Q500, The Weirdest Optical Mouse

i was on ebay a while back and i came across a mouse of a design i didn't recognize and after trying and failing to find any information on it i decided to order one and what i received was not remotely what i expected this here is the boy and i've sussed out most of how it functions through sheer deduction since there ain't jack or about it online there's only two numbers anywhere on the entire device and neither turn up any results uh the model number on top appears to be q 500 but that search just gets you ebay listings and forum posts from other people who have one it also says a model dmp3 on the bottom and it mentions a shaker's ink none of which get any results either so this thing's a total mystery we can know at least that it is an optical mouse and from the overall design language i would guess that it was made before optical mice were common although i could have stuck the landing on that piece of conjecture by just looking at the interface it uses cereal which was more than on its way out by the mid 90s and we didn't really start getting cheap optical mice on consumer desks until the 2000s i'm just going to guess that it came out in let's say 1996. now that would make this a 25 year old design which is not to say that it was made 25 years ago the factory making these probably got in order to make 30 million of them in 95 and are still churning them out today however the reason uh that it's important that it's old is that this probably predates the popularity of the non-typical optical design and i'd love to explain how that works so when i come back and explain how wacky the inside of this one is it'll really make a splash so let's make a short story long every optical mouse you've ever used probably depended on the exact same technology this is a typical example of the sort it uses a mechanism that's very cheap very unimpressive and it works really really well let me open this up and i can show it to you it's uh not very impressive this is the usb cable you got the wheel there a couple switches and then that guy and that's pretty much it there's actually one major component you can't see if we lift up this board here on the bottom here there's an led this used to be a discrete part and all that does is just illuminate the surface underneath the mouse the rest of the magic all happens inside this little black rectangle here that black rectangle actually contains a camera inside there's a little tiny image sensor little tiny lens actually not even the lens is right there it's this cast piece of plastic when you set this mouse on a surface the led illuminates it at a grazing angle which highlights all the imperfections little bumps and divots then the camera on the bottom of the mouse takes a picture of all those imperfections now when you move the mouse just a little bit it takes another picture and it compares the two it finds all the little bumps and divots and scratches and it sees that they were here and now they're here and it does a bunch of complicated math and determines that the only way that could have happened is if you move the mouse like that and it calculates a motion vector from that information seems very complicated but apparently it's really easy to do because every single mouse like this is doing that i don't know hundreds of times a second thousands i don't know it's a lot and it seems to work really really well like i don't know about you but this doesn't feel like it should be a reliable method it's a neat trick but i mean come on it's just a trick for the real world you're gonna have to use something different right well of course we all know optical mice are basically bulletproof they basically just work so my intuition is clearly worth dick the modern optical mouse works so well in fact that nobody is nostalgic for what came before i've almost never seen anyone still using a ball mouse on purpose and that has to be the most impressive accomplishment of any advancement in computer technology history congratulations inventors of the cheap optical mouse you made something so unquestionably good that it didn't spawn a sub-community who think it ruined computing please collect the gold silver and bronze trophies because there is no competition in this category naturally by the end of the 90s some semiconductor company was producing these by the bushel and anyone could buy one for pennies and they're completely fully integrated they even have the usb and ps2 interface on board so anyone who wants to make a mouse using this obnoxiously high-tech and seemingly over complicated technique just has to fabricate a really basic pcb by a couple other simple components cast some cheap plastic and they're off to the races now while the concept of an image tracking mouse had actually existed for decades prior to the invention of those all-in-one modules optical mice were too expensive for the mass market which is why we all just had ball mice however there was at least one optical design that was available even in the early 80s at relatively consumer-esque prices that was the mouse systems design which a lot of people will recognize particularly because it was used for a lot of early sun machines now it wasn't just for suns uh got this one for a texas instruments explorer here for instance and these really came out for every platform under the sun no sun intended sorry i promise that's the last sign now you'll notice that the bottom of this has two holes in it instead of the one that we're used to on modern optical mice if you had eagle eyes you might have noticed that the q500 has two as well that's for similar but different reasons we'll get to it later there's a lot more going on inside one of these first off we'll see that the reason there are two holes is because there's actually two leds and they're different colors i can't quite tell from here but one is red and one is actually infrared like the modern design these of course illuminate the surface below the mouse and then the image is projected upwards through two little plastic lenses however unlike the modern design you'll notice there's no sensors here in fact the sensors are all the way over here and the image gets there by way of a pair of mirrors mounted to the top shell the image actually comes up at a 45 degree through these lenses hits these mirrors and gets reflected back down onto the sensors it's really wacky i guess they just couldn't figure out how to get these guys mounted upside down like they do nowadays now these are not like ccd image sensors they don't generate a complete image they're basically just simple photo detectors which is a type of device that produces a voltage based on how much light is hitting it you can tell immediately they have an orientation to them however one is obviously horizontal and one's very clearly vertical and this starts to hint at how the device works clearly this one handles the horizontal movement and this one handles the vertical movement but since each one gets the light from one of the two leds that means that one of these gets red light and one gets infrared light and that seems kind of weird most surfaces like this one for instance they don't react differently to different wavelengths of light in a consistent way based on which direction you're pointing at them that's not really a thing so this seems pointless but the reason for it is that this mouse does not work on most surfaces this mouse actually takes a special custom mouse pad and a real exotic one that's actually a piece of aluminum and this grid printed on here let's get up in there yeah see it's a little tiny fine pitch grid printed on there and that grid is essential to how this works at first glance it looks like the entire grid is printed in blue but if you look closer you'll find that half the lines are blue while the other half are sort of a grayish color this is where the two wavelengths come into play of course the blue lines would appear black under red illumination while the gray lines would appear black under infrared i could have that backwards but the point is each led here is only going to get its matching color of grid line and the other ones won't show up so the image cast on each sensor will only show the information it's interested in and so now we have two sensors getting two different axes of information which all sounds very reasonable now to make that count for anything these photo sensors have to be pretty sophisticated for the era your typical off-the-shelf photo sensor just detects a single point of light but these ones are more sophisticated than that if you look closely you can see that they actually have four separate segments that's why they've got eight legs on them they're actually four photo sensors in one trench coat and each one detects four points of light in a row side by side now you could put four photodiodes in a row but then they'd be spaced out by the size of each photodiode these ones get much closer i don't know if these are custom parts or what so that basically means that each one of these is taking a little 4 pixel linear photograph of the space under the mouse one vertical one horizontal what do you do with that well i read mouse systems patents to understand more of how this worked and basically when the mouse is sitting on the pad an image of the grid is projected onto the two sensors and the magnification ensures that adjacent grid lines will always cast shadows on exactly two cells with an empty cell in between them now suppose that we convert the output of the sensor into binary so we'd have four bets each one indicating whether there is or isn't a grid at that location so for this one we'd have 0 1 0 1. when the user moves the mouse the output will change for instance if they move to the right the grid lines will move left and land here creating the pattern 101 0. if they continued moving it might change to 0.100 and eventually 1001 as another grid line overlaps the end of the sensor of course if the user had moved the mouse to the left you would have the inverse bit sequence you might have picked up that this is really no different than the modern motion tracking approach just using far fewer pixels but since there are only four the amount of information it's collecting is so minimal that they needed to use the special pad to give the mouse a stark enough image that it could reliably identify the motion if it were just the imperfections of a table the mouse wouldn't have enough context information to know what it was looking at given the era it's impressive that they actually found a solution that would fit into the technology available at the time with the horsepower and imaging technology that was available in something of this size and cost it's really very impressive mouse systems really pulled it off nonetheless it seems like very few consumers had these i certainly hadn't seen one until the late 2000s and from the apparently quite high build quality in the metal mouse pad i'm guessing they weren't cheap so with all the context established let's move on to the new kit on the block which certainly was cheap it weighs almost nothing as the shiny look of market dumped odm trash and the buttons feel squeaky and that way you only get in the worst plastic products by all right picking this up you would think that it was a dollar fifty ball mouse until you turned it over and saw that where a ball should be there's a curious diagonal strip of pinholes you know what they aren't even diagonal it's an optical illusion i got got they're so strange and captivating so i just assumed that this was how it worked i was picturing some sort of brilliant approach with three staggered sensors and a lot of math but i'll show you what those holes are for later but they actually do not help the mouse figure out where it is or even where it isn't well perhaps