29

In the top answer to Why were animated screensavers used instead of a black screen on CRT monitors? the main reason was said to be that it avoids users accidentally turning off the machine, thinking it is turned off and trying to turn it on.

Well, that's indeed not a good thing to happen, especially back when it posed a real risk to hard drives. But... who had the idea of using power buttons that allow this issue to happen in the first place? It seems utterly trivial and obvious to avoid it by using flip power switches with a built-in light, like those found in many devices (guitar amps, fridges, power strips...).
In more modern computers there are of course real reasons for having push buttons (which anyways aren't directly connected to the power), but back when switches really were just mechanical switches this doesn't apply.

So: why use push power buttons?

Clarification: I'm specifically talking about the early mechanical push switches used in the 80s, which actually did power the machine hard-off with just a simple press, not only send a signal to the OS to shut down gracefully. (I thought this was obvious from the original questions, but seemingly not.)

Peter Cordes
  • 3,207
  • 17
  • 26
leftaroundabout
  • 600
  • 6
  • 12
  • 1
    I wonder whether "real estate" was a factor. For a toggle switch, you don't want the switch to protrude from the case, as it'd be too easy to bump. So it has to be recessed, as in the PC Big Red Switch that Raffzahn's answer shows. But that means the switch and its housing now occupy several cm of depth in the case. Moreover, the recess, and therefore the housing, needs to be large enough for the operator's finger as well as the switch itself. A pushbutton switch can easily be flush with its housing, or even slightly recessed, and take up quite a bit less space. – Nate Eldredge Apr 23 '22 at 17:39
  • 18
    @BrianH: The sound of a fan is not so "obvious" in a noisy office or industrial environment. – Nate Eldredge Apr 23 '22 at 17:42
  • 1
    All of the answers so far assume IBM products (including PC). But there were other unrelated computers with push-on-push-off switches. I'd like to see an answer address the TRS-80 Color Computer, which had such a switch. – DrSheldon Apr 24 '22 at 03:48
  • 1
    With the current title, my first impression on seeing the title in HNQ was that it was going to be about soft power switches (ATX style). Perhaps since current machines have a momentary soft power switch at the front, and a rocker / flip hard power switch built-in to the PSU if at all. The bold clarification is still necessary, but perhaps we can clarify the title without making it excessively long, perhaps using the word "hard"? – Peter Cordes Apr 24 '22 at 04:33
  • Maybe "When/why did early computers have a push-button hard power switch instead of flip or rocker to mechanically cut main power"? More technical phrasing so I'm not making this edit myself. "Push-button" is apparently the technical term for a non-momentary switch that works like a ball-point pen where you can click it in / out, and talking about "cut main power" reinforces the point that it's not a soft-off. But your phrasing reinforces the actual problem it creates of off being the same action. – Peter Cordes Apr 24 '22 at 04:35
  • Perhaps just "push-button power switch" instead of "power button" would be clearer. "Button" may have some implication of momentary and/or software-based switching, especially for modern readers used to such buttons instead of hard switches. – Peter Cordes Apr 24 '22 at 04:37
  • I missed the point in history where computers had a pushbutton whose action was a hard power cut (or restore). Unless I miss my guess, those push buttons generated a low voltage signal. This would still have to go through a relay to cut power. – Walter Mitty Apr 25 '22 at 02:20
  • Your fridge has a power switch? – Jacob Krall Apr 25 '22 at 02:45
  • @JacobKrall just checked... it actually has a thermostat wheel with a 0-position that turns the motor permanently off. But I'm pretty sure my old fridge has a separate flip switch, and it turned off both the motor and the light, so very clear what was on and what was off. – leftaroundabout Apr 25 '22 at 08:54
  • A better related question would be "Why did so many manufacturers of tower PCs put the fricking power switch protruding at exactly knee height with no cover? The answer seemed to be that no one thought that solving that problem would ever have any effect on their sales. Sadly, they were probably right. – RBarryYoung Apr 25 '22 at 15:53
  • @BrianH: also, the LEDs are a technical indicator (i.e. possibly non-meaningful to many users), small, and can be obscured. No need for pompous “foolish user” comments. – Reid Apr 25 '22 at 16:05
  • The push button switches I remember tended to feel different when you were pushing to turn on vs pushing to turn off, so, by feel, you could easily tell whether you were turning the device on or not. When the device was on, the button had a lower spring constant due the fact that the underlying mechanical switch was pushed in, but when the device was off the mechanical switch was pushed out. – Kyle Miller Apr 25 '22 at 23:40

