• this error appeared a while back intermittently, Blair Welsh from Coherent came for a service visit and found that a circuit board was attached incorrectly in the laser, fixing the position of this daughter board took care of the error
• error returned after 8 months, again intermittent - restart makes it disappear. However, after a video call with Blair it seems like the daughter board is fine this time, suggests factory repair or replacement of laser
• adjusting board during video call made the error disappear and not come back after several test runs, decided to use it until error reappears

• various changes by Bruno Pichler (INSS), detailed description of the full system added here
• upgraded to NI DAQmx 18.0 - to control DAQs straight via Matlab Data Acquisition Toolbox e.g. for LN manipulator or custom piezo timing

• Channel 1 PMT was exposed to a magnet while adjusting the superfusion
• very high baseline noise was traced back to the amplifier unit
• changed it to one from 2P2, the image is nicer than ever, zero baseline noise on either channels
• weirdly channel 2 (GCaMP) has now a lower threshold - gain ~340 looks like the earlier gain 420
• ordered a new amplifier https://www.hamamatsu.com/eu/en/product/type/C7319/index.html

• added Multiclamp 700B from Jonny's / Erika's rig; 700A with headstages put there
• Marco lent us their Picospritzer, to be triggered from ITC-18 TTL2
• added a Luigs Neumann manipulator to target pipettes (the other one has been in use for delivering mechanic / taste stimuli to flies, also it has a very suspicious Y axis, handle with care)
• installed Multiclamp 700B software + Janelia's wavesurfer on PC to control electrode (only one channel)
• Primary and Secondary signals from Ch1 electrode go to Dev1 (NI PCI6323) AI6 and AI7 respectively; Command to electrode comes from same NI board's AO1
• Piezo disabled (would need a second breakout box, or even better a second NI card - see below) to have enough AO channels
• Scanimage Machine Data Files for Piezo changed to have AO3 as Piezo control channel to avoid initialising Scanimage with AO1 as Piezo control and kill the headstage - to access AO3 we need another breakout box
• Recorded baseline noise with model cell in wavesurfer; power cables to the two power meters for laser intensity caused 10mV noise each, these will run from battery
• After this one gets a fairly nice baseline with the imaging box closed, but turning on the PMTs adds high frequency noise

• Piezo set up on Dev1 (PCIe-6323), command sent via COM-port12; needed an RS232 to USB adapter and null modem cable (ordered by Andy Howe) - configured in MDF
• Pockels Cell set up on Dev1 (PCIe-6323) based on Scanimage page http://scanimage.vidriotechnologies.com/display/SI2019/Pockels+Cell+Concepts; unipolar positive input with 1000 Ohm impedance. Initially tried to monitor power with thermal sensor, like the one used after the half wave-plate but this is too slow, changed it to S120C from Thorlabs, Scanimage automatically uses this to calibrate input voltage / output power curve at startup.
• limitation: even though PCI3-6323 has enough AI / AO channels to control and read feedback from both Piezo and Pockels, the timing of these fast processes can't be done at the same time (communication with Jacob Franklin at Scanimage: “Pockel cell control and Piezo control are both hardware timed buffered tasks, as is the galvo control. Due to a limitation with NI hardware these need to be run either all on one 4 AO board, or you will need a separate board for each task, regardless of the number of AO available. The reason for this is that when you create one of these tasks it reserves the necessary timing resources for the entire sub system. So when you try to create another task, such as the piezo task, it will fail because the timing system is already reserved for the pockel cell task.”
• possible solution: moving imaging from PCI-6110 to PCIe-6323 by buying another BNC box + cable; or buying another PCIe-6323 to have one for the Piezo, one for the Pockels (would prefer this latter option); currently two separate MDF exists, one controlling the Piezo, one the Pockels
• external trigger coming from Mac / ITC-18 rerouted to PCIe-6323 'PFI0' instead of PCI-6110 'PFI0' wired to internal trigger. This happened automatically when PCIe-6323 was added as auxiliary board in the machine data file (to control Piezo / Pockels)
• the pick mirror used to measure power after the Pockels needed to be adjusted to have the smallest angle with the laser, to avoid forming two parallel images

• 2P moved to a new, bigger airtable (2500x1500x210+700 leg) - T1525P from Thorlabs / Nexus
• moved by Mark Walling from INSS (Bruno Pichler)
• optical path simplified on 2P1
• PCIe-6323 and breakout box installed again to drive Piezo and Pockels (RTSI cable configured in NI MAX)
• further info on main 2P page

• Changed PC
• Scanimage 5.6 (2019-06-03); DAQmx 15.5; Matlab 2019a
• Hardware configuration the same - NI 6323 not used currently, disabled to run Scanimage - RTSI cable needs to be configured in NI MAX

PC

   Core i7-9700K - processor
   Corsair H110i - fan
   GTX 1050 ti - GPU
   H370 HD3 - motherboard - NI6110 to PCI; NI 6323 to PCIe
   Corsair Vengeance 16gb DDR4 - RAM
   Samsung 500gb SSD sata - storage
   EVGA Supernova 650w PSU - power supply
   Fractal Design mATX - case
   1 PCIe slot receives 2 serial port inputs via adaptor
   Windows 10 x64
   

• Set up LEDD1B Thorlabs Cube (from Andrew Pattons) to drive 617nm LED for optogenetics from Igor - DAC channel 1 on ITC-18

• cleaned objective thoroughly
• aligned mirrors in light path to re-center laser illumination
• 1/2 waveplate controlled by apt - dc servocontroller (Thorlabs cube)
• replaced motor cables to run farther away from PMT cables - decreased structured noise on the image

• LED illumination field re-centered
• Objective home position was way far from what should be ideal for best-performance imaging. Re-set to (-4500, -4500, 0) as explained here Set MP285 Home

• Tried numerous ways of trying to trigger the piezo at the correct time, with minimal delays. This turned out to be impossible, as slices fire at the right time and wait for the piezo but involve a long ScanImage delay, and frames do not wait for the user function to finish, meaning the the first few lines are in the wrong z position as the piezo is still moving. The latter situation was deemed the lesser of the two evils for calcium imaging and was adopted.
• Made a few new configurations for fast calcium imaging, fast anatomical stacks, and detailed anatomical stacks, all with the piezo.
• Set up Igor to trigger acquisition in ScanImage, rather than the other way around, rewiring the NI board as appropriate.
• Set up Igor to fire a ScanImage trigger at the start of a wave, wait 3 seconds, fire an odour trigger, and then wait for 12 more seconds. This is repeated 12 times, with 12 lines in the odour config file.
  • ScanImage was told to wait for an external trigger and do 30 frames, with the piezo resetting to 0 every five frames. 12 repeats were set, and so ScanImage listens for a trigger before starting each acquisition, ensuring maximum synchronicity.
• Set up a cycle in ScanImage, using the aforementioned config files, to do a fast acquisition of 20 calcium frames (10 slices per 'stack') followed by a fast anatomical stack of 10 slices.
  • This worked pretty well, but it would be even better if we could have the config control the piezo settings, such that we can do thinner slice separations in the fast anatomical stack.

• Rotated camera and moved stage to make sample view and movement consistent across wide-field and 2-photon views (see Sample view/movement).
  • Changed shift slow to 0 and slow scan angle multiplier to -1 (to flip in y), in ScanImage (ajdm_piezo_config).
    • The slow scan angle multiplier should be set back to something positive before an ROI is drawn, otherwise ScanImage becomes unhappy. The developers are apparently looking for a fix.
  • Roughly calibrated wide-field view to 0.58 microns per pixel.