Zeiss 5-LIVE and Timing Issues
Locked
 
|
Holly L. AARON |
Zeiss 5-LIVE and Timing Issues
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Dear Confocal Community – This question may be very specific to the Zeiss community
and even more specific to the 5-LIVE users out there. We find the 5-LIVE to be unreliable/unpredictable/unrepeatable
in time intervals shorter than 100msec. By this I mean that if we set up a
time-series in which an image should be taken every 500ms (let’s say the
time required for the image is 30ms), it works well: an image is in fact taken
every 500ms. However, if we then decide we want a shorter interval, 100ms or
less (which should be fine given only 30ms for each frame), the images are
taken at random times, some greater than 100ms, some less:
This gets worse when we go to shorter and shorter intervals,
for example, 50ms:
And even worse for shorter… I am wondering: Does anyone else see this phenomenon? Have
you been able to get around it? We are doing electrophysiology in concert with
imaging and timing is crucial. So far we have not been able to set it up to
trigger each image because that takes too long and is even less predictable. We
would be very grateful for anyone who has found a work-around for this problem.
Thank you! __________________ Holly L. Aaron Molecular Cancer Research Laboratory 447 LSA #2751 510.642.2901 510.642.5741 fax |
|
|
Craig Brideau |
Re: Zeiss 5-LIVE and Timing Issues
|
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Hi Holly! This may be dependent on the computer hardware running the `scope. If it's Windows, then all bets are off for timing when you get down to the low hundreds or tens of ms. Basically the OS can't reliably shuffle data around fast enough at those speeds; you need a real-time operating system for that. In the case of most multitasking operating systems no single process can 'bogart' the CPU, and the time to switch between process threads can vary. What this boils down to is that the computer will take its own sweet time handling the data coming in from your microscope depending on what else is going on with the computer and the time scale we are talking about.
Craig On Fri, Jul 11, 2008 at 12:14 PM, Holly Aaron <[hidden email]> wrote: Search the CONFOCAL archive at http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal |
|
|
Lambright, Geoffrey |
Re: Zeiss 5-LIVE and Timing Issues
|
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Hi Craig, and list, So if it is the case of the computer hardware “shuffling”
the data that is causing the imprecision in the data timing, does that mean
that the timing of the actual image acquisition is precise and the variance
that Holly sees comes from the computer’s ability to process and record
the captured data? Any idea how one could check for that if that was the case? Geoff From: Confocal Microscopy
List [mailto:[hidden email]] On Behalf Of Craig Brideau Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal Hi Holly! This
may be dependent on the computer hardware running the `scope. If it's
Windows, then all bets are off for timing when you get down to the low hundreds
or tens of ms. Basically the OS can't reliably shuffle data around fast
enough at those speeds; you need a real-time operating system for that.
In the case of most multitasking operating systems no single process can
'bogart' the CPU, and the time to switch between process threads can
vary. What this boils down to is that the computer will take its own
sweet time handling the data coming in from your microscope depending on what
else is going on with the computer and the time scale we are talking about. On Fri, Jul 11, 2008 at 12:14 PM, Holly Aaron <[hidden email]> wrote: Search the CONFOCAL archive at http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Dear
Confocal Community – This
question may be very specific to the Zeiss community and even more specific to
the 5-LIVE users out there. We find the
5-LIVE to be unreliable/unpredictable/unrepeatable in time intervals shorter
than 100msec. By this I mean that if we set up a time-series in which an image
should be taken every 500ms (let's say the time required for the image is
30ms), it works well: an image is in fact taken every 500ms. However, if we
then decide we want a shorter interval, 100ms or less (which should be fine
given only 30ms for each frame), the images are taken at random times, some
greater than 100ms, some less:
This gets
worse when we go to shorter and shorter intervals, for example, 50ms:
And even
worse for shorter… I am
wondering: Does anyone else see this phenomenon? Have you been able to get
around it? We are doing electrophysiology in concert with imaging and timing is
crucial. So far we have not been able to set it up to trigger each image
because that takes too long and is even less predictable. We would be very
grateful for anyone who has found a work-around for this problem. Thank you! __________________ Holly L.
Aaron Molecular
Imaging Center Cancer
Research Laboratory University
of California Berkeley 447 LSA
#2751 Berkeley,
CA 94720-2751 510.642.2901 510.642.5741
fax |
|
|
Craig Brideau |
Re: Zeiss 5-LIVE and Timing Issues
|
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Try doing the same thing with a different microscope but similar image parameters; same bit depth per pixel, number of pixels, and acquisition times. I'll bet you see similar performance issues. We have some homebrew microscopes that I have worked with where we encountered operating system timing problems. I actually had to offload a lot of timing stuff, like the pixel clock generation, to external hardware (pre-programmed timer/counter modules with their own hardware clocks) because windows cannot generate a reliable timing signal in the ms range due to the operating system's event handling limitations. Again, you either need dedicated external hardware or a real-time OS to get reliable control over timing for small timing intervals, and Windows (and actually most desktop OSes) simply don't provide this.
Craig On Fri, Jul 11, 2008 at 2:50 PM, Lambright, Geoffrey <[hidden email]> wrote:
|
|
|
Holly L. AARON |
Re: Zeiss 5-LIVE and Timing Issues
|
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Hi, Craig & Geoff & Everyone – The 5-LIVE has a “real-time computer” or
something it calls a real-time computer that the Windows box talks to. I would
assume all the timing and handling is done with the real-time computer, but
maybe it’s not. The 5-LIVE is capable of 120 frames per
second, but maybe not accurately? Even with a dedicated real-time computer? Or
is there something not working correctly on our system? Thanks for this great discussion! -Holly __________________ Holly L. Aaron CRL Molecular From: Confocal
Microscopy List [mailto:[hidden email]] On Behalf Of Craig Brideau Search the CONFOCAL
archive at http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal Try doing
the same thing with a different microscope but similar image parameters; same
bit depth per pixel, number of pixels, and acquisition times. I'll bet
you see similar performance issues. We have some homebrew microscopes
that I have worked with where we encountered operating system timing
problems. I actually had to offload a lot of timing stuff, like the pixel
clock generation, to external hardware (pre-programmed timer/counter modules
with their own hardware clocks) because windows cannot generate a reliable
timing signal in the ms range due to the operating system's event handling
limitations. Again, you either need dedicated external hardware or a
real-time OS to get reliable control over timing for small timing intervals,
and Windows (and actually most desktop OSes) simply don't provide this. On Fri, Jul 11, 2008 at 2:50 PM, Lambright, Geoffrey <[hidden email]>
wrote: Search the CONFOCAL archive at http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Hi Craig, and list, So if it is the case of the computer hardware
"shuffling" the data that is causing the imprecision in the data
timing, does that mean that the timing of the actual image acquisition is
precise and the variance that Holly sees comes from the computer's ability to
process and record the captured data? Any idea how one could check for that if
that was the case? Geoff From: Confocal Microscopy List [mailto:[hidden email]]
On Behalf Of Craig Brideau Search the CONFOCAL archive at http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Hi Holly! This may be dependent on the computer hardware running the
`scope. If it's Windows, then all bets are off for timing when you get
down to the low hundreds or tens of ms. Basically the OS can't reliably
shuffle data around fast enough at those speeds; you need a real-time operating
system for that. In the case of most multitasking operating systems no
single process can 'bogart' the CPU, and the time to switch between process
threads can vary. What this boils down to is that the computer will take
its own sweet time handling the data coming in from your microscope depending
on what else is going on with the computer and the time scale we are talking
about. On Fri,
Jul 11, 2008 at 12:14 PM, Holly Aaron <[hidden email]> wrote: Search
the CONFOCAL archive at http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal
Dear
Confocal Community – This
question may be very specific to the Zeiss community and even more specific to
the 5-LIVE users out there. We find
the 5-LIVE to be unreliable/unpredictable/unrepeatable in time intervals
shorter than 100msec. By this I mean that if we set up a time-series in which
an image should be taken every 500ms (let's say the time required for the image
is 30ms), it works well: an image is in fact taken every 500ms. However, if we
then decide we want a shorter interval, 100ms or less (which should be fine
given only 30ms for each frame), the images are taken at random times, some
greater than 100ms, some less:
This gets
worse when we go to shorter and shorter intervals, for example, 50ms:
And even
worse for shorter… I am wondering:
Does anyone else see this phenomenon? Have you been able to get around it? We
are doing electrophysiology in concert with imaging and timing is crucial. So
far we have not been able to set it up to trigger each image because that takes
too long and is even less predictable. We would be very grateful for anyone who
has found a work-around for this problem. Thank
you! __________________ Holly L.
