50,000,000 Actuations. Really?

[abdulmuhsee]

So out of curiosity, I did the math.  Being very generous with usage here, if I were to press a key 1,000 times a day, it would take over 100 years for that key to wear out @ 50,000,000 actuations.

Now, part of my brain is saying that's BS, since I've heard of many a person that has a key or two that's started to act up, or perhaps the story about the Topre boards in use by banks which have worn-out spacebars after a year or two.  There's no way they hit 30,000,000 in that period of time.

With this forum, you'd think that keyboards only lasted a month or two with all the new ones coming in, but how is the agreed-upon number of 50,000,000 accurate?

[JPG]

It's probably true for a non-defective switch. Still, they last long, whatever that number is.

[daerid]

I can blow through a thousand keypresses in an hour... although probably not on the same key. But it's probably upwards of 500 or so per key.

Which still.. 200 years. Hmmm... that is weird

[chill1217]

a lot of specifications like these are estimates.  i think to rate hard drives they put them in different temperatures and conditions to simulate harsh conditions so that they wear down faster.  i think the number of estimated actuations is similar.  maybe they push a switch down with really excessive force = 1000 normal button presses, or doing it after a coat of acid to degrade it faster etc. etc. 

you have to take the specification with a grain of salt

[rootwyrm]

Sigh. No.

Cherry has equipment specifically designed to test large numbers of switches in both a destructive and non-destructive fashion. The reliability number is established based on a large sample pool subjected to both direct actuation tests (button pushers) to the point of destruction and scientific examination of switch wear patterns at specific points. All this data is put into a magical number blender to give them a number which is reasonably accurate. And when they fall short, they change things and start over.

So when they say that it is good to 50 million actuations, that is not a number they just pulled out of their butt. They performed a lot of testing and engineering to get to that number and it's considered a generally accurate estimate. It is not a guarantee that a switch will last for 50M actuations - only that under normal conditions with normal use it should last that long.

Also, you fail horribly at math. Like, you need to go back to remedial arithmetic horribly. A hundred years? Not even close. Seriously - off by factor of 100 wrong.

[terran5992]

My cherry keychain is losing its clickyness after about 6 months.

[tbc]

Also, you fail horribly at math. Like, you need to go back to remedial arithmetic horribly. A hundred years? Not even close. Seriously - off by factor of 100 wrong.

i don't get it.

what's wrong with his calculation?

[abdulmuhsee]

Also, you fail horribly at math. Like, you need to go back to remedial arithmetic horribly. A hundred years? Not even close. Seriously - off by factor of 100 wrong.

1000 presses/day * 365 days/year = 365,000 presses/year.

Divide that into the estimated lifecycle of 50,000,000 presses and you're left with approximately 137 years before switch failure.

Would you please correct my mistake, rootwyrm?

[ften]

Like rootwyrm said, they would have machines to do validation on their parts.  Just like how a company that manufactures knifes would validate their blades or a company that manufacturers suspension forks would perform torture tests on their product.

[naokira]

How about the pcb how long they will live?

[tp4tissue]

Also, you fail horribly at math. Like, you need to go back to remedial arithmetic horribly. A hundred years? Not even close. Seriously - off by factor of 100 wrong.

1000 presses/day * 365 days/year = 365,000 presses/year.

Divide that into the estimated lifecycle of 50,000,000 presses and you're left with approximately 137 years before switch failure.

Would you please correct my mistake, rootwyrm?

(highpitch) Awkward///

[Photoelectric]

Also, you fail horribly at math. Like, you need to go back to remedial arithmetic horribly. A hundred years? Not even close. Seriously - off by factor of 100 wrong.

His calculation is correct if he assumes 1000 switch presses a day (which is a fairly high number for the same switch) and 50 million actuations.  He's already demonstrated his math, and if you look at it another way, 50*10E6 actuations / 10E3 actuations per day= 50*10E3 or 50 thousand days, which is approximately 136 years.
--
Doing some more fun analysis, taking an extreme case:

Say you type fast, like 100 words per minute without using punctuation.  And you type for 5 hours a day of pure typing time (=a lot), so 30,000 words per day.  Using some Wolfram Alfa data, average English word length is 5 characters long, and using Wikipedia data, the most common letter is "e" (~12.7%).  Ignoring punctuation, you type ~30,000 * 5 = 150,000 characters a day just for words, 12.7% of which are the letter "e", which translates to ~19,050 keypresses a day for the letter "e".

