Do mechanical switches last 100 years?

[wellington1869]

Whether you're talking about cherries, alps, or BS, etc, most mechanical switches tend to be rated between 10 and 30 million keystrokes.

How much is that in years, exactly?  Lets assume a writer who writes 2000 words a day (which is about what a working journalist might produce, for instance, cough, xsphat, cough).   Lets also assume a switch rated at 20 million strokes.

Average word length = 5 (plus 1 for spacebar) = 6.

2000 words per day means 12000 characters a day. (2000*6)

Given 27 keys (26 plus spacebar), that means on average each key is pressed 444 times each day. (12000/27).

That means 1 million key presses per key in 2,252 days. (1000000/444)

That means 20 million key presses per key in 45045 days. (2252*20)

Which is 123 years, lol (45045/365) before the average switch on that board wears out.

Comments on my guestimation/math?  And if the above is a good enough ballpark, should we never again worry about mechanical switches "wearing out"?

[xsphat]

I'm not a working journalist, anymore, I write humor columns on the web, a blog and fiction.

And I like to write 2000 words a day even though I throw out most of it on some days.

[wellington1869]

I'm not a working journalist, anymore, I write humor columns on the web, a blog and fiction.

And I like to write 2000 words a day even though I throw out most of it on some days.

lol. Well I think 2k/day is a good ballpark for dedicated writers though. As a grad student I'd give my left arm to be able to produce that much each day consistently

[Chloe]

I wrote a script to count key presses.  I worked out I would get two to three years out of the Cherry MLs which are rated at 20 million, although some keys are used more than others.

[xsphat]

lol. Well I think 2k/day is a good ballpark for dedicated writers though. As a grad student I'd give my left arm to be able to produce that much each day consistently

Just takes dedication. All you have to do is start. I write that much just to keep in practice. That way, when an idea comes to me I can just let it pop without any warm up period. But I know the reality of it — my classmates hate me for it.

[xsphat]

I wrote a script to count key presses.  I worked out I would get two to three years out of the Cherry MLs which are rated at 20 million, although some keys are used more than others.

Damn that seems low. Think of how long I've had my Topre 86, and in that time I have written at least a couple hundred thousand words on it.

[Chloe]

I think using a key counter is misleading because it counts total presses (all keys) and not total presses per key.

[wellington1869]

I think using a key counter is misleading because it counts total presses (all keys) and not total presses per key.

chloe, what numbers are you seeing on your key counter that made you estimate a 3 year lifespan?! Thats seems way too low for a mechanical board, no?

[wellington1869]

I think using a key counter is misleading because it counts total presses (all keys) and not total presses per key.

True, tho for ballparking purposes I wonder how much of a difference it would really make (I mean, if the estimate is reduced even as drastically as 100 years to 20 years, I'd argue that still means we should never really worry about "wearing out" our boards (ie, it would be far more likely that we'd get bored with it and buy another board just for fun in that timeframe)).

But out of curiosity, I wonder what the character distribution is in the English langauge. I know the "E" is the most common letter in English, but I'd love to see a distribution chart. Even if we calculate "how soon before the switch under the E key wears out", I still think its going to be measured in decades for "the average writer".

[wellington1869]

I write that much just to keep in practice. That way, when an idea comes to me I can just let it pop without any warm up period.

Thats a great idea.  All writers should do this. The discipline (and habit) is an important part of being a professional at the craft rather than just a noodler I think.

[Chloe]

chloe, what numbers are you seeing on your key counter that made you estimate a 3 year lifespan?! Thats seems way too low for a mechanical board, no?

It does seem low, considering I got four years out of my rubber dome Trust keyboard. If my average is around 18000 key presses a day, 20 million/(18000*365) = 3.04 years. A better way would be to count individual presses per key.

[Chloe]

chloe, what numbers are you seeing on your key counter that made you estimate a 3 year lifespan?! Thats seems way too low for a mechanical board, no?

It does seem low, considering I got four years out of my rubber dome Trust keyboard. If my average is around 18000 key presses a day, 20 million/(18000*365) = 3.04 years. A better way would be to count presses per key.

