Author Topic: Fast Low resolution imaging  (Read 323 times)

0 Members and 1 Guest are viewing this topic.

Offline AkiTaiyo

  • Contributor
  • Posts: 5
  • Country: gb
Fast Low resolution imaging
« on: June 19, 2020, 01:34:43 pm »
Been playing with a Heimann 32x32 HTPA thermopile array, and have had some success with getting high speed images out.
There’s a setting for adjusting the frequency and resolution, which if you tweak them you can get the readout very fast!  On my test I use a micro to poll the sensor to see if a new frame is ready, then when its available download the A+B segments (the readout is strange!), then trigger a new measurement as soon as the readout is complete.
I can get approximately 170Hz out of the sensor, which isn’t bad (32x8 res)!  Of course because of the weird readout, if you want the full frame, then you actually have to perform 4 measurements, which reduces the framerate and adds a lot of tearing..
The image attached shows the I2C readout with a measurement from start of polling of one frame to the start of polling for the next.
The calibration build into the sensor becomes garbage though so you need to do your own cal, which is easy enough to do roughly provided the module doesn't change temperature too much.
I suspect I can go faster but my little EFM32 really doesn't like the I2C running much above 1MHz, also the waveforms on the I2C are starting to look horrible, even though the pull ups are probably as low as I really want to go.
They do an SPI version, I'd like to get hold of one to see how fast its possible to get data out of it without the I2C limitation.

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9541
  • Country: gb
Re: Fast Low resolution imaging
« Reply #1 on: June 19, 2020, 04:27:22 pm »
Do you consider the frequency response of the thermopile pixels to be a limitation in your tests ? I would expect the thermal response coefficient of their design to dictate the maximum useable sample rate, even if the ROIC can operate at a faster rate.

A black body equipped with a rotary shutter wheel would help test the true maximum response rate of the pixels.

« Last Edit: June 19, 2020, 04:29:12 pm by Fraser »

Offline bap2703

  • Regular Contributor
  • *
  • Posts: 110
  • Country: io
Re: Fast Low resolution imaging
« Reply #2 on: June 19, 2020, 05:19:59 pm »
Their older generation 32x31 was "8ms for Xenon or about 50ms for vacuum".

Their 32x32 appears here with an indication about its speed.

Their 80x64 is said here to have a 4ms time constant.

But their 16x4 is 0.8 ms for example.

I guess they decided the user doesn't need to now to pixel speed for their 32x32d.
The following users thanked this post: Fraser

Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo