PDA

View Full Version : New Scoring Panels



Deimos
04-26-2013, 10:44 PM
I think some more mounting options are sorely needed on the scoring panels, and with all the rules changes coming up, what better time than now? Screw mounting would also enable easily swappable colored covers for the scenario games. While there are a lot of things to consider, I whipped up some designs to start with. Most of the measurements/distances are either obtained from the current panels with digital calipers, or random numbers that looked about right (not a whole lot of science there).

Anyways, here is the large panel:
4645
(all measurements are in mm)

Here is the small panel:
4646
Note: The screw holes line up with the holes in the middle of the large scoring panel to make swapping easier.

All screw holes are 3mm in diameter.

In order to prevent the screws from compressing the FSRs, spacers of some kind are needed. I found some washers of the correct thickness (about), but I think some laser cut polycarb (or other material) sheet would make a better spacer. I found some with a width of about .5mm (English equivalent), and the FSRs are supposed to be .5mm (according to the Trossen site). So that works out pretty well. I'd imagine they would be glued to the main panel.
4647

These are slightly more complex than necessary (obviously) but I don't know if increased laser cutting time is an issue.
I also don't know if there are issues cutting very small thicknesses.

Also, covers:
4648

My goal here was to enable people to actually integrate these directly into their robots chassis (flush mounting). That's why the screw holes are so far from the FSR leads. The short dimension is about the width of two scoring panels (1 for the small panel), so it is as small as possible in that dimension.

Obviously a lot of experimentation would be necessary for a new target panel design. If anybody is interested in the CAD files or wants me to offset the lines for beam width let me know.

I've probably forgotten no small number of things that should be integrated, so any input would be great (even if you just think we should keep the current ones).

tician
04-26-2013, 11:41 PM
In order to prevent the screws from compressing the FSRs, spacers of some kind are needed. I found some washers of the correct thickness (about), but I think some laser cut polycarb (or other material) sheet would make a better spacer. I found some with a width of about .5mm (English equivalent), and the FSRs are supposed to be .5mm (according to the Trossen site). So that works out pretty well. I'd imagine they would be glued to the main panel.
Or only use screws on the back/support panel, and continue to keep the polycarbonate protectors on the surface of the FSRs with adhesive alone?

Deimos
04-26-2013, 11:56 PM
Or only use screws on the back/support panel, and continue to keep the polycarbonate protectors on the surface of the FSRs with adhesive alone?

That would be a simpler solution, we'll have to see how Andrew decides to do team colors (Swappable colored covers may be useful).

jwatte
04-27-2013, 01:54 AM
Screws wear out - or more importantly, the threads you put them in do.
How about slotted bevels on the edges, that the colored covers slip into? Someone posted 3d printed frames a bit like that a while back.
Fastening the panels themselves to the mech could be done from the frame, or even by extending the backing plate to make space for holes.

ArduTank
04-27-2013, 10:24 AM
Or, you could make brackets for the sides that would mount to the mech.

Deimos
04-27-2013, 12:07 PM
Screws wear out - or more importantly, the threads you put them in do.
I hadn't considered this, but i suppose we could use threaded inserts, standoffs, or nuts.

How about slotted bevels on the edges, that the colored covers slip into? Someone posted 3d printed frames a bit like that a while back.
That could be slightly more difficult to implement, however if done right it would be a much better solution than screws.

Or, you could make brackets for the sides that would mount to the mech.
Yes, however I think the simplest mounting options would be desirable, allowing people to only make small changes to their design in order to mount the panels.

Update: I found some adhesive backed magnetic sheets of appropriate thickness on McMaster carr. Perhaps these could be used to secure covers, while using screws to secure the panels themselves. I believe the leading hypothesis is that the scoring system issues have been being caused by induced current from fluctuating EMI, so I doubt any static magnetic fields would be an issue. Thoughts?

ArduTank
04-28-2013, 10:23 AM
It would still cause a problem, and I wouldn't trust a magnet to hold it on with the amount of force the AEG's gearboxes put out (shattering acrylic and whatnot).

Deimos
04-28-2013, 01:46 PM
It would still cause a problem
Could you explain your reasoning? I'm not sure I understand.



and I wouldn't trust a magnet to hold it on
Definitely agree here, I've drawn up some designs to implement Jwatte's suggestion with tabs.
4651
The slotted ones would be slightly thicker than the FSRs (about 1mm - 2mm or something) while the thickness of the other ones wouldn't really matter (same as covers to reduce number of needed materials?) Both would then be attached to the main panel and these would slot into the opening:
4652
Thoughts?

ArduTank
04-28-2013, 06:46 PM
A static magnetic field will still cause FSRs to malfunction if placed too closely.

Good deal with the tabs for the covers.