5 Answers5

44

AT power supplies often did have a line-voltage flip switch. Indeed many later machines also have such a switch around the back to totally switch off the system. However ATX and later systems use a "soft switch" which yields a number of capabilities.

First, and most obviously, it enables the computer to switch itself off. Once the software shutdown has completed the "power on" line is allowed to float up to +5V and the PSU drops the power good signal and shortly thereafter all power lines except +5V standby. This could not easily and cheaply be done with a flip switch.

Next, it allows the system to start itself up. For instance a wake-on-lan interface allows a remote "wake up" signal to be sent. The +5Vsb allows a minimal set of subsystems to be active such as PXE. On receipt of the message PXE grounds the power on line and the PSU powers up the main supplies, then asserts power good once things are stable.

It's helpful to regard the front panel switch not as THE power switch, but as A switch amongst a few possible switches. Indeed, it's not even a switch, it is merely a request to the PSU.

  • 10
    Ok. Yes, the automatic capabilities is where push switches make sense (as I already remarked in the question). However, push switches were standard before these features were added, weren't they? The oldest computer I ever used had a push switch on the front, but purely mechanical action (and when pushing, the travel was slightly longer when switching from off to on than the other way around – but you only really felt it when the button was already pushed in and it was too late to reconsider). What was the history behind all that? – leftaroundabout Apr 22 '22 at 21:59
  • This is the correct answer. The power switch has to allow for multiple ways to turn on/off a machine. – NuclearFission Apr 23 '22 at 12:44
  • 1
    One important aspect of the ATX style soft power switch is that it allows graceful shutdown. This started to became relevant around the time of the Windows 95 that had significant amount of background processing and also implemented write caching in certain cases. – vhu Apr 24 '22 at 08:05
  • Exactly, hence "Once the software shutdown has completed ..." –  Apr 24 '22 at 09:27
  • 1
    Soft power-down doesn't prevent use of a flip switch as before long the state will match that of the switch. We're used to this, for example, with motors.When you turn off many appliances it takes them a while to "stop". Soft power on would be a good reason, but how common was it on early PCs? – Dannie Apr 24 '22 at 11:20
  • 1
    You're assuming that shutdown could be completed before the capacitors discharged. –  Apr 24 '22 at 21:45
39

TL;DR:

So: why use push power buttons?

The same reason that it is used in other appliances: it's convenient.

The Works:

Just think, early (IBMish) PC had their power switches at the right side all the way at the back (*2). One had to to feel one's way all around to flip it - of course with no chance to look at its state. The switch was part of the power supply. This made sure that all high voltage switching only happens within the PSU.

enter image description here

_(Picture taken from Wikipedia)

Putting it up front wasn't really a solution as this would break the encapsulation of high voltage within the PSU. Keep in mind, the PC was a device intended to be opened by any average user, thus exposure to high voltage was a complete no - usually prohibited by most electric codes worldwide. The same reason goes for why it had to switch mains, as only that would guarantee that the PC is completely powered off if switched off but not unplugged - which, depending on workspace setup may not always be possible.

Moving the PSU upfront was not really a great idea, as that would mean having the power cord coming out on the side. Only a PSU reaching all across the case would avoid that.

Using a push switch is about the only (*1) solution, implemented by extending the visible push button by a long 'stick', going all the way back to the PSU, pushing the 'real' toggling power switch. Now the user could operate the power switch from the front of the computer while still having a code-legal encapsulated PSU at the back.

The original AT-class (286) HP Vectra of 1985 was one of the first IBM(ish) PCs to offer this.

HP Vectra

(Picture taken from Wikipedia; marking mine)

Long story short: It's all about convenience.

