Tuesday, April 30, 2013

A Warped Idea

One of the issues with trying to roll my own fiberglass rube to replace the metal in the lower cage to get improved stiffness without a lot more weight is the need to get 20" diameter molds.  Car wheels would be expensive, but assuming that I did not get anything on them (since you wrap a plastic sheet around them before you start glopping on the epoxy), they would be resellable afterwards.  And there are plenty of used, cosmetically challenged but still round 20" wheels out there, I think.

What happens if you take an existing aluminum tube, and add a layer of fiberglass and resin?  The resin alone would dramatically stiffen it, and the fiberglass would make it strong as well.  Or will thermal expansion coefficient differences between the aluminum and fiberglass cause it to crack the fiberglass layer?

What is attractive about this warped idea:

1. No need for 20" molds.

2. It covers over all the existing holes, although the one that are still needed could be recreated by drilling from the inside, or by putting something in them to prevent them from being covered over by the fiberglass.

3. Minimal materials costs, since I don't need to make an entire 20" ID fiberglass tube -- just improve an existing piece of aluminum with what would likely be one layer of fiberglass cloth and resin.  The result would be far stiffer than the current aluminum, and would not add enormous weight.

If this seems strange, there is a long practice in amateur telescope making of epoxying cardboard concrete forms tubing to get the cheapness of Sonotube with the improved moisture resistance and improved looks of fiberglass.  Any thoughts?

If you are wondering about lack of interest in politics and current affairs at the moment: I can only focus like a laser beam on one intense project at a time.  And the overwhelming lack of interest by magazines on this topic that should be interested in my research on the ineffectiveness of background checks for reducing murder rates has left me feeling like the effort in that area is a bit wasted.

UPDATE: I meant to look this up this morning, but a reader's comment pushed me to do so.  Aluminum has a thermal coefficient of expansion of 22.2 x 10-6 m/m/K; unfilled epoxy is 55, or more than twice as much expansion.  If I understand the units correctly, this means that a meter of aluminum will expand .00000222 meters for every degree Kelvin (or Celsius) of temperature increase.  The normal range of temperatures that this part will be exposed to is about 70 degrees Fahrenheit, or about 40 degrees Celsius; a part that is .0023 meters thick will therefore expand by billionths of a meter, or 0.0000080409288 inches.  Even with repeated cycles -- I am having a hard time imagining this alone would play any significant role in delamination.  There are doubtless other problems behind the delamination problem that commenter below describes.

I am going to experiment with some Coke cans this evening to see how what a layer of fiberglass does to the stiffness of what is universally recognized as a very flexible piece of aluminum.  This is also a remarkably smooth aluminum surface, much less likely to accept epoxy than the tube that I would need to actually turn into an aluminum fiberglass composite.  If it doesn't just turn into a mold for the tube, the combination will give me confidence that my strategy can work.


  1. It seems in boat and aircraft building the problem with bonding fiberglass and aluminum is not so much the expansion as that over time you find that the aluminum oxidizes, and the bond fails.

    Seems to me that a sandwich/laminate of another aluminum sheet would be a possibility. There is some literature on bonding Alu to Alu, and if you have concentric circles with through-fastenings for brackets, the two parts are going to remain stiffer and together with or without successful bonding between them.

  2. I am a bit confused by this: aluminum oxidizes almost immediately. I would think the problem is more related to failure of the fiberglass to make a permanent bond to the aluminum. Temperature differences over time might be the cause of continuing separation. Epoxy composites have more than twice the thermal coefficient of expansion of aluminum.

    One other possibility, if delamination starts to happen a year or two down the line, is to remove the aluminum when that happens, and use the fiberglass as the exterior mold for the inside.

    Of course, there will be several substantial bolts holding the aluminum and fiberglass layers together.

    Tonight, after filling an order for a customer, I will start experimenting with Coke cans, and see what happens.

  3. One trick I have heard is to apply mixed epoxy to the aluminum, then wet sand through the epoxy to remove the oxide, while avoiding any air contact that would re-generate the oxide coating.

    Do you know someone who could weld large aluminum "washers" to the front and back of the ring? This would effectively give you a circular "c" channel, which should be significantly more rigid that just a hoop.


  4. One other point to be aware of regarding thermal expansion and contraction: it is not the variation in thickness that you have to worry about. It is the variation in circumference. Using your numbers for aluminum and epoxy thermal expansion coefficients, the aluminum hoop will change diameter by about 0.45mm with a 40K temperature swing. An epoxy hoop will change diameter by about 1.1mm over the same temperature swing.

    Best of luck in re-engineering that beast!


  5. Some thoughts on your plan:
    1. The picture you show of the previous version shows a tapered structure. I'm not envisioning the taper in your current proposal.
    2. Aluminum is much stiffer than fiberglass resin, on a volumetric basis. However, a thin aluminum structure like a pop can can be hugely stiffened with a layer of fiberglass.
    3. The link you show to making a fiberglass tube seems like a good way to go. You could cut the wood disks in graduated size to get the taper. The plastic film, which looks like a 5 or 10 mil graphics film, would allow you to make a nice smooth tapered tube.
    4. With the method in number 3 and enough layers of fiberglass (maybe 2?), the tube would be extremely rigid and well damped.
    5. You would need to put features on the ends to mount your optical elements.

  6. Also, roughen the surface to improve bonding (both the increased surface area and the vertical increase in mechanical locking).

    Scratch the hell out of it with a coarse file (or the tail of it) before applying the epoxy and it should improve adhesion substantially.

  7. I think you would be better off if you extended the tubes to attach directly to the baseplate of the mirror. Even attaching them closer to the mirror would reduce the bending moment.

  8. Philabor: There is no taper in the telescope -- perhaps that's just perspective.

    Sigivald: Yes, roughened the painted surface would help.

    Glen: In retrospect, mounting the blocks for the lower end lower down would have been a very good idea. I was so focused on reducing the length (and therefore weight) of the truss tubes that I completely forgot that the lower cage tube would bend as well. At this point, I am tempted to try fiberglass reinforcement first; the cost to order more tubes and shipping, and then the effort to reattach and recut, is more than the cost in time and materials of disassembling the lower cage, and adding two layers or fiberglass to it that tube. But perhaps after enough time fiberglassing Coke cans, I will take that approach. Worst comes to worst, I can fiberglass the lower cage tube so that I have a smooth surface that doesn't look like a shotgun blast got it.

  9. Glen: The more I look at this, the better your idea sounds. I can't mount them where the mirror cell mounts to the tube, but I could mount the blocks below where the mirror cell mounts to the tube. That would be below the center of gravity of the lower cage. I could also buy the tubes locally (although I would have to flat black them myself) and it would likely be cheaper than ordering them from Moonlite. In addition, I could have the local vendor cut me only the length that I need.