lens diameter for camera coupling

classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

lens diameter for camera coupling

Jeff Spector
Greetings,
 I was asked this by a colleague and thought I'd pass it along to the list as it may be of value to others. They are designing a mutli-camera adapter for a TIRF microscope using two emccds. They plan to use 1 lens to collimate the light out of the microscope and two separate lenses to focus on each camera. Their questions was, is there any advantage to using a 2" diameter lens than to a 1" diameter lens, given that the microscope exit port field of view, and the emccd entrance port are both less than 1" anyways about 0.7 inches if I remember correctly. They asked if a larger (2") lens would give less aberration because you are going "more through the middle" than a 1" lens where you are getting closer to the edges.
any thoughts?
 Thanks! 
-Jeff


------------------------------------------------------------------------------
Announcing the Oxford Dictionaries API! The API offers world-renowned
dictionary content that is easy and intuitive to access. Sign up for an
account today to start using our lexical data to power your apps and
projects. Get started today and enter our developer competition.
http://sdm.link/oxford
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lens diameter for camera coupling

Kyle Douglass
Hi Jeff,


On 03/09/2017 04:34 PM, Jeff Spector wrote:
> Their questions was, is there any advantage to using a 2" diameter
> lens than to a 1" diameter lens, given that the microscope exit port
> field of view, and the emccd entrance port are both less than 1"
> anyways about 0.7 inches if I remember correctly. They asked if a
> larger (2") lens would give less aberration because you are going
> "more through the middle" than a 1" lens where you are getting closer
> to the edges.
> any thoughts?

I'm assuming that your colleagues are essentially building a 4F system
onto the exit port of a commercial microscope, i.e. the distance between
the "normal" image plane of the microscope and either of the image
sensors is four times the focal length of the lenses used in the
multi-camera adapter. I'm also assuming that the lenses in the adapter
are all identical.

For these types of systems I have always worked under the guideline that
the f-numbers of the lenses in the multi-camera adapter should match the
tube lens of the microscope. This guideline is essentially stating that
it's the ratio of the lens focal length to its diameter--not the
diameter itself--that is a major factor that determines whether any
additional aberrations will be introduced. Additionally, I suspect that
it's the tube lens diameter that sets the upper limit on the diameter of
the light cone, not the exit or entrance ports of the microscope/camera.

If money is not a problem, I would go with what I see as the safest
option: use tube lenses in the multi-camera adapter that are identical
to the tube lens in the microscope. This at least ensures that symmetry
and f-numbers are preserved. The Nikon tube lens for their CFI60
objectives sells for $265.
https://www.edmundoptics.com/microscopy/infinity-corrected-objectives/nikon-cfi-60-infinity-corrected-brightfield-objectives/58520/

At any rate, the confocal listserv might be a bit better-suited to your
question than this one. I would highly recommend having a look at some
related issues in these threads:

1)
http://confocal-microscopy-list.588098.n2.nabble.com/4f-system-alignment-with-fluorescent-light-td7583200.html
2)
http://confocal-microscopy-list.588098.n2.nabble.com/Reflected-image-gt-astigmatism-tc7586476.html

Cheers,
Kyle

--
Kyle M. Douglass, PhD
Post-doctoral researcher
The Laboratory of Experimental Biophysics
EPFL, Lausanne, Switzerland
http://kmdouglass.github.io
http://leb.epfl.ch


------------------------------------------------------------------------------
Announcing the Oxford Dictionaries API! The API offers world-renowned
dictionary content that is easy and intuitive to access. Sign up for an
account today to start using our lexical data to power your apps and
projects. Get started today and enter our developer competition.
http://sdm.link/oxford
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lens diameter for camera coupling

JonD
Administrator
Jeff Spector wrote
They are designing a mutli-camera adapter
for a TIRF microscope using two emccds. They plan to use 1 lens to
collimate the light out of the microscope and two separate lenses to focus
on each camera. Their questions was, is there any advantage to using a 2"
diameter lens than to a 1" diameter lens, given that the microscope exit
port field of view, and the emccd entrance port are both less than 1"
anyways about 0.7 inches if I remember correctly. They asked if a larger
(2") lens would give less aberration because you are going "more through
the middle" than a 1" lens where you are getting closer to the edges.
any thoughts?
This may in fact be the case but I wouldn't base my decision off of that logic, basically you would be paying for something that you don't need.

You should calculate the required clear aperture to gather all the rays at the location where you are placing the lens.  This depends both on the size of the object (in your case, the location where the single-camera sensor would go) as well as the maximum angle of the rays (which ultimately depends on the objective NA and then the optics inside the microscope).  Then find a well-corrected lens with at least that aperture.


Kyle Michael Douglass wrote
I'm assuming that your colleagues are essentially building a 4F system
onto the exit port of a commercial microscope, i.e. the distance between
the "normal" image plane of the microscope and either of the image
sensors is four times the focal length of the lenses used in the
multi-camera adapter. I'm also assuming that the lenses in the adapter
are all identical.
A splitter device it doesn't necessarily need to be 4F, you just want to put the dichroic mirror in collimated space and then re-image the two split paths onto two separate sensors using separate tube lenses.  But there is no downside to using a 4F system besides compactness.

As an aside, a 4F arrangement of lenses is needed when you need to preserve both positions and angles between the "sample" and "image" planes (in this case where you would put your original camera is your "sample").  Because image sensors to a good approximation don't care about the angle the incident light comes from, you don't usually need a 4F lens system for imaging even though there may be second-order benefits to doing so.


Kyle Michael Douglass wrote
For these types of systems I have always worked under the guideline that
the f-numbers of the lenses in the multi-camera adapter should match the
tube lens of the microscope. This guideline is essentially stating that
it's the ratio of the lens focal length to its diameter--not the
diameter itself--that is a major factor that determines whether any
additional aberrations will be introduced. Additionally, I suspect that
it's the tube lens diameter that sets the upper limit on the diameter of
the light cone, not the exit or entrance ports of the microscope/camera.
This is a good rule of thumb to preserve the full capability of the microscope optics, but it is very possible that you can get away with less for an EMCCD because the sensor size is relatively small.

The critical thing is to have a well-corrected lens with sufficient clear aperture. The lower the f-number, the more difficult the lens design and the more it will probably cost.   The f-number is basically 1/(2*NA).  Conceptually lower f-number (higher NA) means bending the light rays more and doing that over the entire lens without aberrations is increasingly difficult.  For a fixed aperture, lower f-numbers reduces focal length and thus shorter path lengths.  For a fixed focal length, lower f-number means larger lenses.


Kyle Michael Douglass wrote
If money is not a problem, I would go with what I see as the safest
option: use tube lenses in the multi-camera adapter that are identical
to the tube lens in the microscope.
I second this suggestion.  Be aware that there is a preferred orientation of the focus and collimated sides of the lens, and if you do care about a 4F system be aware that the front and back focal lengths are different.

Jon

-------------------------------------------
Jon Daniels
Applied Scientific Instrumentation
29391 West Enid Rd, Eugene, OR 97402
Phone: (541) 461-8181 x118
-------------------------------------------
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lens diameter for camera coupling

Jeff Spector
Hi, thanks for the responses so far! I've added some comments and questions below and also cross posted to the confocal list. I just thought the mm list might have more people using custom scopes:

On Thu, Mar 9, 2017 at 2:17 PM, JonD <[hidden email]> wrote:
Jeff Spector wrote

> They are designing a mutli-camera adapter
> for a TIRF microscope using two emccds. They plan to use 1 lens to
> collimate the light out of the microscope and two separate lenses to focus
> on each camera. Their questions was, is there any advantage to using a 2"
> diameter lens than to a 1" diameter lens, given that the microscope exit
> port field of view, and the emccd entrance port are both less than 1"
> anyways about 0.7 inches if I remember correctly. They asked if a larger
> (2") lens would give less aberration because you are going "more through
> the middle" than a 1" lens where you are getting closer to the edges.
> any thoughts?

This may in fact be the case but I wouldn't base my decision off of that
logic, basically you would be paying for something that you don't need.

You should calculate the required clear aperture to gather all the rays at
the location where you are placing the lens.  This depends both on the size
of the object (in your case, the location where the single-camera sensor
would go) as well as the maximum angle of the rays (which ultimately depends
on the objective NA and then the optics inside the microscope).  Then find a
well-corrected lens with at least that aperture. 


 
 
I'm not sure how one would go about doing this exactly. The sensor is an emccd with a diagonal size of 11.6 mm.
the scope is a nikon Ti-E with a 1.45 NA objective. How would I go about calculating the maximum angle of the rays? 
Do I just take the angle from the microscope size port to where the top of the emccd ship would go? I guess that would require knowing how far from the body of the scope the image plane is, but I guess we can look that up on the Nikon website...
You say a well-corrected lens with at least that aperture, this all came up because we are trying to build out the system using parts from thor labs. are their achromat doublets (which is what we were going to use) not "well-corrected" ?



 



Kyle Michael Douglass wrote
> I'm assuming that your colleagues are essentially building a 4F system
> onto the exit port of a commercial microscope, i.e. the distance between
> the "normal" image plane of the microscope and either of the image
> sensors is four times the focal length of the lenses used in the
> multi-camera adapter. I'm also assuming that the lenses in the adapter
> are all identical.

A splitter device it doesn't necessarily need to be 4F, you just want to put
the dichroic mirror in collimated space and then re-image the two split
paths onto two separate sensors using separate tube lenses.  But there is no
downside to using a 4F system besides compactness.


Yes, it indeed is not a 4F system. The plan is to place and f=100mm lens 100mm from the image plane on the side of the scope (so to collimate the light) and then send this collimate beam through some dichroics and bandpass filter, and then through an f=200mm lens with the cameras placed 200 mm away. 


 
As an aside, a 4F arrangement of lenses is needed when you need to preserve
both positions and angles between the "sample" and "image" planes (in this
case where you would put your original camera is your "sample").  Because
image sensors to a good approximation don't care about the angle the
incident light comes from, you don't usually need a 4F lens system for
imaging even though there may be second-order benefits to doing so.



Kyle Michael Douglass wrote
> For these types of systems I have always worked under the guideline that
> the f-numbers of the lenses in the multi-camera adapter should match the
> tube lens of the microscope. This guideline is essentially stating that
> it's the ratio of the lens focal length to its diameter--not the
> diameter itself--that is a major factor that determines whether any
> additional aberrations will be introduced. Additionally, I suspect that
> it's the tube lens diameter that sets the upper limit on the diameter of
> the light cone, not the exit or entrance ports of the microscope/camera.

This is a good rule of thumb to preserve the full capability of the
microscope optics, but it is very possible that you can get away with less
for an EMCCD because the sensor size is relatively small.

yes, the emccd is pretty small (less than an inch)



 

The critical thing is to have a well-corrected lens with sufficient clear
aperture. The lower the f-number, the more difficult the lens design and the
more it will probably cost.   The f-number is basically 1/(2*NA).
Conceptually lower f-number (higher NA) means bending the light rays more
and doing that over the entire lens without aberrations is increasingly
difficult.  For a fixed aperture, lower f-numbers reduces focal length and
thus shorter path lengths.  For a fixed focal length, lower f-number means
larger lenses.


I guess for the lense chosen the f/# are 100mm/25.4 mm (for 1") =  3.9
and 200mm/25.4mm = 7.8 are these considered "low"

and for 2" lenses   thy are 1.9 and 3.9



looking at the nikon tube lens on the edmunds site it has an f/# = 200mm/36mm = 0.18 = 5.5 

so if i want to "match" to 5.5 which set do I use 1" or 2" ?

 


Kyle Michael Douglass wrote
> If money is not a problem, I would go with what I see as the safest
> option: use tube lenses in the multi-camera adapter that are identical
> to the tube lens in the microscope.

I second this suggestion.  Be aware that there is a preferred orientation of
the focus and collimated sides of the lens, and if you do care about a 4F
system be aware that the front and back focal lengths are different.


 Yes, thanks for the warning. If i remember correctly the flat side points towards where you want to focus? 


Thanks for all the help!

------------------------------------------------------------------------------
Announcing the Oxford Dictionaries API! The API offers world-renowned
dictionary content that is easy and intuitive to access. Sign up for an
account today to start using our lexical data to power your apps and
projects. Get started today and enter our developer competition.
http://sdm.link/oxford
_______________________________________________
micro-manager-general mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/micro-manager-general
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: lens diameter for camera coupling

JonD
Administrator
Jeff Spector wrote
I'm not sure how one would go about doing this exactly. The sensor is an emccd with a diagonal size of 11.6 mm. the scope is a nikon Ti-E with a 1.45 NA objective. How would I go about calculating the maximum angle of the rays? Do I just take the angle from the microscope size port to where the top of
the emccd ship would go? I guess that would require knowing how far from the body of the scope the image plane is, but I guess we can look that up on the Nikon website...
The rays at the plane of the original image sensor depend on the length of collimated space within the microscope.  But by back of the envelope calculations that angle won't be very important and you should be fine with 1" lenses.

I'll attempt to explain, with the caveat that I am not a trained optical engineer; I've just picked up some things in the past few years.  Please someone correct any mistakes.

Consider a cone of light emitted from the corner of the FOV in your sample.  With a 100x/1.45 objective and assuming temporarily that your splitter relay has 1x magnification, the 11.6mm diameter sensor yields 116um diameter FOV at the sample and thus the light cone originates 58um from the central axis.  A certain portion of that cone, determined by the NA, will be captured by the objective and become a parallel bundle of rays in collimated space, with bundle diameter 2*2mm*1.45 = 5.8mm where 2mm is the focal length of your objective and 5.8mm is the back aperture diameter.  The bundle makes an angle of ~58um/2mm = 0.029 radians = 1.66 degrees with the central axis.  That bundle of rays will travel until it hits the tube lens at which point the tube lens will focus them to a point on the original sensor which is 0.029 radians * 200mm = 5.8mm from the central axis just as we'd expect.  This is just restating how optical magnification comes about.  This is all happening within the main microscope body.

The included angle of that bundle of rays being focused on the sensor (or where the original sensor was) depends on the size of the back aperture and the tube lens focal length.  A quick sketch reveals that the half-angle is just the NA divided by the magnification, or 0.0145 radians in this case.

The reason the collimated distance matters is that it is a "bias" for this included angle relative to the central axis.  If the collimated space is less than back focal distance then there will be an outward bias, if it is greater then there will be an inward bias, and if you have a 4F system then there will be no bias, meaning the central ray of the bundle will be parallel to the optical axis as it hits the original sensor.

As a simplifying rough guess, say the your main microscope is approximately 4F-like, so the central ray of the bundle landing on the edge of your sensor is roughly parallel to the optical axis. Then when you remove the original sensor and let the rays keep traveling another 200mm before hitting the next lens, the outer ray will displace an additional 0.0145 radians * 200mm = 2.9mm.  Meaning your rays from the entire FOV are contained in diameter 11.6mm + 2*1.45mm = 17.4mm.  So 1" optic should be fine for the initial splitter lens if it's 200mm focal length, and you have even more margin if you decide to use a 100mm lens.

You can extend the same sort of logic to determine the minimum clear aperture of the second optic, which similarly will depend somewhat on the collimated space distance but I would still expect 1" to be enough.


Jeff Spector wrote
You say a well-corrected lens with at least that aperture, this all came up
because we are trying to build out the system using parts from thor labs.
are their achromat doublets (which is what we were going to use) not
"well-corrected" ?
I can't comment on these particular lenses.  You can often find lens files and simulate them.  The question is how much additional spread the lens adds compared with the diffraction limit.  Obviously "good enough" depends on how much you care about approaching the diffraction limit, how much you care about chromatic aberrations, etc.  Our rule of thumb is that achromat doublets are usually acceptable for 250mm focal lengths and above.  For shorter focal distances we either use an engineered tube lens (e.g. the Nikon tube lens) or we combine two COTS achromats with longer focal length to make a shorter focal length compound element that has much better performance than a single achromat (but a a custom engineered compound lens would be better still).

** Commercial plug:** ASI sells a variety of tube lens assemblies including two separate achromats as described above at 100/125/140/180mm, with glass from microscope companies (the Nikon lens is our default) and also longer focal lengths with single achromats.  Our tube lens assemblies almost all have clear apertures are 30mm.  We also sell filter cubes and all the pieces to conveniently attach together and make the gizmo you are describing (though I have never put our building blocks together this way and characterized the performance).  I know that Cairn sells something like this as a purpose-built product but I have never tried it.


Jeff Spector wrote
Yes, it indeed is not a 4F system. The plan is to place and f=100mm lens
100mm from the image plane on the side of the scope (so to collimate the
light) and then send this collimate beam through some dichroics and
bandpass filter, and then through an f=200mm lens with the cameras placed
200 mm away.
Note that you are introducing another 2x magnification with this scheme, so your FOV at the sample will be correspondingly smaller.

Also note that performance may be slightly improved by adding some extra space between the relay lenses (i.e. make it more 4F-like), particularly with engineered tube lenses that have been optimized assuming collimated space will be approximately as long as inside the target microscope.  For example the Nikon tube lens is intended to be 100-200mm from the objective BFP.

Jeff Spector wrote
 Yes, thanks for the warning. If i remember correctly the flat side points
towards where you want to focus?
Orientation is either obvious or described by the manufacturer.  Usually the flat side is towards focus as you said.

Jon

-------------------------------------------
Jon Daniels
Applied Scientific Instrumentation
29391 West Enid Rd, Eugene, OR 97402
Phone: (541) 461-8181 x118
-------------------------------------------
Loading...