Nixie tube clock... without the nixie tubes!

[cprobertson1]

Another quick update:

The LED filaments arrived today - been playing with them and much to my chagrin they can't be bent more than 15 degrees before the internal connections break.

You can view the teardown here

I believe it's literally just a series of LEDs glued to the metal backing (with holes in the backing) and simple connecting wires going between them (with a limiting resistor in the middle and at one end)

Bending it breaks the connecting wire, which is a pity!

There are closeups of some of the LEDs - note the double-image is due to the reflectivity of the underlying metal.

approximate size of LEDs = 0.3mm

And as a final note - JEEZ are those little things bright! My retinas are still recovering from when I first powered it on!

[Richard Crowley]

The LED filaments arrived today - been playing with them and much to my chagrin they can't be bent more than 15 degrees before the internal connections break.

Exactly as I feared.
But kwass claimed: "They bend very easily and hold their shape once bent."
Maybe the "not breaking" part wasn't important to him?

[kwass]

The LED filaments arrived today - been playing with them and much to my chagrin they can't be bent more than 15 degrees before the internal connections break.

Exactly as I feared.
But kwass claimed: "They bend very easily and hold their shape once bent."
Maybe the "not breaking" part wasn't important to him?

Her, not him.

Those seem quite different from the ones I bought on ebay:

http://www.ebay.com/itm/1W-COB-LED-Filament-Light-Bulb-Globe-Candle-Light-Source-Warm-Pure-White-DIY-/151416910080?ssPageName=ADME:L:OC:US:3160

(the 4200K ones.)

They bend very easily and hold their shape even in a 3mm radius circle.  28 LED's inside each.

[cprobertson1]

Those seem quite different from the ones I bought on ebay

I did wonder that - I think these are cheap-arse chinese things that I have. They work great when in a straight line, but those little wire connections don't hold up to much!

I think I'm going to get some really small SMD parts and try making my own pre-formed digits using copper wire and a combined metal/wooden jig to solder them on: I'll coat them in Hot Dip Wax for the prototype, and a granular-resin for final design if that works (usually used to protect the brittle carbide tools from chipping in their storage drawers at work: nothing worse than an £800 tool getting chipped because somebody slammed a drawer shut!)

Anyway, I'll try making some once I get my SMD parts in (I've yet to place the order as I've got a bunch of other parts to source for the Nixie and the non-nixie clocks in the same order, and I figured I might as well get my clocks right before I do anything else )

Bit of a pity I got inflexible ones though! They would have looked pretty nice if they were formed into digits!

[Richard Crowley]

Bit of a pity I got inflexible ones though! They would have looked pretty nice if they were formed into digits!

How do they perform if you try to run them at "Nixie" light levels?  Do they operate only at "blinding levels"?
Do they even light up at lower voltages?

[cprobertson1]

You can drop the voltage quite a bit and they're still visible

44-46V was when the light became visible - but the individual LEDs were very easily distinguished - at 46-52V you could look directly at them without your retinas melting, but they were bright enough that you couldn't see the individual LEDs.

At 52-60V they used increasingly high current and were too bright to look at. No seriously, those were BRIGHT.

At 110-120V they make a lot of smoke and a brief flash reminiscent of the trinity test. They don't smell nice either. I suggest not doing this.

(Ps - I was just wondering at what voltage they died at, they were surprisingly stable up to 110V, then the current draw shot through the roof and that was the end of the test-piece.)


I'll get a low-voltage picture tonight and show a comparison at the various voltages: though I may need to put a welding visor in front of the camera

[Zero999]

I doubt they have current limiting resistors. It's more likely they're designed for constant current operation. Open any mains LED lamp and you'll find some sort of current limited power supply, whether be SMPs or just a capacitor in series with a bridge rectifier and a couple of resistors to limit the inrush and discharge the capacitor.

Use a 60V to 70VDC power supply and add a current limiting resistor to each LED strip. If you try to run them off a constant voltage, you'll probably find the current shoots up as the temperature rises. This may not cause a problem at low currents but resistors are cheap.

[cprobertson1]

There are two (small) resistors in-line with them (one half-way down the strip and one at the anode).

For the regular tests I had an in-line resistor (I think it was 220 Ohms? You can swap the resistor out of it - its just a male/female 4mm banana jack on a board with metal tabs across that I made to allow you to lock a through-hole resistor across it - cheap, and great for testing individual components.)

For the death-test it was just a 10R, 10W resistor to I didn't trip the over-current on the power supply.

Pretty sure they undergo cascading failure at the high voltage/high current just as you said!I was just death-testing it anyway - I've no real plans to run them at 120V-DC.

Well... unless I'm trying to blind somebody of course


They are surprisingly nice little components - it's a pity they aren't a) cheaper, b) more readily available and in my case c) more flexible (though KWass' looked to solve that issue!)

[Zero999]

They're reasonably priced when you consider how many LEDs you're really buying.

You can build a constant current source from a transistor and a resistor. No base resistor is needed because emitter degeneration makes the input impedance very high: a couple of hundred k.

Powering the MCU from 3.3V would reduce the voltage across the emitter resistor so it can be lower, given the same current but the current regulation won't be as accurate.