Nikon 300mm f/4 tripod collar revisited

Started Feb 23, 2013 | Discussions thread
Flat view
WGHalvorsen Contributing Member • Posts: 613
Nikon 300mm f/4 tripod collar revisited

Some years ago, Bjørn Rørslett posted a review of the Nikon 300 mm f/4 lens on his website, He criticized the stock tripod collar for being dynamically weak and showed convincing evidence of image blurring in photographs taken at 1/15 sec exposure time. He also showed two quick-and-dirty solutions for stiffening the collar by wedging objects between the foot and the lens.

Bjørn's assessment was discounted by several individuals. Notably, Leonard Shepherd said in this forum that "the Nikon collar does everything it should" on 25 Oct 2007 and that Bjørn did not understand "the physical parameters that apply to testing tripod collars", in a post on 27 Oct 2007.

I thought it would be worthwhile to revisit the subject of tripod collar dynamic stiffness, and do a little more in-depth investigation, given the greater sensitivity of today's cameras, with their increased pixel density, to camera shake.

The problem is illustrated in the following two 100 percent cropped photos, one taken with the stock Nikon collar (purchased in 2011) and the other with the Kirk replacement collar for the 300 mm f/4 lens, both at a shutter speed of 1/30 sec. The photos were captured with a Nikon D800 and a TC 14 converter, with the lens collar foot mounted on a Novoflex CB5 ball head on a Gitzo GT5562LTS tripod. Image blurring due to camera shake is quite bad with the Nikon collar but is not evident with the Kirk collar. Blurring was essentially eliminated with both collars when the shutter speed was increased to 1/200 sec.

The next figure shows the time history of vibration measured in the vertical direction at the free end of the lens with a miniature (0.7 g) accelerometer, with the lens mounted on the stock Nikon collar. The raw accelerometer signal was filtered with a zero-phase 5-40 Hz bandpass digital filter. Vibration in the lateral direction was quite similar but with a somewhat reduced amplitude. The period of the oscillation is about 50 msec, so there is more than one complete peak-to-peak vibration excursion during the 33 msec exposure time with a 1/30 sec shutter speed. The peak-to-peak displacement amplitude is about 25 microns, equal to about 5 pixels in the D800 sensor. The corresponding amplitude with the Kirk collar was about 4 microns, less than one D800 pixel, with an oscillation period of about 27 msec. The duration of relatively strong vibration with the Nikon collar is about 300 msec; it was less than 100 msec with the Kirk collar.

The Power Spectral Density function of the vertical vibration is shown in the following figure. When comparing response at different frequencies in such a plot it is important to keep in mind that here we are looking at acceleration, but imaging errors are directly related to displacement rather than acceleration. Displacement amplitude diminishes with frequency-squared compared to acceleration in linear spectra and with frequency to the fourth power in PSD spectra. Thus the 20 Hz peak in the vertical response in the plot is about 3 times stronger in displacement than the 32 Hz peak, although they have nearly equal PSD amplitudes. The dominant peak in the PSD spectrum for the Kirk collar was at about 37 Hz.

To further investigate the dynamic response properties for the two collars, I performed artificial excitation frequency response measurements. This type of measurement involves monitoring vibration to a controlled and measured excitation force, so that a measure of dynamic stiffness is obtained. The following figure shows the frequency response function, in terms of the ratio of acceleration to applied force, for the Nikon collar with excitation and response in the vertical direction at the free end of the lens. The function is dominated by a strong resonance peak at about 20 Hz, which correlates very well with the vibration response with shutter actuation.

The vibration mode shape in the vertical plane associated with the 20 Hz resonance was measured by applying the dynamic force at different points along the length of the lens and on the collar foot, with the results indicated by the heavy red lines in the following photograph (the yellow line indicates static position). The mode primarily involves rotation of the lens at the collar with respect to the foot, which is why Bjørn's quick-and-dirty solutions, adding stiffness between the lens and the foot, were effective.

The corresponding frequency response function for the Kirk collar is plotted with a heavy blue line in the following figure, with the Nikon collar function plotted with a narrow red line for comparison. The dominant resonance mode has been shifted up to about 37 Hz with the Kirk collar and the peak dynamic stiffness is about 6 times greater, correlating well with the vibration response to shutter actuation.

Image blurring due to camera shake can be expected when the following two conditions are met:

1. The vibration displacement amplitude of the camera is high relative to the sensor pixel size.

2. The shutter open duration approaches one-quarter of a vibration period at the high-speed end and is not much greater than the duration of relatively strong vibration at the low-speed end.

With the stock Nikon collar those conditions are easily met for the D800 camera with shutter speeds ranging roughly from 1/160 sec to 2 sec. Those conditions are never met with the Kirk collar, due to the lower vibration amplitude. Clearly, the stock Nikon collar design is deficient as far as camera shake is concerned.

As an additional note, Leonard Shepherd suggested that using proper long-lens technique solved the camera shake problem. Indeed, I was able to increase the peak dynamic stiffness by about a factor of three by tightly gripping the lens and eliminate image blurring at a shutter speed of 1/30 sec. However, as Bjørn Rørslett has shown, LLT can itself induce blurring at lower shutter speeds. Also, LLT is often impractical, such when doing tethered or remotely controlled shooting. So, LLT can hardly be considered an effective solution to the deficient Nikon collar design.


 WGHalvorsen's gear list:WGHalvorsen's gear list
Fujifilm X100V Nikon D300 Olympus E-M1 Nikon D850 Olympus OM-D E-M1X +37 more
Nikon D800
If you believe there are incorrect tags, please send us this post using our feedback form.
Flat view
Post (hide subjects) Posted by
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow