If you’ve ever encountered a rapidly spinning split-flap displays at an airport terminal, it’s hard not to stop and marvel at them in action for a few extra seconds. Because of this same fascination, [M1k3y] began restoring an old one-hundred and twenty character sign, which he outlines the process of on his blog.
Finding documentation on this old relic turned out to be an impossibility; the producers of the model themselves didn’t even keep it off-hand any longer. In spite of that, [M1k3y] was able to determine the function of the small amount of circuitry driving the sign through process of elimination by studying the components. After nearly a year of poking at it, he happened across a video by the Trollhöhle Compute Club, demonstrating the successful use of the same display model. Luckily, they were kind enough to share their working source code. By reverse engineering the serial protocol in their example, he was able to write his own software to get the sign moving at last.
Once up and running, [M1k3y] learned that only eighty of the sign’s characters were still operable, but that is plenty to make a mesmerizing statement! Here is a video of the cycling letters in action:
I am almost positive that flip displays are why God created rust ! Never liked them, especially as alarm clocks, click, pop. Try to sleep . . .
I’ll go with not the best technology for alarm clocks, but other than that….so cool!
It’d be fun to make one of these that tracked a twitter feed…
That’s one of the things I plan to do. (If I find enough of the modules)
Nice, seems to have a bit of a problem with sticking letters though but nice.
“At the moment I’m trying to figure out why the modules are dying one after another. Every once in a while, a module won’t rotate any more and only jitters when getting a command.”
Well, I can make a few guesses.
As far as I can tell, the power requirements of 9VDC for logic and 48VAC for motor were determined from a similar module. Perhaps that’s a bad assumption, and those voltages not appropriate for this module.
The 74HC165 and Z86E08 MCU have a normal supply voltage of 5V. Absolute maximum voltage is 7V.
[M1k3y] says there’s a voltage regulator. Assuming this is correct, I can positively say it’s not on the address board, where power comes in. So power must go from the address board, through the header to the main logic board, and voltage regulator. Then 5V should come back to the address board. The first possibility is therefore:
1) He inadvertently wired 9V to the 5V rail, *after* the voltage regulator, therefore exceeding Vmax of the digital logic.
But I can’t find the voltage regulator on the main board, either. Raising two more possibilities:
2) There is no voltage regulator. In which case the module should be receiving pre-regulated 5V, not 9V. The presence of such a large electrolytic filter cap may be evidence there is no local regulator.
3) It’s a small SMD regulator I can’t spot in the pictures. I see no heavy traces that might clearly act as heatsinks for such a component either. The regulator’s dissipation limit would be small, and if fed too large a voltage, it could fail.
All three possibilities could be tested by measuring the voltage on the 74HC165 supply pins, on a failed module, when being fed 9V. I would rule these issues out before looking into additional possibilities:
4) The 48VAC is too high, causing motor or motor driver failure.
5) The modules are just old and prone to some other failure to be determined, which is not a result of user error.
“So power must go from the address board, through the header to the main logic board, and voltage regulator. Then 5V should come back to the address board.”
This is correct.
“1) He inadvertently wired 9V to the 5V rail, *after* the voltage regulator, therefore exceeding Vmax of the digital logic.”
Would be possible, but I measured the voltage at several points in the circuit always ending up with values between 4.9 and 5.05 Volt.
“But I can’t find the voltage regulator on the main board, either.”
I’m almost sure that the voltage regulating is (when seen from the side) on the left of the MCU.
“2) There is no voltage regulator.”
Checked that
“The regulator’s dissipation limit would be small, and if fed too large a voltage, it could fail.”
Shouldn’t happen, I’m driving the system only with 7 Volt at the moment because otherwise the power supply would get too hot over time.
“4) The 48VAC is too high, causing motor or motor driver failure.”
Possible, but I don’t think this is the cause of the problem. I have checked that my modules are exactly the same as the ones used in the Trollhöhle, and they are using theres with 48 Volt for over half a year now without failures. Also I’m driving mine with only 40 Volt (Just happend to hava a fitting power supply).
5) The modules are just old and prone to some other failure to be determined, which is not a result of user error.
Also possible, but I hope not. Only thing I can say at the moment is that neither the mechanics, the MCU or the address-chip fail.
“I’m almost sure that the voltage regulating is (when seen from the side) on the left of the MCU.”
Not sure if we’re looking at the same part. The 8-pin SOIC chip to the left of the MCU, from the perspective of the photograph showing the MCU, is a serial EEPROM. But the part [k-ww] pointed out may indeed be the regulator. Sounds like you’ve successfully ruled out power problems though.
I like [k-ww]’s suggestions to swap the phototriac and lube the motor. Using heat to draw the oil in is an interesting trick, one I’ve not heard before. Though I’d be inclined to attempt disassembly of a motor, if possible, to see if there’s a simpler method of lubrication; since you’ll likely have to do it many times.
I wish you luck. This project has a ton of charm.
“I’m almost sure that the voltage regulating is (when seen from the side) on the left of the MCU.”
I see, I didn’t made my point clear enough. I was talking about the image showing the complete module. On the image you are referring to it is above the MCU.
“I like [k-ww]‘s suggestions to swap the phototriac”
I will try this the next weekend.
“Though I’d be inclined to attempt disassembly of a motor, if possible, to see if there’s a simpler method of lubrication; since you’ll likely have to do it many times.”
If it has to be done, I will try to open one of the motors, but I’m not sure if this will be necessary. I already disassembled one of the modules that stopped working, and the motor could be rotated easily by hand.
I wish you luck. This project has a ton of charm.
Thanks. I hope I will find the problem this week.
“and they are using theres with 48 Volt:”
Theirs* :)
Don’t sweat it though.
Sorry about that, but as you may have guessed, I’m not a native speaker. I know the differences between they’re theire and there, but sometimes I still make an error :D
But thank you for pointing out that mistake to me.
The small SMD device next to the yellow cap and the bar code label is most likely the voltage regulator. Measure across pins 8 & 16 of an address decoder chip to verify this.
The socketed chip is most likely an opto-isolated triac, used to drive the motor.
Since it is socketed, I would try swapping it between a working and non-working board to see if it makes a difference.
Another possible cause is that the oil in the motor is ‘gumming up’ – this happens in electric clocks also, as they get old. Place the motor in the oven to heat it, then place it in a pan of light [sewing machine] oil, so that the output gear shaft is under the surface of the oil, and some oil will be drawn into the motor and it’s internal gears as it cools.
I’ve been wanting to create a working replica of the Swan Countdown Timer from Lost by making/converting a split-flap display, But I haven’t been able to find any old split-flap clocks around town, and haven’t been able to find any 3d models of split-flap display parts or any good references. Does anyone know where I might be able to find some good references for creating a split-flap display from scratch?
If you make this, please take my money since ThinkGeek refuses to.
I wish that clock (http://www.thinkgeek.com/42.shtml?qty=1&cart.x=88&cart.y=24) was a real product. I would have a lot of trouble keeping myself from buying it no matter what the price.,
Oops. Wrong link. I meant http://www.thinkgeek.com/stuff/41/dharma-alarm-clock.shtml
http://beachlab.org/flip-flap/
They’ve got a good thing going here, and the GitHub is active.
Perfect! Thank you! That was exactly what I was looking for.
I wonder if there is software to emulate this, it should be easy and might be fun for some things.