8

This question is about computers with front panels that use only toggle switches for input of binary data (such as address and data values), as opposed to keypads for numerical input or other methods.

On these front panels, as well as state set by ON-OFF switches as above, there's a need for momentary-contact switches for "commands" such as examine, deposit, reset, start/run, and so forth.

In the 1950s either push buttons, such as on the IBM 1620, or momentary-contact toggle switches (that spring back to "off" when released), such as on the PDP-1, were used, depending on the manufacturer and model:

IBM 6120 front panel PDP-1 front panel

But by the 1970s, it seems that momentary-contact toggle switches were by far preferred over push buttons, both for minicomputers such as [various versions of the PDP-8] and the Data General Nova and [Nova 3], and microcomputers such as the Altair 8800, IMSAI 8080, and Altair 680:

IMSAI 8080 front panel

(Above, the left-hand two banks of eight switches are standard ON-OFF toggle switches for address and data input. All of the right-hand bank of six switches, except for the right-most POWER switch, are two-way (ON)-OFF-(ON) momentary toggle switches for functions like EXAMINE, EXAMINE NEXT, DEPOSIT, RESET, etc.)

In at least the case of the Altair 8800 and the IMSAI 8080, these momentary contact toggles were still (mostly two-way) SPST switches, no different electricially from normally-open push buttons.¹

Was there any particular technical or ergonomic reason for this seemingly wholesale move to toggle switches, or was it just chance or fashon?


¹ See the bottom centre of the schematic on PDF page 25 of the Altair 8800 Theory of Operation Manual & Schematics. S4 is an SPST momentary-contact toggle switch; S5 and S6 are three-way toggles (resting at centre, with momentary up and down) which are effectively a pair of SPST switches sharing a common terminal. For all of these one terminal is grounded by Q̅ from the run/stop flip-flop (upper left) when the system is stopped. A line connected to the other terminal, normally held high by a 1K pull-up, is momentarily brought low when the switch is held. This signal is debounced by half of a 74123 dual pulse generator.

  • 2
    Electronically speaking, switch denouncing may be easier to implement for momentary toggle switches (gates) than a pushbutton (RC circuit). But I'm not sure if it's really a contributing factor to the front panel design. I'll check some schematics and report back... – 比尔盖子 Sep 2 at 7:34
  • @比尔盖子 That seems very odd to me, because I don't see what the difference would be between the two poles of an NO pushbutton and the two poles of an OFF-(ON) toggle switch. Wouldn't you debounce either in the same way, using one of any number of methods? (FWIW, the Altair 8800 used 1/2 74x123, which I guess included an "RC" circuit because that chip wants an external cap and resistor to set the timing constant for the pulse length.) – Curt J. Sampson Sep 2 at 7:38
  • You're right. It's completely irrelevant. I originally thought that a double-throw switch allows one to use the classic SR latch debouncer, but after checking the schmatics of various PDP-8 front panels, I found the momentary toggle switches on some models were single-throw. Also, packaged gates, either latches, Schmitt triggers or single-shots were used in all debouncing circuits, there was really no advantage of using a double-throw, so I am incorrect. – 比尔盖子 Sep 2 at 8:36
  • The switches in your bottom photo don't look like momentary toggles to me. They are very clearly resting in two different positions. – JeremyP Sep 2 at 9:13
  • @JeremyP The address/data switches are not momentary, but the "command" switches for examine etc. are. I've updated the post to better describe this. – Curt J. Sampson Sep 2 at 9:32
7

There are usually multiple contacts made when a switch is operated (it's called bounce). Debouncing a large number of (SPST) pushbutton switches is difficult (requires lots of components, or even a short computer program), BUT if you use a double-throw switch (SPDT), this can be accomplished with a S-R flipflop. The S-R circuit is simple, early integrated circuits could incorporate several such (the SN7400 NAND chip makes two S-R flipflops). More info hereset-reset flipflop debouncing.

So, a toggle switch with a buckling-spring mechanism (a "microswitch" of the SPDT type) with a bit of electronic support, made a noise-suppressed logical input, ideal for the front panel switch application.

Only a few (strobe-generating) functions really required good debouncing, but once you designed such switches into the front panel, it was convenient to use lots of identical ones, for noncritical functions as well.

