Galvo Integration in Two Photon Microscope

[P. Muller]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hey,

we use a custom-built two photon micrsocope for imaging biological samples. Most of the optics is currently mounted on an optical table. The collimated beam is the guided into the rear port of a leica dm irm body where it is focussed using a 100x NA 1.4 objective lens (altough we consider going to a 63x/1.4 lens). We use a coherent chameleon laser source.

We would like to implement a galvo scanner in order to scan free trajectories at high velocities. What are the main design aspects I have to take into account when buying a xy-galvo scanner? For our application, high repeatability and high velocities are of prime importance. Also, I would like to sync the galvo precisely with our AOM to avoid uneccessary photobleaching. And it would be neat if we had a conjucated rear focal plane available somewhere in the optical path (e.g. for the usage of shaded ring filters).
Mostly we use Matlab for our custom software and I would like to drive the mirrors from matlab without having to worry too much about writing my own Galvo-driver. We would not mind changing the setup fundamentally, e.g. getting rid of the microscope body...

I dug a bit through thorlabs' and cambridge technolgy's websites, but perhaps you guys can give me some deeper insights... Thanks!

[Michael Giacomelli]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

The main design criteria are usually the mechanical requirements of your
scanner (scan angle, beam diameter) and your scan optics (focal length and
angular magnification).  If you only want to scan small field numbers (say
half your objective field or less), getting the beam diameters and
magnification correct is usually good enough.  At larger field sizes, scan
optics become complex to design due to off axis aberrations.

Depending on your budget, you might consider buying a commerical kit.
Thorlabs for example will sell you various DIY 2P scope kits that combine
some or all of a scan head (using CT scanners), PMTs, A/D, and
c++/labview/matlab acquisition software (and open source code if you wish
to dig into and customize it).  These incorporate custom scan optics with
performance that is very difficult to match without a lot of optical design
experience while still allowing extensive modification.  Of course, doing
it yourself will be much cheaper (scanners are a few thousand and it sounds
like you can reuse your existing detection optics).

Mike


On Mon, Aug 28, 2017 at 10:55 AM, P. Muller <
[hidden email]> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Hey,
>
> we use a custom-built two photon micrsocope for imaging biological
> samples. Most of the optics is currently mounted on an optical table. The
> collimated beam is the guided into the rear port of a leica dm irm body
> where it is focussed using a 100x NA 1.4 objective lens (altough we
> consider going to a 63x/1.4 lens). We use a coherent chameleon laser source.
>
> We would like to implement a galvo scanner in order to scan free
> trajectories at high velocities. What are the main design aspects I have to
> take into account when buying a xy-galvo scanner? For our application, high
> repeatability and high velocities are of prime importance. Also, I would
> like to sync the galvo precisely with our AOM to avoid uneccessary
> photobleaching. And it would be neat if we had a conjucated rear focal
> plane available somewhere in the optical path (e.g. for the usage of shaded
> ring filters).
> Mostly we use Matlab for our custom software and I would like to drive the
> mirrors from matlab without having to worry too much about writing my own
> Galvo-driver. We would not mind changing the setup fundamentally, e.g.
> getting rid of the microscope body...
>
> I dug a bit through thorlabs' and cambridge technolgy's websites, but
> perhaps you guys can give me some deeper insights... Thanks!
>

[Craig Brideau]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

The hard part is actually all the software and timing hardware necessary to
synchronize the pixel clock and scanning. You can buy scan and tube lenses
off-the-shelf and there are a few choices for galvo vendors who will help
you pick something suitable for the scan angle and speed you need. As
Michael suggests, you can save yourself quite a bit of time by going with a
kit as those come pre-packaged with the necessary software and timing
hardware so you can focus on the mechanical aspects of the design.
If you are determined to do the whole thing from scratch, ScanImage is a
good place to start if you are familiar with Matlab. If you do LabView then
another good platform is Helioscan which includes the software components
necessary to drive all the hardware in that platform. Take a good look at
the suggested hardware for each platform and work out the options and costs
before you commit to a given software package. You may find that by the
time you add up the cost of all the parts a pre-built solution may not be
that much more expensive if you consider all the software arrives
pre-installed and tested for you.

Craig

On Tue, Aug 29, 2017 at 9:43 AM, Michael Giacomelli <[hidden email]> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> The main design criteria are usually the mechanical requirements of your
> scanner (scan angle, beam diameter) and your scan optics (focal length and
> angular magnification).  If you only want to scan small field numbers (say
> half your objective field or less), getting the beam diameters and
> magnification correct is usually good enough.  At larger field sizes, scan
> optics become complex to design due to off axis aberrations.
>
> Depending on your budget, you might consider buying a commerical kit.
> Thorlabs for example will sell you various DIY 2P scope kits that combine
> some or all of a scan head (using CT scanners), PMTs, A/D, and
> c++/labview/matlab acquisition software (and open source code if you wish
> to dig into and customize it).  These incorporate custom scan optics with
> performance that is very difficult to match without a lot of optical design
> experience while still allowing extensive modification.  Of course, doing
> it yourself will be much cheaper (scanners are a few thousand and it sounds
> like you can reuse your existing detection optics).
>
> Mike
>
>
> On Mon, Aug 28, 2017 at 10:55 AM, P. Muller <
> [hidden email]> wrote:
>
> > *****
> > To join, leave or search the confocal microscopy listserv, go to:
> > http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> > Post images on http://www.imgur.com and include the link in your
> posting.
> > *****
> >
> > Hey,
> >
> > we use a custom-built two photon micrsocope for imaging biological
> > samples. Most of the optics is currently mounted on an optical table. The
> > collimated beam is the guided into the rear port of a leica dm irm body
> > where it is focussed using a 100x NA 1.4 objective lens (altough we
> > consider going to a 63x/1.4 lens). We use a coherent chameleon laser
> source.
> >
> > We would like to implement a galvo scanner in order to scan free
> > trajectories at high velocities. What are the main design aspects I have
> to
> > take into account when buying a xy-galvo scanner? For our application,
> high
> > repeatability and high velocities are of prime importance. Also, I would
> > like to sync the galvo precisely with our AOM to avoid uneccessary
> > photobleaching. And it would be neat if we had a conjucated rear focal
> > plane available somewhere in the optical path (e.g. for the usage of
> shaded
> > ring filters).
> > Mostly we use Matlab for our custom software and I would like to drive
> the
> > mirrors from matlab without having to worry too much about writing my own
> > Galvo-driver. We would not mind changing the setup fundamentally, e.g.
> > getting rid of the microscope body...
> >
> > I dug a bit through thorlabs' and cambridge technolgy's websites, but
> > perhaps you guys can give me some deeper insights... Thanks!
> >
>

[Peter Rupprecht-2]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi P.,
>> What are the main design aspects I have to take into account when buying a xy-galvo scanner? For our application, high repeatability and high velocities are of prime importance.
The standard scanners that most people use for scanning microscopy (since many years) are the 6215H (and sometimes the 6210H) scanners from Cambridge Technology. I have not met many people trying out other scan heads, except for special applications (like extremely large mirrors). But maybe this has changed since I last took a closer look. Normally, you should go for a xy-scan head instead of buying two separate scanners, since the distance of the scanners in the xy-scan head is minimized. (Depending on your application, you could also use two single scanners and use a relay system.) For more details, you should talk to somebody of e.g. Cambridge Technology (but not a sales-person!).If you want to go for highest speed, you have to choose the smallest mirrors (the 6215H comes with several mirror sizes). But if the mirror is too small, it will clip your incoming beam - it's always a trade-off.
>> Also, I would like to sync the galvo precisely with our AOM.
This should not be a problem. The galvo drivers are controlling the galvo based on a simple analog input; this is probably similar for your AOM. You can generate these analog outputs e.g. using a simple NI DAQ board (although this is probably not the cheapest solution). It is not really difficult to synchronize the two output signals, but writing the infrastructure code including the GUI is in my experience more time-consuming.
Since you bring up Matlab, there is Scanimage, as mentioned by Craig, which is however not entirely open any more. But it includes an adapter/wrapper library written in Matlab that allows to use NI DAQ boards very nicely. Rob Campbell has put together a nice set of Matlab code examples that should get you started if you are interested: https://github.com/tenss/MatlabDAQmx
One limitation that you should keep in mind if you want to use NI DAQ boards is the internal memory. It is typically limited to a couple of thousand data points. If you want to create a complex scan trajectory that consists of 20'000 points, you might run into problems.
Best,Peter
 

    P. Muller <[hidden email]> schrieb am 8:41 Dienstag, 29.August 2017:
 

 *****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hey,

we use a custom-built two photon micrsocope for imaging biological samples. Most of the optics is currently mounted on an optical table. The collimated beam is the guided into the rear port of a leica dm irm body where it is focussed using a 100x NA 1.4 objective lens (altough we consider going to a 63x/1.4 lens). We use a coherent chameleon laser source.

