|Sanding exposed this crack after Magnaflux determined the rod was good. Methods like Magnaflux only find flaws on the surface. This had to be replaced.
Fortunately GT Car Parts provided a used replacement in short order. All my rods were inscribed with "494." This was not the exact weight in grams of any of the rods assemblies (rods, caps, bolts) but it was close, so I suspect it was a grouping where all of the rods were "about" 494 grams. The replacement was marked 500, and so it took a bit more sanding to get it to match all the others.
Here they are, all sanded to 120 grain and weighing 287.5g. The next step will be to sand them to 400, shot peen them, and polish them, and open the oil hole (barely visible on the top of the rods) slightly with a 45 degree cut.
A few kind souls have written to point out the virtues of end-for-end balancing. I tried to get repeatable measurements by using various techniques to suspend one end of the rod: Threads, hooks; nothing worked. I finally wound up building my own jig from some scrap metal and a couple of bearings.
The bearings worked beautifully. I was able to find the exact size to fit each end of the rod. This enabled me to get fairly repeatable measurements. The other problem that I have not (yet) completely overcome is the effect of drag on the balance scale. When the free end of the rod moves up and down on the scale, it is actually moving in an arc and wants to drag horizontally over the surface of the scale. This friction has prevented repeatable measures to the accuracy I want (0.1g). Adjusting the hight of the jig helps by minimizing the arc. I may solve this yet, but the ends are close.
|The heaviest piston (left) was different from all the rest. Notice that it has a flat surface machined into the bottom while all the other pistons were rounded (right). I suspect this piston is from a different lot.
|Graphing the weights it's easy to see that the heaviest piston is over two standard deviations from the mean. Also, the next heaviest piston looks far from the rest too. If those two are removed from the sample, the remaining 10 are much better statistically behaved. The standard deviation goes from 4.6g to 2g.
Well, here's how it looks now. This photo was taken the day it was shipped off to its fortunate owner. (This somehow makes me feel worse.) See Miura 5096 on the Lamborghini Registry site for more photos of this beautiful car.