PDA

View Full Version : Adding force sensitive resistors to the phoenix.



Xan
01-05-2009, 05:10 AM
Hi All,

Iíve spend some hours of my Christmas vacation in the cold garage. Iíve added LMís FSRís (http://www.lynxmotion.com/Product.aspx?productID=627&CategoryID=8) and Iím pretty happy with the result. The rubber cap is just a little bit bigger and all the wires are perfectly hidden in the nylon sleeve.

I think the pictures will explain it all :)
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030194.jpg

Started with a 8mm aluminium rod and some M3 bolts
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030191.jpg
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030195.jpg

The bumper is a little bit bigger then the FSR but thatís a good good thing because it fits perfectly in the rubber cap.
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030196.jpg
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030197.jpg

The cable is strapped behind the leg so you can hardly see it from the front.
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030199.jpg
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030200.jpg

Here is the endresult:
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030204.jpg
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030203.jpg

The wires are still hanging loose as you can see. Iím making a small circuit to convert the analog signals to digital ones. But I havenít got that ready yet.

Xan

Quantum
01-05-2009, 08:41 AM
Im trying to do the same with just a switch. But the problem Im having is debouncing the switch.

How is the software on this going?

Paul

ooops
01-05-2009, 09:36 AM
Very cool stuff. The pictures are great as well. Will you be sharing the code on making them work?
+rep for sharing:)

jes1510
01-05-2009, 09:46 AM
Wow that looks great man. Does the rubber bumpon affect the response of the FSR?

Xan
01-06-2009, 03:26 AM
Hi All,

Thanks for the reactions!


Im trying to do the same with just a switch. But the problem Im having is debouncing the switch. How is the software on this going?
Just a switch should be enough to measure if the feet touches the ground. Canít you add a switch with a spring to return it to the basic position?

I didnít start on the software part yet. I have to improve some other small parts and get V2.0 up and running before Iím gone include the terrain adaption. But I wanted to get away from my desk for some hours ;)


Very cool stuff. The pictures are great as well. Will you be sharing the code on making them work? +rep for sharing:)
Thanks! Yepp, Iím gone share this code as well. Iím planning to release a major update with release 2.0. There is a lot of speed improvement in there. And Iím gone include the terrain adaption in 2.0 and release it in 2.1. You can check the latest status at my topic on lynxmotion (http://www.lynxmotion.net/viewtopic.php?t=4285 )


Wow that looks great man. Does the rubber bumpon affect the response of the FSR?
I didnít had the time to test the affect of the rubber bumper and the working angle. Iím hoping to get some results later on this week. Lets hope for the best ;)

Thanks, Xan

Quantum
01-06-2009, 08:51 AM
Just a switch should be enough to measure if the feet touches the ground. Can’t you add a switch with a spring to return it to the basic position?


The problem is the switch jumps open to close several times and the leg doesn't stop but studders.

The software runs to fast for the switch so I have to make circuit like this:
Values will be different just have to do the math.
http://hades.mech.northwestern.edu/wiki/images/6/61/Debounce_schematic.png (http://hades.mech.northwestern.edu/wiki/index.php/Image:Debounce_schematic.png)
Ill start a thread about this once I got something finished.

Paul

Xan
02-06-2009, 12:08 PM
Hi All,

I had those pictures lying around for a couple of weeks but didnít had time to post them. :oops:

I did some testing with the setup on a breadboard before I found the right resistor value. The value needs to be very precise to let it work. With other values the rubber caps give enough force to prevent the Schmitt trigger from switching. I came out with some 6k8 resistors and they work pretty good.

Iíve made a home brew PCB with the needed components. I miss the time that I went to school and simply could walk to the lab and made my own ďprofessionalĒ PCBísÖ :roll:

http://forums.trossenrobotics.com/gallery/files/2/0/3/7/p1030206ps.jpg
Any suggestion for a company who can make me a real PCB is welcome. ;)

The 9V battery for the logic is moved to the upper deck to create some space for the PCB. Everything is connected now and Iím pretty happy with the result!
http://forums.trossenrobotics.com/gallery/files/2/0/3/7/img_0041ps.jpg
The PCB is placed upside-down where you can see the green wires sticking out.

No update on the programming yet.

