Software Scoring Machine

[JS Hart]

It's a Saitek P220 Game Pad, about $10 at Fry's (an electronics outlet out here). I use the wireless version (Saitech p2900, $30)for an optional remote controller, as it also still works with the faster (patched) USB poll speed.

http://shop1.outpost.com/product/343...H:MAIN_RSLT_PG

[griffindm]

Sweet. Nice low price also. So, as a remote controller I would imagine you using two of these. One to take strip inputs from the two floor cords (and strip ground). The other to serve as the referee input to award touches and start/stop the clock.

One of the things we have been exploring is being able to drive external scoring lights to improve visibility. Any thoughts there? Unless you use a gamepad with rumble option (and it had enough outputs), I imagine you would have to use another device or port to drive the outputs.

One of the long-term options if this concept proves workable in a competitiion scoring machine (not one of Tom's immediate goals, but if we could do it, so much the better) would be to network the computers to allow for a central display of all bouts underway. In the other thread (scoring with LCD monitor) that was mentioned as a nice extra at some events, though in practice that was done via FencingTime being used in parallel with the real tournament software to drive the monitors. Obviously down the road quite a ways, but it would be really nice for a bout committee.

[griffindm]

Another USB timing changer is explained more fully here

I found the web pickings on the patcher.exe to be a little slim. They should both work, though this one has more info, including an option to restore the original XP USB sampling system file.

[JS Hart]

Yes, I read in the thread the idea of using a gamepad with rumble and replacing the rumble motors with relays to drive external lights. Essentially, using the rumble as output channels while the normal gamepad buttons remain as the input channels from the fencers (the way I've done it.) Sounds like an easy to implement idea. I think there's a Saitek gamepad version that also has rumble, and I'd start there since both the other Saitek controllers I've used so far are unfazed by increasing USB poll speed.

The gamer's basic I used (GLBasic) has commands for turning the rumble on already built in. I actually used it with the remote controller-gamepad as a way to signal back to the ref that the clock had started, and that a touch had been scored just in case the audio cue was missed.

The language also has a set of network commands that I have never looked at, but I assume it would be easy to send names and results back and forth to the bout committee station. Not a big priority for me, though. Like Tim, my interest is to allow more small clubs and schools access to cheap scoring equipment.

On that note, the costs of my adapter to build are about $50 each. Do-it-yourself "Git-er-done" reels (bungee or the similar ones I've built using retractable clothes-line reels from Home Depot) are maybe another $50-$60. So who ever it was that ballparked about $125 for box and reels was right on. So any school/club that has access to a cheap laptop/desktop can easily have their own electric equipment. I have run this on an old Pentium 3 laptop running at 500 Mhz. (i.e. pretty obsolete).

One other idea about repeaters I had was to simply stick some small photocells onto the laptop screen and let them operate relays for the repeater lights. Just an idea...

[JS Hart]

I believe that I tried that other mouse-rate changer without success...

[jabalino]

You know, I'm liking the photoreceptor driven repeater lights idea more and more. They'd be totally independent from the screen lights and pretty darned easy to build.

I'm actually a lot more interested in integrating the scoring box with the bout records/running software. We've been talking for a while now about running a pretty automated tournament in which the laptop acts as the score sheet and record keeper as well as just the scoring box. Van Housen had a fun idea, which was to continuously update all bouts to the central computer so that you could be dynamically seeding the DEs as the pools were going on. Placements would fluctuate wildly at the beginning, and settle in toward the end. In practice I suppose this could cause some problems - imagine knowing that you're likely to finish in someone's bracket, so intentionally losing touches to avoid that. But it even without that, it'd be nice to have DEs ready to go immediately after pools are finished, and having totally accurate brackets published in real time and displayed via a projector onto a wall or something.

BTW, shawn - very cool idea about the force feedback as a tactile indication for the ref that a touch has been scored.

[griffindm]

I was considering the possibility of being able to support the full FIE timings (the 1 ms sabre is problematic anyway. It's their way of saying ANY contact is long enough) in light of the other capabilities we are talking about. The screen clock, extension lights, remote control, networking capabilites are all more useful in a normal USFA competition rather than an unofficial club event. If, as Shawn has indicated, we can beat the sampling threshhold limitations, why not go for the full FIE spec?

