FO Power Meter Calibration Uncertainty

 

FOA is often asked why two different fiber optic power meters differ in readings. To understand this measurement uncertainty, you should start by reading the FOA Online Reference Guide on optical power measurement and calibration of meters.

In 1983, NBS (now NIST) set up a program for FO power calibration in their FO group. All told, they spent five years on the program, working with semiconductor detectors to understand their characteristics, developing a way to transfer their ECPR standard to something portable that could be sent out to people desiring direct traceability, and fighting with everybody about what wavelengths to calibrate at (which culminated in a meeting of the US military standards types in Phoenix where Bob Gallawa and I (Jim Hayes) finally won the case to have calibration only at 850, 1300 and 1550 nm.)

NBS offered good transfer standards with a H-P lab meter as a transfer standard that is shipped to the user's cal lab to establish traceability. But we still have a major lack of understanding on what fiber optic power measurement uncertainty really means. FOA's Jim Hayes was educated as a physicist/astronomer and trained in metrology techniques by an astronomer/mathematician who really understood astronomy as a "bootstrap" science where everything was inferred, not measured, and understanding metrology techniques was mandatory to producing reliable data. So he was able to help set up the NBS standard.

In any power meter measurement uncertainty comparison today (and NIST (new name for NBS) will crucify you if you say accuracy!), you have to understand the trail of calibration. NIST has a primary optical power "standard" in the basement of 325 Broadway in Boulder, CO, a ECPR (electrically calibrated pyroelectric radiometer) that measures optical power by comparing it to the heating power of a resistor, which can be accurately calibrated.

But the ECPR only works at high levels, so a high power laser is needed for an optical source for transfer. To also get away from the uncertainty of coupling through a fiber, they use lasers (HeNe for 1550, YAG for 1300 and gas for 850) and optical splitters to transfer power to a laboratory meter to use as a transfer standard.

The laboratory meter is used to calibrate our working standards which we use for calibrating the instruments we sell. Transferring the standard to each instrument adds to the calibration uncertainty. Each instrument has it's own errors, plus we have the errors caused in the transfer process: source wavelength, spectral width and stability, variations caused by the connection to the source over a fiber optic cable (just bending losses are very critical on a lab bench!) , not to mention  contamination by dirt.

 

Overall we can show the error buildup in a table:

Cal Std	Uncertainty	Total Uncertainty	Cause of uncertainty
Primary: ECPR	±1% absolute	±1%	ECPR
Transfer: H-P	±1% transfer	±2 %	Transfer coupling, instrument error*
Working: FOPM Manufacturer	±1% transfer	±3 %	Transfer coupling, instrument error*
Sold Product	±1% transfer	±4 %	Transfer coupling, instrument error*
Manufacturers spec. (conservative*)		±5%/See below*

* Including other factors:

• Source wavelength
• Spectral width
• Detector sensitivity as a function of wavelength
• Temperature
• Fiber coupling
• Fiber movement

So any instrument calibrated by this process has an uncertainty of + or - 5% compared to the NIST absolute standard. And any two instruments in the field can expect a worst case variation in measured values of 10%, since they could differ from the NIST absolute standard by 5% in opposite directions! In the real world, these error are not usually that big, they RMS out (some + and some -) to make the actual error less, but this worst case scenario is completely feasible with two instruments which meet their specifications!

Note: Another problem in meter readings of power being different comes from meters themselves. Some inexpensive meters offer a self-calibration option. You can push a couple of buttons and calibrate the meter however you want. That is not proper calibration. A properly calibrated meter is traceable to a national standard and sealed with a dated sticker indicating the date of calibration. These inexpensive meters are possible problems when measuring transmitter or receiver power since you do not know if the calibration is traceable and trustworthy.

 

 

Jim Hayes, President, The FOA