Sunday, July 24, 2011

Frequency response test for thermal imaging sensor

I am testing the frequency response of the TPS334 IR sensor. My very crude tests agree with the datasheet, which means that the image scan time will be very long for high contrast images. Many low-cost microbolometer-based thermal imaging cameras have 80x80 sensors (6400 pixels). At a 10Hz pixel clock (-3dB sensitivity from datasheet), the frame scan time would be 640 seconds, or nearly 11 minutes. Yikes! At 100Hz, the sensor would only produce an output level of %10 of its DC capability, and still require just over a minute to scan the whole frame. Obviously, this will not be a "live video" system, but might still produce some interesting still image thermographs.

I am working on other methods of sensing long-wave IR too. More later.

13 comments:

  1. Is there any microprocessor involved in this system?

    ReplyDelete
  2. Hello Ben, congrats for owning an fpga now. If you want, you can check out, if my vga code works with your board, if you have a 1280x1024 monitor. To make it work, you have to open the project in quartus and go to (Assignments->Device) to select your specific fpga device and also (Assignments->Pin Planner) to change the pins for the 50mhz clock input and VGA output(vsync,hsync,green) (look on page 37/38 on this doc: http://www.terasic.com.tw/cgi-bin/page/archive_download.pl?Language=English&No=83&FID=91d2cf46f90973377b1f71bd42040740) only one bit monochrome so use the pin for green[3]

    Tell me if it works:)
    Simon
    http://dr.kerker.zuliah.com/vga_monitor_codepage.zip

    ReplyDelete
  3. Simon, I will try out your code soon. Thanks -Ben

    ReplyDelete
  4. That's really exciting project. Can't wait for some preliminary results and thoughts.

    ReplyDelete
  5. Ben: you have a real interesting blog thanks i have learned a few things. It seems that you have all of the bits for a good TIS. i am sure that you thought of cooling the sensor with LN2 It will be interesting to see your test results at a much lower temp.
    Thanks, ron Cheers.

    ReplyDelete
  6. There are IR sensors that contain array of pixels. Have a look here http://www.perkinelmer.com/CMSResources/Images/44-4345CAT_SensorsAndEmittersInfraredSensing.pdf

    Here is one product based on similar sensor http://www.robotshop.com/devantech-8-pixel-thermal-array-sensor-1.html

    There exist arrays up to 256 linear pixels.

    ReplyDelete
  7. Kris, I looked into pyroelectric array sensors. They cost about as much as cheap complete thermal imaging cameras (about $1200) in single-unit quantities. They also would also provide lower performance than a microbolometer array.

    ReplyDelete
  8. Found this the other day, claims to be a $100 thermal camera.

    http://www.extremecircuits.net/2012/12/low-cost-arduino-thermal-camera.html

    ReplyDelete
  9. Ooops, shouldn't comment when tired. I thought it was more sophisticated like the one you were working on.

    ReplyDelete
  10. Ian, a scanned single-pixel sensor is one of the ways to avoid needing a microbolometer (or other sensor) array. There is also this project:
    http://gizmodo.com/5940238/awesome-thermal-imager-turns-your-iphone-or-android-into-the-predators-eyes

    I am not sure what sensor he is using, but it looks pretty neat.

    ReplyDelete
  11. This comment has been removed by a blog administrator.

    ReplyDelete
  12. I am testing the frequency response of the TPS334 IR sensor. My very crude tests agree with the datasheet, which means that the image scan time will be very long for high contrast images. Fluke TI32 price

    ReplyDelete
  13. The superb highly informative blog I’m about to share this with all my contacts.redcurrent

    ReplyDelete