two possible optic designs for a case

[Karl Bellve]

Attached are two images. A 3D printable case will be designed by a Summer intern based on one of the following two designs.

Optical Pathway with Lens

Advantages:

• Attached directly to the back of the microscope
  
• Compact optical path.

Disadvantages:

• Extra optical elements.
• Requires a precision lens mount.
• Harder to align.
• High Precision placement of the lens needed.
• Optical Pathway might be too short.
• Reduced sensitivity but increased range.

Optical pathway without a Lens:

Advantages:

• No Lens needed.
• Easier to align.

Disadvantages:

• Would not mount directly on the back of the microscope.
• Larger Case.
• Optical Pathway might be too long.
• Increased sensitivity but decreased range.

Both designs will use a precision mount for the first mirror after the laser.

Opinions welcome.

[Karl Bellve]

I think we are going with design #2 (sans lens) due to its simplicity and cost effectiveness. 

A lens adds complexity and an additional $400 for a nice XY adjustable mount but the cost of the lens.

The next thing we are thinking about is the how to make a case and what materials we should use. 

3D print the entire case or laser cut a case?

ABS, Acrylic or something else?

People can get access to 3D printers, but if I provide the plans online, I think it would be easy enough to order a laser cut case online...

The case will be 8" x 8", which is large for 3D printing.  Not that large if you laser cut a case.

So, I am thinking about laser cutting a case, with the appropriate mounting holes built in. Some of the parts will still be 3D printed, while others (such as the mirror mount for the laser) should be purchased.

I will be going with Baltic Birch Plywood for the case since it has a really low thermal expansion, unlike acrylic and ABS. Baltic Birch is in the range of Steel and 1/7 of Acrylic and ABS. One can also engrave a laser cut case.

It would also be interesting to see a wood case next to all metal microscope...a clash of materials. :D 

This case would also replace the current pgFocus case. front plate and back plate and would actually be cheaper...

Something to think about....

Cheers Karl

[Kyle Douglass]

On Friday, May 27, 2016 at 4:48:08 PM UTC+2, Karl Bellve wrote:

Attached are two images. A 3D printable case will be designed by a Summer intern based on one of the following two designs.

Optical pathway without a Lens:

Advantages:

• No Lens needed.
• Easier to align.

Disadvantages:

• Would not mount directly on the back of the microscope.
• Larger Case.
• Optical Pathway might be too long.
• Increased sensitivity but decreased range.

Both designs will use a precision mount for the first mirror after the laser.

Opinions welcome.

Hi Karl,
I've been doing a bit of research lately into TIRF illumination. One of the things that has come out of it is that the beam should be focused to as small a spot as possible and as close as possible to the edge of the objective back focal plane to keep the beam as close to the coverslip as possible. (See my drawing here and an explanation of the spot size in the BFP and laser penetration depth here.) Getting a good focus is hard to do if the beam quality is poor or if the laser itself is not collimated as it enters the tube lens. And if the evanescent light is not well constrained to the coverslip, I've noticed that cells can scatter the return beam and distort it, which makes it more difficult to do automatic scanning and imaging with the pgFocus locked.

Two things to consider:
1) if the laser diode is not collimated already, you may need a telescope to do collimate it (of course, buying a collimated laser diode should fix this)
2) if the beam quality is bad, you won't get a well-defined spot on the BFP. Maybe there are single-mode, fiber-coupled, cheap IR lasers that could be suitable?

Just my two cents. HTH.

Kyle
 

[Karl Bellve]

On Wed, Jun 8, 2016 at 6:42 AM Kyle Douglass <kyle.m....@gmail.com> wrote:

Hi Karl,
I've been doing a bit of research lately into TIRF illumination. One of the things that has come out of it is that the beam should be focused to as small a spot as possible and as close as possible to the edge of the objective back focal plane to keep the beam as close to the coverslip as possible.

Great point. 

Outside edge of the objective (fp at bfp of objective) = steeper angle at the coverslip = better return beam.

 

(See my drawing here and an explanation of the spot size in the BFP and laser penetration depth here.) Getting a good focus is hard to do if the beam quality is poor or if the laser itself is not collimated as it enters the tube lens. And if the evanescent light is not well constrained to the coverslip, I've noticed that cells can scatter the return beam and distort it, which makes it more difficult to do automatic scanning and imaging with the pgFocus locked.

Two things to consider:
1) if the laser diode is not collimated already, you may need a telescope to do collimate it (of course, buying a collimated laser diode should fix this)

Telescope = two lens (with different focal points) + pinhole? This would make the beam smaller and eliminate non-collimated light....

This would be hard to implement due to case space constraints and alignment sensitivity but it would clean up a beam.

Thorlabs offers a 780nm collimated laser for $98 and an 808nm collimated laser for $168.

2) if the beam quality is bad, you won't get a well-defined spot on the BFP. Maybe there are single-mode, fiber-coupled, cheap IR lasers that could be suitable?

One possibility is not to even use a collimated laser but just an IR LED and monitor its skewed PSF reflection. This is what Definite uses. I think they must focus to a spot at the coverslip and not at the bfp. Don't quote me on that. It would be much cheaper, and would be simpler to set up for the novice. Not sure if I should put in the work to design that and it might still be covered by a Zeiss patent.

Cheers

Karl

 

[Karl Bellve]

This is what Definite uses. 

* This is what Definite Focus uses. 

[Kyle Douglass]

In our lab we also have an ASI CRISP which, like the Definite Focus, projects the image of an IR LED onto the coverslip/water interface. A design schematic is here: http://www.asiimaging.com/wp-content/uploads/2015/09/CRISP_diag.jpg. (I can verify it's correct because I've taken the CRISP apart multiple times :) The reflected light comes back to the CRISP module and passes through a relay lens. In the focus of the lens is a wedge mirror that is positioned by micrometer so that it blocks half of the focused spot. This essentially makes the direction of the light passing through the focus very sensitive to the displacement of the coverslip/objective distance. I believe the effect is known as Foucalt's knife edge test (https://en.wikipedia.org/wiki/Foucault_knife-edge_test). To measure this displacement they use a dual photodiode and compute the difference between the two halves, sending a feedback signal to the piezo stage/positioner.

My somewhat subjective opinions on this modality vs. TIR-based autofocus are:

TIRF-based advantages
1) Extremely sensitive to coverslip/objective displacement --> precise locking
2) No problem getting a strong return signal due to use of a laser

TIR-based disadvantages
1) TIR alignment is always challenging and subject to misalignment
2) Any distortion in the return beam shape prevents accurate centroid determination

IR LED-based advantages
1) Not so sensitive to scattering of the IR light by cells on the coverslip --> easier to lock the focus and perform automatic scans
2) Alignment is not as critical because only differences are measured (rather than a fit to a Gaussian)

IR LED-based disadvantages
1) The axial range over which we can of lock appears to be smaller, but I don't have numbers to confirm this, so I might be wrong. With the CRISP, we have a small linear range in the response curve; if the displacement gets driven outside this range, the negative feedback becomes positive and the system drives itself out of focus.
2) The reflection from a coverslip/water interface is quite weak.

I must also add that we have been struggling with the CRISP firmware and control software for two years. I'm not sure if this is because the Foucalt knife edge method is inherently difficult to control or if the software is just a bit buggy to use.

I hope this helps people think more about their autofocus options and the strengths and weaknesses of both techniques.