Joystick testing thread

[Sorgelig]

Dualshock 3 works but too much hassle with pairing. With every pad switch you need to use PC to pair DS3 again.
Fortunately i have Dualshock 4 as well which doesn't need PC to pair - thanks to Sony for DS4 pairing mode.

Besides this, DS3/DS4 have some problem with analog sticks due to hyper sensitivity. After some tweak in MiSTer's input code it's much better now.
I'm not sure is there anything to do with xinput, but -32768..+32867 is lame. -128..+127 is enough to digitize such tiny sticks - and even this is too much.
I will adopt this crazy range.
If there won't be any strange issues, this 8bitdo receiver is really useful - especially for DS3/DS4 as they become reliable gamepads.

Some time ago, I ordered two SF30Pro from Amazon and returned them because the analog sticks were'nt perfectly physically cantered and I could not get full values on diagonals (I checked on Windows). Do your joypad exhibit the same small defects?

i don't see any problems with physical aligning of sticks. As for range: you need to understand that full range of values are measured only on strictly vertical and horizontal. If you draw the circle around the stick, you will see that it's impossible to reach the full range on both axis at the same time. Diagonal will be limited by radius of circle. So, it's supposed to be this way.
If some gamepad allow to reach max values on both axis in diagonal, then this gamepad will have dead zones on maximum horizontal and vertical positions. It's simple rule of math - nothing more.

[Locutus73]

Dualshock 3 works but too much hassle with pairing. With every pad switch you need to use PC to pair DS3 again.
Fortunately i have Dualshock 4 as well which doesn't need PC to pair - thanks to Sony for DS4 pairing mode.

Besides this, DS3/DS4 have some problem with analog sticks due to hyper sensitivity. After some tweak in MiSTer's input code it's much better now.
I'm not sure is there anything to do with xinput, but -32768..+32867 is lame. -128..+127 is enough to digitize such tiny sticks - and even this is too much.
I will adopt this crazy range.
If there won't be any strange issues, this 8bitdo receiver is really useful - especially for DS3/DS4 as they become reliable gamepads.

My point was that, if you check my previous tests, you will see that every pad (DS3, WiiU, Switch) is normalized through 8bitdos retro receivers in xinput mode: same -32768 +32867 range, same axis/codes for same buttons, same VID:PID (045e:028e) that corresponds to the wired Xbox360 controller. I suspect that, if someone tests a real wired Xbox360 controller he should get the same -32768 +32867 range, same axis/codes, so, adopting this scheme in MiSTer, you’re not adopting a strange standard: you’re adopting the de facto standard for modern Windows gaming. I think it’s a good choice. Mayflash DS3 dongle doesn’t filter the events, but I suspect that, in xinput mode, it uses all the same values and codes. You should get the same connecting directly the SF30Pro in xinput mode (turn it on holding start+X).

Some time ago, I ordered two SF30Pro from Amazon and returned them because the analog sticks were'nt perfectly physically cantered and I could not get full values on diagonals (I checked on Windows). Do your joypad exhibit the same small defects?

i don't see any problems with physical aligning of sticks. As for range: you need to understand that full range of values are measured only on strictly vertical and horizontal. If you draw the circle around the stick, you will see that it's impossible to reach the full range on both axis at the same time. Diagonal will be limited by radius of circle. So, it's supposed to be this way.
If some gamepad allow to reach max values on both axis in diagonal, then this gamepad will have dead zones on maximum horizontal and vertical positions. It's simple rule of math - nothing more.

This in theory, but you can adapt values non linearly “stretching” the circle to a square, or you can reach full values on the square inscribed inside the circle (so clipping full up, down, right and left); I’m pretty (not totally) sure Windows does something like this; other consoles may adopt other techniques (see Dolphin emulator analog joypad mapping). I have not a SF30Pro under hand, but, if I recall correctly (again, I’m not sure) when I tested it with DXTweak2.exe (search it online) I got full values on all diagonals except some defective ones (i.e. top-left of the left stick) where I reached max value on X and max-<little value> on Y (and the opposite) when fiddling on the stick. Try DXTweak2 (or Windows built in joy.cpl). My sticks were slightly physically misaligned too (look from the top).
I’m asking because I’d like to try them again (maybe mine came from a bad batch).

