spyder2 lcd filter function

Yes it is an IR filter and Plamas have plenty of IR. So without the filter the S2 would see way more Red than the human eye does.

The S2 has two internal calibration tables that are selected with the CRT mode and all others. As many of the tristim meters on the market do. So in theory it should without the IR filter work on a CRT just fine. The problem with the S2 is they don't seem to be factory calibrated very well with or without the IR filter. So if you have a few of them you will get a different result from each but which one is right.

I had run some measurements with the SpyderTV sensor on a Pioneer plasma in LCD vs CRT mode w/o filter. The link has the HCFR files if you're interested.

http://www.avsforum.com/avs-vb/showt...&&#post9551257

How have tristimulus sensors handled IR and plasmas in the past? I know they were not very good for LCD displays unless calibrated specifically for them, but until recently, I do not remember ever seeing mention of an IR filter being required.

It didn't make sense to me either, till I realized that inexpensive tristimulus meters work by a combination of filter design and calibration (fudging?). So, if the calibration is done with a CRT source, it will not work so well with an LCD and so on. Come to think of it, even the expensive ones are made the same way.

This is why I think the S2 design is great. Because it has 7 sensors to better detect the spectral response over the ones that only have 3 or 4. It has a light pipe to minimized off angle light on a LCD, a removable IR filter for Plasma and a diffuser for ambient, all for under $100. Just wish they had a calibration service and NIST certs or allowed 3rd parties to setup such a service.

derek

I thought the LCD light pipe and IR filter are the same assembly - until you just posted this and I took a tiny screwdriver to pop it out the filter. It was not a new one I received and the filter had water spots on it. I presume the calibration table assumes both the pipe and filter are in place - as it is not really a modular assembly even though it does appear to come apart. The diffuser does not come off - at least I am not going to try to take it off.

zoyd search out the Spyder patent - the calibration process is more complex than you are making it out to be. A calibrated Spyder competes with the similar KM design with 5x more sensor that is sold as a cheap portable substitute for a spectroradiometer.

http://patft.uspto.gov/netacgi/nph-P...S=PN/6,163,377

Not sure what the patent office uses for images - my browser is not showing them - but the claims indicate the response curves are in there.

They use the calibration method in this paper - which I am too cheap to pay for...

http://www.iop.org/EJ/abstract/0022-3735/8/1/016

colorimeters are only as good as their filter set matching the CIE standard eye response - and calibration thereof.

They are, I was just experimenting with mine and removed the IR filter to run with my i1Pro to see what the IR cut is for the S2. So then I started to experiment with and without the IR filter in the light pipe. I'm just not happy unless I'm taking something apart

Well at least now we know it was a plasma IR filter - I thought it was a polarizer for LCD light!

The point of their patent is to use edge rather than bandpass filters. This means their filter stack does not optically make a CIE XYZ response - rather they take the response of the edge filters - and do some DSP math to come up with the XYZ response. XYZ response is the integral of the spectral response under each of those CIE filter curves. In a traditional tristimulus sensor - you would get that from three perfectly CIE matched optical filters - which are hard to make to match CIE. The Spyder design (and the KM CS-200 design wiht 5x more filters) - you are taking the optical responses of multiple filters - applying DSP and calibration tables to come up with the XYZ response thru curve fitting of data.

So according to that diagram - filter 7 is good for >640nm, filter 6 is good for >600nm - take their integral difference and you know the energy in band 600-640 (roughly as it depends on each diode and filter performance) Above 640nm CIE has some X, little Y and no Z response - yet the Spyder filters pass all this spectra. This mean deep red/IR band is digitally filtered rather than optically - thus the external filter which optically converts all of the edge filters into essentially passband filters to avoid the digital noise of doing a rolloff.

The whole point of XYZ filters concept is that it can measure something that is not spectrally matched to your reference source and tell how they appear different to the eye - it is the integral response that matters. The error comes where the reference and actual curves maximally differ only if there is significant spectral energy for the source at those wavelengths such that the integral response is off.

They key in all this is do you get the same integral response when you apply the resultant filters - and that depends on the calibration tables in the Spyder - which Derek has investigated and found many samples all had the same tables. It is unlikely that individual diodes and filters have the same response - so the error lies in the poor factory preset - it needs calibrated to be accurate. They are going to charge way more for the Platinum sensor even though it is the same hardware - you pay for that calibration. It does not matter how good the sensor design is if it is not calibrated - just like your TV.

Keep in mind those patent charts are demonstrating an example - who is to say your sensor is over/under the IR part of the X curve - the point is - your sensor is different.

No you can indeed measure different spectra - that is the whole point of simulating standard eye response rather than a standard CRT response. The eye sees different spectra that have the same integral response as the same color. So only if the source spectra has significant energy where the XYZ curve approximation is large integral error would this be a problem.

Based on this patent example - certainly large IR spectra would not match very well because that edge is digitally created without the external filter to roll it off. And the peaks of the XYZ in this example have some error - but the source spectra would need to have its maximum energy at those wavelengths for this to show. Surely for this example lasers at 460, 560, 600 would have a lot of error - but you would read lasers at 470, 570, 620 without error. So the spikier the spectra - the greater possibility of error - but that does not guarantee error. The more broadband the spectra - the greater chance of these wavelength errors canceling out.

The KMCS200 has more on this in their datasheet. With more sensors you get a tighter curve fit - and they qualify it for all kinds of display types - yet this is a colorimeter- it is not a spectroradiometer. But since they have the tighter curve fit even the narrow band primaries will measure very well.

http://www.tequipment.net/pdf/Minolt..._datasheet.pdf

But this is true even of spectroradiometers whose sampling rate is too low - and certainly true of colorimeters with even fewer filters that clearly had no money spent on finding three optically perfect X, Y, Z filters (if they even exist)! Factor in nonlinear optielectronic response as far as working at low light levels...

The Spyder flaws are not one of design - they are ones of default factory calibration.

Well - you cannot have mine - for the lamp I tested it on it fought the KM instrument to the ten-thousandth place - which is good enough for FPTV. And when I reviewed the Ruby I got same xy primaries from a PR650 based review. Hopefully Spyder Platinum solves the calibration - the patent points out that their technique allows the sensor to be manufactured cheaply (diodes, filters, plastic) - the cost is in calibrating it - especially if you want traceability. Not something you will find in the fleabay $99 special. People should now understand why calibrated sensors are expensive - especially if you want to measure dark narrow band primaries.

The cross sensor corrections you get in the software is not going to be good as calibrated XYZ curve generation in the firmware - for the same reasons you mentioned - it will be good for the similar lamp and brightness.

Wow, this is a great discussion.

Zoyd wrote:

You can really think of it as TWO sensors. There are TWO separate coefficient tables, and one is used with the filter ON, and one is used with the filter OFF. They are BOTH designed to match CIE standard observer curves...it's just that the filter reduces IR emissions and off-axis light, and with the filter removed, we get more light transmission to the detectors.

If you're really curious about how all of this works, check out the calibration FAQ:
http://www.spydertvpro.com/faqs2.php