We would like to implement a galvo scanner in order to scan free trajectories at high velocities. What are the main design aspects I have to take into account when buying a xy-galvo scanner? For our application, high repeatability and high velocities are of prime importance. Also, I would like to sync the galvo precisely with our AOM to avoid uneccessary photobleaching. And it would be neat if we had a conjucated rear focal plane available somewhere in the optical path (e.g. for the usage of shaded ring filters).
Mostly we use Matlab for our custom software and I would like to drive the mirrors from matlab without having to worry too much about writing my own Galvo-driver. We would not mind changing the setup fundamentally, e.g. getting rid of the microscope body...

I dug a bit through thorlabs' and cambridge technolgy's websites, but perhaps you guys can give me some deeper insights... Thanks!

   

[P. Muller]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Thanks Peter, Craig and Michael,

>> the distance of the scanners in the xy-scan head is minimized
as far as I know this is usually being done, altough the rear focal plane is not perfectly imaged in this case. I guess aberrations (which to my understanding is mostly vignetting) are neglible?
I've also heard of those former "TiLL Photonics" yanus scanheads, which they do not manufacture anymore. They claim to perfectly relay the rear focal plane in their system, somehow using concave mirrors. Similarly, there exist a few companies (e.g. Mirrorcle) which provide MEMS-galvo-scanning units. But they seem limited in terms of speed and scan range.

>> The galvo drivers are controlling the galvo based on a simple analog input
According to CT (Sales...), they also provide a digital driver board, which is supposed to ease the integration of the galvos. Do you know something about those? Especially when it comes to sync accuracy and speed?

>> Rob Campbell has put together a nice set of Matlab code
Thanks for this! I had a look into it and it seems really helpful!

>>  One limitation that you should keep in mind if you want to use NI DAQ boards is the internal memory.
Okay so this might indeed be a challenge that I ignored so far. I have to see how I can overcome this limit... if you have any hints or ressources, I'd be thankful if you let me know

>> The main design criteria are usually the mechanical requirements of your scanner (scan angle, beam diameter) and your scan optics (focal length and angular magnification).  If you only want to scan small field numbers (say half your objective field or less), getting the beam diameters and magnification correct is usually good enough.  At larger field sizes, scan optics become complex to design due to off axis aberrations.

Going through the calculations, I see (and agree) that the CT 6215H should be sufficient for the task... (considering a Leica 63x/NA1.4 and Thorlabs SL50-CLS2 and TL200-CLS2). Thanks!

[Craig Brideau]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

On Thu, Aug 31, 2017 at 3:11 AM, P. Muller <
[hidden email]> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Thanks Peter, Craig and Michael,
>
> >> the distance of the scanners in the xy-scan head is minimized
> as far as I know this is usually being done, altough the rear focal plane
> is not perfectly imaged in this case. I guess aberrations (which to my
> understanding is mostly vignetting) are neglible?
>

Most designs put the X and Y galvo as close together as possible and call
it 'good enough'. You get a little bit of distortion in the scan as the
origin point of rotation is wobbling slightly, but this is *usually*
negligible.


> I've also heard of those former "TiLL Photonics" yanus scanheads, which
> they do not manufacture anymore. They claim to perfectly relay the rear
> focal plane in their system, somehow using concave mirrors. Similarly,
> there exist a few companies (e.g. Mirrorcle) which provide
> MEMS-galvo-scanning units. But they seem limited in terms of speed and scan
> range.
>

It is not common, but you can use concave or parabolic mirrors to relay
point-to-point on the galvo mirrors to eliminate the wobble but this
increases the complexity of aligning the system. Alternatively, I believe
Leica uses a three galvo system to correct for the motion in one of their
systems, but then you have a 50% increase in the probability that something
will go wrong.


> >> The galvo drivers are controlling the galvo based on a simple analog
> input
> According to CT (Sales...), they also provide a digital driver board,
> which is supposed to ease the integration of the galvos. Do you know
> something about those? Especially when it comes to sync accuracy and speed?
>

At the end of the day the position signal is usually analog from a
potentiometer. It is possible to use a digital encoder in the galvo design
but this is more expensive. Good analog circuitry will do just as good a
job, if not better, than digital assuming you have good noise filtering
built-in. Digital suffers from quantization error which can limit your
accuracy, requiring either a finer (more expensive) encoder or some fancy
tricks to interpolate position from a coarser sampling.