Locutus73

[Sorgelig]

Unfortunately, if i plug 8bitdo controller to PC, i get BSOD. I forgot when last time i've saw BSOD on my PC. So unfortunately i cannot use this receiver on my PC. When i will have a boring time i will try to understand why it happens.

Fortunately, if i plug receiver with button pressed, BSOD doesn't happen, so at least i can perform firmware update and DS3 pairing.

[Sorgelig]

I have XBOX360 gamepad. It has the same VID/PID and linux sees it as USB device, but unfortunately it's not recognized as input device.

[Locutus73]

I have XBOX360 gamepad. It has the same VID/PID and linux sees it as USB device, but unfortunately it's not recognized as input device.

Wired, wireless with dongle or wireless with sync’n’charge cable? I think they behave differently. Did you try with MiSTer Linux or with a full blown PC variant? Maybe XBox gamepad is waiting for some acknowledgement/handshake from Windows before starting to send events (just like DualShock 3 which, before pressing PlayStarion button, is seen by Linux but doesn’t send events). Under Windows (just like with XBox) the joypads are informed by the operating system about their numbering (player 1, 2, 3 or 4) so that they can turn on the corresponding led (the led circle around the Xbox button). Even DualShock 3 with MayFlash dongle or SF30Pro implements this (SF30Pro has four leds facing the user). If you connect first a XBox360 joypad to Windows it switches on the first led; then if you connect the DS3+Mayflash you see led n. 2 and then if you pair through Bluetooth SF30Pro in xinput mode you see led n.3. Maybe the XBox360 pad firmware is awaiting for this info.
I’m pretty confident that the events you see through the 8bitdo dongle are standard xinput events so you’re degrading the value of your last MiSTer update labeling it as “Input: better support for 8bitDo receiver”; I strongly suspect you coded better support for xinput devices.

Locutus73

[Sorgelig]

I’m pretty confident that the events you see through the 8bitdo dongle are standard xinput events so you’re degrading the value of your last MiSTer update labeling it as “Input: better support for 8bitDo receiver”; I strongly suspect you coded better support for xinput devices.

you don't understand what you're talking about.
Even if you move analog stick strictly vertically on DS3/DS4 pad, you will see a lot of horizontal events in range like -2000..+2000. I don't know your education background but if you have engineering background you should know what is mechanical tolerance and how many digits in 16bit counter are trustful in this construction.

[Locutus73]

Fortunately i have Dualshock 4 as well which doesn't need PC to pair - thanks to Sony for DS4 pairing mode.

Yes, thanks for that and shame to Sony for the nasty cryptographic challenge/response authentication scheme over Bluetooth that PS4 uses in order to authenticate only official and licensed controllers. But I can understand that from a commercial and quality control standpoint.

Locutus73

[Locutus73]

I’m pretty confident that the events you see through the 8bitdo dongle are standard xinput events so you’re degrading the value of your last MiSTer update labeling it as “Input: better support for 8bitDo receiver”; I strongly suspect you coded better support for xinput devices.

you don't understand what you're talking about.
You can fork MiSTer project and make it much better if you are so confident.

Uh? I’m sorry, but I don’t understand what you mean.

Locutus73

[Sorgelig]

Uh? I’m sorry, but I don’t understand what you mean.

I've edited my previous post for better understanding.

[Sorgelig]

Well, i've re-read your answer and probably you mean that my last update is related to all xinput devices instead of 8bitDo.
No, it's not, since i've hardcoded VID/PID to apply changes to 8bitDo (XBOX360) only.
There are many different gamepads/joysticks with so many different interpretations - especially when it comes to analog sticks. MiSTer code already includes tweaks for several different gamepads. And there is no pattern to make it auto configured. Different manufacturers assign different axis numbers. And even ranges are different - i know at least 3 different ranges already.
Probably, it's possible to request USB descriptors to get valid ranges, axis, buttons.. But you never know what chinese manufacturer will "invent" - and then the whole idea will be broken again.

