I've a white balance problem - any ideas?

Are you sure the temperature changes and not just the brightness? I would also be interested to see any references to backup this theory. Thanks.

Interesting. I understand what you are saying about TV screens and so on, but how that relates to a fluoresent tube I'm not so sure. The idea that you would catch both states of such a device behaving in the way described when those states oscillate 50 or 60 times per second is a real stab in the dark IMO. What are the chances that you'd get one of each with two stabs at it? Not 50/50 but I'm not a mathematician. Perhaps someone out there is?

The concept of fluorescent tubes not having a complete spectrum of output is well documented in Physics text books and photographic books over many years.

This idea that the colour changes because of the nature of AC current seems a bit thin to me, but I'd be very pleased to learn where the information comes from and what's behind the theory.

Of course as there is mixed lighting (as in this case) and one of the sources is flickering then the overall colour balance would also be flickering because the mix of the two sources would be changing...

He is quite correct. Tungsten lighting does not have time to 'turn off' when the current drops to zero, which it does 100 times a second if the ac supply frequency is 50Hz. 120 times a second if it's 60 Hz.
Flourescent lighting however does actually turn off, but the eyes can't percieve this 'flickering' when they are trying to resolve an image, but the human brain can in the following scenario.

Some people who use CRT computer screens in a flourescent lighting environment suffer headaches. Quite often, a CRT/computer screen is refreshed at 60Hz. The mains ac frequency in UK is 50Hz. The difference between the flourescent lights (50Hz) and the screen refresh (60Hz) causes the screen to appear to flicker, or beat, at 10Hz. A frequency that some people can't handle, hence the headache (the solution is to change the refresh rate to 70Hz or higher).

A camera however can quite easily take an accurate WB reading as long as an appropriate shutter speed is selected to avoid this 'dead zone' in flourescent lighting.

1 second = 1000 milliseconds, 50HZ = 1000/50 = 20 milliseconds or 1/50 seconds. Choose a shutter speed of 1/50th or slower, and you'll not have a much of a problem getting an accurate WB setting. In 1/50th of a second, UK mains supply has gone from 0 - max - 0 - min - 0. The camera, in WB terms, will record the average.

I'll leave you US users to do the maths for 60Hz supplies.

Of course, at the time an image is recorded, possibly using a faster shutter speed, the temparature of the light may well be different to that which has been used as WB but hey, that's a photographer's lot, use flash.

1 in 4, I think.

Anyway, I am suggesting taking more than two stabs (as many as one can be bothered) - to get a reasonable chance of capturing both states. Not for each shot, but as a one-off job before shooting, and then to apply the information to the images afterwards.

Good point. Thanks.

I guess that this lighting scenario would need a white reference (white card or clothing) in every shot. Difficult.

If the non-flourescent light source was constant*, there would only be two alternating temperatures as the flourescent light flickered, so no need for a white reference for each shot.

* In reality it may not be absolutely constant, but perhaps near enough not to matter.

wow...
talk about "over-thinking" the subject.
I'm not the brightest bulb on the christmas tree, and when it comes to math, I'm pretty much "dumb-as-a-stump".
I've never had these kinds of problems shooting under flourecent or mixed lights.
Just set everything to manual (including color balance) meter carefully, and shoot!
In practice, I lose more shots on auto, than manual.
In practice, I seldom if ever see that much difference between two consecutive frames.
In reality, I'd grab the one I felt was better tune it up, trash the other and move on.

It's not something that I would shoot on AWB, although sometimes mixed lighting sources are handled very well by AWB.

I'm still not convinced - I have never ever seen anything else on this subject anywhere, so I really would like to know where the information is sourced from and what experimental evidence there is for this effect.

Any ideas where to look?

'nuff said.

The flicker of (certain types of) fluorescent lights is well documented, and it seems reasonable to assume the overall colour temperature will change when there is an additional light source of a different type.

http://en.wikipedia.org/wiki/Fluorescent_lamp#Flicker

www.google.co.uk/search?hl=en&safe=off&q=fluorescent+flickering&meta...

I will try some experiments myself just because I think it's an interesting subject. 'Nuff not said!

That should read 'Not 'nuff said'

I've just been looking at http://en.wikipedia.org/wiki/Fluorescent_tube

Modern tubes with electronic ballast are flicker-free, but notwithstanding that, the article reminds us of the way these tubes work.

The charged gas is in plasma state and releases photons in the UV part of the spectrum. These photons hit and excite the phosphors than line the glass tube. The phosphors fluoresce with visible light, just like a CRT monitor. The colour of the light depends upon the phosphors used, not the nature of the AC power.

There is no mention in the article of changing colour balance and I really don't see how there would be.

Hope that helps!

If the scene was lit by both tungsten and flourescent lights, I can quite understand that the colour temperature of the light will be constantly changing from almost 100% tungsten (when the flourescent is at 0 volts, or glowing at its dimmest) to a mix between tungsten and flourescent (when the flourescent is at max voltage, or glowing at it's brightest).

Problem is, a camera will take the AWB reading at a point in time, then some time later, it'll expose the image when the colour temperature of the lighting may well be different, because the mix of tungsten and flourescent may differ.

I can understand how it might happen, but I've never had much of a problem.

PS: I don't believe 'flicker-free' tubes are flickerfree, but I do believe they flicker much less.