Doug Kerr
Well-known member
Some of you may recall that in 2006, I was involved in the matter of the anomalous implementation, in the Canon Speedlite 580EX, of the new feature in which the automatic setting of the beamwidth of the flash unit would take into account the sensor size of the camera on which it was used (with cameras supporting the feature, such as the EOS 20D). I now have an update on this story regarding the behavior of the Speedlite 580EX II.
BACKGROUND
Most serious EX-series Speedlite flash units for some while had a variable beamwidth ("zoom") head. The beamwidth was set automatically based on the reported focal length of the lens. The object was to optimize the beamwidth for the field of view given by the lens. This reduced amount of the luminous output of the flash unit directed to regions not included in the image. This increased the efficiency of the flash system, and increased the maximum available luminous intensity-time product, giving a greater "flash range", for longer focal length shooting.
In such flash units as the Speedlite 420EX and 550EX, there were seven discrete head zoom positions (and thus seven discrete beamwidths) available. The flash unit would use the smallest beamwidth that would adequately illuminate the field of view given by the reported lens focal length.
However, prior to the introduction of the new feature, this scheme was always predicated on the use of the flash on "full frame" cameras. On smaller-sensor cameras, the head zoom position was still set to the beamwidth appropriate for the field of view the lens would have given on a full-frame camera - a beamwidth greater than was useful for the actual field of view of the camera. Thus, optimum performance was not attained on "smaller format" cameras.
With the new scheme, introduced on the Speedlite 580EX (and shortly after on the 430EX), and first supported on the EOS 20D and EOS 350D dSLR bodies, the camera reported its frame size class ("1.0X", "1.3X", or "1.6X") to the flash unit along with the focal length reported by the lens. The flash unit then reckoned the desirable beamwidth for the resulting field of view, and set the head zoom to the narrowest available value from the repertoire of seven positions.
However, there were soon reports of an apparent anomaly in the implementation of this feature on the 580EX. After a review of test results from scores of 580EX owners, it was determined that the algorithm used for reckoning the needed head position was indeed incorrect. The result was that, on a "1.6X" frame size camera (such as an EOS 20D):
- The two widest head zoom positions were never used.
- For lens focal lengths less than about 22 mm, the resulting flash beamwidth was (theoretically) insufficient to fully illuminate the field of view (assuming that the beamwidth was not enlarged by manual deployment of the "wide panel", which in any case was not really intended for use with any but the widest basic beamwidth setting)
- For three other small ranges of lens focal length, the same shortcoming occurred.
(Note that there was one camp that held that this behavior was not anomalous and was actually Canon's intent, based on their interpretation of certain language in the 580EX manual. I reject that premise.)
In the course of this collection of reports and test data from users, the following came to light:
1. On the Speedlite 430EX (introduced shortly after the 580EX, and providing the new scheme as well), the algorithm was correctly executed. Further, that model actually had ten discrete head positions, all of which were available under the scheme of automatic beamwidth control (with sensor size accommodation). (The three "new positions" were not revealed to the user, and were not accessible under manual "zoom" control.)
2. On Speedlite 580EX units produced after a certain date, a different, "correct", algorithm was used. (These units still had seven head zoom positions.)
Eventually, in the face of pressure by users armed with information developed during the "study", Canon (in different regions at different times, eventually including the US) became willing (perhaps grudgingly) to take earlier 580EX units into their Service Centers and retrofit them to behave "properly" (usually at no cost). Nevertheless. there has never been (to my knowledge) any "official" acknowledgement by Canon USA of this "issue".
Incidentally, I never had a 580EX or 430EX - all my analyses were done based on reports from colleagues.
UPDATE
I recently acquired a Speedlite 580EX II flash unit. One of my first tasks was to perform a basic suite of tests regarding its beamwidth control scheme when used on an EOS 20D ("1.6X"). My initial basic findings are as follows:
- Like the Speedlite 430EX, the 580 EX II has a repertoire of ten head zoom (beamwidth) positions. It looks as if the three "new" positions have the same nominal beamwidth as those of the 430EX (although sufficiently-precise measurements to confirm this have not yet been taken).
- The algorithm for selecting the head zoom position is appropriate.
STILL NOT COOL
There are two aspects of the beamwidth control scheme of the 580EX II, when used on a "1.6X" body, that I do not consider appropriate:
1. The reporting of the actual lens focal length on the panel, of the flash unit is rounded to the next lower of the set of focal lengths that are used to "label" the original seven head zoom positions on older Speedlite flash units. (These were originally the focal lengths of lenses, used , for which the beamwidths given by the various discrete zoom positions were suited on a "full frame" body.) Since these numbers are meaningless in this context, rounding the display of lens focal length to them is absurd.
2. The display of the range of valid subject distances for the current situation of the camera is based on the rounded reported focal length of the lens rather than on the selected beamwidth. For example, when the flash unit is used on a "full frame" body, and the lens reported focal length is 51 mm, the head is moved to the "50 mm" position. The display of "range" (which takes into account the ISO sensitivity in effect) is appropriate for the beamwidth produced by that head setting.
However, if we put the same flash unit on a "1.6X" body, for a reported lens focal length of 51 mm, the flash unit sets the head to the "80 mm" position. However, the "range" display is the same as before: appropriate for the beamwidth produced by the unit at the "50 mm" head setting.
BACKGROUND
Most serious EX-series Speedlite flash units for some while had a variable beamwidth ("zoom") head. The beamwidth was set automatically based on the reported focal length of the lens. The object was to optimize the beamwidth for the field of view given by the lens. This reduced amount of the luminous output of the flash unit directed to regions not included in the image. This increased the efficiency of the flash system, and increased the maximum available luminous intensity-time product, giving a greater "flash range", for longer focal length shooting.
In such flash units as the Speedlite 420EX and 550EX, there were seven discrete head zoom positions (and thus seven discrete beamwidths) available. The flash unit would use the smallest beamwidth that would adequately illuminate the field of view given by the reported lens focal length.
However, prior to the introduction of the new feature, this scheme was always predicated on the use of the flash on "full frame" cameras. On smaller-sensor cameras, the head zoom position was still set to the beamwidth appropriate for the field of view the lens would have given on a full-frame camera - a beamwidth greater than was useful for the actual field of view of the camera. Thus, optimum performance was not attained on "smaller format" cameras.
With the new scheme, introduced on the Speedlite 580EX (and shortly after on the 430EX), and first supported on the EOS 20D and EOS 350D dSLR bodies, the camera reported its frame size class ("1.0X", "1.3X", or "1.6X") to the flash unit along with the focal length reported by the lens. The flash unit then reckoned the desirable beamwidth for the resulting field of view, and set the head zoom to the narrowest available value from the repertoire of seven positions.
However, there were soon reports of an apparent anomaly in the implementation of this feature on the 580EX. After a review of test results from scores of 580EX owners, it was determined that the algorithm used for reckoning the needed head position was indeed incorrect. The result was that, on a "1.6X" frame size camera (such as an EOS 20D):
- The two widest head zoom positions were never used.
- For lens focal lengths less than about 22 mm, the resulting flash beamwidth was (theoretically) insufficient to fully illuminate the field of view (assuming that the beamwidth was not enlarged by manual deployment of the "wide panel", which in any case was not really intended for use with any but the widest basic beamwidth setting)
- For three other small ranges of lens focal length, the same shortcoming occurred.
(Note that there was one camp that held that this behavior was not anomalous and was actually Canon's intent, based on their interpretation of certain language in the 580EX manual. I reject that premise.)
In the course of this collection of reports and test data from users, the following came to light:
1. On the Speedlite 430EX (introduced shortly after the 580EX, and providing the new scheme as well), the algorithm was correctly executed. Further, that model actually had ten discrete head positions, all of which were available under the scheme of automatic beamwidth control (with sensor size accommodation). (The three "new positions" were not revealed to the user, and were not accessible under manual "zoom" control.)
2. On Speedlite 580EX units produced after a certain date, a different, "correct", algorithm was used. (These units still had seven head zoom positions.)
Eventually, in the face of pressure by users armed with information developed during the "study", Canon (in different regions at different times, eventually including the US) became willing (perhaps grudgingly) to take earlier 580EX units into their Service Centers and retrofit them to behave "properly" (usually at no cost). Nevertheless. there has never been (to my knowledge) any "official" acknowledgement by Canon USA of this "issue".
Incidentally, I never had a 580EX or 430EX - all my analyses were done based on reports from colleagues.
UPDATE
I recently acquired a Speedlite 580EX II flash unit. One of my first tasks was to perform a basic suite of tests regarding its beamwidth control scheme when used on an EOS 20D ("1.6X"). My initial basic findings are as follows:
- Like the Speedlite 430EX, the 580 EX II has a repertoire of ten head zoom (beamwidth) positions. It looks as if the three "new" positions have the same nominal beamwidth as those of the 430EX (although sufficiently-precise measurements to confirm this have not yet been taken).
- The algorithm for selecting the head zoom position is appropriate.
STILL NOT COOL
There are two aspects of the beamwidth control scheme of the 580EX II, when used on a "1.6X" body, that I do not consider appropriate:
1. The reporting of the actual lens focal length on the panel, of the flash unit is rounded to the next lower of the set of focal lengths that are used to "label" the original seven head zoom positions on older Speedlite flash units. (These were originally the focal lengths of lenses, used , for which the beamwidths given by the various discrete zoom positions were suited on a "full frame" body.) Since these numbers are meaningless in this context, rounding the display of lens focal length to them is absurd.
2. The display of the range of valid subject distances for the current situation of the camera is based on the rounded reported focal length of the lens rather than on the selected beamwidth. For example, when the flash unit is used on a "full frame" body, and the lens reported focal length is 51 mm, the head is moved to the "50 mm" position. The display of "range" (which takes into account the ISO sensitivity in effect) is appropriate for the beamwidth produced by that head setting.
However, if we put the same flash unit on a "1.6X" body, for a reported lens focal length of 51 mm, the flash unit sets the head to the "80 mm" position. However, the "range" display is the same as before: appropriate for the beamwidth produced by the unit at the "50 mm" head setting.
Last edited: