20W Halogen bulb viewed by a photodiode

[StillTrying]

Semi-interesting observations.

Yellow is the bulb's 11.5 VAC, Purple is the relative light level.

[MK14]

As expected. The 50Hz AC, becomes 100Hz (like) because the bulb does not care about polarity.
I.e. 50Hz AC, rises and falls 100 (Hz) times per second.
With some kind of leading/trailing phase lag, because of the heating effect, and the light given off by the filament, as it rises/falls in temperature.
It's interesting to see it happen in practice, thanks!

tl;dr
50Hz AC = 100Hz worth of half sine cycles (180 deg) = 100 on/off flashes per second, for filament bulbs.

[MK14]

Or in pictures:

[StillTrying]

"With some kind of leading/trailing phase lag, because of the heating effect,"

Yep, I was expecting the light peak to be at the far end of the voltage peak, but not passed it by that much. And the minimum light peak is well passed the zero volt point, - over half way towards the next voltage peak.

Here's a 230V 3W LED GU10, its light output is nearly all 100Hz ripple.

[schmitt trigger]

Interesting experiment! 

I would think that the phase lag between peaks would be caused by thermal inertia. Which in turn wholly depends on the filament's thermal mass.

It would be very interesting if you repeat the same experiment with a different type of (incandescent) bulb and compare the results.

Simpler said than done, because incandescent bulbs are becoming harder and harder to find.

[StillTrying]

"It would be very interesting if you repeat the same experiment with a different type of (incandescent) bulb and compare the results."

Yep, I'll do a normal 240V 60W which should have a very thin filament compared with the 12V 20W halogen, I'll try to get around to a 240V 9W CFL as well. Hopefully that will be enough to satisfy mine and others' curiosity so I'll get back to doing something more useful.

[StillTrying]

The thinner filaments of the 240V 60W made almost no difference to the light delays, I even checked the mains against the 12VAC just in case, the mains phase was actually 0.5ms ahead of the 12VAC, the 20W halogen is still powered by it.

The cheapo 9W CFL emits loads of EMI at 50kHz, I think a lot of the noise is actually in the light signal as putting a dark card between the bulb and photo reduces the HF noise - as well as the light.

[StillTrying]

This is what I was using, it's not very sensitive but it gives a reasonably accurate view of the shape of light pulses from DC to about 500ns, not bad for a few junk box components!

[Neomys Sapiens]

Very interesting experiment.
And deeply troubling results.
Noise everywhere, eewww!

[babysitter]

Optical frequency doubler!

[TerraHertz]

You should add a current trace. You may be surprised by what it does at turn-on.

Btw, how come the 11.5VAC ramps up over a few cycles? Not using a mechanical switch?
Edit. Ah! Silly me, I know why. Current limiting transformer. Because of the current effect I'm referring to above.

[StillTrying]

"Optical frequency doubler!"

I did wonder for a minute if a photodiode viewing an incandescent had any practical use in producing a smooth low frequency sine wave, but I don't think it does. The 100Hz light shape is not a sine - just close.


"You should add a current trace."

I'd already put in a 0.1R in and looked and measured!

The 20W halogen's cold resistance seems to be about only 0.38R, from the peaks of V & I the hot resistance is 6.4R, an increase of X17 cold to hot.

With only 0.38R cold resistance, adding 0.1R will reduce the peak currents, but I'll just ignore that error.

The peaks of current at switch on can be as high as 8.3A, before settling to running peaks of 2.45A.
The peaks of Watts at switch on can reach 80W, before settling to running peaks of 38W.

Surely my measurements of this bleedin' halogen light bulb are now complete!
Here's the light pulse from a cheapo photo flash gun (on low power) for a change.

[Zero999]

A full wave rectifier and low pass filter will give a similar result.

Is the lamp being powered from a low frequency or high frequency transformer? It shouldn't make any difference, since with a high frequency transformer it's modulated by the mains frequency and the  high frequency element will be well above the cut-off frequency of the lamp.

[Circlotron]

You could feed the output of the photodiode amplifier into an audio amplifier and listen to the "sound" of various light sources.

[StillTrying]

One of the attempts to detect the 20W halogen's small change in resistance with its 100Hz 12% change in light level, using the scopes V/I.

There's a very slight increase in resistance(white) corresponding with the peaks in light(blue), and a slight dip in resistance corresponding with the dips in light, but that's about the best I got.

[Red Squirrel]

That is really cool, as mentioned in the other thread I had no idea incandescent (or filament based in general) bulbs even did flicker, as I figured the filament stayed lit from the heat during the lower voltage part.  When you turn off a light you can actually see that it takes a few seconds for it to fully turn off.  I guess it still goes down to that dim level for each cycle.

Wonder if CFLs are like this too.     Or neon tubes.      They would probably be a much higher frequency.

Going to have to play with this myself at home when I get the chance.   

[NiHaoMike]

Some CFLs have a very slight "afterglow" for minutes after they're turned off. But that can only be seen in a very dark room.

[schmitt trigger]

Indeed, all phosphors have "persistence".
Which was useful when fluorescent lamps were powered with 50/60 Hz ballast, as a means to reduce flicker.

Nowadays fluorescent ballasts are all high frequency.

[SG-1]

A very cool experiment.

[StillTrying]

Another one for the collection.
Red is the light shape on brightness 1/7, green is on brightness 5/7.

[StillTrying]

A Philips 18W->100W Energy Saver CFL. NOS.
Easier than the cheap 9W CFL - more light and less EMI.

Quite low p-p light ripple, similar to the first plot at www.derlichtpeter.de/en/light-flicker/cfd/

The blue line is the no light level.

[StillTrying]

Short pulses through some 5mm LEDs.
The Top Trace is always the detected light.

RED Led  1us ~80mA pulses, - from long ago.

Green Led  0.5us ~110mA,

Super bright Med White Led  0.5us ~30mA, ~300X more light per mA than the standard green LED.

1us, Red, Green, White, 5mm.
I don't know why Red LEDs usually give a more curved top light shape, I don't think it can be a photodiode thing because the top of the current shape through the LED matches it, - curved for red and flat for green.

[tom66]

If you have a plasma TV, you'll get some interesting effects looking at that with your photodiode.

High speed video of a plasma TV:

[StillTrying]

If you have a plasma TV, you'll get some interesting effects looking at that with your photodiode.

Only got LCDs, the photodiode + 1 transistor hasn't been sensitive enough to get a light shape from a TV or mobile phone backlight or a VFD, by the time I get it (in contact) close enough, the electronic noise is too high.

[StillTrying]

Street lamp outside, light colour is bright creamy yellow, HP sodium I think. Drops down to near zero light in the dips.

Using a big magnifying glass to get an image of the lamp's bulb onto the PD.
Blue line = zero light level.