homemade flex sensor

I've used black antistatic foam as a flex sensor before. The only problem is that each time you bend or unbend it the resistance is slightly different.

James

How do you know when to zero it?

Mitch Berkson

How does this work? Capacitance? Doesn't seem that a sense of resistance would yield much. And how does one attach leads?

-- Gordon Author: Constructing Robot Bases, Robot Builder's Sourcebook, Robot Builder's Bonanza

It changes resistance when compressed. Hooking a piece (1.5"x0.25") up to an Ohm meter by placing 0.5" leads in each end I get about 18k. Compressing the center (with the handle of insulated pliers), or flexing it gives a swing of about 600 Ohms. You can get a larger range by applying pressure over a greater area. When uncompressing/unbending the foam it can take a few seconds for the reading to stabilize as the foam is slow to expand.

The best way I've found to use it is between two pieces of flexible metal (insulated with electrical tape) with each end fixed. It?s a cheap sensor and will degrade over time, but you get what you pay for.

James

You put a foam between two flexible stripes, and thats it? Nice, how do you measure the resistance? With plates on the end or how? Greetings Christian

Check out Forrest Mims' "Sensor Projects"... Basically, one or both ends of the graphite covered acetate is sandwiched in between two pieces of screwed-together copper clad, and leads are soldered to the copper clad.

Works > >

And cheap sensors in disposable applications are a valued resource . I wonder how many commercial sensors began life as hobby inventions like yours ? How do you attach to the foam itself ?

Medical electronics use vast numbers of patient contact sensors . All intended to be very reliable but only having to live a short life. Could not the methods used to make electrosurgery plates and EKG/TENS leads fabricate large common ground sensor arrays ? Price and lifetime again being a juggling act . Now if your 'bot sheds it's sensory layer of "Skin" as a snake does several issues open up . Robotic expenses ?

Quantity pricing and competive economics in the " Skin of the moment" After a glance at cell phone covers 'BotSkins seems more credible eh ? Software to extend the life of degraded skins could be interesting .

One of my battle tank sensors in high school was the water level switch from a discarded washing machine . It's a very low pressure AIR switch . Fender of tank had 3/4 latex tubing from Jerryco on impact edge . Tubing getting pressed raised air pressure in chamber of switch actuating the motor reverse relay . Relay held closed by R/C network . BtW- this was in 1977 . Someday I may unearth those notebooks .

Tossing a concept to the group- HD ribbon cable having antistat foam strips one edge conductor designated as common and the others as scannable - thumbtacked or stapled thru the cable to some flexible flat soft insulating substrate-consider the flat float foam kid's toy sold in "dollar stores " further details left open .

Oren Beck

" Lately I'd be in deep yogurt if my name weren't on my shirts- me looks down and uh... "

I was asking about graphite covering acetate, actually, not the anti-static foam technique. I asked because neither acetate nor graphite are all that compressible, and if the material is between rigid copper cladding as noted in another post, it doesn't sound like much of a flex sensor to me.

Maybe I misunderstood the purpose of the graphite U.

-- Gordon Author: Constructing Robot Bases, Robot Builder's Sourcebook, Robot Builder's Bonanza

James wrote:

If I'm not mistaken, Gordon, the flexure of the U provides a slightly varying length. If you use this as one leg of a Wheatstone bridge or similar circuit, you get the equivalent of a strain gauge. Bending the plastic sheet then reads slightly different resistances on the bridge.

Cheers!

Sir Charles W. Shults III, K. B. B. Xenotech Research

321-206-1840

And somewhere around the time of 06/25/2004 15:37, the world stopped and listened as Gordon McComb contributed the following to humanity:

I've read somewhere that you can take strips of foil mounted on bare phenolic, drawn really tight, and soldered at both ends to make the sensor. Use the same solder pads to attach wires.

Other ideas is to take the plasitc from a gallon mike carton, and with a pencil, lay down some HEAVY traces on it. The more traces that are SERIALLY connected across the axis of the bend, the more of a change you get per unit of movement. Use standard hardware or a pop rivit gun with spade connectors to attach the wires.

Because you are going to get VERY minute readings from these types of sensors, I highly recommend using the wheatstone bridge with a insturmentation amplifier (a fancy op-amp) for the front end. I've done an ASCII wheatstone bridge circuit that I've used.

VCC | | /--o--\ / \ / \ x R1 x R2 x 5K? x 10K? / \ x R3 x R4 x 10K? POT x Sensor / \ o------ + - ------o \ / \ / x R5 x R6 x 5K? x 5K? \ / \ 10K? POT / ---xxxxx--- | R7 | GND

The idea behind the R1 & R3 is to allow adjustment to zero out the bridge when the sensor is not being actuated. This will give an adjustment range of 5K? - 15K?, more than enough for our pourposes. Recomment that you make R3 a multi-turn pot so that the bridge can be finely tweaked because you are going to have some zeroing issues with a single turn pot.

R2 and R4 is the sensor side of the bridge. R2 is to bring the resistance of the sensor up to a managable level and to limit the current flow.

R5 and R6 are the bottom ends of the two halves.

Some people are going to say that you don't need R7. If R5 and R6 are presision resistors, then they are right. Otherwise, throw in the R7 as shown. The logic behind this is to balance out the bridge so that the current flow in the bottom half between the two legs are equal. The 5K values are used because when the 10K pot is centered, both ends to the wiper will measure 5K or so, which will make the leg 10K.

Oh my goodness, that's an old thread you've resurrected :) Let's see if we can put it to rest in peace with some good answers.

One of the most common DIY techniques for measuring physical force is the use of antistatic foam, which has a high resistance that gets lower when it's compressed.

Cheap and available, but not without problems. It's not very elastic, so when the pressure is released, it doesn't quite return to the original shape, or resistance. This makes accurate measurement impossible. Furthermore, with repeated manipulation, it will eventually break down. I seem to remember that this is because most types are made with a conductive coating which slowly flakes off.

There's a far better material that works on the same principle, but strangely, it doesn't get mentioned often. Carbon-filled conductive rubber lowers resistance if compressed, and raises resistance if stretched. It's much more elastic than conductive foam and can withstand higher repetitive forces, returning to its original shape -- and very nearly the same resistance -- each time.

Conductive rubber is commonly used for skin electrodes, like for TENS units. You can find TENS electrodes on eBay fairly cheap. Some are more "hackable" than others, look for ones with a sufficiently large flat area that can be cut off and incorporated into your own sensors. Note that some are backed with a non-conductive rubber on the non-skin side, usually in a different color, and these are generally less useful for our purposes.

Connection from the TENS machine to the electrode is done via a cable terminated in a metal pin, which is inserted into a hole in the electrode and held by friction. But for sensor use, we need more secure electrical connections. The simplest way is to just sandwich the rubber between two metal parts, such as strips of copper-plated circuit board, under pressure.

You can also make electrical connections through the rubber with pop rivets and spacers. Spacers are just a particular type of washer, made so the hole closely matches the diameter of the rivet. Solder a wire, or a spade/ring terminal, to a spacer (make sure it's not a stainless steel spacer, or this will be impossible). Punch or cut a hole in the conductive rubber a little smaller than the rivet diameter (there are tools sold in craft stores, made for punching holes in leather, that work great for this). Slide the spacer with the wire attached onto the rivet, push that through the hole in the rubber, and put another spacer on the other side. Apply pressure to the assembly from both sides while popping the rivet with a rivet tool to get a good solid connection.

The only way to attach things to conductive rubber is via mechanical means. Epoxy, silicone, and super glue don't stick, even if you rough up the surface. It cannot be heat fused; instead of melting, it burns. As far as I know, it cannot be solvent welded either -- I've tried alcohol, gasoline, acetone, toluene, xylene, and a few more exotic ones I can't remember the names of.

Plain conductive rubber sheets can also be found. About ten years ago, I purchased a 12" x 9" x 1.2mm sheet from a neuromedical supply company for $60. It goes a long way -- I still have about half of it, despite numerous experiments and projects. The company I bought it from seems to be gone now, but I'm sure there are others.

Hope this helps, Chris

The reason why it's not mentioned often is because it's usually not proportional. Even with amplification, the best response you can get from conductive rubber is basic on/off sensing, which can be done far more reliably with traditional push buttons. Conductive foam is better at proportional sensing, but as you mentioned, it has other undesirable features.

I once happended upon the company Xilor

They sell both sheets of conductive rubber as well as a conductive liquid rubber. The liquid is especially interesting, as they directly advertise it as a means of bonding bits of metal to the rubber sheets, thus forming solder tabs. The sheets are fairly inexpensive, but the liquid rubber will set you back at least $50. Also, as mentioned in the datasheets, since the rubber is intended for use in cheap key matrixes, the response has a fairly steep transition, making proportional force-sensing difficult. I haven't bought anything myself, but they're one of the few manufacturers I've found willing to sell to customers not buying in bulk.

Dont know if this would work but here goes..... If you took 2 rectangles of tin foil/flexible conductor, put a thin flexible insulator between the two, would this create a capacitor ? If so when the "plates" were drawn apart the capacitance should change. If this worked you should be able to sense than change, and read a flex in you sensor with carfull placement of the plates.

My electronic knowlege is not good enough to calculate the size of plate needed etc.

Would this work ?

Well the reason I ask is that strain gauges are still pretty expensive, but a graphite U on a piece of plastic could be made for pennies. I don't imagine there is a strain gauge cartel working to prevent cheap versions of the product, so I wonder if anyone has specifically tried this, and how accurate/useful the results are.

-- Gordon Author: Constructing Robot Bases, Robot Builder's Sourcebook, Robot Builder's Bonanza

That is because these are not accurate; nor do they remain very useful after many flexings. Great for experimental work, but that's about it. Now, as for a better version, you can etch a very convoluted circuit path on flex circuit board and get the same effect and also retain pretty good accuracy for years. This is my favorite method to use. Use a thin line, try to make it at least 20 cm or so in total length, and coat with another layer of Kapton or clear coat to protect it.

Cheers!

Sir Charles W. Shults III, K. B. B. Xenotech Research

321-206-1840

You are correct that conductive rubber is not proportional -- at least not linearly. But way off on the "best response you can get...is basic on/off sensing" part.

If you want to quantify force, not just detect a threshold, use a look-up table. Apply known weights to your DIY sensor, and take a reading at each weight. Add these readings to your micro. Values inbetween can easily be linearly interpolated, and accuracy increases with the number of data points. If you have the processor power, you can even get fancy and curve fit an equation to the data using linear regression. Each sensor will be a little different, so repeat for each if you want highest possible accuracy.

If you want a perfectly-defined force/resistance relationship, and near-identical specs for each sensor, then you have to go commercial. Otherwise, conductive rubber is in my opinion the best DIY option if you want a durable and repeatable sensor.

Thanks, that's a great resource and I've added it to my bookmarks. Never tried the liquid conductive rubber myself, as the only suppliers I previously found had huge minimum orders. I even tried to make my own conductive rubber once, and though some of the results had interesting properties, conductivity was not one of them :)

Chris

Airmar

supposedly sells piezo-electric based sensors. I can't find where they document it directly (they have a datasheet but it's pretty unspecific) but I believe they mention their PDVF compound could be used for tactile and strain sensing at a "low cost".

Kynar is okay for bend sensors, except they are somewhat noisy and they only register change, rather than flat DC, which is what a lot of people look for in a "strain gauge."

I've used them quite successfully for bumpers, where I was really only interested in some sudden change. As an example, one of the books below (RBB) shows a project using two strips of PVDF taped onto the back of some plastic report cover material. Made into a "bow" that extends in front of the robot the thing senses any deflection of the plastic. There is no attempt at determining the amount of force, though.

Measurement Specialties

has some nice technical info plus example circuits. You can buy the stuff through Thomas Register, though with a $100 minimum. A couple of outfits, like ImagesCo, resells it with a markup.

-- Gordon Author: Constructing Robot Bases, Robot Builder's Sourcebook, Robot Builder's Bonanza