Equally important, adding a certain type of switch is a design decision not necessary based on technical issues, but what look/feel/operation habit a designer wanted to achieve.

Besides, in real life, no user checks the state of a switch before flipping it, when he already assumes the computer being off - after all, he already ignores other easy to check signs like lit LED on keyboard, case or monitor.

*1 - which BTW was the same with the Apple II/III series, undoubtedly the major prototype for the IBM PC, with their enclosed PSU and a switch on the back.

*2 - Of course, like almost always there are other possible solutions. For example using some secondary low-voltage circuit, much like for the later power button, but operated by a switch. Just, this would complicate the PSU considerably, and thus increase the price - not anything compatible manufacturers, usually fighting on cost base, would want.

Equally, if not more important:

It would be a technical divergence from the example set by IBM. IBM had a fully concealed PSU. Having a different design, no matter whether better or worse compared to IBM, has always been used as a point by IBM sales force - and even more in the mind of decision making on buyer side. So it was extremely important for makers of compatibles to withstand that. A compatible had to be the same, or better, without changing anything. It took many years until the market was levered in a way to incorporate new ways.

Toby Speight
  • 1,611
  • 14
  • 31
Raffzahn
  • 222,541
  • 22
  • 631
  • 918
  • 12
    But a) it's not more convenient: a flip switch is exactly as easy to use b) most appliances don't use push-push switches, but instead something with distinct on- and off-directions, even though accidental switch-off would typically have less problematic consequences than it does in computers. – leftaroundabout Apr 22 '22 at 21:22
  • 1
    You asked, I answered. If you want to debate this, a Q&A site might be the wrong platform. Beside, do you really think people look at the state of a witch before flipping when they already assume it's off (due the dark screen)? – Raffzahn Apr 22 '22 at 21:24
  • "even though accidental switch-off would typically have less problematic consequences" - but accidental switch ON is usually worse: accidental blender, accidental power tools, accidental kettle, and accidental oven are all pretty bad. – 9072997 Apr 22 '22 at 21:26
  • 10
    @Raffzahn I don't want to debate this, I want to hear the real reason for this design decision, which “it's convenient” clearly doesn't provide. And yes, people absolutely do check the state of a toggle switch. They certainly look at it (because the eye guides the finger much quicker to the switch than blind touch-searching), and when the switch is glowing that's a very obvious warning not to press it for turning the computer on. Even without a light, one would quickly get used to which orientation of a daily used switch is on and which is off, and pressing the other side has no effect. – leftaroundabout Apr 22 '22 at 21:32
  • @leftaroundabout Interesting how still agrgue after stating not to argue, isn't it? You may need to accept that convenient - and even more design decisions - is an individual value. You may personally check visually first (or believe so), but that won't tell anything about other people or what a common behaviour is or not. Last but not least, remember that screen savers were introduced at a time when the default power switch was a huge lever, located t the side and to the back of a PC. Not exactly a location to look at. – Raffzahn Apr 22 '22 at 21:45
  • “when the default power switch was a huge lever” – was it? Interesting. Well, maybe that's part of the real answer, a design change from back- to front-panel that for some reason also caused a change from flip to push design? – I don't know, thus the question. – leftaroundabout Apr 22 '22 at 22:01
  • @leftaroundabout But that's exactly the point - it's about convenient operation. – Raffzahn Apr 22 '22 at 22:08
  • 9
    Now the answer makes complete sense, with all the context – it is in fact an interesting story. I still disagree with the tl;dr though, because the gist isn't at all that a push switch is per se more convenient than a flip/lever – but instead that this was the only feasible way of moving the switch to the front, and the advantage of having it on the front outweighed the inherent disadvantage of the push design. – leftaroundabout Apr 22 '22 at 22:24
  • @leftaroundabout: From a user perspective, a toggle on the front panel may have been preferable to a push button, but reliably mechanically conveying the state of a toggle switch through a long piece of plastic is harder than mechanically conveying the state of a momentary push switch. If one is using a push switch and the attached rod shrinks by 3mm with repeated use and age, a user will need to push the switch an extra 3mm in order to power the machine on or off, but that would generally not be a problem. If one were using a toggle, however, ensuring reliable operation would be harder. – supercat Apr 22 '22 at 22:35
