Re: [micro-manager-general] Questions on Hamamatsu Flash4

[Mark T. <ma...@op...>]

Hi Thomas,

On Thu, Jan 14, 2016 at 9:27 AM, julou <tho...@no...> wrote:
> Do I understand correctly that when a camera supports adjustable intervals,
> a MDA (with sequencable channels and/or piezo) drives the camera in
> streaming mode with the exposure defined by the channel’s exposure and the
> interval defined in the MDA delay?

No, this is not supported by the MDA engine yet (it certainly will be
at some future time).
MDA always uses software timing when the interval is greater than exposure.

> What are the camera that support this feature? Would it be in principle
> possible to implement it with a Flash4 (by enhancing the device adapter
> and/or the driver) or is it beyond the camera specs?

The only camera adapter that implements it, that I'm aware of, is
Andor (not AndorSDK3).
In any case, it can only be accessed from scripts/code.

>> Also, a new member of our group used to be able to acquire stacks faster
>> (using another Flash4, maybe with more recent firmware…). He's using MM from
>> Matlab (with snapImage() and getImage()) and is currently experiencing the
>> same delays described in my previous email (while he used to be able to
>> acquire 30ms images every 100ms, and move the stage between 2 images).
[...]
> We’ve done further testing with this… When we run the following bsh script
> with 30ms exposure, the average time is ≈330ms.
>
>  for( int i =0; i<10; ++i){
>       start = System.currentTimeMillis();
>       mmc.snapImage();
>       mmc.getImage();
>       print(System.currentTimeMillis() - start);
>    }
>
> This is very puzzling because when we click as fast as possible on the “Add
> to Album” button of the GUI, we reach almost 10Hz!! So somehow MM And the
> camera can talk faster to each other…

Note the difference between "Snap to Album" (in the main window) and
"Add to Album" (in the Snap/Live window). The latter simply copies the
existing image to the album.

> Also we were able to test that this doesn’t come from our old firmware
> (2.03A) since Andreas Durandi (from Hamamatsu Switzerland) was kind enough
> to configure another flash4 (firmware 2.50A) with MM1.4 and run the same
> script: he obtains similar delays (≈350ms).

These delays don't strike me as out of the ordinary.

The only way to (appear to) capture individual frames faster that I
can think of is to have a sequence acquisition running in the
background. In this case, snapImage()/getImage() will return the
latest frame from the sequence acquisition (at least for some
cameras). I wouldn't depend on this behavior without testing with your
particular camera that it returns an image acquired at the timing you
expect, though, since this is an area where µManager's device
interface is not as strictly defined as it ideally would be.

This is not to say that µManager cannot be improved to better support
accurate intervals. It's definitely on our mid-term to-do list.

Best,
Mark