Forum
Register Calendar Latest Topics
 
 
 


Reply
  Author   Comment  
cognoman

Junior Member
Registered:
Posts: 23
Reply with quote  #1 
Plexon Omniplex utilizes a NI DAC (PCIe-8361) to sample up to 32 analog channels along with the electrophysiology data. In the current set-up, therefore, raw voltages corresponding to eye position (from eyelink) and the joystick position is now being routed to the analog channels on Plexon. The computer does not need another DAC. However, would it be possible to run ML without a dedicated DAC card?
0
Jaewon

Senior Member
Registered:
Posts: 312
Reply with quote  #2 
Running ML without a DAC is possible. But the DAC board is for ML to communicate with other devices. Without it, ML cannot send out eventcodes to Omniplex for data sync and cannot detect eye fixation or joystick action. Unless you are doing a behavioral study only, you will need a DAC for ML even though you record analog signals with Omniplex.
0
cognoman

Junior Member
Registered:
Posts: 23
Reply with quote  #3 
I see. Would it be feasible to modify eyejoytrack, for example, to call pl_getAD(the routine that plexon uses to sample analog channel online) and to write events one could call PL_DO commands? (thinking aloud.)
0
Jaewon

Senior Member
Registered:
Posts: 312
Reply with quote  #4 
If you run ML (or any application) on the same computer with Plexon, you can peek a sample from what Plexon is recording. I doubt that Plexon has enough extra digital bits though.
0
cognoman

Junior Member
Registered:
Posts: 23
Reply with quote  #5 
Yes, in this case ML and Plexon are on the same computer. Plexon is interfacing with a NI card and has 32 analog and digital inputs and 32 digital outputs. I'm attaching the plexon analog sampler for reference. Let me know if you have any suggestions.

 
Attached Files
zip codes.zip (51.25 KB, 1 views)

0
Jaewon

Senior Member
Registered:
Posts: 312
Reply with quote  #6 
If you care about accuracy of stimulus presentation timing, you should not run ML and Plexon on the same computer. In fact, you can't even do any keyboard typing or mouse click while running tasks in the old ML versions, due to MLHELPER.EXE. NIMH MonkeyLogic does not disable keyboard/mouse, but it will require a DAC board under its own control for accurate timing and complex behavior analysis. So don't be stingy on the necessary cost. Buy a DAC board and just use it well.

By the way, how is the USB joystick test going?
0
cognoman

Junior Member
Registered:
Posts: 23
Reply with quote  #7 
[smile] I dont know if we are being stingy! The 256 ch. omniplex systems cost a lot and we have 4 of them! I am hoping to make good use of its analog and digital I/O capabilities. Plexon is controlled by a 6-core workstation with 3.5 GHz processor. If ML could share the system, that would be ideal. I am planning on asking Plexon for their mexPlexonline.c to see if that provides any insights.

USB joystick is for the human behavioral set-up. And that set-up does not have a DAC card as you pointed out. We have ordered one and that is shipping soon. We are also getting analog joysticks, btw!

  
0
Jaewon

Senior Member
Registered:
Posts: 312
Reply with quote  #8 
The NI boards in omniplex are used to run the omniplex. You are making good use of them. You may not think so because you are not using all 32 channels, but the number of channels is meaningless here. Each NI board has only one A/D converter inside and a multiplexer quickly switches signal sources for the converter. Therefore, those 32 channels cannot be driven independently and you need one board for one job whether you use one channel or 32.

I think it is not fair if you try to save one grand from the behavioral control system after spending tens of thousands for the recording system. In my opinion, behavioral control is as important as signal collection. What is the use of expensive cookers if you don't have any ingredient?

When it comes to timing accuracy, two cheap computers are much better than one expensive computer. It is like having two guards is more secure for watching two posts than just one guard running back and forth. Windows is not a real-time OS and you don't get to decide how much time Windows spends for one job and another, so neither the number of CPUs nor their clock speeds can guarantee precise timing. You just have to run the minimum number of programs so that each thread can be active more frequently.

I understand now why daqmex.mexw32 couldn't find NI-DAQmx in your computer. Maybe I should separate the USB joystick part from the NI-DAQmx in the toolbox. Thanks for your help!
0
LulaNord

Junior Member
Registered:
Posts: 1
Reply with quote  #9 
HI...i am a new user here. As per my knowledge the DAC board is for ML to communicate with other devices. Without it, ML cannot send out eventcodes to Omniplex for data sync and cannot detect eye fixation or joystick action. Unless you are doing a behavioral study only, you will need a DAC for ML even though you record analog signals with Omniplex.
0
ryklin

Administrator
Registered:
Posts: 230
Reply with quote  #10 
I believe it is possible to get real-time data from a plexon using a digital interface. I personally tested it many years ago, but never fully developed anything with it. I know there was a plexon server that can provide all data over tcp/ip, and if you're running behavior on the same computer as signal processing, you could communicate by shared memory too. This would allow you to utilize plexon's daq card for behavior. However, this is certainly not implemented in Monkey Logic, and probably never will be.

While I agree with everything stated earlier, thinking prospectively, it would indeed be advantageous to build multiunit-electrophysiology and behavioral testing systems that can run on just one computer that was also cheap to build. That may not be possible yet, so you're going to need a dedicated daq board, and split your analog signals to sample them independently, exactly as Jaewon said.

PS, there are integrated stimulus presentation and electrophys systems used in vision science (http://diagnosysllc.com/products/espion-pattern-profile/), but they are very different from plexon+monkeylogic.  


0
Previous Topic | Next Topic
Print
Reply

Quick Navigation: