Actual Light waveform, ARC4+ Rev.2

Here is the actual light waveform, based on the actual light output on the ARC4+ Rev.2:


.
.
.

And just to prove the measurement system is fast enough, here is a shot on the fastest light source I had sitting around within reach:



BTW, the H block on the bottom with the 500nS after it, means that each division is 500nS long.

OK this is totally cooool

How did you measure this? with what equip?

Thanks cy.

With one of these, 54852A Infiniium Oscilloscope and InfiniiMax 1131A Probing System, with additional options added (heheh, check out the list price for it at this URL):
http://we.home.agilent.com/USeng/nav/-11496.536894341/pd.html


And a rather accurate and quite fast light meter circuit I built using some NPN 5 GHz fT wideband transistors, a very low noise and low offset voltage, low offset bias and low offset current, very high input impedance, ultra low noise, high-precision, 180 MHz op-amp rigged as a transimpedance/transconductance amp. And a very low capacitance (high speed) photodiode, which is reverse biased to further reduce it's capacitance. Back when I designed it, it would follow a light pulse with a 200nS rise time, rather accurately, no waveforms still kicking around on that one. FYI, 200nS is about 200 billionths of a second, and to see something that fast and follow it's edge, it has to be even faster. I never did test how fast it really is, because it was adequate for what I needed, and I didn't have time to create a faster light source.


FYI, the waveform above is for level 5, if I recall correctly.

This one here, I believe which was level 1, I'll try another measurement when I get a chance, and find out (it could have been another level):

Okay, here is another actual light output measurement, on one of the more stable levels, but "magnified" just to show a few cycles. It is at 10uS per division, and the frequency of these fine "comb" pulses is about 19,463.9 Hz. I had to freeze it, because the waveform is very jumpy, in both amplitude and frequency (see the waveforms above for details on this).

Proves how well (or how slow) the eye integrates the pulses. The jitter on the pulses are probably due to the microprocessor based timing, and the clock source (an RC circuit)

Yes those graphic representations mimic what we have observed (and heard!) here as well.

Many light sources are far from CW. As a matter of fact, 60Hz fundamentals can be realised from a standard incandescent bulb running on house power! (in the US of course)

Fluorescents driven with standard tar ballasts are the worst offenders. The pulses from the Arc even on level 16 are only apparent to the eye when viewing a moving object or when the light is waved quickly in hand.

Cheers!

Kewl.

Now, you -know- that we want to see light waveforms for all 16 light levels. /ubbthreads/images/graemlins/smile.gif

[ QUOTE ]
sharkeeper said:As a matter of fact, 60Hz fundamentals can be realised from a standard incandescent bulb running on house power! (in the US of course)

[/ QUOTE ] An incandescent bulb 'flickers' with 120/100 Hz.
If there is any problem with flicker in an Arc4 it should be easy to fix with an software update. Peter, what was the reason to take this relatively low frequency?

Actually, 120vac has a 60hz sinewave.

The frequency selected was a constraint of the package size. Even so, it is sufficient to avoid annoying flicker. Level 16 is right on the edge though.

Also, the burps you see at level 16 are not the switching speed but rather a different artifact that becomes apparent at that level - the effect of the microprocessor's normal operation on the power buss.\

They are interesting waveforms aren't they?

Peter

[ QUOTE ]
Gransee said:
Actually, 120vac has a 60hz sinewave.

[/ QUOTE ] Yes, buit I was talking about an incandescent bulb's flicker (which is only very slightly due to the filament's thermal mass, but visible under some circumstances) and this is at 120Hz.

[ QUOTE ]
The frequency selected was a constraint of the package size....
Also, the burps you see at level 16 are not the switching speed but rather a different artifact that becomes apparent at that level - the effect of the microprocessor's normal operation on the power buss.

[/ QUOTE ]
Ok, so it is not indentional. I thought it has something to do with tint control. Thanks.

[ QUOTE ]
PeLu said:
Ok, so it is not indentional. I thought it has something to do with tint control. Thanks.


[/ QUOTE ]

I'm assume that "tint control" is accomplished by using differing pulse frequencies and durations to achieve the multiple output levels. The traditional method of dimming the Luxeon would be to simply drive it at a lower current. However, at different drive currents it tends to exhibit slightly different colors.

If you always drive it at the same current, your tint should be fairly consistant. To get less light, leave the pulse on for a shorter duration. As long as the switching frequency is sufficiently high, the human high won't notice the flickering (too much).

Which begs the question: At low power levels, are you still stuffing the emitter with ~700ma pulses (with a relatively long off time in between)? I guess the diode doesn't really care. They probably get killed off by heat, not current.

Yes, that is how tint control has been done with white LEDs for several years now.

Check out LumiLEDs camera flash application notes for the Luxeon. Somewhere on the board here, there is also a post by me, which showed some applications which LumiLEDs covered in the presentation paper, of whacking it with very short 5.3A pulses