CD Robot, Version 1

Joining two households into one when I married my wife was a simple affair; she had all of the dishes, furniture and towels, and I had all of the tools. But, we both had CD’s. Even on the day after our wedding, when our apartment was clogged with wedding gifts and necessities for our new house, the stack of CDs towered over it all.

We are proud of the music we have collected, and naturally, we like listening to it. But storing and protecting it is a complex problem. And there are discs that we have not listened to in years, because we’ve forgotten we have them. The emergence of MP3 pointed to a solution, and in 1998 I started planning a Digital music server.
The first problem to solve was ripping the discs. I figured that with diligence, I could manually rip 4 or 5 a day, consigning me to a year of slavery. Clearly I needed a robot. The original idea, sometime in 1998, was to lift CDs from one stack into the CD tray in a computer, and after ripping, put the CD in another tray.

Good idea, but I could do better, since I should not have to open the CD cases myself either. Here begins a several year detour.

The first experiments used a Lynxmotion 5-axis robotic arm These robot arm kits use RC servos and a serial servo controller to connect via a serial port to a computer. This arm was a lot of fun, but it did not have the power required to remove a CD from a case.

My second idea was to build a tall CD rack to hold cases. Then a vertical robot arm would remove each case, open it, remove the CD from the case, and put the CD into a CD-ROM drive. The vertical axis was initially built from the lead screw from a garage door opener. The linear slide opposite the lead screw uses a stainless steel shower rod and six in-line skate wheels, glued side-to-side in pairs, as bearings. This assembly worked – the carriage moves up and down – but with the feeble steppers I have it is so slow that I would be better shuffling CDs about manually.

A later version of the same idea used an entirely custom assembly composed of aluminum C-channel extrusion, a 3/4” threaded rod, and a bit of Teflon shim stock. This linear slide is also functional, but slow and under powered. IN this view, you can see an oblique view of the horizontal arm with the gripper that grabs the CD case. The gripper is powered by an air cylinder that is connected to the gripper by a cable.

The horizontal axis of the robot is made from a linear bearing from a printer, and the acme lead screw from a cheap Chinese drill-press table. Like the other linear slides, it is functional but slow.
This assembly opens the CD case. This is essentially an exoskeleton that fits around the CD. Hidden inside are two spring-loaded washers that engage the slots on the side of the cover of a CD case.

This smaller robot would remove the CD from the case. It is made from telescoping brass tubing and RC servos. The end-effector is a two-part hook that fits in the slots in the rose in the middle of a CD case that engages the hole in the CD. Each hook on the end effector is attached to a square tube that has a rack that engages a pinion on a servo. When the pinion gear turns, the hooks move in opposite directions, either coming together or moving apart.

Although the end-effector is dodgy, the remainder of this mini robot works surprisingly well.

After a year of building bits of the robot and not coming any closer to assembling the whole, I realized that I was tackling too much. I’d only be able to finish this project if I abandoned the idea of having the robot open the case. Within a few hours, I had settled on a design that was functionally equivalent to my original sketches. In the intervening years between the original idea and the beginning of actually progress I had learned a lot about machine shop work, and more importantly, the price of hard drives had dropped to where I could afford to store all of the data I was about to collect.

The final design is a cylindrical robot with a suction-cup effector. It simply shuttles CDs from one stack to the CD tray and finally to an output stack. Nearly all of the parts were salvaged from around the shop. The vertical axis is build from the horizontal axis on earlier versions. The rotary table is made from a cast iron timing gear. The vacuum pump is one of several that have been hiding in dark corners of the shop for years.

If I had bothered to do any time and motion studies, this design would have been obvious from the start; it takes 20 hours for this robot to rip a stack of 100 CDs, and only a few minutes for me to remove the same CDs from their cases.
With this much effort in the project, I did not ever want to do this again. So, I saved all of the CD data as WAV and losslessly compressed the files with FLAC. In the end, it took about a week to rip my entire collection to 250 GB of hard drive space.

Of course, music data on a hard drive is not a music server. Now I needed something to play the CDs. I figured if I had all of the data as WAV files, I should play them that way – no need to suffer the degradation of MP3 conversion. I just need to hack the SliMP3 server to decompress FLAC and send the WAV dat to the SliMP3 head unit, right?

Er, no. The SliMP3 decodes MP3 in hardware.

Next, I tried to hack an old i-Opener. These failed Internet appliances run Linux and have USB ports, so I could connect it to the net, use a Stereo-Link to actually play the music, and use XMMS as a decoder. Even describing the ordeal exhaust me, so I will quit by saying that none of the guides for hacking an i-opener were any help with getting my two devices to boot. By the time I did load Linux on them, through a combination of unusual hardware, trial and error and luck, I decided that the SliMP3 sounds fine playing MP3 as 320 kb/s.

So, all I have yet to do it pull Ethernet cable to the stereo and I will be in digital music bliss.