My question is, how will you get the covers in due to the fact that the covers don't bend.

byi
04-28-2013, 06:57 PM
Note about the 3D printed frames:

I downloaded them and printed them through shapeways, and was told by Andrew during the event that they were not considered legal becasue the bent the covers and prevented proper scoring. better watch out for that.

Deimos
04-28-2013, 07:39 PM
My question is, how will you get the covers in due to the fact that the covers don't bend.
The polycarbonate covers used now are quite flexible because they are so thin. So one would simple bend it and slide it into both sides at the same time.


I downloaded them and printed them through shapeways, and was told by Andrew during the event that they were not considered legal becasue the bent the covers and prevented proper scoring. better watch out for that.
Somebody would just have to experiment with tolerances until they found a good balance between tight fighting and laying flat. If there's enough interest in new panels and we get the design nailed down that is.

jwatte
04-28-2013, 08:49 PM
I think we can have a pretty fair amount of slop on the sides (at least 0.5 mm each side) to make sure there's no "out-bending" happening. That will allow the covers to be fully flat and still well attached to the 'bot.

When it comes to the designs, I find that I would be better able to figure out how well they'll work if there's a 3D rendering of them, rather than the white-on-black outlines only.

As this is all new design, it might be even easier to re-design from scratch. We'd need to assume some particular kind of manufacturing (laser cutting? 3d printing? injection molding?)
For example, if we assume 3D printing with 0.1mm precision, or injection molding, then we'd have something like this:

4653

The idea is that this is both the "back panel" on which you tape the FSRs, and the frame for holding the polycarbonate cover.
The cover can be fully square, and slide in from the top. The screws that hold the panel, also make sure the polycarb doesn't slide out upwards (there's no open slot in the bottom, so it won't slide down anyway.)
Finally, the holes on the sides (should be two holes on the far side, too) would be for the flex connectors to the FSRs to exit.

FWIW: I think I could make this with the 20-ton injection molder at the workshop, although it'd take many hours and a couple of hundred of dollars of material to make the mold...

Another thing that would be cool would be if the full-vs-half panel FSRs could somehow be modular, so you somehow snapped in two half panels to get a full panel. In that case, the half panels might even be fully manufacturable as a unit.

Deimos
04-28-2013, 09:54 PM
I think we can have a pretty fair amount of slop on the sides (at least 0.5 mm each side) to make sure there's no "out-bending" happening. That will allow the covers to be fully flat and still well attached to the 'bot.

Yes, agree here.


When it comes to the designs, I find that I would be better able to figure out how well they'll work if there's a 3D rendering of them, rather than the white-on-black outlines only.
Definitely, I'm just lazy :D.

Here's a 3D render for you:
(it's a little bit difficult to see whats going on - i'm slightly CAD challenged :p)
4654
I think a major benefit of bending the panels and slotting them in is that no screws need to be removed, however it could end up being more difficult than just using screws.

As this is all new design, it might be even easier to re-design from scratch. We'd need to assume some particular kind of manufacturing (laser cutting? 3d printing? injection molding?)
I hadn't considered any manufacturing process other than laser printing, but now that I think about it injection molding would be perfect for something like this. While 3D printing would be great for prototyping, i doubt it would be cost-effective in the long run.


Another thing that would be cool would be if the full-vs-half panel FSRs could somehow be modular, so you somehow snapped in two half panels to get a full panel. In that case, the half panels might even be fully manufacturable as a unit.

If the half size panels had enough screw holes, they could just be screwed together with some laser cut tabs (like the ones I posted earlier for spacers). I'm sure somebody else could come up with a more elegant solution though.


FWIW: I think I could make this with the 20-ton injection molder at the workshop, although it'd take many hours and a couple of hundred of dollars of material to make the mold...
Assuming
A: there's enough interest in new panels
B: we get a design that everyone's happy with
and C: we get Tybs to approve it
I'm sure it wouldn't be a problem to fund this.

DresnerRobotics
04-29-2013, 02:56 PM
Few things-

I'm totally up for improvements to these. They've been low priority as we've focused on getting the actual transponders 100% operational and reliable. We're at that point aside from improving code.

I would prefer to keep these laser-cut in design. We might be getting a bit too complex here, simply expanding the footprint slightly to add for some holes on the back-plate might work just fine. I like the idea of making them universal for half or full sized panels too (everything is half panel by default, can be connected to make a full panel).

I'm not opposed to the 'holster type' panel designs, but again- injection molding or 3d printing are beyond the manufacturing capabilities I can offer for these. Ultimately if we choose a new design, I need to be able to manufacture them in-house. We also had two competitors show up with 'custom target panels' this year, which is technically illegal.

I didn't DQ them based upon it because their bots were otherwise sound (well, one of them), but it did in fact cause problems with scoring as they were improperly designed and caused the lexan to bow out. This is EXACTLY why the rules state that you cannot modify the scoring system in any way. In the future, if competitors show up with modified scoring systems that are not pre-approved, you will be automatically disqualified. We have to keep a standard here, it's one of the most important aspects.

Bottom line is, simplicity should be a key factor when proposing a new design.


I'm sure it wouldn't be a problem to fund this.

Heh. Mech Warfare has cost me approximately $2400 in upkeep costs over the last 12 months, with all but a couple hundred (donations via our paypal) of that coming out of pocket. With the arena coming to Chicago the upkeep costs will go down, but we'll invest more in R&D. Funding is always a problem.

jwatte
04-29-2013, 04:48 PM
So, if we can make a mold, I can actually run a short injection molding run. I even have 10 pounds of PP pellets sitting here with nothing to do with them ;-)

A super simple design would be to extend ears to the panels ("H" style) and have screw holes through the panels. Make the backing panel and the lexan have the exact same outline. Put the screws through both at the same time. Optionally, put a thin washer between them to compensate for the thickness of the FSRs.
Lasers can also cut puzzle-like interlocking shapes into the materials, although to get strong interlocks, you'd need more margin than you'd want on a single half-size panel.

jwatte
04-29-2013, 07:11 PM
Here's what I mean. Assuming FSRs need 50x50mm of space (else, adjust sizing as needed.)
This panel shape would be cut both in ABS for backing and PC for cover. It has a 2.5mm fringe around the FSRs, except for the "notch" in the lower-right which goes exactly to the edge of the FSR.
The idea is that this panel can be rotated 180 degrees and matched up with another one, for a full-size panel.
There would be a 2.5mm space between the FSRs of the two panels; hopefully the force distribution of the PC would register a hit even if it was dead center.

46584659

A strip of plastic could be drilled (or laser cut) with the hole pattern for the top, two-lug, and bottom, single-lug, sides, to perfectly position two panels side by side. You'd use one at the top and one at the bottom. With thicker material, that could be a bracket mounted to the 'mech, and could space to fit the PCB on the back side (which should use horizontal connectors, btw.)

byi
04-29-2013, 09:39 PM
Hmmm, stringing half panels to make full panels leaves that gap, which means a dead-on hit will count as a miss.

Also, I would like to mention that my mech was in fact functional excluding the microcontroller breaking. Nothing to do with the soundness of the mech itself.

CogswellCogs
04-29-2013, 10:16 PM
>> We also had two competitors show up with 'custom target panels' this year, which is technically illegal.
I had no idea those frames would constitute a 'custom target panel'. The panels themselves do not allow for easy mounting and I was trying to find a good way to firmly attach the target panels to the mech. The panels are, of course, original equipment direct from Mech Warfare Global Headquarters.

My plan is to tweak the design with a slightly larger slot and longer side frames. I hope this will eliminate the bowing. Afterward, I planned to test with a chronographed Wal-Mart Defender shooting .12g pellets from 25'. If I can show that target panels mounted in this way reliably register hits, can we consider that an approval process ? I can send a set of frames to the Royal MechWarfare Testing Lab for their own testing if desired. I'll also send a set to BYI because I feel bad that he had to pay ShapeWays for frames that may cause problems.

Of course, if we redesign the panels to make them mountable that will obviate the need for mounting frames. I went down that route originally because the electrical tape/Velcro route just wasn't cutting it.

Deimos
04-29-2013, 10:21 PM
Heh. Mech Warfare has cost me approximately $2400 in upkeep costs over the last 12 months, with all but a couple hundred (donations via our paypal) of that coming out of pocket. With the arena coming to Chicago the upkeep costs will go down, but we'll invest more in R&D. Funding is always a problem.
I did not mean trivialize a couple hundred dollars, or to imply that you had abundant amounts of money burning a hole in your pocket. I was under the impression that myself and anyone else willing would fund the materials out of pocket.


injection molding or 3d printing are beyond the manufacturing capabilities I can offer for these. Ultimately if we choose a new design, I need to be able to manufacture them in-house.
I won't pretend I understand all your reasons for requiring in-house manufacturing, but if Jwatte really wants to... 10lbs of scoring panels is a lot of scoring panels :D. That would probably last for a few years at least.


Bottom line is, simplicity should be a key factor when proposing a new design.
Pffff, using simplicity and design in the same sentence...


which means a dead-on hit will count as a miss.
I'd think it would still trigger the FSRs, the PC might still distribute the force (maybe).
However, I don't really see the advantage of interlocking panels if you still need to screw them together.

I was thinking something more like this:
4660
For a full sized panel, just put two side by side, and screw them to some kind of back plate.
The covers could then be slotted on top of the screws: (screw head go in big holes, then panel slides down and rests in small holes)
4661
This particular panel is just the other one + holes, so it's not quite what the end product would look like.
There would probably also be need for spacers, to prevent covers from compressing the panels.
As for simplicity, if we're adding holes we might as well go the whole mile :).

byi
04-29-2013, 10:43 PM
dont worry about it, cogs.

DresnerRobotics
04-29-2013, 11:05 PM
>> We also had two competitors show up with 'custom target panels' this year, which is technically illegal.
I had no idea those frames would constitute a 'custom target panel'. The panels themselves do not allow for easy mounting and I was trying to find a good way to firmly attach the target panels to the mech. The panels are, of course, original equipment direct from Mech Warfare Global Headquarters.

My plan is to tweak the design with a slightly larger slot and longer side frames. I hope this will eliminate the bowing. Afterward, I planned to test with a chronographed Wal-Mart Defender shooting .12g pellets from 25'. If I can show that target panels mounted in this way reliably register hits, can we consider that an approval process ? I can send a set of frames to the Royal MechWarfare Testing Lab for their own testing if desired. I'll also send a set to BYI because I feel bad that he had to pay ShapeWays for frames that may cause problems.

Of course, if we redesign the panels to make them mountable that will obviate the need for mounting frames. I went down that route originally because the electrical tape/Velcro route just wasn't cutting it.

It was no big deal as we caught it pretty early on.

It's technically illegal because the rules specifically state that you may not alter the plates in any manner.


The plates should not be altered, other than to apply velcro or similar to the back for attachment to the Mech. You may color your target panels so long as the paint product does not interfere with sensitivity of the sensors. Due to technological constraints, plate design may change from time to time, however event organizers will to be consistent from year to year with the plate design.

Your mounting frame caused a gap between the lexan and FSRs, resulting in them not registering hits. That's the exact reason the rule is in place. Like I said though, it's not an issue at all as the impact it had was minimal.


I have no qualms with us coming up with a better solution for mounting them. We just need something that's uniform and well tested so we don't run into problems such as what happened with your printed frame.

ArduTank
04-30-2013, 07:52 PM
We could make mounting tabs on the edges, and then, say, have plastic film in place of the electrical tape (on the sides).

Deimos
05-09-2013, 09:10 PM
Pressure-Sensitive Conductive Sheet (Velostat/Linqstat): http://adafruit.com/products/1361
Has anybody heard of/used something like this? I just saw it on Adafruit, but this could be a much more cost effective solution than the current FSRs. It would also allow for a lot more variation in shape and size. I might pick some up and test it out if nobody has already done so. Thoughts?

(http://adafruit.com/products/1361)

ArduTank
05-12-2013, 07:03 PM
I wouldn't use it. Minor variations in panel designs would cause the scoring system to have to have a "no hit" calibration setting. Too much work. Stick with the FSRs. We can vary the panel shape and size as it is with the FSRs, the only thing preventing it is the rules.

Deimos
05-12-2013, 08:56 PM
Minor variations in panel designs would cause the scoring system to have to have a "no hit" calibration setting.
Do you mean that the transponder wouldn't pick up hits because they would show up at a slightly different voltage? The transponder boards have a programming header... It probably wouldn't be that hard to calibrate them. The idea would be to have a couple (maybe 3) sizes (this could also contribute to implementing classes).



Too much work.
Hmmm, I'd think the main issue with a new solution is added complexity. I doubt most people around here would consider gluing a couple of things together too much work :D. However, a more complex solution is harder to get consistent, and could be a problem for an application that requires consistency. From a cost standpoint, the current FSRs are something like $30 a panel, whereas this would be under $10 per panel (most likely). It'd be a shame not to even try it out though, and if it won't work... well, it was only $4!

jwatte
05-13-2013, 01:23 AM
If we need to change the number of FSRs on panels, or change the materials, I think there's sufficient signals to the scoring panels that you could put a resistive divider and an opamp on them, and maybe even a trimpot. That way, sensitivity would be very easy to calibrate, and hits would be clear and easy to read by the microcontroller (without needing to change the current code, even.)

Note that I'm not arguing for changing the current design -- only suggesting how we could do that in a compatible manner, if we needed to.

DresnerRobotics
05-13-2013, 12:37 PM
We will not be switching from 1.5" FSRs anytime soon, but I do appreciate the suggestion and people keeping their eyes peeled for newer tech. In order for us to make a switch, they would have to be 100% compatible and identical in function with the existing FSRs in the wild as I can't ask people to repurchase the most expensive part of the scoring system.

Again, I do appreciate people doing research and looking into viable alternatives. If someone wants to pick some up and prototype a Target Panel alternative with it, I'd love to see your documentation and results. I currently do not have any spare R&D cash to invest in this though as we have a working solution standing.