this photo from the ebay listing will clarify things this too uses a custom mouse pad but instead of a grid it uses two one-dimensional patterns each one only covering half the pad now when i saw this i was immediately upset i mean what if you move it too high or too low and one of the sensors comes off its respective half you lose that axis right well once it arrived it made a lot more sense because it turns out the pad is incredibly small i don't know if it comes across in the video but this is the smallest mouse pad i've ever seen and that's the trick the sensors are separated far enough that you can't actually get them off their respective areas if you move this all the way up see you'd come off the top of the pad before this guy could come off of its area and likewise down here same deal so the pad has to be small because it has to be exactly twice the height of the mouse that's janky ain't it okay but what are they doing here okay so they've taken a grid and split it conceptually in two so instead of the x and y axes occupying the same space and being separated by wavelength they're instead in two separate physical spaces and using the same wavelength it seems worse probably and i suspect they just did it to save on leds which implies on its own that they've probably skimped on the design but you'll be astonished at just how much skimpage has occurred this is the cheapest optical mouse design ever invented bar none and it accomplishes that by almost not being an optical mouse at all let's open it up and i'll explain what the hell i mean by that its first unique quality is that there are no screws at all which is a new level of cheap to me the whole chassis is just two plastic halves with no hardware just uh put a screwdriver in here and there it is all right are you ready is everyone ready i doubt it i'll give everyone a few seconds to uh catch their breath what the hell is this are those fiber optics what's this big gray monolith is there an adult here what you are looking at is by far the most impressive achievement in the field of saving money the folks who design this deserve a zinc trophy for cost cutting because this thing has accomplished the nearly impossible this is an optical mouse that has no specialized components the modern optical design uses that special little black rectangle a dedicated chip with a whole computer and image sensor in it and even the mouse systems design uses those funky quad photo sensors which i'm sure they ordered custom but this this uses only off-the-shelf components widely available at the time and every single custom part is just cheap molded plastic while these are for instance genuinely fiber optic strands they're very thick very simple closer to the acrylic light pipes you'd find in a cheap network router these can be made by almost any factory and have no particular tolerances the leds here and here are both infrared which is interesting knowing how this works i feel like it could have been done just as well using plain red leds but either option was still very inexpensive besides the leds and the unknown chip here we have some capacitors some resistors and diodes and one transistor and that's it the monolith in the center into which the fiber optics go can actually be disassembled by popping these tabs and that reveals two plain photodiodes and both the fiber strands go into those diodes and that's it that is everything that we know about this device so this design is completely unique different than all other approaches to making a mouse in history and since the chip resists all attempts to identify it all we can do is guess how it works but i think it's self-evident if you stare at it long enough so here's my interpretation okay so the leds illuminate the surface under the mouse and the light is carried by two strands back to both photo sensors i was confused by this at first because if both fibers are coming out of the same spot here then they're capturing the same information so why get redundant info on both photodiodes doesn't seem to make any sense and how would the mouse know which direction it's moving by sampling only one point anyway it also seemed like this would be really imprecise since the mouse would only see a transition when you moved onto or off of a stripe and since there's only about 100 stripes on here a trip from one side to the other would produce about 200 pulses which isn't really enough for smooth mouse movement i think so convinced that this thing would be unusable i hooked it up to my pc to see just how poorly it performed and lo and behold it worked pretty well it felt a little rougher than a normal mouse maybe but nothing you'd notice if you weren't paying attention i think the only real criticism i found is that if the pad is rotated relative to the mouse it starts to get weird you get like diagonal movement which is just unavoidable with this design i think you really have to keep it straight which is harder than it might seem also if either one of the sensors manages to slide off the top or bottom of the pad you will indeed lose that axis even though you weren't fully off the pad yet which is kind of a bummer but not as bad to deal with as i expected and neither of these are show stoppers i couldn't figure out how performance this decent was possible there just aren't enough lines on the pad i puzzled over this for a while but after studying the pickup assembly i think i figured out what's going on it's pretty cool most optical mice have a lens over the sensor assembly but this one doesn't if you notice the tube fiber strands are just hanging out right there and they are side by side so while i was thinking that they were sampling the same spot in truth they capture two points right next to each other if we take the pad and put it right next to the sensor you can see that the width of two strands equals exactly the width of one line and now it starts to make more sense since it samples two bits it offers about four times the resolution you would expect from this pad while there's only about 100 lines each one produces four state transitions as the mouse moves across the pad each fiber strand hits each stripe twice one on then the other on then one off then the other off so in the full width of the pad there will be about 400 transitions which is a reasonable number to produce smooth motion the reason that both pairs of strands terminate into both photodiodes is to save more money while the mouse systems design has dedicated sensors for each axis this one shares resources by strobing the two leds turning on only one at a time so it can sample the horizontal then the vertical and so on instead the sensors have to be the size shape and orientation of the data being sampled the designers figured out that they could reshape the information move the components of the image around to make it fit these cheap generic components and also use the same components for both orientations it's brilliant honestly they could have gone even more hardcore with this there could be a whole matrix of fibers carrying a rough image to a grid of photodiodes you could make a device that works like a modern optical mouse but without a specialized sensor just off-the-shelf parts i wouldn't be surprised if that product exists out there somewhere in fact i suspect it might so that's basically how i think it works i think it's super awesome and ridiculous and i have a further hypothesis that i don't really have the skills to prove but it seems plausible and here it is i think that this in truth is a ball mouse let me explain here's a ball mouse let's open it up here is the ball when you put the mouse on the surface and move it the ball rotates it rides on these two plastic rods and the rods rotate a couple of plastic discs looking closely you can see that those discs have notches in them and when the discs spin they pass between a thing called an opto-interrupter an opto-interrupter looks like this it's an led and a photo sensor held a certain distance apart and the disk passes between the two and when the notches come around they interrupt the beam this produces a series of pulses as the disc rotates this combination of parts is called a rotary encoder and they are very commonplace even in modern devices like the wheel in the mouse you're using now dq500 is in essence a pair of rotary encoders that have been unrolled and turned into linear encoders you're moving the sensor past the light and dark instead of the light and dark past the sensor but it's the same principle and it's so similar that it makes me wonder if this mouse could actually be using an ordinary ball mouse controller chip meaning this mouse would contain no parts that were not already available off the shelf if this is true then this is a masterpiece of cheapness a magnum opus of cost cutting if you have any remaining doubts about just how cheap this thing is let me show you now what the holes in the bottom do okay there's the three holes and they line up with um right about there so they do nothing i mean i don't know for sure why they're there but i'm guessing that the people who made this were just not above a little bit of fraud and decided that the three holes would make it look a little more interesting i don't know either way i am astonished at the sheer hutzpah of attempting to move us out of the dark age of mechanical mice into the future with something this unfathomably low cost my hat is off to these folks now while i said that this hardware was unique really this sort of unspeakably cheap product with no attached brand name always gets sold under a thousand names and a thousand models and it seems like that is the case earlier this year lgr covered a quote optical laser mouse that has a totally different brand and design yet is definitely made by the same people since we can now compare these two designs although i can't open his up and see what's inside of it let's take a moment and look at it from a technical perspective in clint's unit there are still two sensors but they aren't offset by the full length of the mouse they're right next to each other now although they are still oriented in two different directions the pad while still a custom design uses a grid of dots instead of the q500 split grid or the mouse system's full grid this seems problematic to me because the dots are all the same color so how could the sensors tell them apart if you were moving the mouse vertically for instance and both sensors were sliding over dots they'd both be seeing pulses of black and white so how would the horizontal sensor know that those weren't supposed to be horizontal pulses my guess is that this is a later revision where someone figured out that if you space the sensors at a half multiple of the grid size it becomes mechanically impossible to get the vertical and horizontal sensors to travel over dots simultaneously when moving in only one axis i admit i have a hard time illustrating that but i think it makes sense in my head so clint's mouse looks like a later better design but i wish i could open it up because while i suspect it has largely the same parts as the q500 there is one curious difference that could turn out to be substantial if you look closely at the sensors they appear to have eight fiber strands instead of two so much like i had hypothesized earlier it seems like this might be using a whole grid of photodiodes to assemble an actual low-res photo it could even be using an 8x8 pixel ccd clint wasn't available to get me inside shots in time for this video so i can't really be sure but if it turns out to be interesting enough maybe i'll do a follow up now despite these looking like they have different mechanisms they're still clearly made by the same people since the special button on mine offers the exact same weird features as the special button on his this button is on