blanket-lined) bucket, carefully tilt it back, bring it down, and deliver it right to the open garage door. Success!
The initial inspection showed the machine arrived in pretty good shape. Now with the pressure off, and parts unbolted and the machine in a controlled environment, I could take my time. The exterior panels were in good shape. Nothing was damaged internally that I could see. The top will need refinishing due to years of the Flexo being dragged across it, likely during its display tours. The power supply unit (mounted behind the air filters at the base) is a monstrosity that will require some going over. I pulled the cover on the ‘logic card’ and carefully unplugged it from its edge connectors for inspection. This PCB-sandwich contains the 1000+ diodes that bring the smarts to the machine while the bottom half contains some serious resistors. All mounted in the airflow channel for even operating temp. Overall assessment: Positive.
Note: Work had been ongoing on module cleaning and recap, Flexowriter restoration and control panel cleaning/rebuild while waiting for the machine to arrive. Updates for those topics can be found in their respective sections.
Jan 19: Chasing the Details
The air filters for this machine are real oddballs, but the closest that make sense are two 1x9x14’s and one 1x9x18. I can order these “custom made”, which is really nothing more than cutting them down and regluing the cardboard frames, or I can do it myself. I’ll check stock first locally to see what I can modify.
Update: Local box store to the rescue; now to find time to cut them to size.
I also had a chuckle the first time I saw print #312613 in the released schematics for the LGP-30. It’s a drawing of the power cord, and that’s it. But the laugh was on me since I was able to reference 1956 Hubbell part number #7313 which is the twist-lock end. Ebay to the rescue, I found a seller with the exact bakelite and steel part, NOS. A modern plastic version just wouldn’t be the same.
Locating heavy duty ‘rubberized’ 12/3 flex cord also proved difficult as everything today either has molded connectors or a shiny PVC coating. I ended up ordering the stuff from a contractor supply house that sold on ebay and it did not disappoint. Mated with the bakelite receptacle and a steel-bodied plug, the assembly looks factory-correct.
While waiting for the blower motor start cap to arrive, I turned my attention to the wiring harness. Translucent plastic clamps of various size are used throughout the machine to carefully secure and route the various harnesses, and this plastic has become brittle to the touch. Luckily, I was able to find 1/4″ and 3/8″ plastic clamps locally that did the trick. All are secured with allen head cap screws and washers, of course.
Jan 28/29: Panel Fit-Up While Waiting on Parts
I spent a little time this weekend working on the ventilation blower motor (see Blower Section) and aligning panels. The motor start capacitor finally showed up late in the day Saturday, which left me with the morning to play with the brackets and fasteners that the exterior panels attach to. The alignment wasn’t so hot by the time the computer arrived by freight, and many fasteners were loose. After getting the four exterior panels set with a minimum, but reasonable gap, I installed the top cover. The only fixed points for the top is where the front pivots attach. Everything else is adjustable around this. Next was getting the control panel set so it appeared flush, but wasn’t touching the top cover, which can move ¼” in any direction due to tolerance stackup. There are screw pairs on each side of the controls for fore/aft, left/right and panel angle (12 in total). With that reasonably set, I installed the Flexowriter ‘table’ and noticed yet another serial number (375) that was penciled in on the bottom.
The Flexo table is trimmed in ribbed stainless like a 50’s Formica countertop and this trim matches the band around the top cover of the computer. Unfortunately, there were a couple unsightly dings in it. In the end, I removed the couple dozen screws that held it attached and massaged out the glaring imperfections with a nylon-tipped mallet and wood blocks. The top surface looks like deep brown linoleum, which was a popular material for desk tops. I cleaned this well to remove years of adhesive, stickers and packing tape residue, then treated it with leather cleaner and buffed it to restore some of its luster. I think it’ll present well as-is.
An update on the entire system…we are up and running! With the hold-in contacts changed out in the sequencer, we have reliable startup. News of the bearing replacement was covered in the blower section, and an update in the mag drum section covers the greasing of the noisy end bearing (which is thankfully staying nice and quiet). The issue now is nonsense on the scope tube. Just a squiggle of a line rather than 3 distinct square-wave waveforms. Replacing the old mica caps on the vertical and horizontal boards did not improve the situation.
I found time to pull out the Tek scope yesterday and play around a bit. We have a good signal from the timing head on the drum, which sets the machine’s cadence (left pic-beautiful waveform). A clock generator board takes this signal and provides ‘T’ and ‘T-bar’ pulses to the machine’s logic (rt pic). These signals also looked acceptable.
Probing further, the signal ‘t3’ (shared by both vert and hor modules), which I assume is the result of some combination of logical signals from one or more of the three timing tracks looked…terrible. The next move is to get the timing heads more accurately set. I’ll need to pick up and modify a couple small hex wrenches to do this with the assembly mounted on the drum. In the mean time, I pulled the diode logic board and went through all the diodes. Two bad ones were found and replaced, but no change to t3.
Cory Heisterkamp 2017