1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.
Dismiss Notice
Like RaceDepartment on Facebook.

Why you don't need many pedal calibration steps

Discussion in 'Sim Racing Hardware' started by Niels_at_home, Sep 6, 2012.

  1. Niels_at_home

    Niels_at_home
    Reiza Studios

    As I mentioned it in a sub forum/tread, it makes sense to say it here too.

    Years ago when Huttu ruled GPL, I believe he, or other aliens, used the Microsoft steering wheel. It had pedals that gave 64 steps of calibration. Nowadays I read on more forums that more is better. Some pedal electronics allow up to 4096 steps of calibration!

    In our world, big numbers sell and this will probably never change. It doesn't mean it is better though as with everything, there is a point where it doesn't matter anymore.

    Our feet and legs are not super sensitive or accurate. A long pedal may have 4 inches or 100mm of travel. If I ask you to press 86mm then 53mm, then 97mm, will you hit these targets? Even Huttu probably won't! :) So really 100 steps of calibration is fine. For brake pedals with proper load cells, 100 may be on the low side. In an extreme case of 100kg (220lbs) that would mean 1kg steps and we are more able than that, so higher resolution matters there.

    Now of course you are pressing the pedal at a certain speed and surely having more steps makes throttle aplication smoother right? Yes it does, but does that matter? Its not like 100 steps are so little that each step is large. Plus, big plus, when pressing the pedal down at a constant speed, the pedal calibration steps *AND* USB check time interval play a role. If there are 10000 steps of calibration but the pedal is only read ever second, and you apply from 0 to 100% throttle in one second, the game will see an instant full throttle readout, not a smooth application.

    Standard usb devices usually have a 125hz update rate. Lets make this 120 for convenience and apply throttle from 0 to 100% in 0.6 seconds. Every % input then takes 0.006 seconds. At 120hz USB, the pedals are only checked every 0.0083 seconds. If your pedal has 100 calibration steps, in this example the USB frequency is already not high enough to 'get' each position. On top of that, the USB device may be updated faster, but sims may still check the pedals only at the frame rate which isn't uncommon although new sims should hopefully be better than this!

    Finally there is signal filtering that tends to be a part of commercial mass market pedals such as the Logitech products. They know the potentiometers will wear out, so they filter its signal. For example the pedal electronics spend time looking at 10 readouts, then throw out the highest, lowest, and average the rest and send this to the computer as the pedal position. This filtering tends to add latency to the pedal. Changing from logitech electronics to a 'Bodnar Box' gets rid of this latency. So if you are better with the Bodnar Box, it is not so much because of extra calibration resolution, but from reduced latency.

    So in short, except for extreme load cell brake forces, 100 steps of pedal calibration is enough. You will not get faster or anything if this is increased to 4096. Our feet / legs are not accurate enough instruments. USB polling frequency soon becomes a limiting factor, or worse, games that only read new positions at the framerate (may only be 60hz) are a weak link. There is nothing WRONG with more resolution, and technically it is not hard to do, so you may as well do it, but don't forget its just a nice big number, it doesn't really matter!
     
    • Like Like x 5
  2. What about when holding a certain throttle position? As you've stated, it's hard for a humans leg to be accurate and holding your foot at a certain position is pretty much impossible.

    With that said, surely having more steps smooths out the small changes in the foot and that can only be a good thing to throttle control?
     
  3. Niels_at_home

    Niels_at_home
    Reiza Studios

    I'm quite sure that there isn't really a situation where it would really do anything noticeable. Sure when you have to drive 60.000 miles per hour and you may need 40.5% throttle for this, with 100 steps you either do 40 or 41%, but that is so close, your foot probably moves a tiny bit occasionally anyway.

    There is a difference between 100 and 4000 steps, but when driving / racing, I'm convinced there is no meaningful difference.
     
  4. Someone back in 80s said that 640kb of memory will be enough for computers forever. I'm not buying it. I'm sure with enough practice better resolution will provide better results.
     