The rumble for the touch being scored is a great idea. How about a similar rumble when the clock runs out? That would be in addition to a pc audio signal.

Can you explain the photoreceptor idea a little more? It sounds like hardware on the monitor that recognizes the lights going off and repeats them. How does this affect the cost?

I'd still like to see someone's logged output of the timing registers for debounce time to see if the patched USB really gets us to the detail we need. The other issue that could still potentially get us is the OS thread control for multitasking. The log file should show if it's really an issue or not.

[griffindm]

Originally Posted by JS Hart

My intentions thus far are to get the design stabilized and publish them with "basic" software free. "Advanced" software with the flicks and instant replay I was thinking of asking a modest $5 or $10 for... Is that reasonable?

I'm corresponding with Tom out here in SoCal and hoping we can cooperate on this project, and maybe fence and drink beer, too.-- Shawn

Shawn & Tom,

First, I know Tom has already said he is not in this to sell a product. Shawn, you have said you might be interested in doing something that would get a little return for your efforts.

I think that offering the basic software for free would help those that have the skills and interest to build the interface themselves. Where I think you could make a little money, and build the interest in this as an alternative to hardware boxes, would be if you and Tom could come together in a unified design for hardware and be able to offer it bundled with either of your software designs, or join forces and make a single product with basic and advanced options. That would lower the technical expertise for people to try the product. While there would not be much money in this at the start (relative to the cost of making the interfaces), it would be an excellent way to gain traction in the fencing marketplace. Selling the USB adapter with the basic software for something in the range of $50 would spark some interest, especially in the smaller private clubs or college club teams where the price of systems is of highest importance. Even if you decide NOT to sell an adapter, having detailed "How-To" plans to help others to make their own would definitely aid in the acceptance of the product as a hardware box alternative.

My interest in helping the development and acceptance is in assisting the creation of a lower cost scoring solution to fencing, and leveraging the capabilities and availability of increasing cheaper pc systems (especially older ones). Software is cheap, and if the hardware is ubiquitous then the entire system is affordable and capable of easy upgrades and expansion.

I don't know if you two have thought this far ahead, but you may be onto something significant for the worldwide fencing community.

[Jvanhousen]

The rumble feature would be great. How about a rumble 'countdown' of say 3 rumbles one second apart which would alert the ref of the impending audio signal so he/she would be prepared and not have the need to glance away from the action to check on the stopwatch/machine display?

Originally Posted by griffindm

The rumble for the touch being scored is a great idea. How about a similar rumble when the clock runs out? That would be in addition to a pc audio signal.

[oiuyt]

Originally Posted by Jvanhousen

The rumble feature would be great. How about a rumble 'countdown' of say 3 rumbles one second apart which would alert the ref of the impending audio signal so he/she would be prepared and not have the need to glance away from the action to check on the stopwatch/machine display?

If there's a built-in countdown clock it should be able to lockout lights anyway, as all of the commercial S/T machines do currently. No need to try to watch the clock.

Besides which, the lights should already be within the referee's field of vision, even while watching the action.

-B

[Wafath]

(This may be a threadjack, and I may be exposing my ignorance here, but...)

I am curious if anyone considered using a sound-card for the application instead of other i/o systems?

There may not be enough configuration options with it, but the sound card naturally can do A->D sampling at 40,000 Hz, more than enough to measure all timings. It can also create an impulse signal to be detected on the various lines.

I haven't come up with a crude wiring diagram that would work here, but then, I am not an EE.

Just a thought,

W

[JS Hart]

Using the sound card as input is kind of an interesting idea. Hacking a gamepad was the way I chose because all the event inputs from the fencers were then easily accessible in GLBasic as gamepad data.

I think that the rumble/tactile feedback to the referee already includes time expiration and is built into my current software. However, I quit using the wireless gamepad-with-rumble as my remote control when I pushed the usb polling speed to 1 ms... the rumble-pad I was using couldn't hack the new speed, so I switched to a Saitek p2900 wireless gamepad-as-remote which did run at the faster polling speed. There are probably other types of wireless-rumble-pads that can run at 1 ms, but I didn't want to have to buy several in order to test them.