Aaron Molecular
Cancer
Research Laboratory 447 LSA #2751 510.642.2901 510.642.5741
fax |
|
|
Michael Weber-4 |
Re: Zeiss 5-LIVE and Timing Issues
|
|
Search the CONFOCAL archive at
http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal Hi Holly and list, the scan head is controlled by an external computer, that is correct. Using Nobody would use Windows for this task, also not for a point scanner, if you see the time scale for the x/y scan mirrors. Time steps for Windows and Mac OS are in the range of 60 ms, if I remember correct. Unfortunately companies think that customers only buy systems that run Windows. One won't get this problem with Linux (but maybe others). These external "real-time computers" are usually controlled by Linux. The faulty timing in the final image is added on the way from the external computer to the Windows machine, since incoming data appears faster than the OS can stamp it. Now when it comes to triggering, as you plan it, Holly, I am not sure which computer controls and handles this. In principle this needs to be done directly via the RTC, but I recommend to get in touch with your local Zeiss application specialist. Michael > Search the CONFOCAL archive at > http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal > > Hi, Craig & Geoff & Everyone - > > > > The 5-LIVE has a "real-time computer" or something it calls a real-time > computer that the Windows box talks to. I would assume all the timing and > handling is done with the real-time computer, but maybe it's not. > > > > The 5-LIVE is capable of 120 frames per second, but maybe not accurately? > Even with a dedicated real-time computer? Or is there something not > working > correctly on our system? > > > > Thanks for this great discussion! > > -Holly > > __________________ > > Holly L. Aaron > > CRL Molecular Imaging Center > > http://imaging.berkeley.edu > > _____ > > From: Confocal Microscopy List [mailto:[hidden email]] On > Behalf Of Craig Brideau > Sent: Friday, July 11, 2008 1:56 PM > To: [hidden email] > Subject: Re: Zeiss 5-LIVE and Timing Issues > > > > Search the CONFOCAL archive at > http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal Try doing the same > thing with a different microscope but similar image parameters; same bit > depth per pixel, number of pixels, and acquisition times. I'll bet you > see > similar performance issues. We have some homebrew microscopes that I have > worked with where we encountered operating system timing problems. I > actually had to offload a lot of timing stuff, like the pixel clock > generation, to external hardware (pre-programmed timer/counter modules > with > their own hardware clocks) because windows cannot generate a reliable > timing > signal in the ms range due to the operating system's event handling > limitations. Again, you either need dedicated external hardware or a > real-time OS to get reliable control over timing for small timing > intervals, > and Windows (and actually most desktop OSes) simply don't provide this. > > Craig > > > > On Fri, Jul 11, 2008 at 2:50 PM, Lambright, Geoffrey > <[hidden email]> wrote: > > Search the CONFOCAL archive at > http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal > > Hi Craig, and list, > > > > So if it is the case of the computer hardware "shuffling" the data that is > causing the imprecision in the data timing, does that mean that the timing > of the actual image acquisition is precise and the variance that Holly > sees > comes from the computer's ability to process and record the captured data? > Any idea how one could check for that if that was the case? > > > > Geoff > > > > From: Confocal Microscopy List [mailto:[hidden email]] On > Behalf Of Craig Brideau > Sent: Friday, July 11, 2008 1:03 PM > To: [hidden email] > Subject: Re: Zeiss 5-LIVE and Timing Issues > > > > Search the CONFOCAL archive at > http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal Hi Holly! This > may > be dependent on the computer hardware running the `scope. If it's > Windows, > then all bets are off for timing when you get down to the low hundreds or > tens of ms. Basically the OS can't reliably shuffle data around fast > enough > at those speeds; you need a real-time operating system for that. In the > case of most multitasking operating systems no single process can 'bogart' > the CPU, and the time to switch between process threads can vary. What > this > boils down to is that the computer will take its own sweet time handling > the > data coming in from your microscope depending on what else is going on > with > the computer and the time scale we are talking about. > > Craig > > > > On Fri, Jul 11, 2008 at 12:14 PM, Holly Aaron <[hidden email]> wrote: > > Search the CONFOCAL archive at > http://listserv.acsu.buffalo.edu/cgi-bin/wa?S1=confocal > > Dear Confocal Community - > > > > This question may be very specific to the Zeiss community and even more > specific to the 5-LIVE users out there. > > We find the 5-LIVE to be unreliable/unpredictable/unrepeatable in time > intervals shorter than 100msec. By this I mean that if we set up a > time-series in which an image should be taken every 500ms (let's say the > time required for the image is 30ms), it works well: an image is in fact > taken every 500ms. However, if we then decide we want a shorter interval, > 100ms or less (which should be fine given only 30ms for each frame), the > images are taken at random times, some greater than 100ms, some less: > > > > > Image # > > Time Expected > > Time Actual > > > 1 > > 0 > > 0 > > > 2 > > 0.1 > > 0.0987 > > > 3 > > 0.2 > > 0.1998 > > > 4 > > 0.3 > > 0.3009 > > > 5 > > 0.4 > > 0.4008 > > > > This gets worse when we go to shorter and shorter intervals, for example, > 50ms: > > > Image # > > Time Expected > > Time [ s] > > > 1 > > 0 > > 0 > > > 2 > > 0.05 > > 0.0209 > > > 3 > > 0.1 > > 0.1064 > > > 4 > > 0.15 > > 0.1273 > > > 5 > > 0.2 > > 0.2104 > > > > And even worse for shorter. > > > > I am wondering: Does anyone else see this phenomenon? Have you been able > to > get around it? We are doing electrophysiology in concert with imaging and > timing is crucial. So far we have not been able to set it up to trigger > each > image because that takes too long and is even less predictable. We would > be > very grateful for anyone who has found a work-around for this problem. > > > > Thank you! > > __________________ > > Holly L. Aaron > > Molecular Imaging Center > > Cancer Research Laboratory > > University of California Berkeley > > 447 LSA #2751 > > Berkeley, CA 94720-2751 > > 510.642.2901 > > 510.642.5741 fax > > [hidden email] > > http://imaging.berkeley.edu |
| Free forum by Nabble | Edit this page |