[xsphat]

Thats a great idea.  All writers should do this. The discipline (and habit) is an important part of being a professional at the craft rather than just a noodler I think.

I've spent years studying habits of the greats, and this one is used by most of the more prolific ones. But in the end, it all boils down this — you have to love the language you write in or you're dead in the water from the start of the voyage. I love American English, as do most of my peers, and I love telling stories. That's why I do what I do.

[wellington1869]

It does seem low, considering I got four years out of my rubber dome Trust keyboard. If my average is around 18000 key presses a day, 20 million/(18000*365) = 3.04 years. A better way would be to count presses per key.

Definitely it would have to be presses per key; thats sort of what I tried to ballpark when I divided key presses by 27. If you do the same, you'd get:

18000 presses per day/27 = 666 presses per key (average)

20000000 presses per key/666 presses per key per day = 30,030 days, which = 82 years.

This ballpark could be made just a little more accurate if we know the average number of occurences of "E"'s in the average 2000 daily vocabulary words (I'm certain somewhere that information exists!). If we only count the E key (most common alphabet) that would give us the worst case scenario for the board.

I still think that number would still be in the decades.

[andb]

But out of curiosity, I wonder what the character distribution is in the English langauge. I know the "E" is the most common letter in English, but I'd love to see a distribution chart.

http://en.wikipedia.org/wiki/Frequency_analysis  <- Theres your chart. I liked cryptoanalysis as a kid, and the only stuff I was able to handle was of course based on Frequency analysis.

I think Chloe might have confused switch fail and keyboard fail when making her analysis. Each key has an estimates 20 million stroke life. And "Failure" in the terminology of the manufacturer doesn't mean catastrophic inability to ever work again, it means it isnt working the way it should and likely a geek could find a way to repair it.

So, given that the E is used  about 14% of the time,  based on Chloe's rate, it would still take over 20 years for that single key to fail, if we base it purely on rated life.

In a practical application, if you are willing to work on a board and sacrifice Printscreen, Scroll lock and Pause, maybe right ctrl and alt, then resolder these 5 switches, I think you could effectively triple the life of the board if we are talking about actual failure rates as opposed to estimated lifespans. Math is just off the top of my head, of course.

Chloe, if you want to really script this, record each keystroke then call a script at night to tally how many of each you really pressed. Otherwise, the simple way is to take your total count, assume that 10% are non letters, so take 14% of the remaining 90% and you have the number of 'e's and can calculate your failure date.

[EDIT]Hmm, just realized, the first key to go will be space. DUH. Then for me, Backspace , maybe also enter, and maybe shift[/EDIT]

[Chloe]

Switch failure to me is the point at which it ceases to operate as designed. In the case of the MEI switches in my Datalux, some keys became very sensitive, making the keyboard unusable.

[lal]

This ballpark could be made just a little more accurate if we know the average number of occurences of "E"'s in the average 2000 daily vocabulary words (I'm certain somewhere that information exists!).

If you have access to a Unix machine (e.g. Linux or Mac OS X), create a text file named 'countletters' and copy the following into the file:

Code: [Select]


#!/usr/bin/awk -f

{ for (i = 1; i <= length($0); i++) letters[substr($0, i, 1)]++ }

END { for (l in letters) print l &quot;:&quot;, letters[l] }


Then open a terminal, change to the the directory where countletters is stored and make it executable with "chmod a+x countletters". If you have one or more text files to study, process it/them with
"cat file1.txt file2.txt file3.txt | ./countletters | sort -k 2 -n". Or you can count the letters in some text currently saved in the clipboard buffer by just executing "./countletters | sort -k 2 -n", pasting the clipboard contents to the terminal window, press Enter and Ctrl-D.

Yes, I'm bored

[Chloe]

Chloe, if you want to really script this, record each keystroke then call a script at night to tally how many of each you really pressed.

You could just count each key with separate variables and write this to a file on shutdown. I'd rather not use a keylogger even if I wrote it myself.