  • 1
    It could also be argued that convenience was the reason for placing some USB ports on the front of computer cases instead of all at the rear. – Fred Apr 22 '22 at 22:45
  • 4
    Many of IBM's early PS/2 machines had power supplies that reached all the way to the front, with a rocker switch for the power supply. (Examples I've personally encountered include the Model 50, 60, and 80) – Kaz Apr 23 '22 at 08:16
  • 2
    Re, "early (IBMish) PC...power switch...no chance to look at its state." Once you got your fingers on that monster, you did not need to see it to know its state. – Solomon Slow Apr 23 '22 at 13:24
  • 1
    "It's all about convenience" ... until you had a PC with the power switch level with the keyboard, such that accidentally pushing the keyboard against the case would turn the power off. My first Linux box had that feature, and several times I had to hold the power button carefully with one hand while typing shutdown magic with the other – scruss Apr 23 '22 at 18:19
  • 8
    Was encapsulation such a strict requirement in the 1980s? The 1990s AT-type power supplies I've seen tended to have the pushbutton power switch on a length of mains-voltage flex running back to the PSU, rather than a mechanical linkage to a button on the PSU itself. – john_e Apr 23 '22 at 19:04
  • 2
    @leftaroundabout You seem to want some definitive canonical reason why some computers used button-switches and some used toggle-switches, but you're not likely to find one. It's probably just that one type was cheaper, or that's what they had available, or just because the designer felt like it. They're just two styles of switches, without one being obviously better than the other. Button-switches are older than computers, so the answer to your main question is "they always have". – BlueRaja - Danny Pflughoeft Apr 23 '22 at 20:58
  • 1
    @john_e: Certain industries and fields require the use of equipment which is approved by various ratings agencies, and some stores will only sell such equipment, but if John Q. Hobbyist wants to build his own computer such requirements wouldn't apply. If a company that wants official safety approvals for a number of computer designs that share a power supply, using an encapsulated supply and not having any mains voltage outside it would avoid the need to get each computer design individually tested for such approval. – supercat Apr 23 '22 at 22:26
  • 1
    I had one of the clone PCs with push button at the end of mains voltage cable that went through the case. I wouldn't be surprised if the reason would be just that since IBM uses push button (for a technical reason), serious clones must have one too (to look like IBM). The nice part was that the button had different feel when switching on and off, so you could actually notice when you powered off accidentally and hold the button down. Or even press it quickly again and hope that PSU capacitors kept the machine running. – ojs Apr 24 '22 at 20:23
  • I think the tl;dr is more about how "front switch" is more convenient than "back switch" or "side switch", @leftaroundabout, rather than the switch types themselves. It's just that the switch type was usually determined by the position. – Justin Time - Reinstate Monica Apr 25 '22 at 04:04
  • 1
    @supercat FWIW, my early-90's Compaq DeskPro 486 has a front panel flip switch. Can't find a photo, but it's quite similar to this 386 version. So it's not just hobbyists hacking something together, but business PCs sold by major brands (in my case in Germany, which isn't particularly known for lax safety codes). – TooTea Apr 25 '22 at 11:06
  • @TooTea: I've seen designs which use a mains-connected switch on the front panel. They require extra safety approvals, but some companies decided the benefits were worth the cost. I've also seen designs which mechanically connected a front-panel flip switch to a supply in the back, but that requires a higher degree of mechanical precision than a push-rod-connected button. Again, sometimes the extra cost is worth it. – supercat Apr 25 '22 at 13:09
  • 1
    Designs with front-panel switches actually switching full mains power were very common (usinfg big bulky wires), to the point of probably being in the majority. Indeed, one of the main motivations for changing the way newer power supplies switch on was to not have to route mains through the computer case. – Remember Monica Jul 20 '22 at 22:16
5

It's traditional :-)

Think back to when computers were large, lived in air conditioned rooms, and had a few dozen switches on various cabinets. Push buttons were common for "toggle" functions: online/offline, load/unload, power on/off.

