Analog Logic for Tube Based "Computer"

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have developed an unhealthy interest in tube-based analog logic. Can anyone point me to links or even books which explain the prior art?
Specifically I am interested in a "multiplier" which would allow the use of two input signals, with the output correlated to the product of both. The dual-control 5915 seems like a candidate, https://frank.pocnet.net/sheets/137/5/5915.pdf
but the datasheets seem to assume the use of either G1 or G3 as control, but not both at the same time...
 
I expect you will have to use either a dual section tube, or two tubes, in order to have the same sensitivity on both analog inputs.

You will have to decide whether you want a linear response or a logarithmic response.

I then thought of the 6386 logarithmic dual triode.

Sorry, I can not be more help than that.
 
Moderator
Joined 2011
Last edited:

PRR

Member
Joined 2003
Paid Member
There are SO many ways to multiply with tubes.

Most suck at the limits. All multipliers do, but the early transistor jobs just blew-away all but the slowest and most tweaky tube multipliers.

Korn And Korn was a classic in the day, though I found it assumed much common knowledge which is no longer common.

Why not go back to Ball And Disk?
 
Slow is OK. I actually want to multiply fluctuating DC voltages for fuel injection of my 1949 Flathead. So I need linear multiplication of engine rpm and manifold pressure, with non-linear comp. for engine temperature. Once I figure out the logic side I will start another inquiry into single-ended driving of low-voltage solenoids without an output transformer... (single point injection, continuous flow, via a solenoid to draw a tapered needle out of the fuel orifice (Solex style)). Yes, I know that tubes are not the obvious solution, but I want to be historically authentic, at least in terms of what could have been done in 49, even if no one did.
 

PRR

Member
Joined 2003
Paid Member
Carburation is not naturally a job for electronics. There's no naturally electrical details. While we are surrounded by electric injectors, this can be done all mechanical. As was done on late WWII aircraft, and the late 1950s GM injection. As a contrast, look at the history of the Mopar electronic injection:
1958 DeSoto Electrojector - First electronic fuel injection?
All the cars built with Electrojectors were changed back to carbs. (One was much later gotten working OK with the fuelie.)

Don't multiply voltage. Multiply in time. Ramp waves and comparators.

Don't think in RPM. What the engine really wants is one lump of fuel for every intake stroke. The size of this lump a function of manifold pressure. For non-sync injection (works fine): at each spark trigger, a one-shot makes a ramp. A comparator takes this ramp and manifold pressure, naturally gives the Product. When the engine is cold, reduce the current to the ramp generator, you get a longer/larger lump.

Swapping-around: the ramp generator target could be proportional to manifold pressure. Now a fixed voltage comparator does about the right thing. The "fixed" voltage could vary with temperature.

There also wants to be separate trims for idle (air leakage foils accurate computation) and full-power (we normally over-compress for economy, then suppress knocking by going rich at WOT).

All of which dodges the question: why the heck injection on a flathead?? Ignoring inverted flight and hi-alt carb icing, the usual reason is that a carb must extract energy from the incoming air to do its computations and make fuel flow; the injector does not obstruct the intake air (but must get energy from shaft or dynamo). However a flathead already has such a convoluted air intake that the carb does not dominate the intake restrictions. Yes, I know: millions (well, hundreds) of Hilborns on flatties say it works. And made sense when old Fords were the only engines so abundant that you could afford to blow them up weekly. But the Caddy OHV breathes so much better it aint funny. Chevy V8s too, and are super abundant.
 

PRR

Member
Joined 2003
Paid Member
You want this book:

Handbook of Industrial Electronic Circuits, 1948, J Zeluff, V Markus
Amazon
ABE

I'm not sure how much is directly useful. But it is a good show-and-tell prop when people meet your flathead. And it may teach you some about how they thought back in the day.

Fer example: here is an aircraft turbo control.

Some things you never see today. The power stage works on AC not DC. So while it looks like the 7C5 power tubes are driven parallel, actually only the one with positive plate conducts. This puts AC into a 2-winding motor which screws the wastegate knob.

Down right: all those curlies are on the main transformer and deliver small floating AC voltages. The four pots tap-off fractions of winding voltage. Some of them go in-phase to out-phase, which when amplified drives the wastegate in or out. A bal-pot closes the the NFB loop. (Everything is AC except the 7Y4 makes a little DC for the 7F7 gain stage.)

Power is 115VAC. Hey, a B-17's core is about the size of a house, and the heated flight-suits pulled power like a small house, not to mention a radio-rack as big as some hometown radio stations. However it was 400Hz which is awkward on miles-long lines but makes transformers and motors much lighter.

The book is full of stuff. Some quite simple, some quite arcane. But even railroad splice detectors sometimes have odd details which may be of use elsewhere. (And yes, some "audio", little of real interest.)

Oh-- the "Selector" pot is grounded because that's at the pilot. All the rest is out in an engine pod. This way minimizes wiring. While this is more likely for a B-29, here is how you service the "spark vibrator" on a B-17. (The black thing above the left corner of the ladder is the hub of the turbo.)
 

Attachments

  • turbo-1948.gif
    turbo-1948.gif
    178.7 KB · Views: 67
  • B17-legs.jpg
    B17-legs.jpg
    80.2 KB · Views: 69
Last edited:
My copy of the handbook is on order. Thanks!

I admit that logic plays no role in my fascination with flatheads.
For all of his tragic flaws, I respect Henry Ford for two things; the democratization of the auto via the Model T, and the democratization of racing via the flathead. Both were previously the privilege of the aristocats and he made them available to all.
As a manufacturing engineer I also love innovations in manufacturing, and the casting process which enabled the flathead was pure genius. V8s had been known for years, but could only be produced at great expense. But "Cast Iron Charlie" at Ford revised the casting process to enable mass production.
In addition I love grassroots innovation. And as the flathead was the first low-cost engine with reasonable power it became a magnet for shade-tree innovation. Hilborn was one of the few who became famous, but others dropped flatheads into P-38 drop-tanks for low cost streamliners (or lakesters) to race on the flats (the flathead has less frontal area than OHV engines), supercharged them with military-surplus roots blowers, and invented the quick-change rearend so that the only team you needed was a buddy and a girlfriend help to reconfigure your car from drag-racing in the morning to circle-track in the afternoon. You might not win, but racing was more like playing golf or tennis. The event was generally more social than bitter competition.

My roadster has a Model T body, a Flathead V8, and a home-made quick-change. So All I am missing is a 6-71 blower and the tube-based injection that should have been built by one of those GIs who came back after working with both fuel injection and tubes. I have a pair of 6-71s on the shelf in the garage, so now let's talk about tubes...
 
A previous reply took me here:
Audio Synthesis via Vacuum Tubes

It seems really elegant to do this with a single tube, with one signal on each of three grids.

The only tube I have found which seems to be intended for this is the 5915.
https://frank.pocnet.net/sheets/137/5/5915.pdf

Are there others?

The data sheets for the 5915 provide curves for using either grid 1 or grid 3 for the signal, but not for using both at the same time. Are there references out there for doing so?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.