the side while mine is in the middle which is actually very confusing since it never actually sends a middle click if you just press and release it it sends an automatic double left click which i think showed up on a few other mice around this time just to make double clicking easier however if you hold the middle and right button for about a second the mouse switches to high dpi mode where the cursor moves half as fast for precision operations i'm sure they intended this for graphics to get better precision when doing cad or whatever but when i try to actually do that i find that while it does make precision motion a little easier trying to draw smooth curves doesn't get any easier which seems odd i'm not sure what's going on there but if i try draw a circle it always comes out kind of squarish no matter how slow i go i guess there's some non-linearity with diagonal motion on this mouse but damned if i know how to measure it now you could also use this for gaming for instance here in quake if i switch to high dpi mode it lets me move very gradually to line up a shot this is a feature offered by a number of modern gaming mice usually labeled as a sniper button but this was unavailable in the mid 90s when i assume this one came out so it's pretty ahead of its time in a sense it's kind of cool but since you have to hold two buttons for a full second to get in and out of it it's not really as useful as it could be i still think some people would have liked it though the other special feature on here can be reached by holding the middle and left button for about a second this puts us in joystick mode in this mode the mouse controller keeps track of how far you've moved it from its starting location and continuously sends updates as if you're continuing to move in that direction the further you are from the imaginary starting point the higher the velocity will be as if you were gradually pushing a joystick further and further from center naturally this is probably most useful for games the immediate thought would be a flight simulator where you might want to perform a slow bank or climb but don't want to have to keep picking up the mouse and moving it over and over well that's sort of valid if we fire up descent and try it there it actually works pretty well descent is kind of a flight sim and it uses absolute input meaning that moving the mouse a certain distance normally results in that much pitch or yaw and no more if you want to turn all the way around you normally would have to pick the mouse up and reset it a few times to get there putting this mouse in joystick mode means that as you push it around you're kind of sloshing a motion vector around gradually it's very sloppy and organic it's hard to describe but it makes perfect intuitive sense if you're using it and although i don't know if i'd want to play like this i can see why someone would it is sort of like using a joystick however while this is cool in descent specifically i'm not sure how many other games implemented absolute motion in this way well it seems like it might be useful for a genuine flight simulator as far as i know pretty much all of those had implemented this sort of movement on their own already microsoft flight simulator for instance was already doing this kind of sticky input when you flew with the mouse 10 years earlier so the joystick mode doesn't offer much there conceivably you could have used this in an fps in doom or quake for instance some people were playing pc shooters with joysticks and gamepads at this time and again i can sort of see the appeal since for instance you can both move and rotate smoothly at whatever speed you like not just the one speed permitted by the keyboard thing is these games are so fast that you really don't want to move any slower than a dead run usually and by the quake era i think people were rapidly deciding that absolute mouse movement was preferable anyway so this was probably not going to be a hit with the gamers in any case when you're done with all this you can reset the mouse to normal motion by just holding the center button for a second these weird extra options seem typical of the kind of strange stuff that came out of taiwan and in this case korea in the 90s when odms were trying to come up with some way to distinguish their product in a space where that was virtually impossible all a mouse does is move and click there's not much you can add to make yours stand out from all the rest most manufacturers just accepted that but a few wanted to try to think outside the box anyway so this kind of wacky value add cropped up from time to time things like the double wheel mouse for instance and the wild thing is that most of those wacky features didn't go anywhere were just silly and unusable but these folks were so close to inventing the sniper button they could have made a mouse that was a killer gaming peripheral 15 years before anybody else had the idea despite all the cut corners this undeniably offers a simple affordable optical design and these mice easily could have sold like gangbusters just on that basis but since i've never seen one before i have to assume they didn't maybe just because they weren't imported by a well-known retailer maybe because they came too late and got steamrolled by better designs or maybe just because they were tangibly very very cheap anyway that's all i have to say about it if you enjoyed this please subscribe so i know you like this sort of thing remember turn on notifications because i upload kind of regularly if you really enjoyed it consider supporting me on patreon like these folks who are really making this possible couldn't do it without all of them to everyone else though thanks for watching me talk for 30 minutes about a four dollar mouse

Channel: Cathode Ray Dude [CRD]

Views: 195,654

Keywords: technology, retroelectronics

Id: Cd6lxwjX2Bk

Length: 27min 3sec (1623 seconds)

Published: Sat Sep 11 2021