[wellington1869]

Ok, so if spacebar is used 15% of the time (slightly greater than E which is about 14% of the time), then in Chloe's scenario, for the single switch under the most used key (spacebar):

18000 presses per day, of which 1200 are the space bar. (15% * 18000)
20000000 key presses/1200 per day per key = 16,666 days before failure, = 45 years.

So it appears even in this scenario, we shouldnt really worry about mechanical switches "wearing out"!

[incidentally chloe, 18000 characters/day = about 3000 words per day = about 3.5 pages per day.  Working on a novel are you? ]

[wellington1869]

Does anyone know what metric switchmakers use?
One interesting point might be that I vaguely remember reading about application limits, is that given numbers are valid for an ideal laboratory environment. For example, for switches, it won't take into account all the corroding crap that has been accumulating in your keyboard. Also, about a decade ago they finally got the machines to mechanically test something for, say, 100 million times in a reasonable amount of time, and it turned out even in laboratory conditions theoretic lifetime is hardly ever reached.

webwit, I like your new avatar!

[bigpook]

Doesn't look like the anti-anxiety pills are working so well : (

[wellington1869]

Ok, so if spacebar is used 15% of the time (slightly greater than E which is about 14% of the time), then in Chloe's scenario, for the single switch under the most used key (spacebar):

18000 presses per day, of which 1200 are the space bar. (15% * 18000)
20000000 key presses/1200 per day per key = 16,666 days before failure, = 45 years.

So it appears even in this scenario, we shouldnt really worry about mechanical switches "wearing out"!

[incidentally chloe, 18000 characters/day = about 3000 words per day = about 3.5 pages per day.  Working on a novel are you? ]

45 years of novel writing... thats a whole career on a single keyboard

[itlnstln]

Was it picture of the actual Pac-Man?
 
Or was it a picture of Crack?

[itlnstln]

I can't see the picture, but that is hilarious.  Is that a metaposter?

[slowe]

Ok, so if spacebar is used 15% of the time (slightly greater than E which is about 14% of the time), then in Chloe's scenario, for the single switch under the most used key (spacebar):

18000 presses per day, of which 1200 are the space bar. (15% * 18000)
20000000 key presses/1200 per day per key = 16,666 days before failure, = 45 years.

Just read this thread after it got bumped back to the top...

Actually, 15% of 18,000 is 2700
So if the lifespan of the space bar is 20,000,000 key strokes, that makes it 20,000,000/2700=7400 days or 20.3 years which seems more realistic...

[itlnstln]

Firewalled?

My company uses Websense, so a lot of the pics get blocked here.
 

PS: Ok I admit I had a bigger WTF recently, but still...

Whiskey.  Tango.  Foxtrot.

[FKSSR]

I don't understand the thing about the DeviantArt picture at all.  Can someone enlighten me?  I feel like I'm missing some reference or something.

[o2dazone]

Well, what do you catch? Do you see the Pac Man reference? What about the pills he's munching on?

[FKSSR]

Ah, nope.  I didn't get the Pac Man reference at all.  Now I see it, of course.

Does the title "The Madness of Mission 6" mean anything?  I have to admit that I never played Pac Man that much.

[lowpoly]

The yellow guy grabs the orange guy's crotch but the orange doesn't notice. :smile:

I didn't get the PacMan reference, that's brilliant.

[o2dazone]

:tongue:

hgahahahahahhahahahahahaha
**** I hate deviantart...

[lowpoly]

Frame it like it is!

[DesktopJinx]

Interesting discussion. Looking at and tweaking the base assumptions leads to quite different results -- the realization that the spacebar is probably the limiting factor, for instance. Of course, even if you fixed the spacebar, the vowels are probably halfway to failure...

I had a conversation with a longtime coworker the other day (just hours before I found this thread, actually) musing about a related estimation... how much typing will you do in your lifetime? He immediately said "billions of keystrokes" and I balked, but upon further reflection, it's not so absurd. An information worker who puts in long hours, who uses email for a lot of personal as well as professional communication, and who socializes online in boards or forums or blogs or whathaveyou... in fifty years, that's a lot of typing. Probably a whole lot more than 2,000 words a day. If you did 30,000 keystrokes a day, you're over a half-billion keystrokes in fifty years, and your spacebar is approaching a hundred million.

