Problem reaching 4700K with Solux lamp [solved]

[torger]

I just bought two MR-16 35W Solux halogen black back, 4700K.

According to the data sheet it should provide 4700K in the central bright circle at 12V. Both my lamps provide about ~3900K, and I need to step up to 16 volts to get ~4600K. This is a bit of a disappointment as the idea was reaching 5000K on mild overdrive, and my variable dc power supply can't output more than 16 volts.

So I'm wondering if the marking is exaggerated? Maybe these new black back lamps (before they just let red light out the back) has lowered the color temperature of the lamps?

Surfing the net I get mixed messages if they can do 4700K or not, but it seems many have had problems measuring the lower temperature outer disc or the light going out the back, but I'm just measuring the central disc.

I've attached spectral plots for 12 and 16 volts. From the shape of the spectra I think looks like it's losing power too fast from 550 and down, not enough blue is coming out.

Comparing to the spectral curve Solux presents themselves it doesn't match: http://www.solux.net/ies_files/SoLux%20Spectral%20Data.xls.pdf
The difference between 450 and 600 nm in my 12 volt spectra is 2x and only 1.46x in Solux data.

EDIT June 15: the problem was shown to be an issue with my cables(!), the lamps are fine. What seemed to be 16 volts over the lamp was much less. To drive the Solux lamps in overdrive make sure you have a power source that can deliver not just the volts, but also the amperes required, and make sure that the lab cables (often design for low power) can handle that without power loss.

[digitaldog]

So I'm wondering if the marking is exaggerated?

Maybe, what you report I've also seen as well. Do a search, it's been discussed here a number of times. Here's a screen capture of what I saw and it was shown in the posts on here:
http://forum.luminous-landscape.com/index.php?topic=79604.0

[torger]

Wow... that was a heated debate in that thread... surfing the net it seems like everything around solux is kind of controversial, I wonder why, it's just a lamp.

I just want to get a D50 simulator for some camera profiling testing, but so far this does not cut it.

[digitaldog]

Wow... that was a heated debate in that thread... surfing the net it seems like everything around solux is kind of controversial, I wonder why, it's just a lamp.
I just want to get a D50 simulator for some camera profiling testing, but so far this does not cut it.

Yes, heated. I think most of us are using Solux due to their spectrum and light qualities and as I think I mentioned, the numbers vary depending on the device used and calculated and I personally don't really care about the numbers. For your needs, I don't know of a better man made illuminate but if you need to hit specific numbers, you may have to try some different Solux lamp options.

[GWGill]

According to the data sheet it should provide 4700K in the central bright circle at 12V. Both my lamps provide about ~3900K, and I need to step up to 16 volts to get ~4600K.

Almost sounds like they have been mis-labelled.

Note that lamp life is aprox. proportional to Voltage ^ -12. So a 1% increase in voltage reduces life to 90% of normal, while 16 Volts would be expected to reduce the life of a 12V lamp to 3% of normal.

[torger]

Almost sounds like they have been mis-labelled.

Note that lamp life is aprox. proportional to Voltage ^ -12. So a 1% increase in voltage reduces life to 90% of normal, while 16 Volts would be expected to reduce the life of a 12V lamp to 3% of normal.

Thanks, I'll contact the seller. Yes I'm aware of that I'm burning the lamps fast, but for camera profiling they won't run for long, and I bought two not to light with two lamps (I use flatfield correction to even out uneven light) but to have one extra when one burns out :-).

16 volts was way higher than expected though, according to the overdrive data sheet it's considerably lower voltage required to reach 5000K, but that assumes the lamp starts at 4700K at 12V, not 3900K

Looking at the data sheet of how the spectral shapes for the lower temperature should look it doesn't seem like it is a regular mislabelling, and the lamps reflector looks really blue and it's black back. The spectrum shape seems to match the 4700K lamp data from 500nm and up (it's almost flat at 12 volt, as it should be), red is rolled off the way it should, but it doesn't output enough blue.

My guess is that they have manufacturing issues, some series are good (I doubt they have made up their measurement data), but some are bad and doesn't meet specs. Or this whole black back series is bad, AFAIK all measurements I've seen that meet 4700K is done on the older unpainted model.

The best man-made illuminant currently commercially available to simulate real light is as far as I know Image Engineering's programmable multi-channel LED (http://image-engineering-shop.de/shop/article_iQ-LED/IQ-LED.html), I would be all over it if it cost 400 euro, but it's 6000... way outside my budget. Solux is super-cheap, now if it just worked as promised.

[torger]

If someone happens to surf to this thread, adding this link for reference:
http://www.liulabs.com/Daylight.htm
A filtered halogen lamp that depending on configuration can do D50, D55, D65 and D75. It's $300 though so it's not for everyone.

Haven't decided yet, but I might go down this path, it would be nice with a proper D65 simulator for my camera profiling software development.

[JRSmit]

This is not my experience. I use a regulated power supply with dc out. I regulate it such that I have 12volts on the bulb terminals. I get 4600 - 4700 kelvin .
i use two to illuminate my prints and one has a Lee blue filter. Together it gives 5100 kelvin.

[digitaldog]

I get 4600 - 4700 kelvin .

What instrument and software are you using?

[AlterEgo]

What instrument and software are you using?

may be there is also a difference between batches - say they used to be better and now Solux decided to cut some corners, no ?

[torger]

may be there is also a difference between batches - say they used to be better and now Solux decided to cut some corners, no ?

I've talked to my dealer that sold me the lamps, he tested one from the same batch with a i1Pro and got 4600K I think it was, ie on spec (4700K +/-200K or so). I don't think I've done any mistake in measurement, but I'm getting kind of paranoid.

When you ask why you get a lower temperature than the spec you get this long list of all the mistakes they think you have made. The only thing I have left on that list is that my instrument would be faulty. Unfortunately I have not managed to get my paws on a second instrument to doublecheck. Other lamps measure as expected though, so I have no reason to suspect that it's faulty.

So my best theory is that there is some huge sample-to-sample variation going on. I'll probably ask the dealer to send me the lamp he tested, but I've kind of lost the apetite for Solux, and I what I really want is D65, not D50.

[WayneLarmon]

What instrument and software are you using?

One thing is that undiffused Solux bulbs are directional.  This affects measurements if you aren't aware of this.  I just measured (using ArgyllPro ColorMeter with a ColorMunki) a 36 degree 4700K Solux bulb that was mounted in a home made fixture, pointing down at my copy stand base.



When I put the ColorMunki in the center of the beam I got




4458K.  When I moved the ColorMunki so it was outside the beam's spread I got




3820K.  When I moved the ColorMunki farther from the beam's center I got




3334K.  I tried this several times and it was repeatable (Well, as repeatable as would be expected with an ad hoc measuring system.)

I've been measuring CFLs and LEDs and they don't act like this.  When the ColorMunki is off axis, the spectrum and color temp. don't change appreciably.  

Wayne

p.s.,  I love ArgyllPro ColorMeter

[torger]

I've measured in the same way, using a Colormunki. I have two lamps, one is 3900K, the other 4100K. Undiffused. It's very obvious that the highest temperature is had in the center, one sees that it projects a yellow disc outside the whiter center.

4450K like you get is also a bit on the low side, but I guess acceptable. The +/-200K stated together with these type of lamps should rather be stated up to 4700K, possibly 200-300 less. That they output higher temps seems to be extremely rare.

[WayneLarmon]

It's very obvious that the highest temperature is had in the center, one sees that it projects a yellow disc outside the whiter center.

Yeah, it is obvious, but Andrew asked about measuring techniques.  Apparently the issue of how to measure a Solux bulb has been contentious in the past.  I wanted to get the ball rolling by showing how I measured. 

So we could at least resolve how to measure a Solux bulb: show your measuring technique like I did.

Wayne

[torger]

It was not meant as criticism, just for other readers to understand. If one reads about Solux and has never seen one it seems like it's super-difficult to measure it, but really it should not be that difficult. So I do think there is a real problem with these lamps, that there are sample variations which is considerably larger than the official tolerance. I don't think everyone that has reported low temperatures have measured wrong.

1) don't use diffuser or a shade that may add reflections and alter the temp
2) measure in the center of the projected disk
3) make sure there's not a lot of other light bouncing around in the room that can disturb the measurement
4) ideally feed it with a variable DC power supply so you can check voltage

I'm still waiting to get my dealers tested sample, will probably get it next week.

[digitaldog]

Yeah, it is obvious, but Andrew asked about measuring techniques.  Apparently the issue of how to measure a Solux bulb has been contentious in the past.

I think that's certainly a good possibility, that's why I asked. I wasn't at all anal about how I measured my Task Lamp (which has a diffuser too). I got a value that didn't jive with what it 'should be' but I've said from day one, I don't really care and would expect differing instruments and software to provide differing values too. I used an i1Pro-2, not a ColorMunki, I used BableColor, the method was admittedly sloppy compared to Wayne's. In the end, I don't know we can come to a good conclusion about differences in values reported from single samples of lamps and differing techniques of measuring and I wonder if it matters.

[JRSmit]

What instrument and software are you using?

i1pro2 and i1 display pro. I used i1profiler and also a french measuremet program. The programs gave a difference of 100 kelvin .

[torger]

It matters if you're going to use it as a D50 simulator. If you don't need UV output a working Solux on slight overdrive is both the best and the cheapest one can get in that regard.

For typical viewing applications, probably less so, but those that bothered getting a 4700K lamp instead of a $4 standard 3200K lamp is probably not thrilled if it shows 3900K like one of mine...

[JRSmit]

The solux are directional for sure. I measure the central part. I played with diffuser but that significantly lowered the temperature.
I would love to have a lighting setup that has a broeder area of illumination.

[digitaldog]

It matters if you're going to use it as a D50 simulator.

The one light source that can really do that is 93 million miles away, and measuring the CCT value of a man made illuminant with varying instruments and software isn't isn't going to convince me that Solux can do so.