  • Yes, this is what I was originally thinking in the comment, and later discarded it. After checking the schematics of PDP-8, I saw some versions of front panels use SPDT but other use SPST for the same button, although I did see the flipflop debouncers were used for the SPDT switches, but it seems they just used whatever was best available at the same, so it's probably not a major factor. Still upvoted as the answer is useful. – 比尔盖子 Sep 3 at 2:20
  • This is exactly the kind of thought I was looking for. However, the toggles are SPST, electrically identical to pushbuttons. These were debounced using a 74123 pulse generator. I've updated the post to mention this, and added a footnote giving these details and a link to the schematic. Also, I'm not sure that there were any "noncritical" momentary contact swiches on the panel. (The address/data switches didn't need debouncing, but they were not momentary contact and could not be NO or NC push buttons.) – Curt J. Sampson Sep 3 at 3:18
  • Actually, I should be clear that it's the toggles on the Altair 8800 that are effectively NO switches; if you can provide examples of machines that do use the technique you suggested (using DPST switches to get break-before-make so one can debounce with an SR flip-flop), that would be great! – Curt J. Sampson Sep 3 at 7:16
  • @CurtJ.Sampson: Lee Hart's 1802-based computer has SPDT switches that go between VDD and VSS, and uses non-inverting buffers with feedback resistors to latch their state when floating. That's a recent design based on components that would have mostly been available in the 1970s (except a 32Kx8 CMOS SRAM); I don't know if/when machines would have used that technique historically. – supercat Sep 9 at 22:27
  • @supercat That's the 1802 Membership Card? Interesting design to eliminate some of the usual momentary switches. I note that for the one momentary switch that remains, he uses a push button, not a momentary toggle. – Curt J. Sampson Sep 10 at 2:45
5

On these front panels, as well as state set by ON-OFF switches, there's a need for momentary-contact switches [...]

Don't forget three way switches as well.

Was there any particular technical or ergonomic reason for this seemingly wholesale move to toggle switches, or was it just chance or fashion?

It's all of the above (and more), with technical/design being the lead.

(It helps to keep in mind, for all of these points, that basic on/off have to be toggle switches to begin with - using arresting push buttons would be ratehr expensive)

  • (board) Design - using switches from the same series/family requires only one mechanical layout. Why making your life harder a it must.
  • Availability - designers are usually requested to order from an catalogue of already contracted suppliers. Doing otherwise is shunned at by logistics.
  • Reliability - switches of the same design tend to have the same operational margins.
  • Economics - buying more switches from the same series from the same manufacturer will usually result in a lower per piece cost.
  • Management - only one contract has to be made covering several part.
  • Production - only one tooling for all switches due having the very same measurements.

And yes, it's as well about design and ergonomics:

  • Design - doesn't it look neat if they are all the same?
  • Ergonomics - there is only one movement direction (up/down), not mutiple (up/down, push(pull)). Also having all of the same series/family leads attention to grouping and deliberate marking insted of different forms of switches.
  • Main reasioning, I guess: Momentary switches are much harder to operate by accident than push buttons. – tofro Sep 2 at 8:35
  • @tofro I'd take that as a minor part. After all, a front panel does need flip switches to setup data and address (push buttons holding position are expensive and need much depth), so making the momentary the same just makes sense. – Raffzahn Sep 2 at 8:44
  • I post this comment in a spirit of clarification. You say, "Don't forget three way switches as well." I did not; I was thinking about them throughout the question. You may want to go back and read again with that in mind to see if you missed anything. For example In the quote following "Didn't you write..." you removed the criticial phrase "as well as" before "ON-OFF switches," which was there specificially to make it more clear that the ON-OFF switches were not the momentary ones, they were the address/data switches described just before that. – Curt J. Sampson Sep 3 at 2:25
  • 1
    I'd give more weight to the ergonomics of it. Given you're flipping in an address or data, flipping a further switch to examine/deposit values is just more... natural. Older PDP-11 consoles were beautifully usable; I never did get on with the 11/34. – another-dave Sep 3 at 12:05
  • 1
    Plenty of flippity-switch experience with various PDP-11s, but no real equivalent for push-buttons. The 11/34 used a calculator-style keypad so it's not really the same thing. Either way though, I'm not sure my comment counts for more than "opinion" - mind you, maybe opinions is all there is. – another-dave Sep 3 at 23:27
0

I used a PDP-11/45 a bit when I was an undergrad. I also used a few computers with some manner of push buttons, including keypads, on their front panels.

The one thing I liked about the 11/45's front panel was that eventually I was able to change multiple switches at once, or sequence switches.

There were also likely design decisions. If you look at the IMSAI 8080 front panel, the on-off and momentary contact switches are similarly designed.

Keep in mind that fingers aren't all the same length, and trying to press multiple buttons is really only possible of the buttons are on a flat surface, like a modern keyboard. With with toggle / paddle switches, you can (and I learned to quite well) change multiple switches with much less effort.

So ... while not one of the people who actually made that design decision, that was my experience.

  • For a question like this, personal experience is certainly good information to have in answers, so thanks for this! In my question, though, I deliberately separate "state" switches such as the address and data inputs, where it would certainly be convenient to change them in sets, and "command" switches such as "examine," "deposit next" and "run," where it seems to me you would never be actuating them at the same time. Did your experience lead you to believe that momentary toggles rather than buttons would work better for the latter? If so, why? – Curt J. Sampson Sep 8 at 1:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.