Now how many keyboards do you need?

[lowpoly]

whatpulse.org keyboard/mouse statistics

[ozar]

When switch manufacturers quote the life of a switch in keystrokes, I wonder if they take into consideration the typing style of different typists?

I really have no idea what kinds of testing they might do with a switch before they come up with a number, but I'm guessing that they have a machine that simply presses a switch over and over until failure occurs.

[keyb_gr]

I really have no idea what kinds of testing they might do with a switch before they come up with a number, but I'm guessing that they have a machine that simply presses a switch over and over until failure occurs.

Actually the Cherry guys do it just like that, at least as far as complete keyboards are concerned.

[IBI]

The 50 million operations life is for the standard keys, the larger keys may have a shorter life.

And 50 million isn't reserved for mechanical keyboards alone, Cherry's FTSC plunger + rubber dome keys also acheive 50 million operations, even their cheap £15 cymotion expert acheives 20 million, which is equal to the figure for the non-linear MXs.

[keyb_gr]

And 50 million isn't reserved for mechanical keyboards alone, Cherry's FTSC plunger + rubber dome keys also acheive 50 million operations, even their cheap £15 cymotion expert acheives 20 million, which is equal to the figure for the non-linear MXs.

The rubber dome Cherrys seem to be about as tough as rubber dome 'boards get. But still that's just the number of actuations until failure, it doesn't say anything about gradual degradation like worsened key feel and such. In that regard I'd have more faith in mechanical or mechanics-over-membrane switches. (MY/FTSC belongs to the latter, no rubber dome involved.)

[wellington1869]

Actually, 15% of 18,000 is 2700
So if the lifespan of the space bar is 20,000,000 key strokes, that makes it 20,000,000/2700=7400 days or 20.3 years which seems more realistic...

Good point, thanks for the correction.  
That would be 20 years for someone who actually writes that much (2k words a day) which is a LOT. I think for the average person the average is  going to be much less (after all there are many days when all i'm doing on my 'puter is checking email and web surfing). So I think you have to figure a great many people are going to average, say, 1k words per day or less, and are going to see 40 years or more with their boards.  And thats easily a 'lifetime' (or a 'career') with one board, if you wanted.

My own 1391401 is 28 years old, going strong...

[keyb_gr]

My own 1391401 is 28 years old, going strong...

I didn't know that there were time machine gateways on the interweb... 18 years would seem more like it for a '91 sample.

[IBI]

In that regard I'd have more faith in mechanical or mechanics-over-membrane switches. (MY/FTSC belongs to the latter, no rubber dome involved.)

I thought FTSC was plunger over rubber dome? If that's not the case then what gives the feedback for FTSC?

[keyb_gr]

If that's not the case then what gives the feedback for FTSC?

The corresponding patent (+ a link to images of the mechanics disassembled) is found in Chloe's Cherry switch wiki. What do you mean by "what gives the feedback" - as shown in the MY force displacement graph (no idea whether we have one here), these are stiff linear with a small nonlinear increase in force as you bottom out, so neither tactile nor auditory feedback up to this point.

EDIT: I found a 2004 Cherry catalog that wrongly states "Elastomer keys with membrane contact switch", for 2006 this was corrected to "Mechanical individual keys with membrane contact switches", as it should be.

It's not a very exciting mechanism with merely springs bending/compressing, but just this also makes for good reliability.

Maybe you were thinking of the Fujitsu Peerless switches, which are clicky and use a rubber dome (and apparently don't hold up that well).

[wellington1869]

I didn't know that there were time machine gateways on the interweb... 18 years would seem more like it for a '91 sample.

lol, clearly my math skills take a wild dip in mid afternoon.

Thats good news, so I'm looking forward to another 22 years with it. (Wait, did I subtract that correctly?)