[Locutus73]

Uh? I’m sorry, but I don’t understand what you mean.

I've edited my previous post for better understanding.

Ah, ok, I see what you mean, but probably I didn’t expressed myself clearly. What I meant was that, regardless of the good sense (or lack of) about using 16bit values for representing such a small finger movement (but it represents 1080degrees rotation steering wheels too for racing enthusiast gamers), what you’re observing through the “brickwall” dongle is not a weird implementation from a Chinese brand (8bitdo), but probably is the standard HID event mapping Microsoft itself chose for xinput.
As an example see https://github.com/khanning/android_ven ... ct_0719.kl the button mapping corresponds to what I observed some posts ago.
So I suspect that, with your last update, you arleady implemented the Microsoft xinput standard HID mapping, not just the 8bitdo controller support.

Locutus73

[Locutus73]

Well, i've re-read your answer and probably you mean that my last update is related to all xinput devices instead of 8bitDo.

Indeed

No, it's not, since i've hardcoded VID/PID to apply changes to 8bitDo (XBOX360) only.

You hardcoded VID:PID 045e:028e right?
If you’ve done that you didn’t hardcoded 8bitDo (XBOX360) identifier; you hardcoded standard wired Microsoft XBox360 controller VID:PID. 8bitdo, Mayflash and others, when implementing xinput mode, actually mimic the standard XBox 360 controller both in VID:PID and in events mapping.

There are many different gamepads/joysticks with so many different interpretations - especially when it comes to analog sticks. MiSTer code already includes tweaks for several different gamepads. And there is no pattern to make it auto configured. Different manufacturers assign different axis numbers. And even ranges are different - i know at least 3 different ranges already.
Probably, it's possible to request USB descriptors to get valid ranges, axis, buttons.. But you never know what chinese manufacturer will "invent" - and then the whole idea will be broken again.

Yes, but I think that Microsoft introduced standard mappings with xinput
see https://msdn.microsoft.com/library/wind ... 05052.aspx
so I suspect you just coded the most widespread modern gamepad implementation, mimicked by many Chinese brands for the sake of Windows gaming compatibility. I.e. with old dinput joypads I had to define button mapping for each game (button and axis mapping changing from pad to pad); with new xinput joypads (real XBox360, 8bitdo, MayFlash) the games just work without any configuration with the same layout you have on XBox and PlayStation games.

Locutus73

[Locutus73]

Unfortunately, if i plug 8bitdo controller to PC, i get BSOD. I forgot when last time i've saw BSOD on my PC. So unfortunately i cannot use this receiver on my PC. When i will have a boring time i will try to understand why it happens.

?!? Strange !?!
Which version of Windows are you using? Does it happen both plugging the receiver and the pad? Have you tried selecting different modes (powering up with start + A or B or X or Y) before plugging the joypad? If I remember correctly powering up only with start the joypad uses dinput mode, but if you power it up with start+x it goes in xinput mode.

Fortunately, if i plug receiver with button pressed, BSOD doesn't happen, so at least i can perform firmware update and DS3 pairing.

This is coherent. If you plug the receiver with the button pressed, it goes in “firmware upgrade” mode and it doesn’t mimic any joypad.

Did you try other joypads on your Windows? Did you install any strange pad driver (like the homebrew ones used for DualShocks)?

Locutus73

[Locutus73]

you don't understand what you're talking about.
Even if you move analog stick strictly vertically on DS3/DS4 pad, you will see a lot of horizontal events in range like -2000..+2000. I don't know your education background but if you have engineering background you should know what is mechanical tolerance and how many digits in 16bit counter are trustful in this construction.

Modern game engines apply a lot of filtering to these 16bit input values. Low pass filters, complex deadzones, bias, input shaping (I.e. interpreting inputs non linearly). Some games are evaluated upon these filtering (I.e. the first Project Cars was considered one of the best racing games using a steering wheel, but an unplayable game on joypads due to bad filtering).
Time has passed since 8bit gaming

P.S.: sorry for the wall of text

Locutus73

[Newsdee]

In any case controller support on MiSTer is so much better than MiST.

