Hard Drive Persistence of Vision (HDPOV)

The platter of a hard drive spins well over sixty times a second. If a narrow slot was cut into the platter to allow LEDs to shine through, we can achieve flicker fusion and trick the eye into seeing a stable image. This phenomenon is known as persistence of vision (POV). There are many examples of LEDs being used for POV, building an image by moving the LEDs or having the observer move relative to them. The LEDs used in this project do not move, and the image is built using the interference of the slotted, spinning platter.

The system works by timing the slot in the platter. The Arduino uses an internal timer to clock each revolution. It achieves this using an infrared gate, which triggers a hardware interrupt on every full revolution of the platter. The Arduino uses the revolution time and phase to schedule a second internal timer. This second timer uses an interrupt to schedule the timing of the LEDs, firing tens of thousands times a second to build a stable, visible image.

This work is based on Ian Smith's version of the same. His work is excellent, and his website on this topic will help provide a complete picture for this instructable.

Step 1 Gather Materials

Hard Drives - You will need to gather a few unused hard drives. It's no problem if they no longer function, but it's imperative that the candidate hard drive can spin indefinitely when it's powered.

Arduino - If you don't have an Arduino board, this link will show you where you can buy one. I developed this with a Diecimila, but I would expect the code to work on a Duemilanove without changes.

ULN2803A - This is responsible for switching the high current 12V LEDs.
Digi-Key: ULN2803APG-ND

RGB LED Tape - The best source I've found for tricolor LED tape is from Super Bright LEDs. A single 19.5" strip costs $19.95.
Super Bright LEDs: NFLS-RGB15

270 Ohm Resistor - Limits current to your sensor source.
Digi-Key: 270H-ND

10K Resistor - Pull-up for your sensor output.
Digi-Key: 10KH-ND

You can build the platter sensor from either an infrared gate or a Hall-effect sensor. The infrared gate uses the slot cut out of the platter to disrupt the beam. The Hall-effect sensor will require that you glue a small but strong magnet exactly 180 degrees from the slot in the platter.

If you decide to use an infrared gate, you can either build your own gate from scratch or you can buy a manufactured gate. If you choose to build your own, you can use any old pair of infrared emitter/phototransistors (mine are from Radio Shack), or you can purchase the following:

Infrared Emitter - Used to provide an infrared beam to time the platter.
Digi-Key: 475-2648-ND

Infrared Phototransistor - Provides a pulse on every platter revolution.
Digi-Key: 475-1080-ND

I did not build my rig with a Hall-effect sensor. I chose to use infrared because I had the parts lying around, but a Hall-effect sensor is less obtrusive. If you choose to go the Hall-effect route, you will need to adjust a few details in this tutorial. For example, you will need to glue a magnet directly opposite of the slot. This will require changing the code loaded onto the Arduino to account for the 180 degree change in phase.

I have not tested this Hall-effect sensor, but I would expect it to work fine:

Hall-Effect Sensor - Provides a pulse on every platter revolution.
Digi-Key: 620-1001-ND

Comments

[JXplicits]

This sketch is just the test pattern and not display the hours like the video, Right?

Sorry for my english if there is any mistake.

[Cobra_Phil]

There is a project at kickstarter about a POV hard drive clock...
http://www.kickstarter.com/projects/680714141/hard-drive-ticktock-pov-clock

[jberwick1]

anybody else having a hard time with python/serial/mac?

its giving me a bunch of trouble, and id hate to give up on such an awesome project, right at the end.

[velaron]

Hello all I am wondering, can i use hall sensor ?

[teknobudd]

The ANODE is the longer and the cathode is the shorter leg of just about any L.E.D. A phototransistor's shorter leg is the collector and the longer leg is the emitter. You can look at any LED from either top or bottom and you will notice its base has a flat side - that is the cathode side. In the case of a phototransistor, the flat side indicates the collector side. Your instructions should work, but the terminology is incorrect.

[Bozza]

Hey, just wondering what brand and how old the hdd u used was. ive opened up two seagates, which at first seemed to spin without stopping, however once the reading arm is removed, they only spin for bout 30secs and then stop. Is there any way to mod the drive to prevent this?

[DJNASHA]

remove the read arm!

[Derin]

Try to gut the arm and keep the inner parts(coil,for example) plugged in.

[zh11107]

http://www.lotusinvention.cn/

[get to da parakeet]

OK I WANT TO POWER MY MOTOR IN THE HARD DRIVE! how do i do that/??

[aiden120000]

capslock makes you sound rude

[vishnubob]

I use ATX power supplies harvested from old machines.

[Jimwright]

My friend Richard M and I finished one today, and it's great! I don't know if my one small problem was the result of using a LED / sensor gate from a floppy drive (see comment below), but it wouldn't sync until I widened the disk gap a little (less than 1/8" total). So, be forewarned.

