You spin me right round, baby...

December 25, 2005 12:40 AM

MTBFFilter: Why is Microsoft's mouse scrolling mean time between failures different for scrolls towards the user vs. scrolls away from the user? As seen here. [PDF]

I realize this question is effectively crossing the threshold into complete nerddom, but I was looking at mouse specs for an answer to this thread, and noticed the difference on some of Microsoft's mice. I can't easily determine if this is the same with other brands, considering some of Microsoft's mice apparently have the same MTBF for scrolls.

Obviously, it makes more sense that a mouse would stand up to higher "down scrolling" stress, since we tend to scroll down more than up... but does that mean they engineer the mouse to stand up to more down-scrolling? Is there any rhyme or reason here?

Thanks in advance... engineers, start your engines!

posted by disillusioned to computers & internet (16 comments total)

no idea. especially if that's read via an optical encoder!
posted by andrew cooke at 3:42 AM on December 25, 2005

Is it not by design? My MS optical mouse, is the same way and does scroll three times faster towards me as away...
posted by stratastar at 6:01 AM on December 25, 2005

the spec is revolutions. if it really does move three times faster on the screen, that's because the screen "jumps" more for the same turnof the wheel (otherwise you wouldn't say faster). so your factor of 3 applies to how things appear on the screen, but the figures in that document apply to movement of the wheel (as far as i can tell).

but maybe that's the source of the confusion.
posted by andrew cooke at 6:30 AM on December 25, 2005

I use my scroll wheel to scroll down a lot more than up. I bet most people do and that's probably why they have different failure times listed. I would guess Microsoft designs the mice so the more heavily used function lasts longer.
posted by 6550 at 10:07 AM on December 25, 2005

6550—my problem with that logic is that it's a wheel.

Are there perhaps two actuators, one to detect downward scrolling and the other to detect upward scrolling?

And perhaps the upward scrolling one is a cheaper unit?

I'm not familiar with the scrolling mechanism's... mechanics, so I don't know if this is practical, and some sort of money saving element.
posted by disillusioned at 10:31 AM on December 25, 2005

It may have to do with the way that force is imparted upon the wheel. When you roll the wheel towards you, your finger may be imparting force tangent to the axel. When you roll the wheel away, people may tend to push down (towards the axel) more than tangent. This would cause more wear on the axel when rolling forward than back.
posted by b1tr0t at 11:02 AM on December 25, 2005

The wheel is a rotary encoder, which has a series of aperatures around the perimeter of a disc. Generally there are two optical sensors per disc, which look though the aperatures, but 90 degrees out of phase to one another. From the two signals, it is possible to derive the direction of rotation.

All of which is to say that there is no inherent mechanical or electrical difference in the way the up/down signal is created. Presumably the difference is in the use of the thing; I like the hypothesis about the difference in finger strength.
posted by Triode at 12:11 PM on December 25, 2005

by the way, MTBF calculations for electronic components are generally bunk. in fact, i'll bet the manufacturer of the encoder/wheel gave different numbers just to make it sound like they had actually put some thought into how they computed the MTBF.
posted by joeblough at 2:44 PM on December 25, 2005

I'm going to second that it is based on actual use and people move the cursor down a lot more than moving it up, especially when scrolling. If I have to scroll down a long page, I move the mouse down - pick it up and place it at the top of the mouse pad - and then move down again. In browsing, there seems to be much more call to moving down the page and starting another from the top, than moving up pages.

So, they probably us a mouse in typical fashion till it fails and note that 100K were up strokes and 300K were down strokes. That's the only reason that makes sense to my little brain.
posted by qwip at 2:44 PM on December 25, 2005

I'm right with Triode up till the 'no mechanical difference'. I only have logitech wheel mice to compare to, so I can't be sure, but...

There is a spring which rides along little bumps molded to the inside of the wheel. This is to give the click-click feel, which actually has nothing at all to do with the signal transmitted to the computer. Normally this spring is asymmetrical, you should be able to feel or hear the difference, so there could feasibly be a difference in MTBF. joeblough has a point here though, I have no idea how they could actually come up with a meaningful number, or why they would publish it.

I'm not exactly sure why they are asymmetrical, but consider... Perfect symmetry is hard, and users do a lot more scrolling down than up, so making the down-scroll action nice and smooth is important. I think the up-scroll is made specially stiff to make the wheel-button easier to use. Imagine trying to press the wheel-button if there was very little friction, it would be hard to keep it from spinning. By making the less used direction more stiff the user can 'lean' on that direction as the button is pressed to keep the wheel from spinning.