Square buttons with a light behind them were common. Check out this paper tape reader, and the other box behind it.

enter image description here

dave
  • 35,301
  • 3
  • 80
  • 160
  • Mainframe CPUs and peripherals had a real power switch, fully cutting off mains power, somewhere at the PSU, while (push) button switches at the front only served as secondary for convenience. – Raffzahn Apr 22 '22 at 23:29
  • 5
    Of course; my point, somewhat tongue-in-cheek, was that computers and push buttons have an association going way back. – dave Apr 23 '22 at 01:31
  • Ok, this is a possible alternative explanation to Raffzahn's. But can you tell which was more important? Any references? – leftaroundabout Apr 23 '22 at 20:43
  • 3
    I don't think this is 'really' an explanation; about the only thing it's good for is to suggest that maybe there was no expectation that people would press buttons without knowing what pressing the button would do. I think @Raffzahn's tale sounds plausible. – dave Apr 23 '22 at 22:05
4

To answer the 'When' part of the question: At least as early as 1982, when the Sanyo MBC-550 was released with a push button power switch on the front. The MBC picks up its design cues from contemporary video recorders (compare the Sanyo VTC-5000) on which the power switch was also a front-mounted push button.

Internally, the power supply of the MBC occupies the right-hand edge of the case from front to back, so there is no need for a mains-voltage wire between the power supply and a separate switch.

john_e
  • 7,263
  • 20
  • 44
3

I know nothing about what was going on in the product planning here. But I do note one thing about the machines I bought that had a power switch. These machines had the power switch directly connected to the power supply. If you hit the switch, you cut the power. Makes sense. Except that it doesn't make sense. If you cut the power without an orderly shutdown of the operating system, you leave the disk in an internally inconsistent state. That means that, the next time you start up the operating system, you have to run chkdsk to clean up the data state of the disk before you can safely run appps. The first computer I bought with a pushbutton also had the following feature. The pushbutton didn't cut power. Instead, it generated an event for windows to field. Windows shut itself down taking several seconds to do so, or maybe even more. Once windows was shut down, windows signalled the power supply to cut the power.
All of a sudden the rule that only dummies hit the off switch was reversed. The only people who didn't hit the power button were old timers who remembered that cutting power would cause chkdsk to run at startup.

Walter Mitty
  • 6,128
  • 17
  • 36
  • 5
    Getting power supplies to talk to the BIOS and getting the OS to listen to the BIOS took Intel and Microsoft pushing a standard (APM) on to PC makers. I think the bait was that you had to support APM to get to use the "designed for Windows 95" logo. – dave Apr 23 '22 at 12:10
  • 1
    As was already discussed both here and in the screensaver thread many of the early push-switch computers did not have any safeguards against turning off, so this answer is rather missing the point. And, even given that we have a safeguard, I'd argue that a flip switch would be preferrable: switch it “on”, and it always means on. Switch it “off”, and it means the OS tries to shut down in any way possible, and then stay off. – leftaroundabout Apr 23 '22 at 20:33
  • 2
    @leftaroundabout While I agree this style of button predated this behavior, I disagree that a flip switch would be more useful for a soft shutdown. The way these buttons normally work nowadays is you tap the button to initiate a soft shutdown, and hold it to force a hard one. You can't make that distinction with a two-way switch. – Kef Schecter Apr 24 '22 at 05:26
  • 1
    @KefSchecter true. (Though again – the long-press-kill is itself a bit of a hack to get a third feature out of the single button, and it's not particularly intuitive. No doubt a similar trick could have also been done with a soft flip-switch, e.g. that flipping on and off again rapidly three times in a row would cause hard shutdown.) – leftaroundabout Apr 24 '22 at 08:51
  • It's possible that the pushbutton predates the function being to alert the OS. But they are certainly related. The early off swithes were simply in line with the power cord. The pushbutton always generated a signal, I think. Where that signal goes is another story. – Walter Mitty Apr 25 '22 at 02:00