Now using that value versus the advertised 50 million actuations for the switch in "e", it will take (50,000,000 / 19,050) / 367 years = 7.2 years.  So if you type that much, your "e" switch will wear out in about 7 or fewer years, given other wear factors (environmental factors like temperature, humidity, friction due to accumulated dirt, contact oxidaction, solder joint cracking, etc.).

[MOZ]

Humidity and dust will ensure you never reach the numbers

[tp4tissue]

Humidity and dust will ensure you never reach the numbers

duhn duhn duhn.....

then again... maybe people are overestimating how much they actually type.

[morpheus]

50 million strokes.

[Danule]

only 1000 a day? you've never played starcraft have you?

[damorgue]

Their testing rigs can of course press at a far higher rate than 1000/day. Try 2 times per second for now, ie 172800/day. They can also test a large number of switches at the same time. If they test a hundred switches, when they have undergone 10 million actuations, they will have a data set of 100*10,000,000= 1x10^9 actuations.

If that amount of actuations yielded 20 failed switches, then you will have 20 failures in 1,000,000,000 actuations. 20 failures per billion actuations is the about the same as 1 failure per 50 million actuations.

This is one way to get such a number. They might also assume that it will be a certain type of probability distribution and extrapolate to get some less accurate results. MTBF is a tricky field. Their rated value is probably a value where a certain percentage will fail.

Things to take away from this:
-MTBF of 50 million actuations does not in any way guarantee that they will last 50 million keypresses.
-It is relatively easy to get values that high with a larger sample size
-Their testing is just an analogue of real world conditions. I would guess real world numbers are lower because of factors not percent in their tests.
-Average is not the same as median (sometimes referred to as mean)! Once the median has been reached, 50% of the samples will in this case have failed and the rest will not have. The average on the other hand is how long they will last on average. Just to illustrate the difference, here with a very small sample size:
Six switches last: 1, 2, 2, 3, 4 and 8 million actuations each.
On average, these will last 3 million actuations. (Sum of all divided by the number of samples)
The median is 2 million actuations. (Half the switches will have failed by then.)
This very small sample size tells us that: 100% will last 1million clicks 83% will last 2 million, 50% will last 3 million, 17% will last 8 million clicks in this case.

Edit: It is early in the morning and it was a while since I read statistics, but hopefully this helps someone.

Edit2: So if you want to be sure that 99.9% of the switches last X, that the number of needs to be far lower than what the MTBF states since it is most likely a value where half the switches have failed.

[terran5992]

Watchout we got a badass over here ^^

[planet36]

Imagine 50,000,000 new switches.  If they were all actuated once, there is a statistical probability that one of them will fail.

[Polymer]

Also, you fail horribly at math. Like, you need to go back to remedial arithmetic horribly. A hundred years? Not even close. Seriously - off by factor of 100 wrong.

1000 presses/day * 365 days/year = 365,000 presses/year.

Divide that into the estimated lifecycle of 50,000,000 presses and you're left with approximately 137 years before switch failure.

Would you please correct my mistake, rootwyrm?

Your math is right...I've seen Rootwyrms math..it is the way he calculates it that is wrong..he estimated 50 million for the entire KEYBOARD..which is wrong..

[Findecanor]

The numbers are 50M actuations for linear Cherry MX. Blues, Greens and Whites are rated at only 10M actuations.
Note also, that this is for actuations. Any tactile switch will have its tactile bump on the slider worn down long before that.

[Grim Fandango]

Maybe 50 million actuations. But jamming on the with regular use, that includes not only careful actuation? Who knows.

[mapple]

The numbers are 50M actuations for linear Cherry MX. Blues, Greens and Whites are rated at only 10M actuations.
Note also, that this is for actuations. Any tactile switch will have its tactile bump on the slider worn down long before that.

Please do me a favour and show me where it's written? i'll sue them all........ seriously....... each page is showing 50M for each switch.....

[rootwyrm]

The numbers are 50M actuations for linear Cherry MX. Blues, Greens and Whites are rated at only 10M actuations.
Note also, that this is for actuations. Any tactile switch will have its tactile bump on the slider worn down long before that.