    • Like Like x 1
  5. Niels_at_home

    Niels_at_home
    Reiza Studios

    This would call for a blind test, though that is hard to arrange. I can adjust the resolution of my pedals with the Plasma Lite USB board that I have, but we'd need a few guys and quite a lot of time to test it. One must agree though that there is a point where it no longer matters. 50, 100, 200, whatever it may be. I really believe people overrate their ability rather structurally. It often happens that people, convinced of whatever they are, can't proof this in a proper double blind test. I strongly suspect pedal resolution to be one of those things, and that people would be surprised how little you need.

    Here is a situation, slowly applying throttle on corner exit. I've analyzed what the weak link is in the controller chain. This is purely technical, I still think humans are an even weaker link, but it still shows an interesting point. You need a physics engine of 700hz in a sim that also checks USB devices at this rate, plus a USB device that updates at this rate in order for a 1024 step pedal to make sense. More is just not noticed! Most physics engines run in the 100 .. 400hz range as far as I know. Any 4096 step pedal is in this example FAR too high resolution to be picked up by any USB device or physics engine currently available.
     

    Attached Files:

    • Like Like x 2
  6. Derek Speare

    Derek Speare
    DSD FTW!

    Is this the offical opinion of Reiza Studios? The OP's avatar leads a reasonable person to conclude he's representing them with this post. The reason I ask this question is that the opinion is potentially at odds with the components some makers manufacture.

    For a sim racing software publisher to post such an opinion - directly or obliquely - about what some sim racing hardware manufacturers make may be seen by some as ill intended or in poor form. It's a curious matter.

    Derek Speare
     
  7. Niels_at_home

    Niels_at_home
    Reiza Studios

    Unless in the Game Stockcar forum, I post as myself.

    Its great that we've come such a long way since the dreaded gameport days, and there is nothing wrong with more resolution in our controllers. There just isn't any real NEED for more resolution at a certain point.

    I did the math and that shows that 1024 steps is already more than the USB hardware and sim can really use during a slowish 1.2 second consistent application of throttle. Why? Just to show that us humans are always thinking more = better, and I like to point out cases where this isn't the case. But as said, it not worse, just more than good enough at some point.

    If upgrading your USB device results in better laptimes or consistency, it is likely in reduced latency as a result of less signal filtering than increased resolution.

    The only opinion I have is that even 100 steps would be enough, which is hard to prove. The math shows 1024 is more than required. You can argue about the opinion, but not really about the math. Unless there is counter-math and an explanation proving me wrong of course, in which case I'm listening.

    When I see unfounded discussions as I have seen numerous times on the 'super resolution USB devices' (on a number of forums and subforums, unlucky for you I happened to read it in the DSD tread yesterday) all I want is to explain with some analysis why it is or isn't worth it. There are always more factors at play. The Logitech signal filtering causes latency, so it makes sense to use a 3rd party pedal solution, things will be better. Also if you use a tiny bit of potentiometer rotation, you may only end up with a small % of theoretically available calibration steps. These things do have to be considered.

    The facts wil vary from sim to sim but I believe rFactor 2 may or may at some point in the future use 400hz USB, the fastest of any sim I know. The pedal resolution limit for this (1.2 second from 20% to 100% throttle) is 600 calibration steps on a USB device at least running at 400hz. USB devices running at 250hz would need no more than 375 calibration steps in this example. Games that poll the controller at the framerate (say 60fps) wouldn't be able to read more detail than 90 calibration steps.

    In the end, we just have to accept its infinitely more likely to be our driving skill that needs work rather than going from 512 to 4096 calibration steps, taking into account the disclaimers that I mentioned above. I would say that is good information rather than an attack at manufacturers, as it still makes sense for most people to upgrade. Probably 8, definitely 10 or 12 bit just likely doesn't matter.
     
    • Like Like x 3
  8. Ok my opinion.

    I had a very shitty wheel laggy and like 10 more faults on it horible to drive. 180Degrees of rotation etc formula force ex logitech.

    Now that i got a g27 by insidesimracing as a prize.

    Better pedals because i have no lagg but if no lagg my old 125 steps was working fine for me. Dont now g27 it is just better because of no lagg.

    New pedals is exactly the same to drive dont now how many steps in g27

    Also with such a greater wheel i cant see any real improvements on laptimes just on how much nicer it feels to drive.

