Generic Filightsim Interface (8 axis)

[PoDunkMoFo]

Anyone see this yet?

http://www.fly-by-night.de/index.php?option=com_content&task=view&id=25&Itemid=32

It claims the following specs via USB port.


    * 8 analog inputs (axis)
    * 64 push buttons
    * 16 toggle switches
    * 1 coolie-hat


Looks like it's coming out of Germany and I don't see any pricing of availability info yet.

Could be interesting if the price is right.

What do you guys think?

[PoDunkMoFo]

Found the pricing.

Looks like 99 euros?

$118 I think.


I was hoping for something cheaper.


Guess that leaves the A-Pac or DaveB's products.

Unless Randy wants to wade in on this one 

[rdagger]

Here is the link to the site of the guy who built it.  I was looking at his site today.  He shows you how to build your own.

[Kremmit]

You like that one, check this out:
http://forum.arcadecontrols.com/index.php?topic=41992.0

Nobody's bought one yet, that I know of.  $95 Canadian, that's what, $80 bucks American?

(Check their gallery page, too.  You think we build neat looking stuff!)

[muzland]

Came across this listing on ebay a shortwhile ago.

Here is the description with some pictures attached.



<sup>Joystick Controller
8-axis, 32 buttons and a hat switch
    
Unique USB joystick controller chip that handles 8 axis, 32 buttons and a hat switch and compatible with Windows and Mac OS X.

Unlike commercial joysticks with 8 bit resolution this high precision controller has 10 bit (1024 positions) for each axis reported back to PC 200 times every second.

Why do you need one? To convert old gameport joystick to USB, replace cheap controller in USB joystick or finally build that flight simulator cockpit panel. Buying consumer joysticks to rip them apart for electronics is wasted money. Probably, you alrady know that?

If 8 axis and 32 buttons is not enough, use more - each device has serial number for Windows to tell them apart and remember when plugged in a different USB port. No more driver reinstalling!


   Specs
•   8 analog inputs, 10 bit resolution each
•   x32 oversampling - smooth, no-jitter action
•   32 buttons
•   8-way ‘point-of-view’ hat switch
•   Full-speed USB connection
•   28 pin DIP chip

   How to use
•   Works in Windows XP, Windows 2000, Windows 98 and Mac OS X
•   Will work in Windows Vista
•   No drivers or support software needed. It just works
•   Use in any game that needs joystick - FS2004, X-Plane, racing, etc
•   Connect two or more if 8 axis or 32 buttons is not enough
•   Buttons and switches are wired in a 6x6 scan matrix
o   ..with diodes if there will be 3+ depressed simultaneously
o   ..no diodes if all buttons are only momentarily pressed
•   Connect to USB socket or directly to USB cable
•   You receive chip only - assemble the schematic yourself (and save a hundred dollars)

To assemble this simple circuit you have to know how to handle electronic parts and know basics of electronics.

I hope it will make a great weekend project for you!

Analog inputs accept any reasonable value pots from 1kOhm to 100kOhm. Buttons/switches are wired as 6x6 scan matrix: 32 seen by software as normal joystick buttons and 4 last ones (33-36) making a hat switch. Buttons can be left unused but pin 1 needs pull-up resistor even if not used. Unused pot inputs should be grounded to prevent random readings.

This chip is based on a controller from Microchip Technology Inc.</sup>


Would this make a good ipac or keyWiz substitute with the added bonus of analogue inputs? Will it handle multiple button presses correctly without any issues?

[muzland]

Sorry I missed off the price........

The guy wanted

[PoDunkMoFo]

Interesting

Here is a guy that sells the completed product.

http://www.lbodnar.dsl.pipex.com/joystick/

$29.99

Seems as if all the buttons are in a Matrix so I don't know if ghosting is an issue or if that's a keyboard only problem.

[refusnik]

The guy wanted

[muzland]

Hi Refusnik,

Glad you took it in good spirits. I did not realise that it was special delivery rates.

Have you any information on any ghosting or blocking issues that may occur with this?

Also, I would probably have bought one (would have been interesting to build up on a bit of veroboard) but it was the thought of having to source all of the caps, resistors, 4Mhz Clock, USB socket and pin outs separately that put me off. It can cost a fortune to get just a small number of each of these parts. Anybody know of good sources for this??

As an example, I ordered 100 .187" female push on crimp connectors recently for my button and joystick wiring. They ended up costing me just over

[refusnik]

Have you any information on any ghosting or blocking issues that may occur with this?

Also, I would probably have bought one (would have been interesting to build up on a bit of veroboard) but it was the thought of having to source all of the caps, resistors, 4Mhz Clock, USB socket and pin outs separately that put me off. It can cost a fortune to get just a small number of each of these parts. Anybody know of good sources for this??

I think the built up board is a good option. Is this really dollars or GBP as I though Pipex was a UK ISP? Even in pounds it is comparably prices to an IPAC with the added extra of analugue inputs.

I have nothing to hide 

The scan matrix is a scan matrix.  If you don't use diodes in series with buttons than three buttons pressed simultaneously in such a way that they form three corners of a box (L-shape)  will short together two row and two column scan lines and there is now way to tell whether the fourth corner button is pressed or not.

There are two ways around that:
1. Use diodes - guaranteed to work on wires up to few feet long and no cross-influence between the buttons/switches.
2.Put some thinking in arranging the buttons so that those, that are most probably pressed at the same time does not create an L-shape, e.g. putting 6 buttons on either line or diagonal of the scan matrix will make them completely independent and not causing any ghosting even if they are closed at the same time.  However, press anything else and there will be L-shape created.

I did not put scan matrix diodes on the PCB on purpose due to its size.  It is easier to mount them directly on the buttons/switches themselves.  And many people just don't need them at all as they use onl momentary presses (e.g. many flight simulators, weapon firing, etc...)

I really wanted to give all the options to the builders by providing chip-only option - on a shoestring you don't need a USB B socket - you can slice and solder USB cable directly to the proto board.

The cost IS in USD and flat over the world - so boils down to

[muzland]

Nice one Refusnik,

So what you are saying if I have got this right is put the diodes on every switch on the scan matrix and there are no ghosting or blocking problems anymore. Each input is completely independent.

Pretty cool if that is the case and only for the cost of a few diodes. I would still have to wire up the scan matrix as well. I take it that with this configuration I would not be daisy chaining the ground connections but running two wires direct into the appropriate matrix point??

Regards
Muzland

[muzland]

Ooops! Missed out another question.

What is the difference between a Hat Switch input and a normal input? They look the same electrically in the circuit.

Cheers
Muzland

[refusnik]

...if I have got this right is put the diodes on every switch on the scan matrix and there are no ghosting or blocking problems anymore. Each input is completely independent.

Yes, this is correct, each button becomes completely independent.  This is tested and guaranteed.

Pretty cool if that is the case and only for the cost of a few diodes.

I have quite a few small diodes left so I can throw in 40 diodes for an extra quid.  In fact, you can use any diodes you like apart from exotic types (Zener, tunnel, etc.)

I would still have to wire up the scan matrix as well. I take it that with this configuration I would not be daisy chaining the ground connections but running two wires direct into the appropriate matrix point??

You can do it either way.  I think daisy-chainig is easier as total length of wires is less. As usual, it's better to plan it on paper first before commiting to soldering and drilling. 

[refusnik]

What is the difference between a Hat Switch input and a normal input? They look the same electrically in the circuit

Electrically they are the same but logically those four buttons are decoded into 8-position point-of-view thingy.  The reason for that is two-fold: 1) some people simply need it as many joysticks have one and 2) it is pretty useless to have more than 32 buttons on a joystick (even though USB HID device can have practically unlimited number) because very few games will make any use of them without remapping.  I was trying to create most standard hardware so 32 buttons is the maximum that "plain vanilla" joystick can have so 4 leftover buttons were sent to become a hat switch!

[PoDunkMoFo]

This seems like a very interesting possibility.

Correct me if this is wrong but when people were initially (back in the day) having issues with keyboard encoders weren't there inherent issues in using diodes.  In other words didn't other problems crop up?

Not saying that if that was an issue for kb's it would pertain here.

[refusnik]

Correct me if this is wrong but when people were initially (back in the day) having issues with keyboard encoders weren't there inherent issues in using diodes.  In other words didn't other problems crop up

I think I know what you mean.  PC keyboards were not designedto handle more than a fistful of pressed keys whether controller could theoretically tell them apart (with diodes) or not (without).  With arrival of USB keyboards even keyboard HID interface itself encourages developers not to expect many keys pressed - USB keyboard report structure fixes the maximum number of simultaneously depressed keys to six (in addition to modifier keys like ALT or SHIFT).

Buttons on a joystick are completely independent and there are no bottlenecks anywhere on the interface level to deliver their status to the application. 

I have tested the hardware to make sure high speed matrix scan works correctly on long twisted wires (worst case scenario.)  If you consider that all 36 buttons need to be scanned at up to 500 times per second in addition to digitizing 8 analog channels and doing other stuff, we are already talking about microsecond-long pulses sent down the wires. 

I can point out quite a few reasons why an improperly written  code or "re-designed" controller can misbehave.  One of them is not taking wires' crosstalk and capacitance/inductance into account.  What works on a compact tiny keyboard will not necessarily work on a loom of wires few feet long.  This is why I took it seriously and spent quite some time developing and testing this bit.

Sorry for long posts - I tend to get into too much detail... 

[muzland]

Cheers for all the info,

It looks more and more promising. Some more questions though as my curiosity has been piqued!

Any diode will do? What about an LED? Would it illuminate as you pressed the button??

Have you tried this out in a mame control panel setup?

You mentioned that the length of wire is important. This would be very important for my cab as all of the buttons and joysticks are modular. Each stick or button cluster plugs into a wall jack via CAT5 cable. The back of the wall jack runs some more CAT5 to my sidewinder joypad hacks. All in all about 3ft of cabling. With your board I could of course mount it closer to the wall jacks (probably on the back of the box) eliminating a lot of cabling.

I guess there is no need to remap buttons, etc like a keyboard controller as MAME would see this as a 8-axis, 32 button joystick with hat switch under windows. It would then be a matter of configuring all of the axes, and buttons to the correct item in MAME.

Anybody know what issues we could have with other emulators. I am only familiar (to a small degree) with MAME although I plan on adding Daphne and a ZX Spectrum emu.

Is this a generic chip that you have custom programmed?

If so all it is missing for the MAME community is the ability to add optical devices such as trackballs and spinners to be a one stop solution. I doubt anybody needs 8 axes for MAME stuff but I stand to be corrected.

Either way it sounds like a great product for $30.

Daisychaining: I take it you mean to daisy chain a complete row or column of 6 inputs?

Cheers
Muzland

[refusnik]

Any diode will do?

Yes, from few Amp rectifier diodes to small signal diodes, Schottky diodes, etc.  I suggest to select by size and cost but not necessarily in this order 

What about an LED? Would it illuminate as you pressed the button??

No, it will not - the pulse length through a diode is in the order of few hundreds nanoseconds and it is too short to cause it to light up.

Have you tried this out in a mame control panel setup?

Please PM me with a short instruction and I will test this for you.

You mentioned that the length of wire is important. This would be very important for my cab as all of the buttons and joysticks are modular. Each stick or button cluster plugs into a wall jack via CAT5 cable. The back of the wall jack runs some more CAT5 to my sidewinder joypad hacks. All in all about 3ft of cabling. With your board I could of course mount it closer to the wall jacks (probably on the back of the box) eliminating a lot of cabling.

You should not have any problems with that length.  Any wiring is unique but I wouldn't expect any glitches until you reach 10ft or so.

I guess there is no need to remap buttons, etc like a keyboard controller as MAME would see this as a 8-axis, 32 button joystick with hat switch under windows. It would then be a matter of configuring all of the axes, and buttons to the correct item in MAME.

Anybody know what issues we could have with other emulators. I am only familiar (to a small degree) with MAME although I plan on adding Daphne and a ZX Spectrum emu.

Is this a generic chip that you have custom programmed?

Yes, this is a Microchip's PIC18F2550 + 200 hours of my time + hardware errata and a workaround I found for them   http://forum.microchip.com/tm.aspx?m=138617

If so all it is missing for the MAME community is the ability to add optical devices such as trackballs and spinners to be a one stop solution. I doubt anybody needs 8 axes for MAME stuff but I stand to be corrected.

If you are talking about optical encoders, I have this code working but it needs a nice configuration utility to be able to set up the controller, specify movement limits, etc.  I want simplicity and ease of use so I'm holding this off a bit.

Daisychaining: I take it you mean to daisy chain a complete row or column of 6 inputs?

Yes, what I meant is wire all the buttons between themselves and then bring one wire (6+6=12) altogether to the controller rather than bring 72 wires to the controller board - 2 from each button.  It's purely a convenince choice.

If you have any reservations, I have absolutely no problems with returning the board/chip without any explanantions as we are all engineers and we aften fall for stuff we end up not using which is not good to anyone.