Optical Guns question

[shadowdrak]

This might sound jumbled, can't think of a good way to say it.  Does anyone know how Arcade optical guns(like the ones from happ) send their data?  I don't just mean the pinout.  I assume it is some sort of digital pulse, but I don't rightly know.  I've done a search on google and turned up nothing useful.  I also don't have access to an oscilloscope.  I'd like to hack together some sort of interface (serial, parallel doesn't matter)  but I can't find any info.  I found a premade interface by a company called ror3 but their distributor, happ, doesn't seem to have them in their catalog.
I can't say I care much for the idea of using the act labs guns (I have heard they don't work well on some arcade monitors) because they are not arcade real.  I could probably put the PCB in new gun halves but I am still wary.  Any input is welcome.

[u_rebelscum]

I would be interested, too.  What data is sent where, when, and why; and how the different boards process and output the different types of data.


As for happs (guess):  they might be making ror3's interface under their own brand, like "dell" brand printers are now made by lexmark. [shrug]

[mairsil]

Such a simple device, yet no one has managed to set up a USB interface for a Happs gun yet. I have an ActLabs gun, and I just do not think it even comes close to the accuracy (and feel) of my pair of Happs. I tried contacting the R0R3 people several times to try to get more information on how their device worked, but they never responded to my emails. Personally, I think that their price for such a device is ludicrous considering all it should be is a simple USB interface and a driver. If I had the knowledge to write a device driver, I would do it myself.

As far as I have ever been able to tell, squeezing the trigger causes the gun board to send a single "Trigger Signal" pulse to the game PCB telling it to paint the screen. Assuming that you are not playing Duck Hunt, the game PCB paints the screen horizontally until the gun's optical sensor sees the white horizontal line. As soon as that occurs, the gun board sends an "Optic Signal" pulse to the game PCB letting it know the current position. The game PCB then paints the vertical lines until the optical sensor sees the line and sends a second "Optic Signal" pulse to the game PCB. If their are no "Optic Signal" pulses recorded, the game PCB assumes that the gun was pointed off screen (usually to signal a reload).

Specifically, from Happ's optical gun data sheet (http://www.happcontrols.com/images/pdf/opticalgun.pdf):

"The gun is powered by +5 vdc. The optical output signal is a negative going pulse from a LM311 voltage comparator pulled up with a 4.7k resistor. This output is given at the horizontal frequency and is present whenever the gun is aimed at the screen. The trigger signal is a single negative going pulse pulled up with a 4.7k resistor and is active as long as the trigger is pulled."

I hope this information is what you were looking for. If you ever get a working interface, I would be interested in hearing about it.

[u_rebelscum]

From the happs info, sounds like the gun does almost nothing.  (Which is what many of us thought it did.)  And sounds like all your "how it works" is done by the PCB (or like how the actlab gun works).  Either way it's a little exaggerated (all normal monitors draw lines horizontally only).  

Anyway, so the gun sends two things: a trigger pull on one wire, and a "timed" pulse on the other (other two wires being pwr and gnd).  Notice no X,Y location is sent by the gun; all that is processed on the game's PCB.


But one part of happs info is confusing to me.

"This (optical) output is given at the horizontal frequency and is present whenever the gun is aimed at the screen."

This makes it sound like the gun optical output signal pulses once per line drawn if the gun is pointing at the screen?  I guess the pulse is sent when the optical sensor sees the quick brightening of the new line scan, which might be seen over a couple consequetive lines?

Good thing about this is that the gun sounds to be able to track real time, and the trigger can be held down.  (For the acrade game to track real time, however, the PCB needs to be able to also.)


Sounds to me like the hardware you need to get these signals working on a PC is a circuit more complicated than what's in the box the actlab guns have, not a just a simple USB driver.  The added circuit needs to do everything the actlabs' box does*, plus track real time and allow held down buttons.  Calibration would be a ---smurfette---, unless you have a calibration switch on the added circuit.

*I'm assuming the actlabs gun is much like the arcade gun except it has extra wires for button 2 and calibration.
FWIW (I'm guessing) the actlabs box can time things at least to within ~ 1/20,000,000 of a sec (or ~20mHz: horizontal refresh ~55kHz * 1024 horizontal res / 3 horizontal pixel accuracy).  That fine timing accuracy can't be done over USB 1.1 (max 10 megabit), let alone the slower mouse USB (~1 megabit) or a multiple device loaded USB root hub..


Maybe "just" hack the actlab's box and attach the arcade gun?  You'd be limited to the actlab limits though: no real time track, no "machine gunning".

[shadowdrak]

ok.  so the optical signal is timed based on which pixel it is pointing at.  The data file seems kind of vague.  I would assume that the gun's PCB can sense the refresh rate(if not where does it get the sync?).

But per my understanding of this I conclude it would require:

A PCB to decode the tracking info and accept trigger signals. the board would receive the trigger, optionally flash the screen, then pass the tracking info on to the computer as a mouse click at the co-ordinates.  I guess it could also have an adjustable auto repeat interval until the trigger debounces(but that wouldn't be in the first prototype).

My understanding is still lacking on the timing of the optical signal.  I would probably have to buy one to mess with it.  But I still run into the problem of not having any data analysis tools at my disposal.  If I had defined specs, I could probably design a circuit for it, but that seems like it might be impossible unless we could convince someone to "take one for the team" and buy one to analyze it.  That would be a lot to ask of relative strangers.  Any takers though?
Would I be right in the assumption that the data rate on all standard Pc buses would be too slow to write an acurate virtual waveform analyzer for a computer?  Cause if it is possible, I would personally offer to waste my time to try to hack one together.
I guess I could try to talk to Happ to get some more specific answers about their output signals.  Are the tech support ppl that knowlegable anyway?
If I made a device it would probably be a serial interface so I hopefully wouldn't have to use eeproms -- might not be able to avoid this    
I will try to post updates to the situation, I was hoping this would have any easy solution.  Might be easier to hack a console gun to work on a pc.  Do they have better accuracy than act labs guns?

[u_rebelscum]

(BTW, it looks like actlabs lightguns work pretty much the same as the arcade games.  The major difference is the box that gets plugged in the video output.  So look at how the actlab guns work for more details.)

ok.  so the optical signal is timed based on which pixel it is pointing at.  The data file seems kind of vague.  I would assume that the gun's PCB can sense the refresh rate(if not where does it get the sync?).

Easy for the arcade gun: from seeing the screen being drawn.  
When the optical sensor sees the up-spike of brightness of the beam drawing the new line, it sends the signal.  (Since a white screen has the biggest spike, it's the easiest for the gun to see, so that's why happs suggests the arcade draw the screen white at trigger pull: for a more reliable seeing of the spike.)

It's the arcade PCB that needs to know the frequency to calculate the location.  And since it's the source of the video signal, it knows the frequency.

But per my understanding of this I conclude it would require:

A PCB to decode the tracking info and accept trigger signals. the board would receive the trigger, optionally flash the screen, then pass the tracking info on to the computer as a mouse click at the co-ordinates.  I guess it could also have an adjustable auto repeat interval until the trigger debounces(but that wouldn't be in the first prototype).

My understanding is still lacking on the timing of the optical signal.  I would probably have to buy one to mess with it.  But I still run into the problem of not having any data analysis tools at my disposal.  If I had defined specs, I could probably design a circuit for it, but that seems like it might be impossible unless we could convince someone to "take one for the team" and buy one to analyze it.  That would be a lot to ask of relative strangers.  Any takers though?
Would I be right in the assumption that the data rate on all standard Pc buses would be too slow to write an acurate virtual waveform analyzer for a computer?  Cause if it is possible, I would personally offer to waste my time to try to hack one together.
I guess I could try to talk to Happ to get some more specific answers about their output signals.  Are the tech support ppl that knowlegable anyway?
If I made a device it would probably be a serial interface so I hopefully wouldn't have to use eeproms -- might not be able to avoid this    
I will try to post updates to the situation, I was hoping this would have any easy solution.  

Since the actlab and the arcade guns work so similarly, with the black box that comes with actlabs guns doing what the arcade boards did, I'd suggest checking out what's in the black box that comes with the actlabs gun.

AFAIK, that black box:
a) gets the vertical and horizontal refresh rates from the video signal (and probably calibration, too)
b) intersepts the video signal and sends a pure, too bright, white signal starting from the next new screen after the trigger is pulled
c) takes the time it takes from the start of the new screen until the optical signal from the gun comes and
d) calculates the location
e) using the values from the calibration
f) and sends it USB mouse style to the computer.

That's all the board you're trying to make needs to do, except for arcade level results, you want to calculate the location even if the trigger isn't pulled (ie: add real time tracking).

Might be easier to hack a console gun to work on a pc.  Do they have better accuracy than act labs guns?

Other way around.  The actlab guns have the highest accuracy* of consumer lightguns, ATM.  If an HD TV lightgun comes out, then the actlabs might have competion.  
*(Assuming you calibrate the actlab gun correctly, and the video exactly fills the screen.)

Actlabs might have a higher accuracy than the arcade guns, too.  Due to that the actlabs can work at 150 Hz refresh rate, while arcade monitors run at ~32 Hz, plus the higher resolution of PCs than arcades, it easily has a higher precision.  Whether the higher precision comes out as higher accuracy needs to be tested.

[mairsil]

Other way around.  The actlab guns have the highest accuracy* of consumer lightguns, ATM.  If an HD TV lightgun comes out, then the actlabs might have competion.  
*(Assuming you calibrate the actlab gun correctly, and the video exactly fills the screen.)

Actlabs might have a higher accuracy than the arcade guns, too.  Due to that the actlabs can work at 150 Hz refresh rate, while arcade monitors run at ~32 Hz, plus the higher resolution of PCs than arcades, it easily has a higher precision.  Whether the higher precision comes out as higher accuracy needs to be tested.

On paper, the ActLabs gun may have a higher accuracy than any other gun, but it practice it seems that a lot of people (myself included) are not happy with the accuracy (possibly related to issues with calibration). Personally, I would rather have a real Happs gun.

On a side note, the Pelican Silent Scope Rifle supports HDTV's. It even works on some RPTV's as well.

[shadowdrak]

I don't quite understand why calibration is necessary.  Forgive me if this sounds dumb.

My brain tells me this:

If the control pcb can read the VGA signal it has horizontal and vertical sync, resolution, etc.

So why wouldn't it be enough to know such things. you can count the lines down, by dividing the time by the horizontal retrace interval, correct?
the pixel on that line would be: (time to pulse)mod(retrace interval)*(horizontal resolution)/(retrace  interval)

I guess you could replace the mod part with (time to pulse)-(horizontal lines)*(retrace interval) as it might save time (i dunno really)

as long as you sync the timer to the beginning of the frame, it should have no problem finding the pixel as long as you know the resolution and the proper refresh rates.

Act labs doesn't say why you need to calibrate, only that you do.  I also found nothing on google(maybe I am a sucky searcher).

You have been most enlightening up to this point, and I promise this is the last question.  

[u_rebelscum]

Other way around.  The actlab guns have the highest accuracy* of consumer lightguns, ATM.  If an HD TV lightgun comes out, then the actlabs might have competion.  

...the Pelican Silent Scope Rifle supports HDTV's. It even works on some RPTV's as well.

Pelican Silent Scope Rifle is a consumer gun?  ...Googling...

Found a couple pre-release paper hype-views, a more realist reveiw saying it's not very good, and a game / gun reveiw that likes it but notes some problems.

Pluses:
tracks real time
riffle or shotgun
HD TV

Minuses:
Xbox, not PC
glowing green

neutrales:
seems to want a screen flash


I wonder how this one (and other xbox guns) works (the timing issues, ect)?

[shadowdrak]

I think I am going to try to figure out a theoretical circuit to do what i think needs to be done.  If i can come up with something I am pretty confident about, I will buy one of the 45 happ guns and mess around.  I was reading my previous post and i thought of something.  Does the calibration correct for the photodiode alignment -- depending on the angle it could get pretty far away from the right place.  It tells the software that this pixel represtents a corner hit then offsets it maybe.  Still a bit confused... am i too far off?

[mairsil]

Pluses:
tracks real time
riffle or shotgun
HD TV

Minuses:
Xbox, not PC
glowing green

neutrales:
seems to want a screen flash


I wonder how this one (and other xbox guns) works (the timing issues, ect)?

I actually have both the MadCatz gun and the Pelican rifle, along with the only two Xbox gun games (House of the Dead 3 & Silent Scope Complete). The MadCatz works just like you would expect a lightgun to work complete with screen flashes to get positioning. It is not bad, but feels a little lagged at times.

The Pelican rifle is by far the better of the two guns. Since it is able to work with HDTV's, it has a much tighter accuracy than the MadCatz gun. The rifle functions differently depending on which game you are playing. In HotD3, the gun uses the standard screen flash. In SSC, it does have realtime tracking, but because of how the game designers implemented it, it requires the TV brightness to be so high that the game gets washed out. Even with the TV and game brightness at max (nearly pure white), the game still cannot track the gun very well. This is the complaint that most people have.

Does the calibration correct for the photodiode alignment -- depending on the angle it could get pretty far away from the right place.

For the Happs guns, I am not sure if the calibration has any correction for a misaligned photodiode. From my experience using pretty beat up guns in the acrades (like Dave & Buster's where they are more likely to replace a game with a new one than fix a broken one), you can get some really off shots. The design of the gun helps to compensate for this some. If you look at the exploded view for the Happs gun, you will see that the photodiode is set back a couple of inches from the end of the barrel (it appears to be roughly above the trigger and I am not going to take one of mine apart to find out). Also, the actual barrel opening is only 5/8" in diameter. This really helps to limit the amount of screen that the gun can see at any one time, but is not absolute if the photodiode is far out of alignment.

[u_rebelscum]

I don't quite understand why calibration is necessary.  Forgive me if this sounds dumb.

My brain tells me this:

If the control pcb can read the VGA signal it has horizontal and vertical sync, resolution, etc.

So why wouldn't it be enough to know such things. you can count the lines down, by dividing the time by the horizontal retrace interval, correct?
the pixel on that line would be: (time to pulse)mod(retrace interval)*(horizontal resolution)/(retrace  interval)

I guess you could replace the mod part with (time to pulse)-(horizontal lines)*(retrace interval) as it might save time (i dunno really)

as long as you sync the timer to the beginning of the frame, it should have no problem finding the pixel as long as you know the resolution and the proper refresh rates.

Your calculation misses some things:

Time to move beam from bottom of sceen to top of screen; this time is usually equivalent to ~30-50 lines.  Also different video cards (and sometimes different video card drivers) set a different number of those "blank lines" at the begining the new screen or at the end of a screen.

Time to move the beam back to the begining of a new line.  This usually is a little less than the time needed to draw the actual visible line, but there is also the time the beam spends "straightening out" before the image starts and after the image ends.

Latency between sensor and chip. Very important, as this can be many lines, but even half a line (call it ~1/100000 of a sec) is bad.

Act labs doesn't say why you need to calibrate, only that you do.  I also found nothing on google(maybe I am a sucky searcher).

To account for any and all of the above variences.  The calibration done in the original arcade games also does this.

Note that the actlab calibration method does not account for bent sensors, letterbox or other black bars, or overscans.  But the original arcade calibration does.  Anyone want to guess? (ask and I'll try to make sense tomorrow, it's past my bedtime)

[shadowdrak]

I hadn't considered the vertical retrace.  My brain tells me calibration in this case would have to be hardware driven (other wise it might screw up the timing.  I have no experience with VGA stuff, but might it be possible to "intelligently detect" the amount of blank lines before the frame? The number of blank lines probably changes depending on the refresh rate too I would suppose so an ad-hoc adjustment for your hardware probably wouldn't work.  I guess a simple solution to deal with sensor latency would be to have some sort of tuning device (maybe a pot with some logic associated with it) to adjust the timing offset.  I think the most I will do is probably try to hack the actlabs gun to fit into a guncon style case.  I am having too much trouble tracking down a suitable timer IC(most specs I see are way too slow to be of much use).  I figure for the timer to differentiate between pixels, it would have to operate at the horizontal frequency times the horizontal resolution AT LEAST.  this means I need to count at like 30-60mhz depending on the resolution.  I'll look into it a bit more, but I think I may be out of my league when it comes to the amount of time, effort, and money this will take.

[MoonDog]

If I had defined specs, I could probably design a circuit for it, but that seems like it might be impossible unless we could convince someone to "take one for the team" and buy one to analyze it.  That would be a lot to ask of relative strangers.  Any takers though?

what about taking a collection?  you might be able to get enough people who are willing to pitch in.

[Generic Eric]

If I had defined specs, I could probably design a circuit for it, but that seems like it might be impossible unless we could convince someone to "take one for the team" and buy one to analyze it.  That would be a lot to ask of relative strangers.  Any takers though?

what about taking a collection?  you might be able to get enough people who are willing to pitch in.

.50 to a paypal?  I bet there would be enough intrest within the whole community (make post on mameworld.net and here of course of your intentions.) From that wide range of a community, I bet you could raise enough scratch to raise at least part of the money.

[Minwah]

what about taking a collection?  you might be able to get enough people who are willing to pitch in.

Heck I'd pitch in, I'd love to be able to help get an arcade gun working .

[shadowdrak]

I don't really have the equipment necessary to run the kind of tests I would need.  I am pretty certain that I know how they work after some correspondence with Happ; but as posted earlier, I am not familiar with different IC numbers and can't find what I consider to be sufficient electronics for such a project.  I probably wouldn't have enough time to devote to building anything until fall semester is over(assuming I don't take intersession classes).  Does anyone know chip numbers for something like a quartz timer IC and sychronous counters and such that would work in the 50-100mhz range?  Maybe there is some sort of PIC that could be used instead.  But the bottom line is, anything I make would be pretty simple(calibration would be awkward and accuracy would be questionable) at least in the first version.

[Billkwando]

Hi! I got a very useful peice of info about XBox guns from another message board. Quoting post:

What is the yellow wire used for? Allow me to quote "Hacking the XBox" by Bunnie Huang:

"The video sync signal is a 3.3V CMOS or TTL-compatible signal. It is a basic 15.734 kHz positive polarity pulse train synchronized to the horizontal line time of the composite video output, with a single longer pulse at the beginning of every video field. This signal enable peripherals that are pointed at the TV screen, such as a light pen or a light gun for shooting games, to derive position information."

What does this mean? A device that as of yet does not exist would use this signal combined with a lens & photodetector to determine where at the screen you are pointing. I doubt that it will ever be utilized at this point, unless the Linux guys write something to use it.

from url: http://forums.xbox-scene.com/index.php?showtopic=6134&st=75

Hopefully this helps anyone looking into using an XBox light gun.

[NoOne=NBA=]

The trick with the Act-Labs guns is to make sure the game size matches the calibration screen size.
If that is done properly, the gun will hit almost EXACTLY where you aim it.
The only problems I've had at all, regarding accuracy, are driver/software based.
On several games, you need to use the mouse to calibrate the cursor on the calibration screen by moving it to (0,0) and (256,256) manually.

By contrast, I don't remember ever seeing an arcade gun that shot properly.
My experience with arcade guns has been that you always need to use "Kentucky Windage" on them, based on where the previous shot hits the screen.
My thought is that most of them needed to be calibrated, and would probably have worked fine, had that been done.

My point is that a properly calibrated Act-Labs gun is as accurate as you are going to get.
I don't see the point to paying the HUGE money we are talking to interface an arcade gun if ACCURACY is the goal.
My only complaint about them is that they are ugly, as compared to arcade guns.
I can see someone ELSE spending the money to interface the arcade guns if LOOKS are the goal; but I think you are setting yourselves up for severe disappointment if you think the arcade guns will be any more accurate.

[Kremmit]

I don't see the point to paying the HUGE money we are talking to interface an arcade gun if ACCURACY is the goal.

Unfortunately, my goal is just to get something that works, since I didn't have the foresight to spring for a pair of Act Labs guns before they discontinued them.