Nope. Those were all revised to 50M a few years back. So there's a reason for it. Actuation is "you press the switch and it generate electrical signal as designed."

Dug this up for another thread, but will share here too.

Your math is right...I've seen Rootwyrms math..it is the way he calculates it that is wrong..he estimated 50 million for the entire KEYBOARD..which is wrong..

LRN2USESEARCHFUNCTION

I have elaborated on the math that gives me an expected lifetime of 187 days for Cherry MXes, at great length, and holy crap I actually know how to unit test AND measure these things! I mean there's no WAY someone who cares about their keyboards would do silly things like research, RIGHT?

[Polymer]

The numbers are 50M actuations for linear Cherry MX. Blues, Greens and Whites are rated at only 10M actuations.
Note also, that this is for actuations. Any tactile switch will have its tactile bump on the slider worn down long before that.

Nope. Those were all revised to 50M a few years back. So there's a reason for it. Actuation is "you press the switch and it generate electrical signal as designed."

Dug this up for another thread, but will share here too.

Your math is right...I've seen Rootwyrms math..it is the way he calculates it that is wrong..he estimated 50 million for the entire KEYBOARD..which is wrong..

LRN2USESEARCHFUNCTION

I have elaborated on the math that gives me an expected lifetime of 187 days for Cherry MXes, at great length, and holy crap I actually know how to unit test AND measure these things! I mean there's no WAY someone who cares about their keyboards would do silly things like research, RIGHT?

Yeah..your calculation is wrong...I've seen your post on it..the way you calculate it is incorrect because you base is on (let's use 50 million) over the entire keyboard...

I think you did something like your characters per minute and how many hours you worked on it per day. 

Your calculation was something like:

700CPM x 60 x 8 = 336000

50,000,000 / 336,000 = 148 days...

While not the exact numbers you used, it is close..so you can see where the error is in your calculation  700cpm is over the entire (well not really since some keys are used more than others) keyboard.  You don't type by hitting a single key over and over so the way you've calculated it is wrong.  Assuming any one key shows up one out of 10 times, you're looking at 10 times what you've calculated....

[ch_123]

My cherry keychain is losing its clickyness after about 6 months.

The clicky-ness will wear out long before the switch stops actuating correctly, as the clicker in the blues is an ancillary part of the switch that does not affect its actuation.

The numbers are 50M actuations for linear Cherry MX. Blues, Greens and Whites are rated at only 10M actuations.
Note also, that this is for actuations. Any tactile switch will have its tactile bump on the slider worn down long before that.

It used to be 20 million for the tactile and click switches, but as has been pointed out, it has been changed. It's hard to know whether they were being overcautious before, or cavalier now. I'll give them the benefit of the doubt and assume the former.

[terrpn]

stroke me

[daerid]

My cherry keychain is losing its clickyness after about 6 months.

That's probably because there's some dust or hair or something in the switch preventing the little plastic bit that makes the click from moving freely. Much more common in a keychain switch than one that's in a keyboard.

[Photoelectric]

There is a difference between saying "math is wrong" and "reasoning is wrong".  The first implies an error in calculations, forgetting some unit conversion, etc.  The second says that you set up those calculations incorrectly to start with.  I suspect rootwyrm is actually trying to say the latter.

[Polymer]

There is a difference between saying "math is wrong" and "reasoning is wrong".  The first implies an error in calculations, forgetting some unit conversion, etc.  The second says that you set up those calculations incorrectly to start with.  I suspect rootwyrm is actually trying to say the latter.

Yeah...In actuality, Rootwyrm has made some incorrect assumptions in his calculation....

[tp4tissue]

There is a difference between saying "math is wrong" and "reasoning is wrong".  The first implies an error in calculations, forgetting some unit conversion, etc.  The second says that you set up those calculations incorrectly to start with.  I suspect rootwyrm is actually trying to say the latter.

ah.. no,, i don't care if he's your friend. but by his post, he's clearly throwing Factor of 100 around... so.. he means the former and not the latter.

Rootwyrm done goofed.... Plain and simple...

I suppose this emoticon probably portrays his reactions right now

[abdulmuhsee]

Even with dust and humidity, how much of the keyboard's lifespan would that actually annul?  Even if other wear factors cut off a full 100 years of the keyboard's life, that would still leave the user with 36.98 years of service, which I would be quite content with.  However, even that number seems high to me, though there are some Model Ms pushing that boundary.  Others have failed before reaching it.

I doubt anyone here would be in a position to test it out, either.  Stick with the same keyboard for over a decade and use it faithfully every day?

In other words, it is somewhat deceptive to advertise switches as lasting for 50,000,000 actuations, and I think the statement needs to be put in context.  When advertising their keyboards, a company that lists this number should put a disclaimer that reads something like:

Switch life: rated for 50,000,000 actuations!  Warning, there is no chance in hell you will ever get anywhere near 50,000,000 actuations before the keyboard dies on you.

[tp4tissue]

There is a difference between saying "math is wrong" and "reasoning is wrong".  The first implies an error in calculations, forgetting some unit conversion, etc.  The second says that you set up those calculations incorrectly to start with.  I suspect rootwyrm is actually trying to say the latter.

Yeah...In actuality, Rootwyrm has made some incorrect assumptions in his calculation....

No dude... he's just flat out wrong..

What makes it worse is he was quite "haughty" and "arrogant" about it.

[tp4tissue]

Even with dust and humidity, how much of the keyboard's lifespan would that actually annul?  Even if other wear factors cut off a full 100 years of the keyboard's life, that would still leave the user with 36.98 years of service, which I would be quite content with.  However, even that number seems high to me, though there are some Model Ms pushing that boundary.  Others have failed before reaching it.

I doubt anyone here would be in a position to test it out, either.  Stick with the same keyboard for over a decade and use it faithfully every day?

In other words, it is somewhat deceptive to advertise switches as lasting for 50,000,000 actuations, and I think the statement needs to be put in context.  When advertising their keyboards, a company that lists this number should put a disclaimer that reads something like:

Switch life: rated for 50,000,000 actuations!  Warning, there is no chance in hell you will ever get anywhere near 50,000,000 actuations before the keyboard dies on you.

say what?  I'm Immortal...

I don't get many chances to use that emoticon..

[Photoelectric]

There is a difference between saying "math is wrong" and "reasoning is wrong".  The first implies an error in calculations, forgetting some unit conversion, etc.  The second says that you set up those calculations incorrectly to start with.  I suspect rootwyrm is actually trying to say the latter.

ah.. no,, i don't care if he's your friend. but by his post, he's clearly throwing Factor of 100 around... so.. he means the former and not the latter.

That's not what my post was about   I was just trying to clarify for him that the actual calculations (a.k.a. the math) shown here are correct.

Also he says  "I have elaborated on the math that gives me an expected lifetime of 187 days"--that's not a factor of 100.  That's a result of a totally different calculation.

By the way, I've also searched for rootwyrm's posts with any of these keywords:  "actuation", "187", and "lifetime" and a few others and have found nothing prior to this thread.
---
tp, I <3 you, but you need to learn when to stop

[tp4tissue]

There is a difference between saying "math is wrong" and "reasoning is wrong".  The first implies an error in calculations, forgetting some unit conversion, etc.  The second says that you set up those calculations incorrectly to start with.  I suspect rootwyrm is actually trying to say the latter.

ah.. no,, i don't care if he's your friend. but by his post, he's clearly throwing Factor of 100 around... so.. he means the former and not the latter.

That's not what my post was about   I was just trying to clarify for him that the actual calculations (a.k.a. the math) shown here are correct.

Also he says  "I have elaborated on the math that gives me an expected lifetime of 187 days"--that's not a factor of 100.  That's a result of a totally different calculation.

stop defending him ...

let's all just forget this ever happened...

[tp4tissue]

here's how I would calculate it

keyboard...... keyboard..... on the desk.... how long is the longest lasting of them all..

Keyboard's response:  Until you spill soda on me you ffff-tard...

[abdulmuhsee]

stop defending him

Show Image

...

let's all just forget this ever happened...

Honestly, I don't really care that his calculation was based on an incorrect premise, but I do care when someone calls me a horrible failure that needs to go back to remedial arithmetic.

