I’ve hijacked my own thread!
PCT discussion over here: lynxmotion.net/viewtopic.php?t=191
ok, Jim got me thinking about this again. I dug around in the spare parts box and found what might be a perfect solution.
Hopefully someone can identify the buttons and maybe even a source.
Here are some ridiculously large images of what I’ve got:
I reposted here…
Those are the same ones I’m planning to use for the SES tubing foot pressure sensor. Still waiting for my samples but it looks promissing. Here is an image of what I’m doing, although the Interlink one isn’t in the picture. The idea is to affix the sensor to the aluminum hub using it’s adhesive backing, then the adhesive backed rubber bumper goes on top of the sensor, then the rubber foot slides over the whole thing.
So taking a page from Brooks, I’m going to try to solve my changing values when the bot steps by recalibrating every step. The way I see it, I know when the foot is in the air and can zero-my-altimeter when I know it’s off the ground each time I pick a leg up.
So I swung by home depot the other day looking for something that would function like Matt’s solution above. I failed. So instead I took home some tin snips, a seamer, and a sheet of aluminium.
The sheet on the bottom provides a plate on which the housing above can slide. The housing above has a short channel cut where the screw is mounted that allows it to slide along that base plate.
The sensor is affixed to the end of the lexan foot. The rubber pad at the end of the foot is attached to another small piece bent at 90 degrees. On the inside is another small piece of a rubber that presses against the sensor.
My hands are a mess and there are aluminium shavings embedded in the pads of every finger. Hopefully this version stands up to the miles.
They work pretty well. Most reset to zero every time the foot is lifted. I think the ones that do not will reset properly after a bit of filing. My fabrication skills are pretty pathetic so there’s little consistency among them.
Very odd picture, Jim.
Unless, of course, you meant to make those feet miniature flashlights.
I don’t suppose that you’ve got any plans for adding a similar bracket to the SES?
Speaking of that, I need to figure out how I’m going to attach my FSR’s to my biped.
(I’d rather not revert to glue.)
Any ideas?
me != Jim
At least I think you were talking about my flashlight looking feet. 
I did find in the plumbing department what look now to be the same copper connectors Matt used, but couldn’t find the little rubber balls to fit well inside them. My other concern was on how to attach the copper tube to the leg without compromising the ability of the ball to move freely. Is that heat shrink?
Ugh!
Sorry, Andy.
I guess that I should look at the name next to the posts.
All parties, please ignore my above post.
I’m not a robot guy (yet!) by any stretch, nor am I much of a software guy (yet!), but for some reason I’ve been giving some thought to this particular problem and I’ve got an idea that might be a solution. When I get home tonight and have the time to go into more detail I’ll try spelling it out as best I can.
The short version theory involves some variable resistors and spring loaded feet. I’ve seen variable resistors that are long and rectangular instead of round. These had a little tab that sticks out the side, which when moved changes the resistance.
Suppose you mount the resistor inside a hollow leg and let the “tab” protrude through a slot in the leg. Next you slip a sleeve over your hollow leg and have it spring loaded. This sleeve would capture the tab in some fashion. When the weight of the bot is placed on this foot (sleeve) the spring compresses somewhat and consequently slides the tab, changing the resistance. It seems this would be a decent solution in that as one or more feet are lifted, down pressure on the rest of the feet increases and resistance/pressure changes throughout the system of legs. Wouldn’t this give you more than a “down or not down” reading for your legs, and help eliminate some of the problems with sensors being either too sensitive or not sensitive enough?
A little bit of spring tuning and it seems you would have nice viable system for foot position/pressure feedback.
Of course, I don’t know squat about robotics or software problems so I might be sounding like a complete idiot here. Don’t be afraid to say so.
I have revisited my foot sensor idea, and it now works reliably!
The original idea was to use the Interlink 0.20" FSR. It has adhesive on one side. It is attached to the flat of the HUB-09, a rubber bumper goes on top, and a rubber end cap goes over the whole thing. The original try used a 0.50" ID end cap on the 0.50" OD tubing. It would work, sorta, but it wouldn’t zero out reliably when the foot was lifted. I just received some rubber end caps with a 0.563" ID to try. I used a 4-40 button head screw to hold the tubing onto the HUB-09. There is enough friction to hold the rubber end cap onto the end of the leg, but due to the slightly larger ID it no longer has the issue of not zeroing out when lifted. It starts to register at about 150kohm with as little as 1.6 oz. of pressure. The sensor is has a range of about 5lbs I think. The pressure can be at an angle as well. The rubber end cap protects the FSR and the leg still retains it’s rugged properties. 
Note: Image does not include the Interlink FSR.
That’s great. Will you be adding the 0.53" end-caps to the catalog?
I’ll post some pictures of my solution to this problem soon. Basically I used the existing alumininum tubing for the leg, a spring, magnet, and a linear hall effect sensor. Think of an inverted ball point click pen. I am getting consistent voltage readings from the HES and am accurately predicting how much force/weight is on a given leg at a particular point in time.
I am considering making a batch of these for those who do not have machining tools at home.
Mark K
Mark,
I’d be interested to see what you have got working. I’ve been experimenting with a conductive rubber solution, but I am open to other approaches. My expertise is mostly on the software end.
WalkThisWay,
Here’s a pic of what I have put together. The components on the right top is the hall effect sensor. And below that is a high intensity magnet
http://www.stelk.net/DSC01191.JPG
I’m down! Depending on the cost, I think I’d be very interested in getting in on that. Seems a very elegant and robust solution.
Stalk
I’m very interested in your solution. Let me know what you need.
That looks nice! What are the interface requirements and what would it cost?
I too have interest. This is a well thought out solution. How exactly are the internal components arranged? - a simple sketch would suffice. Thanks in advance
Chris
Thanks,
I still have a couple of kinks to work out first. Such as tightening up the tolerances of the prototypes and trying some different springs to vary the air gap.
Here’s the datasheet for the HES: melexis.com/prodfiles/000475 … Rev007.pdf
Here’s a good document explaining the operation of the device: melexis.com/prodfiles/0003715_hallapps.pdf
And here’s a screen capture from wildfire showing the components internal structure:
http://www.stelk.net/crosssection.JPG
I also need to work on simplifying the structure as right now it would run about $40 / per to manufacture. Almost half of which is the HES since I do not have the Melexis programmer and have not invested the time to design my own.
As far as to the interface I simply connect the wires to a MAX 127 ADC and measure the voltage drop as the air gap varies. The HES is programmed at 100MV/MT @ 2.5V
Let me know if you have any questions
Thanks
Mark Klunder
I failed to mention that the Max127 has an I2C Interface.
Which connects to my Gumstix “Brain” Like Andy my bot is rather “dumb” in that I controll it over an 802.11 connection.
Mark K