    Game stock car v8 sry but this car is to easy to drive for me to even tell if i have improved or not.


    Formula simracing is like 4 times harder to drive and i am about 1-2 tenth faster and i dont crash in braking because of lagg.

    End result cant see how new pedals can make a difference unless a harder spring or softer. Same end result for me
     
  9. And how does changing USB polling rates affect all this?
     
  10. Dave Stephenson

    Dave Stephenson
    Technical Administrator Staff Premium Member

    Not sure how it relates directly to what's being discussed other than explaining possible methods of dealing with low device steps couple with low latency.
     
  11. Really good application of rule of thumb vs. marketing.

    Many premium equipment is sold based on useless specs, prominent examples can be found in the hi-fi world, where you can buy all those cables made from silver that 'improve so much the quality of the sound'. Meanwhile, my humble copper wire can carry ADSL signal for a couple kilometers at up to 1 MHz, but if I don't connect a silver wire to my stereo I can hear how the under 20kHz waves get horridly distorted in less than 5 meters.

    Yeah, sure.
     
  12. Divided by 4096, each step has 25 um. A human ovule is 4 times bigger than that (and that is about the limit of what eyesight can resolve), a skin cell is 30 um.

    I've used micrometric screws with worse resolution than that! If I had a tenth of that resolution with my feet I could write with them better than by hand.
     
  13. I am sure that just as everyone's ears are capable of hearing different qualities of sound, so are everyone's legs/feet capable of lesser or more control depending on the individual. I know for a fact that I hear more in audio than a lot of people do, but I can't say that my leg/feet is any more sensitive than others.

    That being said, the games we play or sims if you will most likely have limits that are lower than what we can actually feel so pedal, and likely wheel resolution will have a diminishing point of return after a certain point. What that point is I have no idea because I have no real world data to compare to sim hardware. I do know however that I own a Leo Bodnar USB cable for my G27 and it is supposed to give a better resolution to the pedals, but I can't tell the difference to be honest.

    One thing though that I should probably point out is resolution and it's effects on other devices on a computer. For my example I will use my mouse. I own a MS Sidewinder Gaming mouse. This mouse has the ability to change the "resolution" of the mouse to 3 different programmed levels of resolution. With my mouse set to 400DPI the pointer moves painfully slow but is pretty accurate, mainly due to the amount of movement per movement of the actual mouse. If I set the mouse to the highest setting which is 2000 DPI then I am able to cover my entire 5200 pixel width of my display without having to lift the mouse, but I am a little less accurate due to how fast the mouse moves.

    How does this apply to a pedal set? Simply that at 2000 DPI the system is still capable of detecting even the slightest movement of my hand so I would imagine that a pedal set that has a higher resolution, such as 4096 DPI, would be able to detect movement at an increasingly smaller amount. You may not be able to accurately say I am going to place the pedal at 97mm of 100mm of available travel, but the pedal will be able to detect you are at 97mm more accurately and more consistently than it could at 100 DPI.

    Will this have an effect on your driving? If you were required to open the throttle to no more than 76mm at launch or you would break your engine and if 70mm would make the engine stall, you would have a higher chance of being able to hit that mark with the higher resolution every time you tried than you would with a lower resolution pedal, you would just need to train your leg muscles to be as accurate as possible as well.

    And just to mention this, but I have seen an interview of a woman who was born with no arms, and she can write better than I can and she uses her feet obviously. :)
     
  14. It's funny that you mention that switching from the stock G25 or G27 pedals interface to the Bodnar box gives better results. That's exactly what I did. I even switched out the potentiometers for Hall Pots.

    Something else I find interesting is that humans will believe almost anything. We are gullible creatures.

    On top of that, our brains lie to us constantly! Optical illusions are a great example.
     
  15. On the same topic (sort of), why do you need a mouse with massive amounts of DPI?

    If I understand how a mouse operates correctly and how a monitor works (resolution), the mouse pointer can only travel within the resolution constraints and not "in between" those constraints.

    For example, let's say that a computer monitor has a resolution of 1920x1080. Your mouse pointer only moves within those constraints. Meaning that it can't move half, a third, or even one quarter of a pixel. If it could, you'd see a very very smooth mouse pointer!