Okay, so actually I am exactly sure... I just don't have any reference to back me up.
posted by Chuckles at 8:59 PM on December 25, 2005

which actually has nothing at all to do with the signal transmitted to the computer

On second thought, that might not be true. The apertures that Triode talks about could be aligned with the bumps that the spring rides on. This would keep the electrical signal in sync with the click-click feel...
posted by Chuckles at 9:33 PM on December 25, 2005

(Random terminology filter: the click-click feel is called detent. Or rather, each place the wheel likes to stop is a detent, so the wheel might have a detent every 22.5 degrees, or something along those lines. I'm just mentioning this because I assume that other people, like me, enjoy knowing the names for things.)

I like Chuckles' theory — the usage pattern and the biomechanics of the scroll wheel are both asymmetrical, so it's reasonable that it would be constructed asymmetrically to match, and that could give it an asymmetrical MTBF.
posted by hattifattener at 1:12 AM on December 26, 2005

the only wheel in a mouse ive disaassembled was not asymetrical in any way that would give a factor of 3 in the mtbf. chuckle's theory doesn't convince me. what about a (circular!) wheel is so asymmetic you get such a difference? what is the main failure mode for such a wheel anyway?

the argument seems to be that the "click bumps" are asymmetric, requiring move force in one direction than another. even if that were true, why does it cause a signifiant difference in failure rates? the argument at the moment is no more than "oh, the numbers are asymmetric, so something must be asymmetric, and, well, i guess it could be asymmetric, so, yeah, it's asymmetric and that explains it". hardly convincing.
posted by andrew cooke at 2:41 AM on December 26, 2005

Here are a couple of pictures of the spring in question on a Logitech mouse: wheel assembly in mouse body, wheel assembly out of body. The wheel on an iogere mouse is quite different. Microsoft's wheel is very different again. Here is an unbelievably thorough optical mouse review site and/or hacking guide with tons of large internal pictures of all kinds of recent mice - very slow to load.

The detent spring in the first two links is clearly tilted back, but only slightly. I would like to assert that the bumps on the inside of the wheel are symmetrical, although we don't have any clear pictures of that - I have opened mine up, and without actually performing a measurement I can say that they certainly appear to be. symmetrical bumps plus tilted back spring means less force for down-scroll than up-scroll. Further, on the Logitech mouse I am sitting beside, the up-scroll clicks sound very different (I would say harsher) than the down-scroll clicks. My girlfriend, asleep on the bed beside me, agrees - she sometimes stirs after a vigorous series of up-scrolls, but the down-scrolls never bother her. Finally, I am quite convinced that I can feel the expected difference in required force, although it is subtle.

unfortunately this data is basically all from Logitech mice, which could be taken as meaningless, if you were so inclined. Assuming Microsoft uses a comparable design for this very specific aspect of mouse wheels we still have a lot of problems when attempting to actually answer the question at hand. There are several design aspects to consider: electronics, molded bumps, spring, every other aspect of the mouse wheel mounting mechanism and the finger to mouse wheel interface... Consider, for example, that dirt accumulation from down-scrolls will be much higher, because the top body of the mouse will scrape dirt from your finger and the wheel will carry that dirt down into the mouse body - dependant on exactly how one uses the wheel, of course (which is another way of saying this last point is complete bull shit, now that I have played with my wheel a little bit).

The thing is, any attempt to calculate MTBF on a device that will be in contact with a human must assume some kind of average human usage pattern. A strong grip will naturally cause more stress, and the effect of one particular friend's sweat on the finish of her mouse is really astonishing! Technically, this makes the distinction between the two directions quite ridiculous regardless of mechanism. I guess the people who wrote the spec. were just swayed by some particularly clear mechanism that they just couldn't neglect, something like a spring that has to buckle to scroll one way, but only flex to scroll the other.

Err... preemptive call out time...

symmetrical bumps plus tilted back spring means less force for down-scroll than up-scroll

I guess it is possible, with some exotic material, that the force required to buckle and then flex to clear a particular obstruction might be less than the force required to flex and clear a similar obstruction. It seems like that would be pretty odd to me...
posted by Chuckles at 4:41 AM on December 26, 2005

Seems to me that toward scrolling involves pulling across the top of the wheel, while away scrolling involves the slightest push coming from my hand with potential full muscle strength. I never thought about this, but I can believe it (though I agree the figures are probably more of a wild-ass-guess).
posted by dhartung at 2:57 PM on December 26, 2005

wag: could the failure be in the electronics? and perhaps for backwards compatibility (no idea how/why) the faster scrolling down is done in hardware rather than software. and that particular bit of hardware just happens to be the weak link that causes typical failure.
i don't really believe it either, but i'm not really sure how scroll wheels fail.
posted by andrew cooke at 3:51 PM on December 26, 2005