[vishnubob]

Congratulations! That's awesome! If you can, please post pictures of your rig, I would love to see it. The sensor required the most tuning, I'm not surprised you had to make this adjustment. Did the code work well for you?

[Jimwright]

Well, mine looks mostly like yours. The wood blocker is one piece, due to the small size of the IR gate. (Pix #2 is the gate, before I removed it from the PCB of the junked floppy drive.) I cut the RGB tape very short and scraped the traces bare to solder on a compact 4-lead ribbon cable. The RGB tape stuck OK. I alcohol-wiped the surfaces and washed my hands. Oh, and 2 coats of varnish on the wood. Otherwise, I'm sure the tape wouldn't stick. I chopped a segment out of the spacer ring below the disk to counter-balance, but it's too much. (You can see the removed segment in the pix.) I wound up adding a lot of screws/washers to the hub to un-compensate. It runs smooth. I didn't get a pix of the completed unit with the IR detector installed. It clamps right in the shallow wood slot. The software runs just like your video. Mine has a very slight jitter. No big bother. I want to make an arduino-less version. I think it is feasible to trigger the RGBs with 3 disk-edge slots and adjustable delay RGB triggers. My electronics basics is Nike Hercules school, but I'm much encouraged by your pioneering. THANKS.

[Jimwright]

I ripped open a 3 1/2" floppy drive and found a LED / sensor. (Lots of them in printers too, as I recall.) It even had a diagram on the back of the pcb that matched the schematic here. I rigged it up with the 270 and 10K resistors, and it put out a great 4.7V when the gate is blocked. What's nice about it is that it's very compact (and free).

[frollard]

Instead of using the slot as a thru-device could you mount the ir led AND photo-ransistor up-facing under the platter, with the platter painted ir-reflective 'white', then the slot would pass overhead, causing a flicker on the transistor... You might need to use an analogue line to read it, as it wouldn't be nearly as defined, especially for the slot size. If the slot could not be detected, simply paint an infrared-black dot on the underside :D AWESOME PROJECT! I'm just waiting for a drive to die now :D

[matt_sawyers]

I saw this on YouTube and I've been waiting for you to make an instructable of this! I'm stoked! Thanks for this instructable!

[shmolky]

I'm in the process of building one of these myself... very cool. One thing I'd add that I had to learn is that older, smaller drives often use glass based platters that have the magnetic coating/film on top which can be sanded away. What's cool is you don't lose much in the way of balance, but on the negative side it's a little harder to sand away the magnetic coating on both sides (to make a clear slit) and I don't think you let quite as much light through as compared to: Newer, more higher capacity drives use aluminum based platters which you can cut through easily, but also would unbalance the spinning platter.

[GBMorris]

This is really cool. What would you say is the total time and cost of this project?

[vishnubob]

Thanks for your comment. If you have a few drives on hand, you will spend less than $50 in parts. It took me twelve hours to fabricate, but that was an ad-hoc process. With parts and instructions, you could build this in four hours.

[incorrigible packrat]

That's hella slick. I'm not so sure about cutting the slot all the way through the platter though(and I'm definitely not the most safety conscious one around). If it was to come apart, that thing could have an eye out, or, worse yet, bust up such a sweet project...

[Guy.Fawkes]

Hard drives normally come in 4 speeds: 5400 RPM, 7200 RPM, 10,000 RPM, and 15,000 RPM. The latter two speeds are "expensive" and "insanely expensive," so you will most likely be using 5400 or 7200 RPM drives that equates to 90 or 120 revs/second (exactly 1.5 or 2 times the AC flicker rate in North America, 1.8 or 2.4 times the European flicker rate).

[fwjs28]

brilliant!

[qwertyboy]

this is so cool! i'd make one if i had any old hard drives and an arduino, but i don't have either.

[killerjackalope]

It creates some interesting effects in the video, if you play around with multiples of frame rates you'd get some interesting stuff, also if you can find a rough speed that's just about below POV effects then you could do the same effects in front of people, but it may case motion sickness...

[merseyless]

now add a dvd capabilities and it would be the GREATEST gadget in the universe!

[S1L3N7 SWAT]

This may have been helpful. http://www.sparkfun.com/commerce/product_info.php?products_id=247

It's a prebuilt version of the IR trip gate.

Very Cool though! 5 stars!

[JakeTobak]

Wow, that's trippy. I want one.

[jeff-o]

Nifty! Did you use 4800rpm or 7200rpm hard drives? Does it matter?

[vishnubob]

I used a 7200 RPM drive. As long as the drive spins well over 30 times a second, it should work fine.

[karossii]

he said well over 60 times per second; there are 60 seconds in a minute (the 'm' in RPM); so 60x60=3600... I would assume that even 4800 RPM would be considered well over 3600... so it should work fine if you time it out as described.

[jeff-o]

It's more an issue of whether the microcontroller can keep up, but considering that the Atmel can do, at least 1 million ops per second, I guess it doesn't matter and my question was pointless. ;)