>>  One limitation that you should keep in mind if you want to use NI DAQ
boards is the internal memory.

> Okay so this might indeed be a challenge that I ignored so far. I have to
> see how I can overcome this limit... if you have any hints or ressources,
> I'd be thankful if you let me know
>

If you are performing simple raster scanning the onboard should be
sufficient. You can also use DMA (not all NI boards support this so check!)
which gives the card direct access to the RAM in the host PC, thus nicely
sidestepping the limit of the card's onboard buffer size.

> >> The main design criteria are usually the mechanical requirements of
> your scanner (scan angle, beam diameter) and your scan optics (focal length
> and angular magnification).  If you only want to scan small field numbers
> (say half your objective field or less), getting the beam diameters and
> magnification correct is usually good enough.  At larger field sizes, scan
> optics become complex to design due to off axis aberrations.
>
> Going through the calculations, I see (and agree) that the CT 6215H should
> be sufficient for the task... (considering a Leica 63x/NA1.4 and Thorlabs
> SL50-CLS2 and TL200-CLS2). Thanks!
>

Sounds reasonable. Best of luck, and tell us how it goes!

Craig

[Zdenek Svindrych-2]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Hi listers,
just FYI, this is an example of a scan head with a telecentric relay between
the galvos (Biorad Radiance 2100):

http://i.imgur.com/YJeJmvF.png

You need a 4-f relay, you can't achieve that with a single lens (or
mirror)... But, as pointed out already, two galvo mirrors close to each
other will do a good job as well; you won't see a relay like this in modern
confocals. Slight increase in mirror sizes (and beam diameter) will avoid
any vignetting.

The situation is more complicated with polygon scanners, as the mirror
trajectory is more complicated...

Also, there is the 4-galvo scanner operating near the image plane (quad
scanner, from Stefan Hell's lab, commercialized by Aberrior), where you can
avoid these issues, but the price is more complicated setup and -more
importantly- quite large mirrors, hence slow scanning!
Best, zdenek
--
Zdenek Svindrych, Ph.D.
W.M. Keck Center for Cellular Imaging (PLSB 003)
Department of Biology,University of Virginia
409 McCormick Rd, Charlottesville, VA-22904
http://www.kcci.virginia.edu/
tel: [phone]

---------- Původní e-mail ----------
Od: Craig Brideau <[hidden email]>
Komu: [hidden email]
Datum: 1. 9. 2017 13:28:34
Předmět: Re: Galvo Integration in Two Photon Microscope
"*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

On Thu, Aug 31, 2017 at 3:11 AM, P. Muller <
[hidden email]> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Thanks Peter, Craig and Michael,
>
> >> the distance of the scanners in the xy-scan head is minimized
> as far as I know this is usually being done, altough the rear focal plane
> is not perfectly imaged in this case. I guess aberrations (which to my
> understanding is mostly vignetting) are neglible?
>

Most designs put the X and Y galvo as close together as possible and call
it 'good enough'. You get a little bit of distortion in the scan as the
origin point of rotation is wobbling slightly, but this is *usually*
negligible.


> I've also heard of those former "TiLL Photonics" yanus scanheads, which
> they do not manufacture anymore. They claim to perfectly relay the rear
> focal plane in their system, somehow using concave mirrors. Similarly,
> there exist a few companies (e.g. Mirrorcle) which provide
> MEMS-galvo-scanning units. But they seem limited in terms of speed and
scan
> range.
>

It is not common, but you can use concave or parabolic mirrors to relay
point-to-point on the galvo mirrors to eliminate the wobble but this
increases the complexity of aligning the system. Alternatively, I believe
Leica uses a three galvo system to correct for the motion in one of their
systems, but then you have a 50% increase in the probability that something
will go wrong.


> >> The galvo drivers are controlling the galvo based on a simple analog
> input
> According to CT (Sales...), they also provide a digital driver board,
> which is supposed to ease the integration of the galvos. Do you know
> something about those? Especially when it comes to sync accuracy and
speed?
>

At the end of the day the position signal is usually analog from a
potentiometer. It is possible to use a digital encoder in the galvo design
but this is more expensive. Good analog circuitry will do just as good a
job, if not better, than digital assuming you have good noise filtering
built-in. Digital suffers from quantization error which can limit your
accuracy, requiring either a finer (more expensive) encoder or some fancy
tricks to interpolate position from a coarser sampling.

>> One limitation that you should keep in mind if you want to use NI DAQ
boards is the internal memory.

