Cassar S vs. El Nikkor 105mm Comparative Focal Shift

[bvf]

In a couple of recent posts I presented some experimental results about how the point of focus (i.e. at the subject, in front of the lens) of the Cassar S and El Nikkor 105mm lenses changed with different wavelengths from UV to IR.

 

In this post I am presenting experimental results about how the focal point (i.e. behind the lens) changes over the same frequency range. This is probably a better metric, and easier to understand. A couple of differences in how the information is being presented here compared with the earlier posts:

1. Results for the 2 lenses are being presented in the same post to make it easier to compare.
    
   
2. Each graph now covers the whole spectral range, rather than having separate graphs for UVA, Vis, and IR.
   

The vertical y axes in the graphs have the same scale for both lenses to allow comparison, but the horizontal x axes may have different scales.

 

The filters used for the different measurements were:

 

UV:

315BP25

345BP25

380BP20

 

Visible:

Midwest Optical BP470 (Blue)

Hoya X1 (Green)

Hoya R25A (Red)

 

IR:

Hoya R 72 + MidWest Optical BP735

MidWest Optical BN850

Midwest Optical LP1000

 

1. Change of Focal Point against Wavelength for various Lens-Subject Distances

 

In these graphs, wavelength is on the horizontal x axis, focal shift (in centimetres) on the vertical y axis. The different curves are for different lens-subject distances: the values of these lens-subjct distances are different for the two lenses.

 

The Cassar generally has a lower focal shift than the El Nikkor, except at very close up and at 315nm, where it goes a bit wild. However, the Cassar S focal length is half that of the El Nik 105, so changes in focal point have a greater effect.

 

One of the strange things about the El Nik 105 is the 345nm measurement. At this frequency the focal shift is negative (i.e. the focal point is closer to the lens), whereas for all other measurements it is positive. This is not an error of measurement, it is really happening.

 

 

2. Change of Focal Point against Magnification for the various Wavelengths

 

These graphs effectively take account of the difference in focal length of the two lenses by relating the focal shift to image magnification/size: magnification=0 represents a subject at infinity.

 

Magnification is on the x axis, with high magnification to the left. The y axis shows focal shift in cms. The differenrt curves are for the different wavelengths.

 

Both lenses perform well in the visible region that they were intended for. Looking at the full wavelength spread, the Cassar is better at lower magnifications (larger lens-subject distances) and the El Nikkor is better at higher magnifications (i.e. closer up) - this reflects the purposes that the lenses were designed for. Both have problems at 315nm - especially the Cassar S.

 

Both lenses perform well in the visible range.

 

The x axis scales are slightly different in the two graphs.

 

 

 

3. Change of Focal Point against Lens-Subject Distance for the various Wavelengths

 

These graphs show how the focal shift changes at different lens-subject distances. (Note that the x axis scales are different in the two graphs.)

 

Lens-Subject distance (in cms) is on the x axis, and focal shift (also in cms) is on the y axis. The various lines are for different wavelengths.

 

The Cassar S looks better, but again its shorter focal length means that focal shift will have a greater effect on sharpness. The Cassar 315nm curve goes ballistic at arround 15-18 cm lens-subject distance: the El Nik 105 graph does not cover the range below 20 cm, but it too is showing signs of reaching for the sky.

 

These graphs again show how the performance in the visibile range is good for the two lenses.

 

[OlDoinyo]

The focal point chagnge graphs show homologous features at similar magnifications despite coming from very different lenses. Why should this be? It is unclear from this post what is used as a reference when computing focus shifts. Could these features be artifacts of the reference itself?

[bvf]

The focal point chagnge graphs show homologous features at similar magnifications despite coming from very different lenses. Why should this be? It is unclear from this post what is used as a reference when computing focus shifts. Could these features be artifacts of the reference itself?

 

All the data was derived using the same full-spectrum Sony A6000. The reference focus position was derived using a UV/IR cut filter on the camera, i.e. where you would focus in normal visible light photography. The Red, Green, Blue visible mesurements used the filters I indicated also stacked with the UV/IR cut filter to avoid any influence from the high IR sensitivity of the camera (esp. with the Red R25A filter).

 

In terms of features that are common to both lenses, some of these you would expect. For example, the good performance in the visible range would be because both lenses were designed to be used in visible light. The higher focal shift at short distances and higher magnifications is going to be down to laws of optics - I'm sure someone here would have chapter and verse on that. I suspect the high shift for 315nm can also be explained by optics theory.

 

But you are right, there are other common features which are surprising. For example, in the first pair of graphs both lenses have bumps at 750 and 380nm, and both have a trough at 345nm (although far more pronounced with the El Nikkor).

 

On the other hand there are a number of significant differences between the two lenses - for example the direction of shift at 345nm, and the vertical order of lines in the first two graphs.

 

There is room for error in all the measurements because they relied on my assessment of when the best focus was achieved and what the position of the helicoid was to achieve this. (I have tick marks at 1 cm spacing around the helicoid and estimate the amount of helicoid turn to the nearest mm. The helicoid turn is converted to forward movement using a factor calculated from the total extension of the helicoid and the number of cms of rotation to achieve this. If that all makes sense.) So where a single reading doesn't fit in with the others (e.g. the trough in the 345nm line at a magnification of about 0.175 on the fourth graph), that could be a measurement error.