Thanks, Dave -- that is where I am interested in going with this. Tom currently has a copy of the "How 2" I wrote, and I am hoping we can get a "unified design" going, then I'll post it to a website for download.

As for "debounce logs," during development and testing I included a data capture loop to debug what was going on in the interface, especially with saber. My "Virtual Scoring Machine" program currently has a "speed test" function where it checks the time per loop to run through the program's main event detection loop 10,000 times. It then reports the average iteration time, the number of "fast" (< 1ms) cycles, the number of cycles between 1 and 5 ms, and the number of "slow" (> 5 ms) cycles. Typically on my modest AMD powered laptop I get 9999 "fast" cycles with 1 cycle between 1 and 5 ms. Occasionally I'll get a "slow" cycle or two. The average cycle time is around 25 microseconds per loop. I considered that pretty usable. I assume the variations in speed are due to multi-tasking threads, so the less other tasks on your taskbar, the faster and more consistent the speed should be (theoretically).

The interface as it stands now costs about $50 to build (not including an (optional remote-controller wireless gamepad) based on parts from various online electronic component supply houses like "Jameco". A "kit" of all the parts should be easy enough to put together.

[Jvanhousen]

Optimally, yes. But many clubs operate under less than ideal situations where it's not always safe/convenient or even possible to position the lights and fencers in the same field of view. Add strips with back to back machines calling out the same beep tones, ambient noise and hard of hearing refs and some extra heads-up notice wouldn't always be a bad thing.

Originally Posted by oiuyt

Besides which, the lights should already be within the referee's field of vision, even while watching the action.

-B

[oiuyt]

Originally Posted by Jvanhousen

Optimally, yes. But many clubs operate under less than ideal situations where it's not always safe/convenient or even possible to position the lights and fencers in the same field of view. Add strips with back to back machines calling out the same beep tones, ambient noise and hard of hearing refs and some extra heads-up notice wouldn't always be a bad thing.

So adding three bursts of rumble (the first two serving merely to alert the referee that a third is coming soon) is somehow better in this case than just one set?

Using the rumble seems like a nice addition. Using it to let the referee know that time is ABOUT TO expire seems more likely to increase error than decrease it.

-B

[Jvanhousen]

Originally Posted by oiuyt

So adding three bursts of rumble (the first two serving merely to alert the referee that a third is coming soon) is somehow better in this case than just one set?
-B

Only when the optional 'Saber Rumbler' is used :
http://www.fencing.net/forums/attach...1&d=1197509630

[griffindm]

Originally Posted by Jvanhousen

Only when the optional 'Saber Rumbler' is used
http://www.fencing.net/forums/attach...1&d=1197509630

(Still rolling on the floor from that image)

[Joe biebel]

I have 3 desktop computers that I am replacing that run various flavors of windows XP. I would like to use them (and add a fourth) as scoring machines in the new salle I am opening. I would be happy to a "guinee pig" for this system.

[jBirch]

I've been thinking about this problem for a while now and what I think you want is a digital timing chip to buffer the inputs, then you want to send the data through the USB to the system as discrete data.

On the leading edge of the pulse from weapon 1, start the clock. On the trailing edge of the pulse from weapon 1, stop the clock, notify the computer that a weapon light event has fired and tag both sides with the clock tick. Repeat for any other weapons in the circuit.

That way, you're not subject to timing events and thread context on the computer and you get a high resolution, simple, signal down the line. The computer's task then becomes applying the game-resolution logic to the timing events.

Something like: http://www.kpsec.freeuk.com/555timer.htm

[JS Hart]

Here's my Virtual Scoring Machine:

http://www.swashbucklersfencing.com/VSM.htm

I called it Ver 1.0, but that's just a number; it may be closer to Ver 0.1.

Joe, I promised you this in Richmond, but it took me a while to get a few things cleaned up and a website put together. Let me know what you think.

-- Shawn Hart

[Joe biebel]

Originally Posted by JS Hart

Here's my Virtual Scoring Machine:

http://www.swashbucklersfencing.com/VSM.htm

I called it Ver 1.0, but that's just a number; it may be closer to Ver 0.1.

Joe, I promised you this in Richmond, but it took me a while to get a few things cleaned up and a website put together. Let me know what you think.

-- Shawn Hart

This is amazing, your amazing! This is so cool
Thank you thank you thank you.