Machine Vision Lens Comparison: Focal length range 23 mm to 25 mm. From Schneider, Ricoh Pentax, Tamron, Edmund Optics and Fujinon

I typically use Fujinon HF sereis lenses.  They are good quality construction and cost under $150 new.  I don't have any complaints, but how do they compare to the rest of lens manufacture's products?  There are more expensive lenses, such as the Schneider Xenoplan at $900+.  What does that extra money get you?  There are some new designs, such as the Edmund Optics C series and UC series.  Do the new lens designs learn from and improve on the old designs?  Is there a better lens than that I use?  To answer that I tested lens performance with two targets.

To measure distortion I used a dot grid and LabView to calculate lens distortion.

To measure sharpness I used a star target.  The maximum edge strength vs line pair per mm is used to compare the center to the edges.

http://stores.ebay.com/vitamall

The targets can be purchased from the ebay store Vitamall 

ebay is a great place to get used test equipment on the cheap.  All the lenses were from my old stock, or purchased for this test on ebay.

The same camera was used for all tests.
Camera: JAI BM-500GE   2/3" CCD     3.45 um pixels   2456 x 2048

All Lenses were tested at F/4.0


*Note: Bottom Right lighting had a problem.  The contrast was low on all bottom right star target inspections.  It is represented as the cyan curve in the graph.  This was an error I failed to correct for.


Edmund Optics 25 mm/F1.4 59871

Focal Length 25 mm
Working Distance: 860 mm
$295
Distortion: 0.05911 %
30% contrast @ 2.3 lp/mm to 2.7 lp/mm

Note, right side lighting had a problem.  The over-all contrast on the right stars is lower than it should have been.

Fujinon HF25HA-1B

Focal Length 25 mm
Working Distance: 860 mm
$120
Distortion: 0.07101 %
30% contrast @ 2.0 lp/mm to 2.4 lp/mm

Ricoh FL-CC2514-2M  (Pentax brand)

Focal Length 25 mm
Working Distance: 860 mm
?$150
Distortion: 0.06061 %
30% contrast @ 2.2 lp/mm to 2.6 lp/mm

Schneider Xenoplan 1.4_23-0501

Focal Length 23 mm
Working Distance: 770 mm
?$900 (Discontinued)  ($1,275.00)
Distortion: 0.07062 %
30% contrast @ 2.3 lp/mm to 2.7 lp/mm

Schneider Xenoplan 1.4_23-0912

Focal Length 23 mm
Working Distance: 770 mm
$1,275.00
Distortion: 0.06831 %
30% contrast @ 1.8 lp/mm to 2.6 lp/mm

Tamron 25mmF1.4 23FM25SP (B201528) 2/3" format

Focal Length 25 mm
Working Distance: 860 mm
$157
Distortion: 0.06801 %
30% contrast @ 1.7 lp/mm to 2.2 lp/mm

How much does an Near Infrared (NIR) optimized image sensor matter?

Basler offers four 1.3 MP sensors in the Ace series monochrome cameras.

acA1300-60gmNIR - Basler ace

Available 1.3 MP sensors are:
Sony ICX445
ON Semiconductor  PYTHON1300
e2V EV76C560
e2V EV76C661  (NIR)

I had an opportunity to compare three of the four, the Sony ICX445 and both e2V sensors.

Since the EV76C661 is optimized for  the Near Infrared (NIR) spectrum the test used IR at 950nm.

I wanted to see if the NIR optimized chip was really faster, and by how much.  Additionally, would the color neutralization of IR be different based on the sensor.

xRite color chart used as a standard for setting exposure and gamma between cameras.

As expected the e2V EV76C661 sensor outperformed the others.  The surprise was that the exposure times were over 2x faster than the others.  It was 2.3x faster than the e2V EV76C560, and 2.66x faster than the Sony ICX445.

The images below have the same brightness, same working distance, same light, the same IR bandpass filter and the same iris diameter was used between cameras.  The only thing that changed was the exposure time.
Note that isn't the same F/stop because the sony required a 16mm lens, while the e2V's used a 25mm lens.  To be fair I used the 25mm lens at f/4 to calculate the 6.25mm iris diameter, then set the 16mm lens to f/2.5 in order to create a 6.25mm diameter.

e2V EV76C661 (NIR)  exposure 7100us  25mm Lens f/4.0

e2V EV76C560 exposure 16600us  25mm Lens f/4.0

Sony ICX445 exposure 19000us  16mm Lens f/2.56

The sensor had no effect on the color sensitivity.  Again, that is expected as the color pigments depend more on the wavelength of light than the sensor, this just proves that.

Here's a comparison of each sensors spectral sensitivity.

And the one not tested:

One last take-away, The Sony sensor did really well compared to the noraml e2V chip.  The Sony pixels are much smaller, 3.75um compared to 5.3um. So with 41% more area the e2V EV76C560 was only 14% faster.