G.E. Slobodzian, Director of Engineering, Retired, Ophir-Spiricon
It seemed that everyone had their own “correct method” for how a laser beam should be characterized. The two measurements requested most frequently were Beam Width and Divergence. A host of other measurements were also requested, such as: Gauss fit, Tophat, Power in a Bucket, Ellipticity, Orientation, Peak Fluence, Centroid, etc. In some cases, only one figure of merit was important; often Gauss fit was highly regarded.
In the late 80’s and early 90’s CID and CCD camera technologies were coming into use as a practical approach for capturing a laser beam’s intensity profile in two dimensions and in real time. Many cameras employed external triggering and variable exposure features and thus were capable of imaging both CW and Pulsed laser beam profiles. This new ability to “see” what their laser was doing caught on and pretty soon became a “must have” in the laser guru’s toolkit.
Of course seeing was not enough to satisfy every need, and thus measurement became equally important. The camera technology led to new algorithms covering many, if not most, of the measurement topics mentioned above. Competing manufacturers using similar technologies employed various algorithms to make one or more of the above measurements. While certain beam width measurements were commonly requested, the algorithms employed were often unique to each investigator and thus manufacturer.
Two problems posed by all cameras were their limited signal to noise and the difficulty in establishing a stable zero, or baseline, from which to measure the beam intensity profile. These limitations imposed different challenges depending upon the method employed for making a beam width measurement. Some of the most popular beam width measurement methods were and still are:
- 13.5% of Peak (actually performed in different ways)
- 86.5% of Total Power/Energy beam widths
- 1/e², also called second moment or D4σ (Dee-Four-Sigma)
- Full Width Half Max, also known as 50% of Peak (also performed in different ways)
- 86.5% of Power in a Bucket, beam diameter, also called encircled power
Methods A, B, and E would each yield C if the laser beam mode was a pure TEM00 and if the percentages were set as shown. If the beam contained higher modes then A, B, and E would not yield C and the degree of error depended upon the mix of higher modes.
In the early days customers would tolerate significant errors in the beam width measurement accuracy. But demands for improved accuracy and repeatability soon grew. During this same period the desire to make accurate second moment beam width measurements also began to dominate. The second moment beam width conformed to laser beam propagation theory, and thus, over time, became the de facto industry standard. However, making this measurement with relatively noisy cameras with undefined (zero) baselines is a challenge and remains so even with the latest camera technologies.
Beam width methods A, B, D, and E all employed a clip level technique where camera pixels above a certain specified value were included in making a beam width measurement. Pixels below the clip level were excluded. The clip levels were typically well above the noise floor and thus measurements were fairly immune to noise so long as a reasonably good baseline could be determined and the beam spot size filled, but did not overfill, the camera imager.
This is not the case when calculating the second moment D4s beam widths. The second moment calculation includes a mathematical term that is the square of the distance that a pixel lies from the beam’s center times the pixels intensity. So not only does the distance from the center increase but the population of pixels located at greater distances also increases. In theory both increase to infinity, but in practice are limited to the area of the imager. If the laser beam is small relative to the size of the imager, these far out pixels can seriously impact the beam calculation. This is true even if offset from zero by a fraction of an intensity count, and even more so if all the noise is made up of only positive values, i.e. when clipped.
In order to tame the possible runaway second moment results, various constraints are employed. Limiting the area over which the calculations are performed is one part of the solution. These aperture-applied techniques were clearly necessary for small beams on a large imager. Various rules for how to establish the aperture size were needed and algorithms had to be defined to create them for many shapes and sizes of beams.
Dealing with the baseline was a more complicated problem. Beam analyzer manufacturers were already familiar with employing clip levels. First attempts involved setting a clip level above or near the noise floor, thus eliminating the noise in the wings of the beam. The problem with this method was that the clip level needed to be constantly adjusted in order to obtain a correct result. As the beam intensity varied the beam width would modify unless the clip level was also adjusted. If the above described aperture increased or decreased the beam width could also change. In effect the clip level approach allowed you to dial in an answer, but it may only work correctly for the beam profile currently being measured. The search for the magic clip level was ruled a dead-end here at Spiricon, however it remained a popular solution for other beam profilers.
In the 90’s, we at Spiricon spent a serious amount of time trying to find a solution to the baseline and the aperture problem. An ISO standard did not exist in those days. Cameras were analog, required external digitizers that were setup typically for 640x480 (width x height) pixels at 8 bits per pixel. Pixel sizes were typically 13- 20μm and the camera imagers were largely CCD interlaced frame transfer designs. The data Dr Roundy used in his first version of this paper were modeled upon this type of camera. This update of that paper will discuss and model newer camera technologies and show some of the results in a somewhat different way; remove some things, and add some new insights.
Today’s modern CCD cameras typically contain millions of pixels rather than a few 100k. Most are progressive scan of interline transfer design. The older frame transfer technology is getting phased out. The pixel sizes are also much smaller, now in the low single digit μm sizes. However these larger imagers can also create new problems. The likelihood and number of bad pixels has increased as has the occurrence of pixels that twinkle. Shading in the output image may be greater and more common. Much larger imagers are also available, making it possible to offer sizes up to that of a 35mm format film.
We have looked at a number of CMOS imagers over the years. The earliest devices were designed more for low-end commercial applications. They were typically rolling shutter designs (not suitable for pulsed lasers); had very low signal to noise ratios; poor pixel-to-pixel uniformity; poor linearity of response; and unstable background black levels. Most of the earlier CMOS imagers did possess one noticeable positive trait; they tended to bloom less, or not bloom at all, when used with YAG lasers. They were also lower cost as compared to their CCD counterparts. We were hopeful that, over time, a suitably improved, lower cost, CMOS imager would be developed.
The most recent crop of CMOS imagers has solved many, but not all of the problems observed in the early examples. They are now triggerable, allowing for use with pulsed lasers. The temporal signal to noise is approaching CCD’s and linearity is also comparable. The lack of blooming at YAG wavelengths has not turned out to be real, but rather just somewhat different than with CCD’s. Perhaps the blooming was there all along and is now easier to detect, or perhaps it is exacerbated by the new buffered designs that enable triggering. Baseline instability remains a serious problem, making it difficult to reliably subtract it out, and pixel to pixel instabilities are still a concern. As these devices have improved their cost has also risen. If they reach parity with CCD performance, perhaps their costs may also reach parity.
During the early 90’s Spiricon developed its first pyroelectric based camera; the Pyrocam. This technology was built on our earlier 1D and 2D pyroelectric imagers. This technology opened up camera quality imaging of UV, NIR, and FIR lasers. Now in its fourth generation it employs a Gig-E interface and has imagers up to 25mm square with 320x320 pixels.
Also during the intervening years imagers constructed using InGaAs focal plane arrays have become more numerous. In their native state these devices exhibit serious uniformity issues. However the camera manufactures have incorporated sophisticated algorithms within the cameras or their drivers to correct these problems. Each camera either ships with or has embedded in its firmware a NUC (non-uniformity correction) file that handles bad pixel correction, flattens the dark field and equalizes the uniformity. They do such a good job that temporal signal to noise is typically in the high 60dB range. They also are designed for the NIR so they handle YAG and Telecom wavelengths without any detectable blooming. Their downside is that they are low resolution, typically 320x256 or 620x512, and the smallest pixel size is in the 20μm range on the higher resolution and 30μm range on the lower. They also operate best when thermally stabilized. They are very expensive, in $15k - $40k range depending upon resolution. Some imagers also exhibit non-linear response that varies with exposure times. Thus one must be careful in selecting these cameras for performing serious measurements. Because these imagers can be used in various night vision applications their sales are ITAR regulated here in the US and come under similar restrictions in other countries, especially Europe.
