Maker Pro
Maker Pro

Lab Grade Accuracy?

R

Roger Breton

Jan 1, 1970
0
What is it that defines a "Lab Grade" instrument, like a colorimeter, as far
as acuracy is concerned? I read on one site that the UDT SLS9400
Tri-stimulus colorimeter is (or was) the first handheld colorimeter to
provide "lab-grade" accuracy and precision.

Is it a function of how well the instrument replicates the 1931 2 degree
Standard Observer sensitivity? And does that translate into a plus or minus
chromaticity accuracy like +- 0.00015, for instance? And do companies go
about characterizing the degree to how well a particular set of filters
makes an instrument adhere well to the 1931 2 degree Standard Observer
curves: with shining the light off a monochromator directly at the
instrument and observing its response at every wavelengths in the visible?

Roger Breton
 
V

Victor Roberts

Jan 1, 1970
0
What is it that defines a "Lab Grade" instrument, like a colorimeter, as far
as acuracy is concerned? I read on one site that the UDT SLS9400
Tri-stimulus colorimeter is (or was) the first handheld colorimeter to
provide "lab-grade" accuracy and precision.

Advertising. There is no standard definition of "lab-grade". It will
mean different things to different people.
 
T

tlianza

Jan 1, 1970
0
Roger,

I disagree "a bit" with Vic Roberts. I've designed colorimeters that have
sold in the 10's of thousands unit per year. I can guarantee to you that
none of those were "lab-grade". While it is true that some people get a
little crazy in the spin about a product, a "lab grade" product will exhibit
at a minimum the following characteristics.

1. A notice of traceability to a NATIONAL standard. That standard will
generally have certification that is NIST (USA), (I'm sorry that I don't
remember the Canadian lab designation), JIS (Japan), DIN (Germany). or
some other standard. This traceability will include a measure of
statistical uncertainty that is associated with the original standard and a
measure of compound uncertainty that relates the measurement instrument in
question to the standard. Current recommendations, seem to imply expressing
uncertainties in terms of a compounded 2 sigma measure expressed at full
scale. NIST publishes a set of guidelines for this presentation of data.

2. If the device is a colorimeter, there is a further set of measures that
relate to the root mean squared difference between the color matching
function and the compound spectral sensitivity of the device as SHIPPED.
This means that each device is chracterized. The UDT is not one of those
cases, although they did publish their filter response and error for in a
paper, bu not for each instrument. I believe that this has a terminology
such as "f1 squared", or something like that. Danny Rich could help me out
here...If he is reading.... The problem with this specification is that it
is VERY difficult to measure and even more difficult to apply a statistical
uncertainty to the measurement. A lab grade instrument would present that
data as well.

The only "lab-grade" colorimeter instrument that I have come in contact with
is the LMT colorimeter from Berlin. All the rest are amateurs...


--
Tom Lianza
Technical Director
Sequel Imaging Inc.- A GretagMacbeth Company
25 Nashua Rd.
Londonderry, NH 03053
 
R

Roger Breton

Jan 1, 1970
0
There is no such definition. Some marketeer came up with this and the
engineers were powerless to prevent it from going into the brochure.


Very often, a reputable vendor will tell you what tests they adhere to
if you ask.

I don't suppose there isn't an easy method of assessing the degree of
conformity of a colorimeter to the 1931 Standard Observer sentivities, is
there? I will often see in the brochure on an instrument a set of curves
showing where the deviation from the standard observer occurs. But I am
intrigued as to how, procedurally, to go about testing for this.

I have Wyzecki & Stiles but I have not found any reference to this in the
whole book.

Regards,

Roger Breton
 
D

Danny Rich

Jan 1, 1970
0
I concur with Tom's statements and assessment. The list of labs is CNRC in
Canada and NPL in England. DIN is a standards organization but not a
calibration or certification lab. In Germany, certfication is offered by
either PTB in Braunsweig or BAM in Berlin.

Danny

tlianza said:
Roger,

I disagree "a bit" with Vic Roberts. I've designed colorimeters that have
sold in the 10's of thousands unit per year. I can guarantee to you that
none of those were "lab-grade". While it is true that some people get a
little crazy in the spin about a product, a "lab grade" product will exhibit
at a minimum the following characteristics.

1. A notice of traceability to a NATIONAL standard. That standard will
generally have certification that is NIST (USA), (I'm sorry that I don't
remember the Canadian lab designation), JIS (Japan), DIN (Germany). or
some other standard. This traceability will include a measure of
statistical uncertainty that is associated with the original standard and a
measure of compound uncertainty that relates the measurement instrument in
question to the standard. Current recommendations, seem to imply expressing
uncertainties in terms of a compounded 2 sigma measure expressed at full
scale. NIST publishes a set of guidelines for this presentation of data.

2. If the device is a colorimeter, there is a further set of measures that
relate to the root mean squared difference between the color matching
function and the compound spectral sensitivity of the device as SHIPPED.
This means that each device is chracterized. The UDT is not one of those
cases, although they did publish their filter response and error for in a
paper, bu not for each instrument. I believe that this has a terminology
such as "f1 squared", or something like that. Danny Rich could help me out
here...If he is reading.... The problem with this specification is that it
is VERY difficult to measure and even more difficult to apply a statistical
uncertainty to the measurement. A lab grade instrument would present that
data as well.

The only "lab-grade" colorimeter instrument that I have come in contact with
is the LMT colorimeter from Berlin. All the rest are amateurs...


--
Tom Lianza
Technical Director
Sequel Imaging Inc.- A GretagMacbeth Company
25 Nashua Rd.
Londonderry, NH 03053

as
 
D

Danny Rich

Jan 1, 1970
0
The procedure is newer than Wyszecki and Stiles. There is a CIE committee
developing a standard but there is some dissention on the "goodness" of the
metric. The test is to illuminate the sensor with monochromatic light of
known intensity. You read the light and compare the reading to that of a
theoretically perfect standard observer move to the next wavelength and
repeat. You would then compute the root mean square difference or the
relative percent difference between an ideal colorimeter and the actual
colorimeter.

Danny
 
Top