PDA

View Full Version : mjbots piezo target panels



jpieper
06-04-2015, 07:11 AM
I realized I haven't actually posted anything specific about the target panels we ended up building for the Boston Mech Warfare expo, so here we are.

We had a piezo design sitting on the shelf for a year or so, which we hadn't finished because of the negative experiences people had here with piezo panels in the past. After seeing RTEAM's mechanical design, we were inspired and finished them off.

First, some pictures of the panel(s) we used for the event:

5987598859895990

Features:


Very hard to trigger with anything but a BB hit or a sharp blow from a hard object, but very reliably registers BBs (and a configurable amount of ricochets too)
Actively driven, so very immune to electrical noise
Compatible with all existing transponders
Full sized and half-sized variants
Emits a digital debugging scope trace with every hit, all you need is a 5V TTL USB dongle to see it
ATTiny85 based, so easily tunable
Relatively cheap: If you make enough for only one mech and outsource nearly everything, including laser-cutting to ponoko, and board assembly to macrofab, the total BOM is $13 a panel, or $52 for a set of four full-sized panels.

Design files/BOM and firmware source:

https://github.com/mjbots/mjscore

Once I get an updated revision of boards and panels back, I'm going to send out some samples to people and teams who are interested in trying them out. Let me know if that might be you!

byi
08-11-2015, 02:33 PM
I have disappeared into labwork for the summer, so I never got a chance to mention it, but damn those are good looking panels.

artans
08-23-2015, 10:22 PM
I got one of your target panels form GhengisDon and tried it during our monthly mech scrimmage. It worked great.

jpieper
09-13-2015, 09:11 PM
It worked great.

"worked great" as in it registered hits when it should have, or just that it provided you with an invincibility shield by never registering anything? :)

I'm curious if it had more false positives than the RTEAM panels? In our testing at the Boston Mech Expo, it seemed to register more ricocheted hits than the FSR panels that jwatte had.

jwatte
09-14-2015, 12:22 PM
In our testing at the Boston Mech Expo, it seemed to register more ricocheted hits than the FSR panels that jwatte had.

I can second that. I think the sensitivity could be turned down a bit and it would still be fair.

jpieper
09-10-2016, 04:32 PM
I've been wanting to be able to verify the performance of target panels for some time, and finally built up a test fixture. Using the pressure regulator on an air compressor it can fire variable speed between 15fps and 150fps and lets you position the panel at various offsets then hit it over and over again at exactly the same spot.

667666776678

As we expected, the mjbots panels are very sensitive with the constants from Robogames 2016. In fact, too sensitive for my fixture. Even at 15fps (which is hard to reliably achieve with this setup), they registered at least 80% of hits across the entire panel. At 30fps it was somewhere between 90-100%. Hitting the screw itself seemed to be significantly less sensitive (something like 0-10% at 15fps), but even 0.5cm away had the same high sensitivity as elsewhere.

I'm curious if anyone would be willing to donate one of the old FSR based panels to the cause so that I could produce a comparison chart. An R-TEAM piezo would be awesome as well if any spare are to be had. My thought is to, if not produce a standard panel, at least document how a standard panel should perform.

Now a little bit about the construction: I started with a CO2 non-blowback airsoft gun (http://www.evike.com/products/53850/), then disassembled and removed the pressure regulator and drilled and tapped a 1/8" NPT thread in the back. The CO2 cartridge in there now is just an empty one to keep air from going out held in place with hardware from an earlier attempt at air conversion. The hopper is just two shapeways 3D printed parts. There is a solenoid for sending air through, connected with a quick connect fitting screwed in, although operating the fixture is still a manual process as you have to pull the trigger to queue up a BB. There is a chronograph mounted inline to measure velocity and a Teensy 2.0+ connected via USB to a laptop to record hits on the panel. The panel mount is just two drawer slides with glides mounted to 2x6's, with the vertical guide undersized so that it can hold position.

jwatte
09-10-2016, 09:22 PM
That's awesome!

Email me your mailing address; I have spare FSR panels I can donate.

GhengisDhon
09-11-2016, 09:09 AM
Very interesting. One of our club members (giantflaw) recently did essentially the same thing looking at the RTeam piezo panels, the FSR panels, as well as a mechanical switch based panel. He didn't look at the Boston piezo variant as I lent all the ones you sent me out and didn't have any available.

He also looked at mounting multiple piezos on the same plate (for redundancy), different methods of mounting the piezos on the plate, as well as different methods of mounting the plates to the Mech.

He wrote up a very nice report that I will either post to github or see if i can make a post out of it.

GhengisDhon
09-11-2016, 05:41 PM
Full report to be found here:

http://forums.trossenrobotics.com/showthread.php?8079-Piezo-Target-Plate-Testing-Results&p=72296#post72296

jwatte
03-27-2017, 05:02 PM
Woo-hoo, old thread! I'm trying to find the R-team design files for the target panel, and found this thread.
Looking at it, ATTiny85 seems ideally suited for a panel design. It has an analog comparator. It runs on any power.
All you need is a Piezo element, a Zener to avoid negative and over-swing, a potentiometer for trimming sensitivity, a resistor and a couple of LEDs for feedback, and the connector out to the scoring board.