Anyways, since my keyboard is apparently supposed to last until after my death and came with a lifetime warranty on the box, I think I'm going to test it out and use only this keyboard for the next ten years and see if it still holds up.  If not, I'm going to call BS on the whole 50,000,000 actuations thing.

That's also assuming I don't lose interest in the whole thing before then...

[tp4tissue]

Let's look at Moby D i c k ,  txt version,  introduction, publisher info deleted.

1,172,140 Characters

190,498 "Space-bar"............ 16.25 %

114,020 "Letter-e" .............. 9.81 %

If ONE key breaks, the keyboard is technically broken.

According to microsoft, keypress is ranked,  spacebar > letter 'e' > backspace


Lets work with space bar.

Assume 2000 words written per day. 10,000 keystrokes.   16.25% spacebar * 10,000 = 1625 spacebars

50,000,000 keystroke * 1day / 1625 spacebar * 1year / 365days  = 84.3 Years


84.3 Years before the space bar kicks it. and your board is technically BROKEN.

[SeriouSSpotS]

That is a long time

[tp4tissue]

Lets look at Charles Hamilton. 

"He is estimated to have written about 100 million words in his lifetime and has featured in the Guinness Book of Records as the world's most prolific author."   -  WIKI

100,000,000 words  * 5 character/word * 0.1625 spacebars = 81.25 million spacebars

Yes this guy can wear out 2x spacebars, but over the course of 85 years (lifetime), idk lets say he started working at 16, that's 69 years.

He averages 4000 words per day using 69 years....

IDK if they counted the space bar in the 100 million words, I would think they did.. but IDK..

[Parak]

The interesting thing is that Cherry is a little inconsistent about their numbers. One spec page I found gives 1,000,000 MCBF (cycles), another gives 1,000,000,000 for the MX switches. Then their current pages don't give MCBF at all, but just a total actuations figure which is not at all as helpful for statistical purposes...

[tp4tissue]

The interesting thing is that Cherry is a little inconsistent about their numbers. One spec page I found gives 1,000,000 MCBF (cycles), another gives 1,000,000,000 for the MX switches. Then their current pages don't give MCBF at all, but just a total actuations figure which is not at all as helpful for statistical purposes...

It's already been mentioned... Cherry is pulling numbers out of their ass.

[berserkfan]

Abdulmuhsee, I don't think we need to theorize so much. Fact is, if we took a survey on geekhack, we probably would be able to determine how many keyboards and how much usage and how many switches were used.

50million is impossible, based on empirical observation.

There are simply too many people who complain that their Razers or something has a jammed key.

Even if Cherry can prove that they tested 50m, these are lab conditions, maybe in vacuums where there was never any dust that could enter a switch or any humidity. 50m done in unrealistic conditions means nothing. I, too, can produce a sculpture out of liquid that lasts for eternity - just make it out of ice and plunk it in Antartica, as opposed to your average sculpture that attracts pigeon droppings with the attendant chemical/ acidic residues that wear down the sculpture over time.

[BucklingSpring]

Here's an idea for the Korean spies - Add a chip on the board to keep counts of key pressed. Make these stats available from the USB port to keep the cost low. On high end bling bling, use a LCD display to show the numbers!!!

I'll take donation for giving up the idea :-)

[tp4tissue]

Here's an idea for the Korean spies - Add a chip on the board to keep counts of key pressed. Make these stats available from the USB port to keep the cost low. On high end bling bling, use a LCD display to show the numbers!!!

I'll take donation for giving up the idea :-)

Buckling Springs, I knew it, You've finally come forward as an NSA spy...

[1391406]

They probably use one of these:

[BucklingSpring]

^This thing don't even bottom out.

[BucklingSpring]

Busted!

Buckling Springs, I knew it, You've finally come forward as an NSA spy...

Show Image

[Polymer]

Abdulmuhsee, I don't think we need to theorize so much. Fact is, if we took a survey on geekhack, we probably would be able to determine how many keyboards and how much usage and how many switches were used.

50million is impossible, based on empirical observation.

The way you're thinking about it is wrong though...one switch, or 100 switches or 1000 switches having gone bad over many many keyboards in a year doesn't mean 50 million isn't the number....

At the same time, do I think their number is based on ideal conditions?  Probably...There are probably a number of other environmental factors involved although I wouldn't say someone's switch going bad because it was in someone's pocket is a realistic example either....