Having XBox one controllers would be nice, but I think they use some proprietary handshake protocol which would be annoying to figure out. I bought a few more 8bitdo receivers for my SNES so that should get me running for MiSTer too.

[Locutus73]

In any case controller support on MiSTer is so much better than MiST.

Having XBox one controllers would be nice, but I think they use some proprietary handshake protocol which would be annoying to figure out. I bought a few more 8bitdo receivers for my SNES so that should get me running for MiSTer too.

Isn’t it ironic I’m arguing with Sorgelig about his code being better than what he says?
Anyway XBox one has different PIDs... but, apart from many Chinese brands just using 045e:028e for mimicking XBox360 controller, Microsoft says that, if device ID contains “IG_”, then it’s a xinput device. See https://msdn.microsoft.com/library/wind ... 17014.aspx
Now, if you have an XBox one controller, can you test it on MiSTer with a debut cable, and report the debug output just like I did some posts ago?
If my hypothesis are correct (default mapping for all xinput controllers), Sorgelig could change the VID:PID filter in its recent code from 045e:028e (XBox360’s one, or, as he says 8bitdo’s one ) to IG_ and automatically supporting other vendors and models of xinput controllers not mimicking XBox360’s one. Otherwise, leaving 045e:028e, he have done a great job anyway covering the most widespread and cloned controller.

P.S.: It should be interesting to test others xinput controller types.
Microsoft defined 10 subtypes
XINPUT_DEVSUBTYPE_GAMEPAD (Subtype 1) Gamepad
XINPUT_DEVSUBTYPE_DANCE_PAD (Subtype 5) Default for any new subtype
XINPUT_DEVSUBTYPE_GUITAR (Subtype 6)
XINPUT_DEVSUBTYPE_GUITAR_ALTERNATE (Subtype 7)
XINPUT_DEVSUBTYPE_DRUM_KIT (Subtype 8 )
XINPUT_DEVSUBTYPE_GUITAR_BASS (Subtype 11)
XINPUT_DEVSUBTYPE_ARCADE_STICK (Subtype 3) Arcade Stick/Arcade Pad
XINPUT_DEVSUBTYPE_ARCADE_PAD (Subtype 19)
XINPUT_DEVSUBTYPE_WHEEL (Subtype 2) Wheel
XINPUT_DEVSUBTYPE_FLIGHT_STICK (Subtype 4)
https://msdn.microsoft.com/library/wind ... 05052.aspx
But just few are interesting, and, for our use it seems to me that the main difference is that some ones (i.e. my 8bitdo SFC30, the SNES perfect replicas, without analogs, and my arcade stick) reports the digital pad as the left analog at the max value, retaining all the mappings, i.e.: the face button at south (A=Microsoft Style, B=Nintendo style, Cross=Sony style) has event code 0x130

[edit]
There’s always the proprietary handshaking issue...
[edit2]
Maybe it’s “just” (easy to say for anyone not helping Sorg with actual coding like me) a matter of using xpad kernel driver https://github.com/paroj/xpad or xboxdrv userspace driver https://pingus.seul.org/~grumbel/xboxdrv/

Locutus73

[Sorgelig]

It's a good idea to use me as a tester of common joysticks. I have nothing to do, yeah

Anyway, Official XBOX360 controller i have is wireless. I have "Xbox 360 Wireless Controller via Play & Charge Kit" and windows plays sound when i plug it into PC. But it seems this adapter cannot be use as a wired XBOX360 even on windows, although i see correct VID/PID.

Cannot tell about xinput in general because i don't have such devices except 8bitDo receiver. I didn't plug SF30Pro to PC by wire yet since using it as a wired controller is not my intention. Anyway, both devices are from the same manufacturer and cannot be as a representative number of different devices. Most likely the firmwares of both receiver and gamepad share many common parts.

[Sorgelig]

It's more funny - it seems 8bitdo receiver doesn't support XBOX360 wireless controller.

[Newsdee]

Xbox has their own receiver for PC... but it probably needs a Windows specific driver. I have to say, one nice and unique thing it has is the ability to send audio to the gamepad so you can plug a headset to it.

I'll try the gamepads in wired mode when I get a chance.

[misterjbam]

xbox controllers
I tore my hair off developing Happi GC
The wireless controllers use a proprietary signal, so do not work with anything other than the official microsoft receiver
The wireless and wired controllers are not managed in the same way by the drivers.
Thank you microsoft ^^

[misterjbam]

for the 8bitdo is this receiver that you use ?
http://www.8bitdo.com/wireless-usb-adapter/

[Sorgelig]

for the 8bitdo is this receiver that you use ?
http://www.8bitdo.com/wireless-usb-adapter/

yes.

[Sorgelig]

I'll try the gamepads in wired mode when I get a chance.

Your modifications didn't include analog stick range handling. You can refer to my 8bitDo tweaks.
Actually not many cores support analog inputs.
I remember Atari 800, Atari 5200 support analog input. There is breakout game for Atari 5200 using paddle input - so you can test there analog sticks.

[Newsdee]

Indeed; I think I just assigned some unusual axis to a couple of VID:PID. As you said some of them have really random wiring or varying dead zones.

The Apple II core uses analog controllers

[Locutus73]

It's a good idea to use me as a tester of common joysticks. I have nothing to do, yeah

Well, on the contrary, my intention is to help and save some time of yours. If you look at my old post with all the tests, you will see that they contains all the infos needed for your recent update. I’m not saying that you should had coded the -32768 +32767 handling as soon as I posted the tests; I’m saying that I wanted to help you providing some pre-collected data (maybe you could have saved some time just performing a fast test to confirm my data). I’m not a native English speaker, sometimes it’s easy to be misunderstood (like yesterday). Again, with my xinput arguments I’m trying to save you some time affirming that your code does more things that you think and that, maybe, with some small changes to VID:PID filtering, it will be more generic and useful.

Anyway, Official XBOX360 controller i have is wireless. I have "Xbox 360 Wireless Controller via Play & Charge Kit" and windows plays sound when i plug it into PC. But it seems this adapter cannot be use as a wired XBOX360 even on windows, although i see correct VID/PID.

I have the same Xbox360 wireless controllers and cable and I confirm that I didn’t receive any event both in MiSTer and Windows8.1. I have to test it on my Windows10 gaming rig. I have to say that I have the Xbox360 wireless receiver and I didn’t test it with MiSTer.

717cEAz8ZTL._SL1500_.jpg

I will test it this week end.

I would like to add that I’d like to try to compile your kernel and the xpad kernel driver https://github.com/paroj/xpad it seems to contain all the custom acknowledgment and handshaking required by some pads. If I have success I’ll perform some other tests and share the kernel module. I’m a Windows dev, with some experience with compilation of unix user space code under native Linux (no cross compiling), OSX and Windows-Cygwin; I never compiled any kernel, driver and never cross compiled any code.

Cannot tell about xinput in general because i don't have such devices except 8bitDo receiver. I didn't plug SF30Pro to PC by wire yet since using it as a wired controller is not my intention. Anyway, both devices are from the same manufacturer and cannot be as a representative number of different devices. Most likely the firmwares of both receiver and gamepad share many common parts.

Again, if you look at my previous posts you'll see that the VID:PID and events you observed are constant among other 8bitdo products and even among other manufacturers, i.e. MayFlash, products mimicking Xbox360 pads. As I said above, I hope to test the Wireless XBox360 controller on MiSTer confirming all my hypothesis about standard event mapping.

It's more funny - it seems 8bitdo receiver doesn't support XBOX360 wireless controller.

It’s not so strange: 8bitdo products are Bluetooth while old Xbox360 controllers use a custom wireless protocol… and then 8bitdo products are meant to “xbox-ize” non Microsoft controllers.

Actually not many cores support analog inputs.

I remember having seen some mouse emulation configuration during controller setup and I assumed it would use analog stick for mouse movement (just like I use to do with WinUAE). Am I wrong?

Locutus73