> Okay so this might indeed be a challenge that I ignored so far. I have to
> see how I can overcome this limit... if you have any hints or ressources,
> I'd be thankful if you let me know
>

If you are performing simple raster scanning the onboard should be
sufficient. You can also use DMA (not all NI boards support this so check!)
which gives the card direct access to the RAM in the host PC, thus nicely
sidestepping the limit of the card's onboard buffer size.


> >> The main design criteria are usually the mechanical requirements of
> your scanner (scan angle, beam diameter) and your scan optics (focal
length
> and angular magnification). If you only want to scan small field numbers
> (say half your objective field or less), getting the beam diameters and
> magnification correct is usually good enough. At larger field sizes, scan
> optics become complex to design due to off axis aberrations.
>
> Going through the calculations, I see (and agree) that the CT 6215H should

> be sufficient for the task... (considering a Leica 63x/NA1.4 and Thorlabs
> SL50-CLS2 and TL200-CLS2). Thanks!
>

Sounds reasonable. Best of luck, and tell us how it goes!

Craig
"

[Michael Giacomelli]

*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
Post images on http://www.imgur.com and include the link in your posting.
*****

Programming a NI board to scan is not hard, but their documentation is not
great, and if you haven't done it before, you'll probably have to put a
fair amount of time into figuring out how to synchronize everything.  I
think most students, upon trying it the first time, fail to synchronize
either the fast axis with the slow axis, or both to the A/D and end up with
a vertically or horizontally scrolling image (the DIO pins on the NI boards
are very useful for triggering an A/D and are easy to sync to the AO pins).
If you can use ready made software that handles this, and you don't need to
do custom scan patterns, it will save you a lot of time.

As for onboard memory, most galvos have a frequency response that is
relatively close to zero by a few KHz (mirrors aren't that light).  At such
very low data rates, most PCIe cards can efficiently stream from memory
without having to worry about onboard buffer or FIFOs. The NI drivers at
least hide this detail from you.

I like those Thorlabs scanner lenses a lot.  We use them in a few systems,
they're very low dispersion, well achromatized, about 5x magnification so
you can use smaller scanner mirrors, and we've done 2P imaging out to a 2mm
field at 10x with surprisingly low vignette even when used with displaced
scanner mirrors.  Just double check that 3mm times 5x mag (from memory,
double check this too) will be enough for your  objective (ideally with
some to spare so you aren't clipping the beam on the mirror).

Mike



On Thu, Aug 31, 2017 at 5:11 AM, P. Muller <
[hidden email]> wrote:

> *****
> To join, leave or search the confocal microscopy listserv, go to:
> http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy
> Post images on http://www.imgur.com and include the link in your posting.
> *****
>
> Thanks Peter, Craig and Michael,
>
> >> the distance of the scanners in the xy-scan head is minimized
> as far as I know this is usually being done, altough the rear focal plane
> is not perfectly imaged in this case. I guess aberrations (which to my
> understanding is mostly vignetting) are neglible?
> I've also heard of those former "TiLL Photonics" yanus scanheads, which
> they do not manufacture anymore. They claim to perfectly relay the rear
> focal plane in their system, somehow using concave mirrors. Similarly,
> there exist a few companies (e.g. Mirrorcle) which provide
> MEMS-galvo-scanning units. But they seem limited in terms of speed and scan
> range.
>
> >> The galvo drivers are controlling the galvo based on a simple analog
> input
> According to CT (Sales...), they also provide a digital driver board,
> which is supposed to ease the integration of the galvos. Do you know
> something about those? Especially when it comes to sync accuracy and speed?
>
> >> Rob Campbell has put together a nice set of Matlab code
> Thanks for this! I had a look into it and it seems really helpful!
>
> >>  One limitation that you should keep in mind if you want to use NI DAQ
> boards is the internal memory.
> Okay so this might indeed be a challenge that I ignored so far. I have to
> see how I can overcome this limit... if you have any hints or ressources,
> I'd be thankful if you let me know
>
> >> The main design criteria are usually the mechanical requirements of
> your scanner (scan angle, beam diameter) and your scan optics (focal length
> and angular magnification).  If you only want to scan small field numbers
> (say half your objective field or less), getting the beam diameters and
> magnification correct is usually good enough.  At larger field sizes, scan
> optics become complex to design due to off axis aberrations.
>
> Going through the calculations, I see (and agree) that the CT 6215H should
> be sufficient for the task... (considering a Leica 63x/NA1.4 and Thorlabs
> SL50-CLS2 and TL200-CLS2). Thanks!
>