[qso]

I think it depends on what kind of typist you are hehe, there are some people like to hit enter key hardly or other keys. And the failure tolerance you have, currently I'm still using my logitech NT100(as u all know it's a cheap keyboard). For me usually I will change this keyboard every 6-8  months not because it's damaged or whatsoever(been using this type of keyboard for 3 years lol), it's just that I'm burdened with the keys(hard to press,sometimes stuck,or just feel weird). So maybe the case will be the same for mechanical keys, although they said it has 20 year life span, the time needed to change it would be cut almost by half or more(sellers love it, no hassle on warranty).  According to typing master satellite my average words are 5460 per day, the highest is 8492 words, and I'm doing 50k-60k strokes per day.

I guess any keyboard would be happy under female's fingers.

[keyb_gr]

According to typing master satellite my average words are 5460 per day, the highest is 8492 words, and I'm doing 50k-60k strokes per day.

So with 15% of that being space, this key would be pressed about 9000 times a day. A switch specified for 20 million operation should then last you about 6 years, one for 50 million even 15 years. (If you change your current boards every 6 to 8 months, they seem to be lasting about 1 or 2 million actuations, which seems reasonable for a cheap rubber dome.) Chances are that even good keycaps (PBT or POM with durable lettering) would show significant wear then.

I think you could pick up a G80-3000 with blacks and be done with it for the next few years. My LPMDE 1€ special was 7 years old, showing significant wear on some keys and had reduced friction in many of the frequently-used switches, not to mention case yellowing, but it was and still is working like a charm. I gave it away to someone with light to moderate keyboard usage but sort of a heavy touch (used to G81) recently, where I'm sure it'll hold up well for years to come.

[qso]

So with 15% of that being space, this key would be pressed about 9000 times a day. A switch specified for 20 million operation should then last you about 6 years, one for 50 million even 15 years. (If you change your current boards every 6 to 8 months, they seem to be lasting about 1 or 2 million actuations, which seems reasonable for a cheap rubber dome.) Chances are that even good keycaps (PBT or POM with durable lettering) would show significant wear then.

I think you could pick up a G80-3000 with blacks and be done with it for the next few years. My LPMDE 1€ special was 7 years old, showing significant wear on some keys and had reduced friction in many of the frequently-used switches, not to mention case yellowing, but it was and still is working like a charm. I gave it away to someone with light to moderate keyboard usage but sort of a heavy touch (used to G81) recently, where I'm sure it'll hold up well for years to come.

I'm waiting for my Filco brown switch... ordered 2 days ago. For the price I'm expecting the Filco will last for 10 years lol or the least 5 years since It's made in Taiwan.

[sixty]

Anyone remembers John Titor? Missed our chance there to ask him how cherry switch boards behave in 2038. Oh well!

[keyb_gr]

One good way of getting trouble with keyswitches must be thorough cleaning.

At least that is what I found out today. I had cleaned up my work G80 with clears before lunch, removing the remains of a historic coke spill or similar from case and (that may have been the mistake) board. At around 6 pm I suddenly noticed my Caps Lock was stuck (in the non-depressed position, fortunately).

Crap. I guess some dirt or cleaning cloth lint got into the switch (tactile ones are more likely to be bothered, remember?) and blocked it. Fortunately we've got some skilled techs around there. I hope it'll be done with unsoldering and a good bath in H2O-dest or alcohol.

Now I do not feel like cleaning up the board I got today.

[keyb_gr]

OK, switch is working halfway well again now. It'll stick when pressed from the right so I may open it up at some point.

Maybe it was greased or something from the factory and that stuff is gone now.

EDIT: Tried soaking with some water and dishwashing liquid and blowing out with compressed air, now it seems to be better.

[keyb_gr]

Last try was with plain drinking water. Caps Lock is still sticking a bit, and Tab, Shift and Ctrl are not quite optimum (Ctrl is squeaking a bit), but at least one can work like that.

Overall I'd say: Keep liquids far away from Cherry switches. Spills should be cleaned up very carefully to avoid contaminating them.