One very convenient feature of modern cameras is that most have eliminated the need for a frame grabber. The newer cameras began by adopting 1394A the B (FireWire) technology, and soon it was challenged by USB2, then USB3, and now by gigabit Ethernet.
One overriding fact that has been hard learned is that good measurements begin with high quality cameras. Not necessarily expensive cameras, but rarely cheap cameras. Spiricon has developed tough specifications and strenuous testing methods and put all cameras through serious evaluations before incorporating them into our beam analyzers. Even with many years of testing behind us, we continue to learn new things and develop new techniques. Not all issues are 100% resolvable, but every effort is made to ship the best value for our customer’s dollars.
When Spiricon developed its baseline correction and auto-aperture methods in the mid 90’s, ISO standards for laser beam measurements were still years off. Having invested many man-hours in laser beam modeling and performing controlled analysis we felt motivated to try and patent the method that has been trademarked as Ultracal™. Two US patents have been issued to Spiricon that cover these techniques.
Ten years after our patent was granted, ISO-11146-3 First edition 2004-02-01 recognized our Ultracal baseline correction method and incorporated it into section 3.2. Additional ISO definitions and methods were also developed and some of the methods for making beam measurements began to become standardized. In some cases Spiricon was ahead of the curve and already making measurements that were ISO compliant. In other cases we adopted the new methods thus expanding the number of options available to our customers.
In Part 2 of this paper we will demonstrate the importance of baseline subtraction by employing our BeamMaker modeling utility. BeamMaker is part of every copy of our BeamGage software. This tool is available to all of our end users, and even to those who simply download our software for evaluation. Thus many of the modeling examples that will appear in this paper can be replicated by anyone interested in doing so.
- Part 2: Baseline Methods and Mode Effects
- Part 3: Beam Aperturing Methods
- Part 4: Scanning Slit vs Camera Based Beam Analyzers
- Carlos B Roundy, PhD; Techniques for Accurately Measuring Laser Beam Width with Commercial CCD Cameras, Spiricon, Inc., Logan, Utah 84341 USA, Presented At 4th International Workshop on Laser Beam and Optics Characterization
- Amiel A. Ishaaya, Nir Davidson, Galina Machavariani, Erez Hasman, and Asher A. Friesem; Efficient Selection of High-Order Laguerre–Gaussian Modes in a Q-Switched Nd :YAG Laser, IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 39, NO. 1, JANUARY 2003
- CCD IMAGE SENSOR NOISE SOURCES, August 3, 2012 Reference Document Rev 1.0 PS-0047, TRUESENSE imaging, inc.
- Albert J.P. Theuwissen, Kleine Schoolstraat; Influence of Terrestrial Cosmic Rays on the Reliability of CCD Image Sensors, © 2005 IEEE
- ISO 11146-1:2005(E), Lasers and laser-related equipment —Test methods for laser beam widths, divergence angles and beam propagation ratios — Part 1:Stigmatic and simple astigmatic beams.
- ISO/TR 11146-3:2004(E), Lasers and laser-related equipment — Test methods for laser beam widths, divergence angles and beam propagation ratios — Part 3: Intrinsic and geometrical laser beam classification, propagation and details of test methods
- ISO/TR 11146-3:2004/Cor.1:2005(E), Lasers and laser-related equipment — Test methods for laser beam widths, divergence angles and beam propagation ratios — Part 3: Intrinsic and geometrical laser beam classification, propagation and details of test methods
- Ram Oron, Nir Davidson, Asher A. Friesem, Erez Hasman; Chapter 6 Transverse mode shaping and selection in laser resonators, E. Wolf, Progress in Optics 42 © 2001 Elsevier Science B. V
- A. E. Siegman; How to (Maybe) Measure Laser Beam Quality, Tutorial presentation at the Optical Society of America Annual Meeting Long Beach, California, October 1997 10. Siegman, Sasnett, and Johnston; Choice of Clip Levels for Beam Width Measurements Using Knife-Edge, IEEE Journal of Quantum Electronics, Vol. 27, No. 4, April 1991