    Here's a simple test for you to confirm this. Try to move your mouse extremely slow. What's happening is the pointer is moving from pixel to pixel and it looks very blocky when moving. That's because it can't move in between pixels.

    So, since your mouse can only move from pixel to pixel, the only thing massive amounts of DPI is going to do for you is give you a much slower moving mouse. I really don't see the point in that unless you are looking to make very precise mouse movements for some sort of photo editing, and even then most people use a Wacom Pen Tablet which comes standard with a "Precision mode" that essentially halves the movement of the pen according to your input.

    When it comes to how smooth your mouse moves across the screen, you don't want a high DPI setting on the mouse. What you want is a computer monitor with an extremely low latency (2ms or below). Your mouse is still moving from pixel to pixel, but the monitor is feeding you this information at a much higher rate and therefore it appears to be moving smoother.
     
  16. If you read my post again, I didn't say using the Bodnar cable gives better results, I said it increases the resolution and that I am not able to really tell the difference. Big difference on that statement and what you said. I will say however that if someone were to use the G27 without the cable for quite a bit of time to where they have gotten used to the pedals accuracy and then switched to the Bodnar cable, they might notice a difference. I have never used the pedals that much with either, so I don't really notice.

    The Wacom tablet thing would be nice, but unfortunately, some of us don't do graphics often enough to warrant spending the $500 or more US on one. Especially when a mouse with a higher resolution can perform the same thing for a lot less money. Now other than graphics, one reason for having a mouse with a lot higher resolution, as I stated in my post was so that I don't have to pick my mouse up and move it to keep from running off my mouse pad.

    With my mouse set at my normal resolution, which is 800 DPI, I am able to navigate everything across my screen without the hassles. If I wish to have faster responses, then I set my mouse to the higher of my presets which is 1600 DPI. I rarely use the lower 400 DPI as that is just too slow for most things I do, though it does come in handy when I do some of the skins I create. The only thing I can't do that a pen tablet can is I can't use pressure to determine width of a stroke, but I am not that accurate anyways with my strokes so a pen would be a hindrance there.
     
  17. Is it just me or are you not getting Niels' point about the limitations?

    If because of software (sim), or hardware (usb speed), a maximum of 256 (generic value) calibration steps can be achieved, no matter what your pedals can put out, that 2048 (generic value) value will be normalised to the 256 calibration step range anyway.*

    In the world of guitar rigs and effects, your signal is only as good as your weakest link (noise and feedback are the enemy). In this case, it's the usb hub the pedal set is connected to, and the software it is used with.

    * The only difference you would feel in that case is between your ears. It's the same thing when sound techs are tired of a moaning guitar player and pretend to turn some knobs, and get a thumbs op from the guitarist saying the sound is now exactly how he wants it. (while nothing has been changed..)
     
  18. My point in all that was that with a mouse, the device is capable of at least 2000 DPI and the difference is noticeable on screen. This would indicate that the limitation isn't hardware at all so if there is a limitation, then it would have to be in the software, and I have no clue, and not really sure that Neils has one either as to whether the software has this limitation set in it. It would be interesting to hear from the sim developers to find out if they have such a limitation put into their sims. Seems unlikely to me.
     
  19. You are talking about spatial resolution and Niels about rate (so you are both right, but not talking about the same).

    When you switch the DPI of the mouse you are changing the scaling factor 'mouse displacement distance' -> 'screen displacement distance'. At 800 DPI you need to move the mouse half compared to 400 DPI to cover the same distance on-screen. No discussion there.

    What Niels is talking is about how accurate and smooth is the movement. In the case of the mouse, do you see the cursor making ugly jumps depending on the DPI setting? (regardless of how far you need to move it to cover a screen) I doubt it. Mainly because the biggest cause for the 'jumpiness' of the cursor on the screen is the screen itself, that is typically under 100 PPI, much less than the lowest setting of the mouse.

    Anyway the mouse analogy is dangerous, because you can actually be more precise with the lower DPI settings (it would be like increasing pedal travel while keeping the resolution).