Glass quality and f stop question.

Kennedy McEwen wrote:
In article , Chris Malcolm
writes
Kennedy McEwen wrote:

The central obstruction is around 25mm diameter (estimated by eye - not
measured). To accommodate that and achieve a t/8 solution, the lens
stop would require to be increased to 67.3mm, which would make an 80mm
front aperture very marginal indeed, with no allowance for tolerances
and other effects at all.

Yet that appears to be what Sony/Minolta have done. I went outside and
aimed a refracting zoom lens set at 250mm focal length and f8 up into
a featureless white sky. No matter how I waved it around, and whether
I set it to spot metering, centre weighted, or matrix, it nearly
always gave 1/100th sec as the appropriate shutter speed, and
sometimes 1/80th sec.

I then did the same experiment with the Sony 500mm reflex lens. It
nearly always gave 1/125th sec as the appropriate shutter speed, and
sometimes 1/160th sec. Of course we have a digitisation error, since
those are the smallest step changes in shutter speed the camera will
make. But the fact that the zoom sometimes dropped down to 1/80th
suggests that the exposure value was close to the lower limit of
1/100th sec, and the fact that the reflex sometimes moved up to
1/160th sec suggests that the exposure value was close to the upper
limit of 1/125th sec. So it's clear that the nominal aperture of this
lens at f8 must have been adjusted to give a transmission factor
compensated stop, rather than the technically accurate stop in terms
of focal length divided by entry pupil diameter. In other words the
aperture has been calculated in terms of entry pupil area, with the
secondary mirror obstruction subtracted, rather than the outside
diameter.

It certainly isn't quite as clear as you seem to think it is. All you
have shown is that the transmission of an undefined zoom lens is less
than that of the mirror lens, despite the obscuration. That is one
reason why almost every astronomical telescope used for deep sky work,
where every last photon counts, uses a mirror rather than refractor
design: the transmission loss of refractors is often more than the loss
of the central obstruction.

I went out this morning and repeated the experiment with three lenses,
and took photographs so that I could see small differences in exposure
from the histograms. Today was a better day for the experiment. The
sky was more heavily overcast, more evenly overcast, and varied in
brightness less with time. Yesterday I must have let some light
variations creep in, because today I found the differences in exposure
between a SAL 50mm f1.4 lens at f8, a SAL 18-250mm at 50mm and f8, the
18-250mm at 250mm and f8, and the SAL 500mm f8 (reflex) to be less
than 1/5th stop when matrix metered. The 18-250mm has no noticeable
vignetting, the 50mm slight vignetting, and the 500mm reflex very
noticeable vignetting, at rough guess half a stop. So if metered in
the centre it comes out brighter.

Note too that the camera is a Sony A350 with an APS-C sized sensor, so
the 500mm reflex, which is a full frame lens, would have more fall off
towards the edges and would therefore on matrix metering come out
darker. So my finding that all three lenses matrix metered come out
with pretty much the same transmission values is in part an artefact
of using a 1.5 crop sensor.

Incidentally, I just dug out my old Tamron 500mm mirror lens from the
back of the cupboard. I haven't used it in years but it is a similar
spec to the Sony at 500mm f/8, although obviously not an AF model and I
have to use an adapter to mount it on my Canon cameras.

I measured the mirror diameter, which is what actually matters here, and
it is 62.5mm +/- 0.25mm, exactly as it should be for the lens
specification. The front aperture of that lens is pretty much the same
size as the Sony at 80mm. The central obstruction however is 36mm
diameter, so if the f/# was spec'd against area rather than diameter, it
would be an f/9.8 - half a stop higher. But it is spec'd at f/8 and,
from the measurements it certainly is a true f/8. Guess what? Using
your same test against a 100-400 f/4.5-5.6 zoom lens at 400mm f/8 gives
a similar result. That old 500mm f/8 mirror lens is almost 1/3 of a
stop faster than the refractive zoom at 400mm, but I know it is a true
f/8 design from the dimensions.

The critical dimensions and specs of the Tamron and the Sony are so
similar that is it just nonsense to suggest that the Sony can have a
stop 5mm larger than the Tamron (which is what would be required to
compensate for the area of 25mm diameter obscuration) or more, if it
turns out that the central obstruction on the Sony is closer to Tamrons
36mm. It would need a front aperture closer to 90mm to permit that and
the Sony just isn't that large.

I suggest you make some attempt at measuring the diameter of the primary
mirror on the Sony lens, it isn't difficult to do, rather than basing
your claims on second order effects.

How can I measure the diameter of the stop? Note that in the
Sony/Minolta design the front glass face is slightly convex, which may
be what permits the size of the secondary mirror to be smaller. I can
see the stop at the back of the lens round the edges of the mirror,
but am not sure how to measure it, and how the front lens would affect
that measurement.

The best idea I could come up with was to place a very brightly lit
ground glass screen over the back of the lens, breathe on the front,
and try to measure the diameter of the projected bright circle. My
estimate is that it is 72-74mm diameter.

--
Chris Malcolm

In article , Chris Malcolm
writes
I went out this morning and repeated the experiment with three lenses,
and took photographs so that I could see small differences in exposure
from the histograms.

Whilst I could half expect Sony to "do a Microsoft" and redefine f/stops
to their own proprietary standard, this lens is an old Minolta design
and, as a traditional photographic company, they simply wouldn't do
that. If the lens was over-apertured to compensate for the central
obstruction, say to f/6.5, then that is what Minolta would have called
it: not least because there were a host of independent 500mm f/8 designs
on the market at the time. Offering a faster alternative would have
been a Minolta marketing man's dream. No, Chris, it is called an f/8
lens because it IS an f/8 lens.

How can I measure the diameter of the stop? Note that in the
Sony/Minolta design the front glass face is slightly convex, which may
be what permits the size of the secondary mirror to be smaller.

It is slightly convex on the Tamron too, but that can't change the
physics - the effective stop is the diameter that a bundle of parallel
image forming rays have as they enter the lens. ie. the apparent optical
diameter of the lens as viewed from infinity. It doesn't matter what
sits in front of or behind the stop.

I can
see the stop at the back of the lens round the edges of the mirror,
but am not sure how to measure it, and how the front lens would affect
that measurement.

You really need a collimator to do that properly but, assuming that you
don't have access to one, there are a couple of alternatives which give
good approximations.

The simplest is just to put the lens and camera onto a tripod with a
rule vertically across the front aperture. What you need to do is set
your viewing point at infinity but that would be impractical and
somewhat difficult to resolve the measurements on the ruler. ;-)
However provided that you view from a distance large enough that the
parallax between the rule and the mirror is insignificant, that should
give you a good enough estimate. The mirror sits approximately 100mm
behind the ruler and, since you are trying to differentiate better than
5mm in 65mm, ie. better than 7.5%, you need to be viewing from at least
1.3m from the lens, but further is better.

If move your viewpoint by 65mm or so towards the side of the mirror that
you are measuring on the rule that will also partially compensate for
the parallax and should enable you to get the stop size to better than
+/-1mm without too much effort.

One difficulty that you will have is that the primary mirror will become
obscured by the image of the secondary and its surround when viewing
from a distance, for fairly obvious reasons: the secondary has to be
large enough to capture all of the primary's reflected light. However
if you shift your viewing position slightly to one side, ie. off axis,
and the rule is vertical then the mirror should be clearly visible.

The best idea I could come up with was to place a very brightly lit
ground glass screen over the back of the lens, breathe on the front,
and try to measure the diameter of the projected bright circle.

No, that would only work if it was a point source at exactly infinity
focus. Your extended screen only enlarges the apparent size by the
field of view and the projected distance.

My
estimate is that it is 72-74mm diameter.

It certainly can't be that large as you only have a front aperture of
82mm (the original Minolta version had an 82mm front thread), and there
simply isn't room to get the field of view through that to a mirror of
that size. The front aperture is always oversized for this reason.

As I said at the start of this, if it really was faster than f/8 then
Minolta would not have called it f/8 - that would amount to underselling
their product.
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)

"Chris Malcolm" wrote in message
...
Kennedy McEwen wrote:
In article , Chris Malcolm
writes
Kennedy McEwen wrote:

The central obstruction is around 25mm diameter (estimated by eye - not
measured). To accommodate that and achieve a t/8 solution, the lens
stop would require to be increased to 67.3mm, which would make an 80mm
front aperture very marginal indeed, with no allowance for tolerances
and other effects at all.

Yet that appears to be what Sony/Minolta have done. I went outside and
aimed a refracting zoom lens set at 250mm focal length and f8 up into
a featureless white sky. No matter how I waved it around, and whether
I set it to spot metering, centre weighted, or matrix, it nearly
always gave 1/100th sec as the appropriate shutter speed, and
sometimes 1/80th sec.

I then did the same experiment with the Sony 500mm reflex lens. It
nearly always gave 1/125th sec as the appropriate shutter speed, and
sometimes 1/160th sec. Of course we have a digitisation error, since
those are the smallest step changes in shutter speed the camera will
make. But the fact that the zoom sometimes dropped down to 1/80th
suggests that the exposure value was close to the lower limit of
1/100th sec, and the fact that the reflex sometimes moved up to
1/160th sec suggests that the exposure value was close to the upper
limit of 1/125th sec. So it's clear that the nominal aperture of this
lens at f8 must have been adjusted to give a transmission factor
compensated stop, rather than the technically accurate stop in terms
of focal length divided by entry pupil diameter. In other words the
aperture has been calculated in terms of entry pupil area, with the
secondary mirror obstruction subtracted, rather than the outside
diameter.

It certainly isn't quite as clear as you seem to think it is. All you
have shown is that the transmission of an undefined zoom lens is less
than that of the mirror lens, despite the obscuration. That is one
reason why almost every astronomical telescope used for deep sky work,
where every last photon counts, uses a mirror rather than refractor
design: the transmission loss of refractors is often more than the loss
of the central obstruction.

I went out this morning and repeated the experiment with three lenses,
and took photographs so that I could see small differences in exposure
from the histograms. Today was a better day for the experiment. The
sky was more heavily overcast, more evenly overcast, and varied in
brightness less with time. Yesterday I must have let some light
variations creep in, because today I found the differences in exposure
between a SAL 50mm f1.4 lens at f8, a SAL 18-250mm at 50mm and f8, the
18-250mm at 250mm and f8, and the SAL 500mm f8 (reflex) to be less
than 1/5th stop when matrix metered. The 18-250mm has no noticeable
vignetting, the 50mm slight vignetting, and the 500mm reflex very
noticeable vignetting, at rough guess half a stop. So if metered in
the centre it comes out brighter.

Note too that the camera is a Sony A350 with an APS-C sized sensor, so
the 500mm reflex, which is a full frame lens, would have more fall off
towards the edges and would therefore on matrix metering come out
darker. So my finding that all three lenses matrix metered come out
with pretty much the same transmission values is in part an artefact
of using a 1.5 crop sensor.

Light measurement is not dependent on image sensor size;
A specfic sensor is used.

Incidentally, I just dug out my old Tamron 500mm mirror lens from the
back of the cupboard. I haven't used it in years but it is a similar
spec to the Sony at 500mm f/8, although obviously not an AF model and I
have to use an adapter to mount it on my Canon cameras.

I measured the mirror diameter, which is what actually matters here, and
it is 62.5mm +/- 0.25mm, exactly as it should be for the lens
specification. The front aperture of that lens is pretty much the same
size as the Sony at 80mm. The central obstruction however is 36mm
diameter, so if the f/# was spec'd against area rather than diameter, it
would be an f/9.8 - half a stop higher. But it is spec'd at f/8 and,
from the measurements it certainly is a true f/8. Guess what? Using
your same test against a 100-400 f/4.5-5.6 zoom lens at 400mm f/8 gives
a similar result. That old 500mm f/8 mirror lens is almost 1/3 of a
stop faster than the refractive zoom at 400mm, but I know it is a true
f/8 design from the dimensions.

