TWiki > Rise Web>TWikiUsers > GregKatz>GregsSummerBlog (29 Aug 2008, GregKatz? )
-- GregKatz? - 07 Jul 2008

Week1: June 25 - 27


Week2: June 30 - July 5


Week 3: July 7 - 11


Week 4: July 14 - 20


Week 5: July 21 - 27


Summary of the week:

Embedded Spoons

Worked on cutting out a plastic sheet of many embedded spoons for helping with the problem of distributing a load to many small patches of microwedges. After calibrating the vinyl cutter, I successfully made a decent prototype with some ABS. It seems doable to form these more precisely by thermoforming. The most challenging aspect is being very controlled about the whole process because microwedges are extremely sensitive to alignment and it is very difficult to get all of the spoons exactly aligned the same way. The prototype is, however, a decent proof of concept.

Electrically Conductive PDMS

This week I discovered that the commercially available electrically conductive pdms cannot be used for making microwedges because it is too granular and can't be used to make DPS stalks because it doesn't cure in the presence of vytaflex. Although this was disappointing, I was encouraged to find that carbon doped sylgard is not too granular to make microwedge-size features. I am hopeful that conductive microwedges can be made with this technique.

Electroadhesive Force Testing

I designed some frames for testing the electroadhesive clamping forces. I had to laser cam some acrylic as a frame and then stretch a thin film over the frame. It is tricky to get something that will allow us to measure forces accurately and also have the ability to gimble and conform to a surface which is what the electroadhesion is intended to do.

Week 6: July 28 - Aug 3


Summary of the week:

Conductive Microwedges

I spent much of the week experimenting with the right technique and concentrations for a mixture of carbon, hexane, and sylgard. I learned that:
  • carbon is responsible for the bubbly roughness not the hexanes
  • the ultrasonic vibrator did not help reduce the bubbles
  • hexanes are fine as long is it not more than about 30%
  • at a "normal" thickness, about 3% carbon is necessary for the multimeter to register the material as conductive (about 10 Mohms depending on thickness)
  • even with lots of hexane the mixture is not self-leveling
  • spinning even the thinnest mixture does not make a smooth surface
  • as long as the sylgard makes up about 80% of the total mass, and the wafer is clean at the start, the wedges come out quite well

Hopefully the thinky (electric two-axis mixer) will help make the mixtures smoother with little or no hexane and higher concentrations of carbon.

New Arm Design for Adhesion Test Set-up

Dan suggested that a new arm needs to be manufactured for the adhesion test set-up and since I haven't been able to that much designing and machining, I like the idea of working on it. I would love to be able to make something that can be useful for some time to come. I made a solidworks drawing of my basic idea and researched methods for low-profile fixturing. It would be great if the test surface could be easily swapped instead of the current double stick tape method. I am trying to make the arm as versatile as possible so that any size or type of test surface can quickly, easily, and securely be fixtured to the arm without the fixturing getting in the way of the test.

Force Testing

I drilled and tapped holes in a thick acrylic plate to be used as an insulator for the force sensor and on Friday Noe and I double stick taped my latest frame on to that plate for testing. We saw some normal force but decided that we would be much better off with a smaller glass test substrate that only covers the area we want to test. It was a good start but the set-up needs more work to do full reliable tests. We made plans for making a new arm.

Microwedge Daughter Molds

Aaron and I have been playing with daughter molds of the microwedge wafers. Our goals are to:

  • make a cheap and/or durable mold so that the expensive su-8 wafers do not have to be worn down
  • make polyurethane compatible molds
We have had success with two materials, wax and clear cubes, that make nice wedges and are compatible with both polyurethane and silicone. The weakness is that neither can be heated in the oven to increase cure times. We tried to make super molds with all four quadrants made of good wedges, but that didn't work very well because of problems at the seams, so we went back to the regular molds with only one quarter good wedges. They work fine with silicone, but we ran out of polyurethane to test with.

Week 7: Aug 4 - Aug 10

Daily bullet points: WeekSeven

Summary of Issues for the Week:

New Arm Design for Adhesion Test Set-up

Solidworks model: clamparmscreenshot.bmp

I continued work on designs I started last week for making a new testing arm for the force sensing set-up. The design goals are to make the arm:

  • stiffer by making it shorter and making it out of one piece of solid aluminum.
  • correct-sizing by making the arm smaller to match the size of samples we usually use, but still capable of testing large surfaces if the need arises.
  • modular by using clamps to hold the test substrate so that the test surface can easily be changed AND by making a swappable, easy to produce, end-piece for even more freedom to change the test surface
  • long-lasting by making the end-piece easily adapted to a future two-axis set-up as well as the current three-axis one

Electroadhesion on Stickybot

When the electeoadhesion is ready to be used on stickybot, there will be a couple of issues that I am trying to prepare for.

  • high-voltage switching The easiest solution is to do all switching at low voltage and use a distinct high voltage dc to dc converter for each foot. This is not a good solution down the road because each converter is a cost in weight, real-estate, and cash. Another option is to do high-voltage switching. It is difficult to find previous work on this topic. The best option at the moment seems to be optocouplers, also known as opto-isolators, or photocouplers. They are a pretty simple device that is usually used to isolate a signal from the rest of the circuit by passing a signal to an isolated circuit from an LED to a phototransistor. Most other switches break down at the high voltages we want to use.
  • control The easiest solution is to use the PWM signal originally intended for the toe curling servos and interpret curled as electroadhesion off and uncurled as electroadhesion on. This should work fine by simply conditioning the signal with a low-pass filter to convert the PWM to DC and then using a comparator to convert the signal to a simple high or low. With an oscilloscope I tested the signal that going to the servos and found that the PWM has 20ms between peaks and has a pulse width of either 1.7ms or 1.1ms. A low-pass filter and comparator should easily be able to convert that signal to a high and low. In the future the better solution would be to make four new digital outputs from the PIC and write new separate code specific to this purpose.

Electroadhesion Force Testing

Making an accurate test for the electro and microwedge combined adhesion forces is a challenge. Partially resolved issues:

  • normal-direction flexibility We need to mount the sample on a frame so that it can gimble in the normal direction and self-conform to the surface, thereby engaging all the microwedges. However, this means we cannot accurately pre-load the system because loading the sample means pushing down the flexible frame mounting. Current strategy is to load the sample by eye and just measure pull-off force.
  • sample size and repeatability Since each sample must be mounted on a frame it is not as easy to test a variety of samples. I have helped with this by making tapped holes in the frames so that is easy to remove a frame and fixture a new one. Still it takes time to make a new frame for each sample and there is no way to guarantee the mylar strips that suspend the sample over the frame are always the same.
Resolved issues:
  • substrate sizing The oversized glass on the old test arm was a problem because it would bump into the mylar or frame of the set-up and we would really like a test substrate that fits just over the electrodes we are trying to test. This is has been fixed because I made a new arm.
  • high-voltage insulation I machined holes in a thick acrylic plate to keep the force plate insulated from the high voltage.

Conductive Silicone Material Testing

Currently the major concern is that none of the carbon doped silicone samples seem to offer anywhere near the adhesive force that aluminum offers. Previously I had been using resistance measured by a portable multimeter to get a rough idea of how much "conductivity" was necessary in a material, but this has proved to be far too rough a test. After making some crude "cantilevered" tests where I isolated the conductive material from the aluminum and conductive epoxies I use as leads, I saw for the first time that the conductive backings were almost exclusively responsible for any visual indication of electroadhesion clamping. Even at 10% carbon, which is the highest concentration I used, the sample performed poorly at the visual test of making a piece of mylar move towards a glass slide. We have decided to formalize the process by making a polyurethane cantilever mold in order to standardize the sample geometry. We will be testing the following:
  • sylgard doped with carbon at various percentages
  • elastosil, a commercially available carbon doped silicone which has shown lower resistance than anything I have prepared
  • tap blue with carbon
  • sylgard dipped in conductive silicone ink
  • carbon and silicone oil spray
  • aluminumized mylar

At this point, I am fairly confident that I can make good wedges out of something at least moderately conductive, however, it seems that an aluminum backing will be extremely beneficial and a conductive silicone would act more as a supplementary member of the electroadhesion clamping force.

Week 8: Aug 11 - Aug 17

Daily bullet points: WeekEight

Summary of work for the week:

Electroadhesion Force Testing

  • Finished making parts for better testing on force plate set-up.
  • Got data for the first time comparing adhesive forces of electrostatic alone and electrostatic behind a microwedge path. The hybrid held .060 newtons in pure normal and .060 newtons in pure sheer as well. Without the patches not as much sheer force was held.

Conductive Material Testing

  • Using cantilever test, I discovered that 1%, 3%, and 5% carbon doped sylgard samples were all conductive enough to move a mylar film on a frame at a distance of 1/8". It seemed that 1% was not quite as strong but 3% and 5% seemed the same. This was the first time that I was able to see doped pdms doing affective electrostatic clamping without the influence of any backing layer.
  • Also tested conductive ink dipped and elastosil cantilevers and they didn't seem to offer any huge improvement over the 3% carbon doped sylgard which makes the best microwedges.

Conductive Microwedge Fabrication

  • Using transparencies and heavy weights I was able to make a nice flat, smooth, thin backing layer with the 3% carbon doped sylgard that is too thick to spin. Unfortunately there are some small roundish imperfections that might be bubbles that show up in the areas with good wedges. In these areas microwedges don't form and also leaves the mold dirty. Around these imperfections the microwedges look very good.

Week 9: Aug 19 - Aug 25

Daily bullet points: WeekNine

Summary of Week:

Electroadhesion Force Testing

  • Laser CAMMed 7 identical frames that can be swapped on the force plate with four screws and started making identical aluminized mylar sheets to test various scenarios on the force sensor, but haven't had a change to test yet.

Conductive Microwedge Fabrication

  • Finally got first two successful conductive castings, though one has tissue as a backing and the second is too thick. I think the main problem before was that too much weight was affecting the cure. Also, realized that if you degas after spreading on the mold, though mroe bubbles come out, these can be flattened smoothly with just a small force.

Electroadhesion on Stickybot

  • Experimenting with Laurel's ball joint foot and some quick prototyping with a sheet of mylar on a frame, Sanjay and I agree that electroadhesion works best on a super flexible film. Although electroadhesion could help on a rigid backing with freedom to rotate and move as a single solid piece, electroadhesion really is most effective at conforming a flexible film to a surface and making it cling tightly. Using microwedges attached to a conductive film would probably work well to really pull the microwedges up close to a surface. Notably, this would work well on rougher surfaces like wood and drywall perhaps. On a frame with squeegies on alumized mylar, we were easily holding 50g in sheer on glass and the wooden door with considerable normal force as well.

Week 10: Aug 25 - Aug 29

Daily bullet points: WeekTen

Summary of the week: Got force data for microwedges and flat pdms on flexible conductive backing. Made poster for end of quarter poster session. Finalized process for casting conductive microwedges.

Instructions for making conductive microwedges:

  1. Clean wafer by blowing compressed air over empty wafer and degas empty wafer for several minutes.
  2. Combine 30g of Sylgard 170 with 1g of carbon powder in a cup. Mix until there is no loose powder.
  3. Transfer to the plastic cup made for the thinky.
  4. Mix in thinky for 2 minute and defoam in thinky for 1 minute.
  5. Use popsicle stick to spread a thin layer of the mixture onto the wafer as smoothly as possible, try not to introduce air bubbles
  6. Degas for 30 seconds, let the pressure return to normal, then degas again. Many bubbles large and small will develop and not necessarily pop themselves, but the second degas should have less bubbles and once the pressure is returned to normal most of the bubbles should settle down.
  7. Sandwich the wafer between two transparencies and push down at first by hand, squeezing excess pdms off the wafer and making it roughly flat.
  8. Put something flat over the transparency on top of the pdms and rest a small amount of weight on top of that. Do not use more than 2 pounds and do not leave the weight on for longer than two minutes. This should be enough to get a decently flat backing layer about half a millimeter thick. More weight or more time can cause the pdms to get stuck in the mold.
  9. Remove the weight after just a minute or two and let cure with the top transparency left on.
  10. After at least 12 hours of curing the top transparency should peel off easily leaving a nice flat backing layer
  11. Allow it two cure for another 12 hours (or less if it cured longer than 12 hours in step 9, it should cure for at least 24 hours total).
  12. Do not use any heat to cure.
  13. Peel from mold slowly and carefully, the carbon reduces strength of the pdms.

This site is powered by the TWiki collaboration platformCopyright &© by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback