Structural Integrity of CDs

Prompted by a rather sensationalist article about CD failure in high speed drives, I wanted to see just how likely failure of media in use would be. After several pretty simplistic attempts at back of the envelope calculations to determine their yield failure I had a range of results from 16 kRPM to 250 kRPM. The former obviously* too low and the later would make kinetic battery history.

*Ironically about the right number for average condition CDs with micro fractures near the centre.

Much googling later I found better figures for polycarbonate yield and a few pages describing the derivation of hoop stress. With some help from Kieron we did some more calculations and got a better range from 60 kRPM to 140 kRPM. Still not really approaching the problem correctly, I finally found this independent paper which set us straight. Using their derived stress equation, one can expect yield failure at about 54 kRPM, but long before that any structural defects, like cracks at the centre will cause failure.

Being Geeks we just had to do some practical experiments, especially if there was chances of high energy failure, destroying things and endangering ones body parts! OK, it isn't really high energy, a few tens of joules, but it would be fun none the less.

Step 0: Collect a quantity of test media. Various types were collected, pressed and CD-R of many vintages. A copy of the paper was printed for reference during the experiment, and an old 1x Compaq CD-ROM drive was located for the experiment.

media collection, paper printed copy, and an old CD-ROM drive

Step 1: Mating an electric motor with a CD. The spindle assembly from the old 1x Compaq drive was removed. It was reasonable quality with a magnetic retainer floating in the top part of the drive. Below you can see the brush-less motor rotor removed from the drive, and my MacGyvered coupling to the small motor, using a surplus worm-gear.

CD-ROM drive spindle assembly
motor coupled to spindle

Step 2: Instrumentation and mounting. An optical tachometer was built to give RPM measurements, and the entire assembly mounted to my vice-bench. Note the eye protection and the black mark on the CD to aid optical tachometer pick-up.

drive assembly clamped to bench
optical tachometer setup with marked CD

Step 3: Apply power. The entire set-up was covered with a photocopier paper box, propped up a little so I could see the motor and make sure it wasn't moving. This was not so much for protection from failed CDs, as this was only for initial proof of concept test, but to contain the CD if it left the spindle. In the first experiments the CD jumped off the spindle several times at about 2000 RPM. No harm done, but the CD took off along the floor and ran around in circles until it hit something. Some sticky tape cured this problem.

The coupling is somewhat flexible. I was concerned it would vibrate and fall apart. There was some modeing of the box, the table, and the drive system as I spun it up, most of these effects were gone once more than 1500 RPM was achieved.

Failure! (But not the kind we want) The current motor is not up to the job. It maxed out at 5400 RPM, pulling over 3.5 A from the supply at around 10 V. The current limiting feature of the supply was used to vary the speed. At the full 3.5 A, the motor screamed in pain, it is not rated for this kind of abuse, the plastic end-cap that holds the brushes glowed blue from the brush arcs inside and began to melt. The familiar smell of burning enamel filled the air and the motor began to stutter, slowing to about 3000 RPM. I cut the power. The motor case was too hot to handle, I was surprised the magnets took the abuse, they must be ferrite not NIB, the case was well above the NIB curie point. The brushes were almost gone, and one pole winding had opened at the commutator.

I'll try again next week with a bigger motor. I also want to cut some tensile test samples from the huge stack of CDs and load them up with a bucket of sand to get a real-world figure for the tensile strength of the plastic. Notched and polished samples would be a good comparison. I could try building an Izod test swing as well.