It almost worked!

"Nothing beats a failure like another try." Anon

C-Lab Home

Many things work the first time for us, because for nearly everything, it's not truly the first time.  Experience does that for (or to) you.  However, it sometimes happens that what seems like a good idea at the time, isn't.  This page is dedicated to those kinds of things.  Why admit we make mistakes?  Why not!  Doesn't everyone?  Mistakes are the great teachers in life if you take the time to figure them out.  To avoid cluttering the project pages with all the things we tried before coming up with the best and final, we decided to put that sort of thing here.  This way, if you look at one of our projects, and decide to duplicate it "except for this detail" you might be spared the time it takes you to try something we've already tried that didn't work.  Enjoy!

Coffee-Can Reflow Oven

Boy...I should have paid more attention in my thermals classes (30+ years ago).  Since I had had pretty good luck using aluminum body Dale-type resistors (from WWII suplus) as heaters, we tried that first with some newly purchased ones.  A heat spreader is used to conduct the heat where we want it.  We assumed we would need one without any testing.  At the very least we needed a flat surface to mount the flat bottomed resistors and later a heating element.  Here is the result:

Picture of thermal spreader with scarsTo make the heat spreader: Obtain a 2" disk of .093" aluminum to glue into the little depression in the bottom of the can to make the outside bottom nice and flat. We cut this from some standard chassis material with a jigsaw, and stripped the paint as above. Use the good red high temperature silastic (type 650, auto parts store) in a thin layer to glue this to the bottom of the can, and wait for it to cure. We've not found a better high temperature adhesive than this silastic; in another application, a woodstove repair, it has been heated to incandescence, which didn't help it much (it got crunchy), but it survived. Cut a 3.25" disc from some copper flashing. Drill a 1/4" hole in the center, and clean it and make it flat. We used nitric acid to clean, about 6 normal (eg 20 or 30%, diluted from the 70% commercial grade, exact strength not critical), then rinse, other things would probably work as well, but this is very fast and easy, and time is money. Make it flat by tapping with a hammer while it's on a steel plate. Don't overdo this step, as you can find yourself inadvertantly coppersmithing and making nice compound curves.

You can see what the resistors' footprint looked like before they exploded with considerable force.  Sorry, I didn't save them for pictures.  A call to a helpful Dale/Vishay engineer explained all.  In the old days, they used a ceramic type adhesive inside the aluminum body.  Some time ago, they went away from this, as the different thermal expansion coefficients caused cracking of the adhesive, and poor thermal coupling.  The new stuff won't take high temperatures, and boils.  As the unit is sealed, if pushed, they explode with some violence (as in, dent in the wall, lot of ceramic bits flying at high velocities).  Glad no one was hurt!  The above can has a 2 part heat spreader.  First there is a 2"x.090" aluminum disc glued into the dimple in the can center -- this is where we want the heat to go.  Then copper flashing at nearly the full can diameter is glued to this.  The hole in the center is to let this glue (Silastic 650) cure in a reasonable time.  The copper is not glued everywhere, just at the aluminum center and a thin ring around the outside.  Otherwise, a cure might take years.  This is important as the silastic won't withstand heat until it is fully cured.  Heat it too soon and it decomposes.  After that it will never cure.

Pretty heater from McMaster-CarrOk, so we needed another approach, and at this point even Doug's Scottishness will allow us to proceed with a "real" heater, bought commercially.  The McMaster-Carr part number for the unit pictured below is 3559K25, for about $42. .Here it is.
This was glued with some more red goop in a thin layer.  Pretty, huh?  Well, there was just this one little problem, actually two of them.  One was too much thermal resistance between the actual element and the can bottom, a "feature" of the thick ceramic insulator.  The other was too much thermal mass, which was excacerbated by the first.

  In the first test, we just poured the coals to this heater, full power.  At some point the Fluoroinert began to boil.  We shut the power off at that instant.  Due to the thermal resistance, by the time the can bottom reached 215°c, the internal heater was more or less red hot.  Heat therefore continued to flow and it boiled about $25 worth of Fluoroinert right out of the can before it could be stopped.  Post mortem showed the fiberglass insulation touching the bottom was burnt.  This is a good little heater, just not for this! Now it should be easy to understand the problem.  The high thermal resistance allowed the heater internals to reach a far higher temperature than the "output" face.  The result was very high "effective" thermal mass, since there was so much temperature differential.  We decided if we couldn't get it right on the first try manually, it was going to be a little too interesting trying to get a computer to control it well, so we moved on and tried something that did work, a halogen bulb.  Next thing we will try after that is the bulb with no heat spreader, as it shouldn't be needed.  But bulb + the spreader work fine, and the thermal mass issues are nearly gone.

Copper-fin heatsink attempt Anything that needs plumbing is a pain, less here than most places, since we have a chemistry bench setup, but it's still a pain.  So, after the first working rev of the reflow oven, we decided to see if we couldn't passively cool the top of the can.  We tried sweating on some 2" by 4" copper flashing fins. Here's what that looks like:

Another no-go.  Well, it works "at all" but we're talking about some expensive chemicals here, and letting any of it get away is not good.  For these fins to do any good, they must get to an above-ambient temperature.  Although the Fluoroinert that touches the can sides condenses just fine with these, the convection rotor we need to get the stuff in the middle isn't strong enough, so it just wafts out of the top -- at $50 per 4 ounces.  The drier hose coming in at the right is exausting air at about 300 cfm, and this turns out to be exactly the wrong thing to do.  We tried adding some more fins, and it looks pretty cool, but we want this thing to work.  There's another slight problem here.  The preheat oven sits on top of this can, and produces most of its heat at its base.  Any that conducts down to the lower can heats the fins and is lost, which in this case meant we couldn't quite reach the preheat temperature goal.  With water cooling, the water can always be turned off...So that's what Model 3 will have.  Again.

C-Lab Home