Xan

No0bert
02-06-2009, 02:29 PM
Any suggestion for a company who can make me a real PCB is welcomeI have been using these guys since December of 2007, and everytime I order a PCB, if I order it in the morning, I get it that afternoon, and this is just FEDEX! Of course I live in the same state, but still. Professional quality, very speedy, very cheap, etc!

Most of my PCBS are around $15-20

http://www.procyonpcb.com/

Adrenalynn
02-06-2009, 05:56 PM
Tybs is hyper-fond of http://www.goldphoenixpcb.biz

DresnerRobotics
02-06-2009, 06:26 PM
lnxfergy is actually the one who pointed me in their direction. Let's see... $99 for 1000 sq cm, including free shipping. I placed the order on last Friday and its going to be at my doorstep on Monday. That is absolutely outstanding for PCB manufacturing turnaround. Quality is superb (this is the actual company that Sparkfun uses) and you can't beat the price. That said- not a good place to go to for onesie twosie work, but when you consider its usually $15-20 per single board, that adds up quick. I'm getting [25x] 2.5x2.5" boards for $99.


Tybs is hyper-fond of http://www.goldphoenixpcb.biz

Adrenalynn
02-06-2009, 06:30 PM
It's also worth noting that in many many hundreds of thousands of boards printed, I've found a truism:

I'd rather pay a hundred dollars for one good board than $3 for a hundred boards with split traces, de-lam, lifted traces, etc. Not all facilities are created equal.

DresnerRobotics
02-06-2009, 06:49 PM
Yup, which is why I'm usually cautious when picking PCB manufacturers. I've seen one too many horrible batches.

Fergs has had great luck with the previously mentioned PCB company, and well, they're who Sparkfun uses, so I think they're a pretty safe bet.

Adrenalynn
02-06-2009, 08:23 PM
Sorry - I didn't mean to suggest they weren't - was just advising others to consider recommendations in order to take care to not end up with a bucket of poop.

Xan
02-07-2009, 03:15 AM
Hi guys,

Thanks for the tips about the PCBís. I defiantly gone check out the companies and see what the possibilities are! I hope I can also order them in just a few pieces. Iíll keep you posted!

One of them even got ďphoenixĒ in his name! LOL

Thanks!

Xan

Xan
03-27-2009, 04:04 AM
Hi,

It was time to resume this part yesterday. I had to put it on hold because I heated the heat shrink so much that it melted the two wires from the FSR to each other. LOL The defect FSR is replaced with a new one.

Iíve added the first part of the terrain adaption to the code. This already gives it more life! When the phoenix is lifted from the table all legs will stretch out. Once back on the table phoenix will get back in his ďnormalĒ position. (about 3 cm above the table) If I lift up the front of the body, the front 4 legs will give support. Once I release the body, it returns again! 8)

All legs return automatically to the ground, even with one (or more) legs on a object. Happy! 8)

Ok, now the down sideÖ

The FSR method doesnít turn out to be as reliable as hoped. Sometimes the rubber cap gives enough force to leave it triggered even when itís in the air. And the weight isnít always sufficient to trigger the FSR with the weight spread over 6 legs.

I have to think about what the next step will be. The chosen threshold for the shmitt trigger is tested very well and I donít think this will give me anymore. I could get back to the analog inputs but this will cost some valuable processor time. Although this will give me more flexibility to adjust the threshold I doubt this will solve the issues.

Another issue involves in the FSRís havenít got enough mechanical hysteresis. The legs needs to be perfectly on the floor to stop motion. If a leg hovers just above the ground it will be moved down a bit lifting the other legs just of the groundÖ

Iím thinking of replacing the FSRís to mechanical switches. No idea about the design yetÖ

Xan

ScuD
03-27-2009, 12:33 PM
Xan,

if you're going for switches I recommend using detector switches, something like these :
http://industrial.panasonic.com/ww/news_e/nr200604MC002_e/nr200604MC002_e/nr200604MC002_e.jpg

They're tiny, available in a lot of different shapes and sizes, both smt and through-hole, plus they only require a minute amount of force to be triggered.

Regarding the schmitt triggers, it'll prolly be too costly but you could use digital potentiometers to set the tripping point on some comparators. But how would you know when to adjust the sensitivity if the problems arise from the rubber caps..

Xan
03-30-2009, 05:19 AM
Hi Scud,

Thanks for the suggestion! Iím thinking of a different setup with mechanical switches. Those look pretty small. :)

Have you got a link for me?

Thanks, Xan