The critical dimensions and specs of the Tamron and the Sony are so
similar that is it just nonsense to suggest that the Sony can have a
stop 5mm larger than the Tamron (which is what would be required to
compensate for the area of 25mm diameter obscuration) or more, if it
turns out that the central obstruction on the Sony is closer to Tamrons
36mm. It would need a front aperture closer to 90mm to permit that and
the Sony just isn't that large.

I suggest you make some attempt at measuring the diameter of the primary
mirror on the Sony lens, it isn't difficult to do, rather than basing
your claims on second order effects.

How can I measure the diameter of the stop? Note that in the
Sony/Minolta design the front glass face is slightly convex, which may
be what permits the size of the secondary mirror to be smaller. I can
see the stop at the back of the lens round the edges of the mirror,
but am not sure how to measure it, and how the front lens would affect
that measurement.

The best idea I could come up with was to place a very brightly lit
ground glass screen over the back of the lens, breathe on the front,
and try to measure the diameter of the projected bright circle. My
estimate is that it is 72-74mm diameter.

--
Chris Malcolm

Kennedy McEwen wrote:
In article , Chris Malcolm
writes

How can I measure the diameter of the stop? Note that in the
Sony/Minolta design the front glass face is slightly convex, which may
be what permits the size of the secondary mirror to be smaller.

It is slightly convex on the Tamron too, but that can't change the
physics - the effective stop is the diameter that a bundle of parallel
image forming rays have as they enter the lens. ie. the apparent optical
diameter of the lens as viewed from infinity. It doesn't matter what
sits in front of or behind the stop.

In other words it's the entry pupil rather than the physical size of
the stop I need to measure.

I can
see the stop at the back of the lens round the edges of the mirror,
but am not sure how to measure it, and how the front lens would affect
that measurement.

You really need a collimator to do that properly but, assuming that you
don't have access to one, there are a couple of alternatives which give
good approximations.

How about a laser pointer?

[snip advice about doing it visually]

Thanks for that. I'll return to this interesting investigation later.

As I said at the start of this, if it really was faster than f/8 then
Minolta would not have called it f/8 - that would amount to underselling
their product.

Thanks for the educational discussion. What we seem to have discovered
so far is that despite the conventional wisdom that catadioptric
lenses have an unusually large difference between f# and t#, at least
in the case of the popular Sony/Minolta and Tamron 500mm versions, the
operation of of the law of swings and roundabouts seems to have
arranged that they're very like our everyday refracting lenses in the
relationship betweem f# and t#. In other words, they're f8 lenses
which pass the amount of light you'd expect from an f8 lens. No
special exposure allowances for their catadioptric nature need be
made.

--
Chris Malcolm

Chris Malcolm writes:
Thanks for the educational discussion. What we seem to have discovered
so far is that despite the conventional wisdom that catadioptric
lenses have an unusually large difference between f# and t#, at least
in the case of the popular Sony/Minolta and Tamron 500mm versions, the
operation of of the law of swings and roundabouts seems to have
arranged that they're very like our everyday refracting lenses in the
relationship betweem f# and t#.

One thing that comes to mind is while mirror lenses obviously have the
big honkin' obstruction, they should also have less loss due to actual
hunks of glass and glass interfaces. So who knows...

-Miles

--
Friendless, adj. Having no favors to bestow. Destitute of fortune. Addicted to
utterance of truth and common sense.

Kennedy McEwen wrote:
In article , Chris Malcolm
writes
I went out this morning and repeated the experiment with three lenses,
and took photographs so that I could see small differences in exposure
from the histograms.

Whilst I could half expect Sony to "do a Microsoft" and redefine f/stops
to their own proprietary standard, this lens is an old Minolta design
and, as a traditional photographic company, they simply wouldn't do
that. If the lens was over-apertured to compensate for the central
obstruction, say to f/6.5, then that is what Minolta would have called
it: not least because there were a host of independent 500mm f/8 designs
on the market at the time. Offering a faster alternative would have
been a Minolta marketing man's dream. No, Chris, it is called an f/8
lens because it IS an f/8 lens.

How can I measure the diameter of the stop? Note that in the
Sony/Minolta design the front glass face is slightly convex, which may
be what permits the size of the secondary mirror to be smaller.

It is slightly convex on the Tamron too, but that can't change the
physics - the effective stop is the diameter that a bundle of parallel
image forming rays have as they enter the lens. ie. the apparent optical
diameter of the lens as viewed from infinity. It doesn't matter what
sits in front of or behind the stop.

I can
see the stop at the back of the lens round the edges of the mirror,
but am not sure how to measure it, and how the front lens would affect
that measurement.

You really need a collimator to do that properly but, assuming that you
don't have access to one, there are a couple of alternatives which give
good approximations.

The simplest is just to put the lens and camera onto a tripod with a
rule vertically across the front aperture.

I'm not following all the details here but I tried the ruler idea on a
35mm f/1.4 and it measured a perfect 25mm. Hmm, well only at closest
focus, with the ruler on the back, looking through the front element.


What you need to do is set
your viewing point at infinity but that would be impractical and
somewhat difficult to resolve the measurements on the ruler. ;-)
However provided that you view from a distance large enough that the
parallax between the rule and the mirror is insignificant, that should
give you a good enough estimate. The mirror sits approximately 100mm
behind the ruler and, since you are trying to differentiate better than
5mm in 65mm, ie. better than 7.5%, you need to be viewing from at least
1.3m from the lens, but further is better.

If move your viewpoint by 65mm or so towards the side of the mirror that
you are measuring on the rule that will also partially compensate for
the parallax and should enable you to get the stop size to better than
+/-1mm without too much effort.

One difficulty that you will have is that the primary mirror will become
obscured by the image of the secondary and its surround when viewing
from a distance, for fairly obvious reasons: the secondary has to be
large enough to capture all of the primary's reflected light. However
if you shift your viewing position slightly to one side, ie. off axis,
and the rule is vertical then the mirror should be clearly visible.

The best idea I could come up with was to place a very brightly lit
ground glass screen over the back of the lens, breathe on the front,
and try to measure the diameter of the projected bright circle.

No, that would only work if it was a point source at exactly infinity
focus. Your extended screen only enlarges the apparent size by the
field of view and the projected distance.

My
estimate is that it is 72-74mm diameter.

It certainly can't be that large as you only have a front aperture of
82mm (the original Minolta version had an 82mm front thread), and there
simply isn't room to get the field of view through that to a mirror of
that size. The front aperture is always oversized for this reason.

As I said at the start of this, if it really was faster than f/8 then
Minolta would not have called it f/8 - that would amount to underselling
their product.


--
Paul Furman
www.edgehill.net
www.baynatives.com

all google groups messages filtered due to spam

In article , Paul Furman
writes
Kennedy McEwen wrote:
In article , Chris Malcolm
writes

How can I measure the diameter of the stop?

You really need a collimator to do that properly but, assuming that
you don't have access to one, there are a couple of alternatives
which give good approximations.
The simplest is just to put the lens and camera onto a tripod with a
rule vertically across the front aperture.

I'm not following all the details here but I tried the ruler idea on a
35mm f/1.4 and it measured a perfect 25mm. Hmm, well only at closest
focus, with the ruler on the back, looking through the front element.

Which is exactly what you would expect Paul. 35/1.4 = 25mm.

What you are probably missing is that the discussion was about
catadioptric, or mirror, lenses and whether the stop was larger to
compensate for the central obstruction. It isn't, because focal ratio
is defined irrespective of central obstruction. However, your 35mm
f/1.4 isn't a cat lens anyway. ;-)
--
Kennedy
Yes, Socrates himself is particularly missed;
A lovely little thinker, but a bugger when he's ****ed.
Python Philosophers (replace 'nospam' with 'kennedym' when replying)