ArduTank
05-13-2013, 03:52 PM
Deimos, what i mean is that you'd have to have a calibration protocol for when the target panels are at rest, because a small (say, 2mm) difference in dimensions would change the input value on an analog scale.

tician
05-13-2013, 06:39 PM
The transponder does not use analog for the target panels. It has the transponder ID DIP switches connected to the ADC pins, and instead uses four PORTD pin change interrupts to detect hits. If the hit is hard enough, the resistance of the FSR drops sufficiently that the output of the voltage divider falls below the upper threshold voltage of a logic input low (usually somewhere around 0~2V for low and 3~5V for high). This is why some people would register hits when firing their weapons (brief induced pulse, but indiscernible from a valid hit).


Was going to wait until after testing, but I could not resist chiming in...
Just got my experimental illuminated half-panel target pcbs about an hour ago, but still need to assemble/program them and then test with the 1.5" FSR's that will not arrive for another day or two. It would be super if it works as I hope, but not much of a loss if it doesn't ($20~$25 each prototype including FSRs; boards are kinda expensive in small quantities). Just need to get an airsoft gun and a thinner piece of lexan (methinks the big sheet in the basement is a bit too thick). Wanted 3mm thick boards for better survivability, but my preferred prototype service doesn't do anything except 1.6mm. If the panel design works at all, and Tybs gives provisional approval as an alternative to the ABS target panels, I'd drop the cash to get 3mm boards made and really abuse those (BB pistol point blank) to make sure the SMD components do not fail (at least the visual feedback should ensure it will be obvious if/when it does).

Single-sided board with all SMD components: one ATtiny13, three 10kOhm resistors, two 62Ohm resistors, two small ceramic decoupling capacitors, four reverse mount LEDs, one 3-pin servo/sensor header, four M2.5 mounting holes (on corners of a 3" x 2" rectangle), and six M2 mounting holes (on corners and midpoints of long edge of a 64mm x 50mm rectangle).

First attempt will continuously poll the signal line and two ADC pins with each FSR in a voltage divider. When it detects a drop below the threshold for some number of consecutive ADC reads, then it pulls the signal line low for ~500ms and flashes the LEDs at 10 Hz (alternating between sides) for ~1000ms. If the transponder (or another target panel) pulls the signal line low, all four LEDs flash together at 5Hz until the signal line is released. If the ADC method does not work for detection, there is always the pin change interrupt method.

jwatte
05-13-2013, 11:23 PM
The ADC is kind-of slow -- at best, you can hope for 10 kHz sampling rate.

An opamp as a comparator to a pre-set value would easily be able to generate the 1-0 transition you can read with a pin-change interrupt. I think that's a fine interface -- the controller detects a hit on falling edge.

Btw: the description of your panel sounds sexy! FR-4 is a lot tougher than ABS, so 1.6mm may actually be pretty good. I imagine you can go to 3.2mm simply by building a four-layer board out of the same material :-)

Gertlex
05-14-2013, 12:53 AM
Tician, is the concern with 1.6mm that the boards flex enough to eventaully causing solder joint failure?

I will note that I like having easily removable score panels (screws don't count!). Additionally, the hole pattern should keep in mind that smaller AX-12 bots have smaller footprints for mounting score panels...

tician
05-14-2013, 02:21 AM
Tician, is the concern with 1.6mm that the boards flex enough to eventaully causing solder joint failure? Pretty much. I can flex them a bit by hand and I'm paranoid, so I will be very mean/destructive in testing (airsoft -> bb pistol -> reciprocating piston/hammer). I was actually sort of thinking of using them as structural components in a small-ish bot (not necessarily for MW, but target panels seemed a nice way to test the durability).


I will note that I like having easily removable score panels (screws don't count!). Additionally, the hole pattern should keep in mind that smaller AX-12 bots have smaller footprints for mounting score panels...
Could always make tiny, tiny aluminum brackets to keep them off the surface of the bot and adapt the screw mounting pattern, or maybe just epoxy a couple small magnets to the bare spots on the component side so the panel can be popped off a complementary magnet set on the bot (maybe hidden behind a body panel?). Although it would still be a bit of extra/unnecessary weight and complexity on a bot with really limited payload. Brain is not terribly functional right now (3:20am with ~2 hours sleep), but I want to say the pcb is 96mm by 58mm and could be shrunken a little bit more.

Gertlex
05-15-2013, 10:05 AM
Could always make tiny, tiny aluminum brackets to keep them off the surface of the bot and adapt the screw mounting pattern, or maybe just epoxy a couple small magnets to the bare spots on the component side so the panel can be popped off a complementary magnet set on the bot (maybe hidden behind a body panel?). Although it would still be a bit of extra/unnecessary weight and complexity on a bot with really limited payload. Brain is not terribly functional right now (3:20am with ~2 hours sleep), but I want to say the pcb is 96mm by 58mm and could be shrunken a little bit more.

I too wasn't fully thinking. I could definitely work with attaching brackets that slide onto my bot.

Regarding size, I was referring tot he hole pattern. If you put the holes at the corners of the score panel, you have maximized the mounting area required



| . |
| |
| | panel
| |
| . |

vs

| |
| . |
| | panel
| . |
| |

/too lazy to do fuller ascii illustration

tician
05-15-2013, 03:14 PM
Basically, it is the minimum length required to accommodate two 1.5" FSR and keep the cables somewhat protected as they wrap around the back of the board and soldered onto very long pads. The width of the pcb is a little larger, but there are 10 mounting holes (6 M2 and 4 M2.5) scattered along the entire length (not sure which will go to lexan, or if just using tape and/or contact adhesive to keep it in place). The M2.5 are on an 2" x 3" spacing. The M2 are on 32mm x 50mm spacing. Testing will show which are required for solid mounting, and I'm kind of hoping none will be required for the lexan (double-sided tape?) giving the maximum number of possible easy mounting points. I will probably add another two pair of M2 screws between the LEDs to give five M2 mounts per edge (need to test).

Crappy ASCII drawing that looks a little better before the forum formats it.


___________________________________________
| O_o_______________o_______________o_O |
| | | |
\ | | /
|| ||
/ | | \
| |____________________________________| |
|_O_o_______________o_______________o_O_|

Gertlex
05-15-2013, 07:49 PM
Yup... 5 M2 holes on each edge would be a satisfactory design for me, from a flexibility of designing standpoint :)

jwatte
05-16-2013, 12:15 AM
I think M2 is extremely tiny. I don't use anything smaller than M3 myself. Then again, perhaps that's just my old man's eyes betraying me... (All boards I make have 1/8" mounting holes, which work perfect with M3 screws)
The combination of small "hardpoint" you affix to the mech, and some kind of slide-ey sticky way of attaching the panels would probably be the most all-around solution, assuming it can be made robust.
If that also did the electrical contact, so the attachment point would also be where the wires run, so I could have neat wire runs on the inside of the mech, that would be perfect!

lnxfergy
05-16-2013, 01:51 AM
The ADC is kind-of slow -- at best, you can hope for 10 kHz sampling rate.

THIS is exactly why we used digital. The first prototypes using FSRs were horribly inaccurate using the ADC. The AVR has only 1 actual ADC, so when you have to keep changing the MUX and waiting for it to settle, the picture is even worse. Now attach a scope to your target panel and shoot it (the panel, not the scope) -- it's almost certain you will miss a number of hits because the change is so momentary (the FSR really only detects the BB for a fraction of a millisecond).

-Fergs

jwatte
05-16-2013, 01:08 PM
the FSR really only detects the BB for a fraction of a millisecond

Right. So, the interface we have to live with is: Digital logic-level pulse from panel to controller, duration > 10 us < 1 ms.
Additionally, there is 5V available to the panels through the existing connector.
There are many ways of turning "panel impact" into "digital pulse" when you already have a 5V input, and a logic-level output.
Thus, it should be totally possible for someone who is interested in doing the necessary work to create really robust detectors/panels to invent here.
There's some question of qualification, and whether anyone has enough time to spend on making sure the hit detect level/robustness is the same as that of the existing panels, but if someone actually makes a robust product that is better than the existing panels in some significant way (price, sensitivity, robustness, etc) then it's probably worth it taking them seriously.
That being said, I don't think it would be reasonable to spend any time on someone who wants to build their own panels only for purposes of saving a few dollars on their own bot -- only alternative panels actually offered for sale would warrant consideration IMO.
Those are my two cents :-)

Gertlex
05-16-2013, 03:43 PM
I think M2 is extremely tiny. I don't use anything smaller than M3 myself. Then again, perhaps that's just my old man's eyes betraying me... (All boards I make have 1/8" mounting holes, which work perfect with M3 screws)

Tiny = less weight :D

But really, I use M2 because that's what AX-12s use everywhere. MX-28's use m2 on the servo horn without a problem (besides... y'know... them coming loose...) so they're fine for mounting score panels.

You just skipped straight to MX-64's, which is the real deception on your end, hehe.

jwatte
05-16-2013, 05:22 PM
To be fair, MX-64's use M2.5 hardware, which are still kind-of small. The bits of my bot that I can control are M3. That's more an outcome of choosing aluminum as the material, and certain tapped holes, rather than the servos themselves. (I don't think aluminum is heavier than plastic for the same rigidity.) Hand tapping < M3 very easily breaks taps :-/

tician
05-18-2013, 10:12 PM
bleh. Took three days to drag myself into the basement to solder up the prototype boards and not only did I trigger a headache, but it appears that I also managed to solder every single LED backwards because I did not double check the data sheet (the silkscreen was too small and barely intelligible, and I interpreted it incorrectly). So, might fix them tomorrow when I feel a bit less like there is a large rodent in my skull trying to gnaw its way out through my eye sockets.

Edit: Scratch that. It appears I did interpret the silkscreen correctly, but I drew it backwards in my library.

tician
05-19-2013, 07:40 PM
Actually, I just cut the common trace of the LEDs and reversed the pin state in software. Originally, it was two LEDs and a resistor in series from +5V to a pin (higher sink current capacity than source), and is now an easier to route (and probably safer) sequence of pin -> resistor -> LED -> LED -> ground. I have one board that appears to be fully functional and programmed, but have not managed to drag myself back to the basement to finish soldering the green wire fixes or place the FSRs for testing under fire. Nor have I dragged myself to the outside world to get an airsoft gun, lexan, and other useful project parts (tennis balls, a Size 3 soccer ball, etc.). The other two prototype boards only have half their LEDs working, so I probably need to do a bit of rework and/or LED replacement on them.

tician
08-24-2013, 09:43 PM
Got two of my new boards soldered and programmed, and all appears well after fixing a very stupid programming error (selected wrong ADC in ADMUX, so only one FSR worked). Need to stop by bigbox or the lab to get some thin polycarbonate sheet, since kapton tape definitely is not going to cut it as protection (nice little dimple where the test BB hit - .25 gram point blank from a Crossman Pulse M74DP at a supposed 175 fps) and the spare pcb appears to be a bit too stiff to register hits (have not tested other sensitivity settings yet).

FRONT (http://forums.trossenrobotics.com/gallery/showimage.php?i=5360&catid=member&imageuser=4806)
BACK (http://forums.trossenrobotics.com/gallery/showimage.php?i=5361&catid=member&imageuser=4806)

ArduTank
08-24-2013, 10:32 PM
Nice.

Xevel
08-25-2013, 07:44 AM
What is the bend radius of the FSR leads? From the picture I wonder if it's not a little bit tight...
These guys (http://akizukidenshi.com/download/ds/interlinkelec/94-00004_Rev_B%20FSR%20Integration%20Guide.pdf)sugges t a minimum bend radius of 2.5mm. I haven't extensively tested it but it seems reasonable to me given the rigidity of the leads.

tician
08-25-2013, 03:14 PM
What is the bend radius of the FSR leads? From the picture I wonder if it's not a little bit tight...
These guys (http://akizukidenshi.com/download/ds/interlinkelec/94-00004_Rev_B FSR Integration Guide.pdf)suggest a minimum bend radius of 2.5mm. I haven't extensively tested it but it seems reasonable to me given the rigidity of the leads.
The leads are not bent so much that they are flat against the back of the board and there is absolutely no twisting or 'turning' of the leads, so there is no obvious sign of damage such as the white lines (stress cracks?) that show up when I know I've damaged the polyester laminate on one of the foot pressure sensor kits for a Bioloid foot. The only thing that actually concerns me at the moment is the bit of spotting at the start of the lead where it is still over the rubber reinforcement because I cannot remember if that is normal or not (think it is just polyester pulling up a bit from the adhesive, but does not reach sensor area).

EDGE-ON 1 (http://forums.trossenrobotics.com/gallery/showimage.php?i=5364&catid=member&imageuser=4806)
EDGE-ON 2 (http://forums.trossenrobotics.com/gallery/showimage.php?i=5365&catid=member&imageuser=4806)
SPOTTING (http://forums.trossenrobotics.com/gallery/showimage.php?i=5366&catid=member&imageuser=4806)

tician
08-26-2013, 10:33 PM
Since the ~3/32" lexan from bigbox is still too thick to get a proper reading except from a rather hard finger press and I do not have immediate access to any thinner (or even know what the regulation thickness is), I went with some of the 1/32" adhesive backed textured neoprene rubber for the new DARwIn-OP feet. Works well enough for the moment and there is no worrying about lexan bending/bulging up an keeping the BBs from registering a hit. There is a very narrow deadzone in the very center where the two FSR meet. Placed the neoprene over a layer of kapton for easy removal from the FSRs to check for damage after it takes a few hundred hits.

Going from the ~16mm red spot and ~2mm welt/blister on my palm, it would seem that the M74DP might fire the BB a bit harder/faster under battery power than in manual mode.

tician
08-27-2013, 12:23 PM
...and I apparently stripped or dislodged a gear barely even half way through the 2000 round bottle. Bright-side: not even the slightest visible damage to the neoprene or the FSR beneath it. Might pop down to the basement to test out a couple rounds in the bb pistol.

jwatte
08-27-2013, 02:00 PM
The plastic Eagle gearbox is pretty prone to breaking. Treat it kindly, don't run it too hot.

I wonder if a vacuum formed plastic with divots pointing into the FSR would work well? The force would be concentrated on the contact area of the divots and thus probably register pretty well, and not be as sensitive to the thickness of the protection.

ArduTank
08-27-2013, 03:14 PM
That sounds like it might. Good idea Jwatte!

tician
08-27-2013, 04:36 PM
The plastic Eagle gearbox is pretty prone to breaking. Treat it kindly, don't run it too hot.
Not sure if it is Eagle as it was designed integral to the gun's shell, but it is kinda crap. Still have manual mode, but that's a pain.


I wonder if a vacuum formed plastic with divots pointing into the FSR would work well? The force would be concentrated on the contact area of the divots and thus probably register pretty well, and not be as sensitive to the thickness of the protection.
I was thinking more along the lines of routing/etching into the polycarbonate where it would normally rest on the separator/spacer of the FSRs. It should permit secure screw connections of the polycarbonate to the target panel with none of the flexing/bulging that would prevent valid hits from registering, but it might make it far more prone to false hits if there is any slippage of the polycarbonate (since it will be in near constant contact with the face of the sensor and probably compressing its traces a bit into the resistive ink inside the FSR). One nice thing about the neoprene is that, even though it is adhered directly to the FSR face, it shows no inclination to register false hits from being jostled, smacked on edge, or even hit on the back of the panel opposite an FSR (only impacts to the front face of the FSR register).

jwatte
08-27-2013, 06:54 PM
Yes, Neoprene actually sounds pretty interesting -- if it can stand the beating, it might be a good alternative to polycarb overall!
Are you using foamed or straight?

tician
08-31-2013, 07:53 PM
Yes, Neoprene actually sounds pretty interesting -- if it can stand the beating, it might be a good alternative to polycarb overall!
Are you using foamed or straight?
Did not notice your edit. It is McMaster item #8445K61 (Medium-strength Textured Neoprene Rubber, Adhesive Back, 1/32" Thk, 12" X 12", 40a Durometer) and cuts easily and cleanly with the cheap serrated medical safety scissors I got from electronics goldmine. I could see it being die cut pretty well.

Quickly tested out the 5 mil UHMW static-dissipative film I got from mcmaster, but it is not the same as the adafruit 150 micron velostat. The 5 mil is actually semi-transparent and the carbon is obviously not uniformly distributed within the UHMW-PE film. The resistance does not tend to be very stable either, even though the crap multimeter may be making that even worse (velostat tested with same multimeter is still quite a bit more stable). Maybe the thicker films are better, but probably not going to bother testing them.

tician
11-15-2013, 05:54 PM
Just got the new boards today, but all my soldering stuff is in the lab. Also have no spare FSR's. I was hoping to buy them with a couple CM-904 and other parts from the trossen shop, but they are not listed yet. As I really have to go back to writing, perhaps it's for the best that I do not have access to yet another distraction.

tician
03-03-2014, 09:32 PM
During further chronocide, I decided to add FSR-ish velostat traces to the illuminated half-panel and succumbed a bit to feature creep. Can use either two 1.5" square FSRs or the 88mm x 44mm velostat trace area with multiple mounting points.

Standard MW Half-Panel configuration:
(2) JST-PH 3-pin SMD vertical mount headers [JST B3B-PH-SM4]
(1) 0603/0805/1206 SMD pull-up resistor (triple footprint, value depends a bit on sensor)

Illuminated Half-Panel configuration (utilizes only the internal 10k pull-up resistors; enough for FSRs, but not sure about velostat):
(1) JST-PH 3-pin SMD vertical mount header [JST B3B-PH-SM4]
(1) ATtiny13 [ATtiny13A-SSU]
(2) 0603/0805/1206 SMD capacitors (one 1uF seems to be enough in tests)
(1) 0603 SMD resistor (10k Ohm pull-up for RESET line)
(2) 0603 SMD resistor (62~200 Ohm for LEDs)
(4) 2.5mm x 2.0mm, Z-bend, reverse mount LED [Kingbright AM2520SURCK09]


The feature currently causing issues is the three JST-PH 3-pin header footprints: two for raw FSR daisy-chain access like the old MW panels and the other for the ATtiny13. Should the headers on the half-panel follow the SHARP pinout (1=Vo, 2=GND, 3=5V) to use the existing crossover cables (http://www.trossenrobotics.com/store/p/5447-GP2D12-Sharp-IR-Sensor-to-Servo-Cable.aspx), or should it follow the servo/sensor/MW pinout (1=Vo/Data, 2=5V, 3=GND) for possible future cables or user crimped cables?

The attached board files use the servo pinout. While this board has not been built or tested, all of the pieces have functioned on past boards (except the never tested 0603/0805/1206 triple footprint). Boards should be less than ~$3 each in quantities of 100, but do not currently have loose cash for OSH-Park prototypes (lost ~$330 on a new tire and ticket last month). Was thinking of adding a bit-bang interface on the data line to enable modification of the ADC threshold (and store it in EEPROM) without a programmer, but never got around to implementing that.

I think I need sleep. Lots of not nice thoughts right now.

edit: Forgot the schematic in case needed (hopefully silkscreen is sufficient).