Are you certain about that 8bit thing? 'Cause most joysticks I have used more than 256 discrete values per axis. I have an old Cyborg Evo with me, and it has 1024 values for the primary axes (X/Y) and 512 values for twist handle rudder and throttle. (X/Y range 0-1023, throttle/twist handle range 0-511). If I'm not much mistaken, 8-bit resolution only offers 256 values, but do correct me if I'm in the wrong here.
i know for certain that both the x52 and ch fighterstick are 8 bit joysticks. part of the advertising for the x52 pro was that it had 10 bit resolution on its x and y axes where as the original x52 had 8. ch scripting mostly revolves around 8 bit axis values, and likewise the ads for the new eclipse yoke specified a 10 bit resolution. also the thrustermaster cougar also claims 10 bit. it is often touted as a special feature, and it doesnt make any sense for any manufacture not to claim a 10 bit resolution on a joystick that supports it. the whole sidewinder line had really good resolutions (as well as some epic sensor technology), and this goes back to the original 3d pro. of course newer ms joysticks may have better stats, it depends both on the sensor technology used (optical, potentiometer, hall sensor, etc), and the adc used to read the axis (usually on the joystick's mcu).
at the api level joysticks are usually represented with a 16 bit signed value, and is often scaled up on the mcu or in the driver (which is more likely because you want to keep the data frame as small as possible for transmission over usb) to take up the full range.
And before you ask, no, they aren't software interpolated - each value can be accessed by correct joystick position, albeit the angle difference is really small between values so it's easy to skip values.
interpolation is not necessary. it would only really be needed if the refresh rate was very low, which is not a problem even at usb's slowest possible data rate. like i said you can get 50 updates a second off of a 4800 baud serial connection. ive been able to pull that off with a 12 byte packet, containing 8 10-bit axes a sync byte and a crc, with a theoretical update rate of 50/sec. usb no doubt has more overhead but it also has a huge speed advantage over a basic serial line.
And tactile feedback is overrated unless you actually have a well functioning force feedback stick - centering force just doesn't typically represent the forces on airplane controls. You can get all the feedback you need from visual and aural cues from the game itself, which is the reason I removed all the centering springs from my joysticks - makes it easier to do minute corrections around the center zone. Though this depends largely on centering mechanism used on the stick. Saitek's centering method just sucks, single gimbal plate just doesn't offer even tension right around the deadzone, leading to problems with static friction locking the stick onto one place, then you increase force and the friction shift to kinetic and you overshoot the small movement you wanted to make...
if youre going to have spring tension its better to have independent spring tension for each axis, to give you a feel for where the joystick is in each axis. x52 is horrible in that it uses a single spring tension for both axes, so its impossible to distinguish between how much x and how much y you have applied without looking at the joystick. games do give lots of visual cues about how your ship or whatever responds to your input. and sometimes its enough. id rather use a frictioned joystick in something like a heli sim, where you spend most of the time away from the center. for fixed wing sims, i can see pros and cons to each method. but for something less realistic, like freespace, springs are often better. id love to see a joystick witch lets you dial in the spring tension and friction, but you could rest assured such a product will be in the $400+ range.
