Excerpted from private email from Ed Lyon. Ed explains the operation of the NU multiplier tube in connection with a possible new application for the tube as an alternative to the extraordinary design of a Double Reflex Super-Heterodyne radio by Robert Weaver. An alternative discussion ot Robert's radio is also available at the Radio Museum.
------------------------------------------------------------------------------
From: "Ed Lyon"
To: "'Joe Sousa'"
Date: Tue, 7 Apr 2009 08:53:10 -0400
Joe;
If it's elegance you like, how about this for a single-plate sheet-beam tube doing what Bob Weaver did:
1. Use the National Union experimental tube I think I told you about over a year ago. It had a cathode, control grid, beam-forming electrodes, and deflection plates, like Bob's. But the plate was J-shaped, sort of like the sketch in the attachment here, with a screen grid inside the "J."
2. Undeflected, the electron stream strikes both the plain tantalum part of the plate and the left part, which is PbO coated to make it a dynode, rich in secondary emission. Electrons emitted here go to the screen. The net plate current is zero.
3. Upon deflection, the electron stream moves to the dynode side or the plain side, depending on deflection direction, and the degree of motion depends on deflection voltage magnitude. This makes the plate current vary about zero, with the plate voltage rising above B+ for leftward deflection (in the sketch), and below B+ for rightward deflection. The screen current does the opposite, but only reaches zero for full rightward deflection.
4. So, one could reproduce Bob's circuit, and get the push-pull output (the L.O. and the first IF output) from a balanced transformer, just like Bob's but driven by the plate and screen, which, in the National Union tube, produce complementary currents, upon deflection, but common-mode currents, upon control-grid action. So it looks like a single-ended tube, but actually makes a balanced output. How's that for stumbled-upon elegance?
We used these N.U. tubes for four-quadrant multipliers in the analog computer days of 1954-55. N.U. made about fifty of them for us. I think I may still have one, somewhere.... wish I knew where.
Ed
In the following email, Ed Lyon described the application of the Anodyne in analog flight simulators of the 1950s
------------------------------------------------------------------------------
From: "Ed Lyon"
To: "'Joe Sousa'"
Date: 2-6-2019
Joe:
On page 9 of the February 2019 issue of Lud’s [Ludwell Sibley] Tube Collector, note the short article on one of the National Union deflection tubes. These are like the National Union tube I tried to explain to you years ago, that I had used to develop a four-quadrant analog multiplier back in the 1950s. I used two tubes plus a handful of K2W Philbrick op amps. We needed the multipliers for a flight simulator. The KC-135 simulator, for example, required over 60 multiplication operations, just for inserting the various aerodynamic functions of Mach number. The all-electronic version we were trying to develop was to replace servo-driven precision potentiometers. The Mach servo, itself, mounted 83 such pots, most of which were tapped with up to 12 taps, for the purpose of approximating non-linear functions of Mach number. Since the Mach servo was actually driven by the first derivative of Mach, its only “answering” feedback was the output of a drag-cup generator, determined by the velocity of rotation, so this servo was an integrator. Of course, overall feedback was via the aerodynamic equations of flight, thrust versus drag, etc. The flight simulators were mostly a-c analog computers, most working on 60 Hz, but some were made for 400 Hz, mainly to make them smaller.
Regards,
Ed
------------------------------------------------------------------------------
From: "Ed Lyon"
To: "'Joe Sousa'"
Date: 2-12-2019
Joe, I have the handbook for a version of the F-86D flight simulator. I was in charge of checkout of this version, then I was in charge of design of the F-86K simulator which was made for use in France and Italy since the US sent all F-86K aircraft to those NATO countries. The F-86K used 20-mm cannon for interception of enemy bombers, while the F-86D used rockets, so the computation of intercept paths and tactics was different. The rocket-fired intercept used a lead-collision course of flight while the cannon-fired intercept used a lead-pursuit course of flight. These flight courses were cued to the pilot through his fire control radar, computer, and display. The aerodynamics of the two planes were similar, but locations of the weapons stores and fuel tankage was different, so that the weight and balance computation, and their effect on flight control requirements differed somewhat. There was also a recoil force on firing the guns which was absent in the case of rockets, but then there was an aerodynamic drag and pitch force in the rocket pack extension (the rockets were in an internal belly-mounted pack which was hydraulically extended into the ventral airstream for rocket firing in the F86D. Our company made 60 of the F-86D simulators and 8 of the F-86K models, these latter ones built into semi-trailers. I can send you some pictures scanned from the handbook, once we get out of this current ice storm.
I also worked on the simulators fo0r the F4D Navy fighter, the A3D Navy reconnaissance/bomber, the RB-66 Air Force bomber, the ZP4K Navy blimp, and the KC-135 Air Force refueling tanker, mostly on aerodynamics computation, then later on testing/checkout. All this was in 1953-56.
Ed