Author Topic: Interfacing ISC0601B bolometer  (Read 5411 times)

0 Members and 1 Guest are viewing this topic.

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Interfacing ISC0601B bolometer
« on: October 02, 2017, 09:00:00 am »
As Autoliv car IR camera hacking has not born a fruit, I tried to interface with the sensor directly.
I thought that nice aluminum car camera housing (FLIR Tau/PathfindIR style) with a lens, nice built-in shutter is a good start for a project.
I managed to convert sensor serial output to BMP for single frame shots, but seems that anything faster would require FPGA skill, which I lack  :-//
I even tried to attract Mike to this idea, but I failed. Perhaps there are individuals here interested in making progress in this field ?
Piotr
P.S. Here - hot wire pictured.
 
 
« Last Edit: October 04, 2017, 05:52:15 am by LesioQ »
 

Offline marshallh

  • Supporter
  • ****
  • Posts: 1462
  • Country: us
    • retroactive
Re: Interfacing ISC0601B bolometer
« Reply #1 on: October 05, 2017, 11:31:59 pm »
Is there more information on the pinout/data format available? Doing fpga stuff is very easy for me.

Maybe you have read out the sensor so slowly that the data has changed during that time? I'm curious if the horizontal lines are simply from a lack of correction.
Verilog tips
BGA soldering intro

11:37 <@ktemkin> c4757p: marshall has transcended communications media
11:37 <@ktemkin> He speaks protocols directly.
 

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #2 on: October 06, 2017, 05:46:33 am »
I did it with discrete TTL dual-channel serial-to-parallel converter, then into LWLA 1034- logic analyzer.
At 64MHz clock rate - this is quite a stretch to this circuit, especially when input levels are just 2.5V for a TTL.
I tried various supply voltages to find the sweet spot, but this appears the best sync I could get.

Do You have any autoliv donor camera ? Would be easier to describe what I did so far.

The picture is just a single frame grab. The sensor sends data for each frame upon inquiry, so no need to hurry with processing.   :phew:
Bias (dark) frame calibration is the first step to make image clearer, I guess. Line sync errors are the white pixels on the right.
 
« Last Edit: October 06, 2017, 05:55:45 am by LesioQ »
 

Offline railrun

  • Regular Contributor
  • *
  • Posts: 113
Re: Interfacing ISC0601B bolometer
« Reply #3 on: February 13, 2018, 01:05:56 pm »
Nice project. Did you make some further steps?
 

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #4 on: February 13, 2018, 01:37:48 pm »
Nope. Stuck on TTL to FPGA move phase.
Tried to digest Xula2 programming but ... too old ? Or maybe too lazy ... :palm:
 

Offline railrun

  • Regular Contributor
  • *
  • Posts: 113
Re: Interfacing ISC0601B bolometer
« Reply #5 on: February 18, 2018, 12:10:33 pm »
So do you have the pinout for this sensor? Or you try and error?
 

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #6 on: February 18, 2018, 12:20:33 pm »
I split some signals with sensor connected to camera body, so original body is handling the power (and some auxiliary DC voltages) and the digital pins go to my interface.
As simple as one input (single pulse as a trigger) and two outputs (serial data out, likely odd and even lines).
 
 

Offline oddbondboris

  • Contributor
  • Posts: 38
  • Country: us
Re: Interfacing ISC0601B bolometer
« Reply #7 on: March 05, 2018, 03:05:49 pm »
is that 64mhz serial or 64mhz parallel
if its 64mhz serial which would make sense for 640x480x30fps*14 bit on two pins that's maybe managable just barely in a pic32 spi peripheral albeit a little out of spec
on any fpga or cpld of almost any description you can do that data rate without difficulty.

i'd be very tempted to use a lattice fpga breakout board since they're cheap, include a programmer and arewidely available to at least get the serial stream properly deserialized and perhaps generate a derrived pixel clock to make keeping sync a bit easier on the parallel side or to output in a ccir656 style waveform and a dedicated ccir656 to usb webcam chip
 

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #8 on: March 05, 2018, 03:45:23 pm »
Yeah, then comes dark-frame-ing (incl. shutter operation), then some more post-production filters.
Yes - data is 2-channel of serial, single Clk input. The sensor is 320 X 240 (c.a.)

« Last Edit: March 05, 2018, 04:07:04 pm by LesioQ »
 

Offline oddbondboris

  • Contributor
  • Posts: 38
  • Country: us
Re: Interfacing ISC0601B bolometer
« Reply #9 on: March 06, 2018, 04:58:23 pm »
well if you ever come across one of these in the sub $100 range let me know, this is exactly my kind of project
 

Offline elbartek

  • Newbie
  • Posts: 3
  • Country: ca
Re: Interfacing ISC0601B bolometer
« Reply #10 on: April 20, 2018, 09:45:08 am »
Hi guys, I am very interested... after spending 5 night trying to restore some kind of firmware to an i7 with bricked firmware... reverse engineering with IDA, countless reboots and firmware upgrades with hacked config files... I managed to somewhat install firmware of the wince op sys to the device and the flash files found from flir, but I can't get the original files and the old versions which I need... so anyways... the device boots up with my hackish sd card autoload hack method but given some config errors or whatever the heck there is... only displays a blue or checkerboard pattern... it seems not to be able to interface to the sensor or calibrate it or whatever.. I am just tired of trying and FLIR is very IP hence I have no access to source and they won't send me the flash file... anyways...
The good news... I got a ISC0601B sensor to play with... and not wanting to reinvent the wheel and probe myself the bolometer... would you share the pinout? I am okay with FPGA development and have some virtexes at hand... I could interface the sensor to an FPGA + MCU
 + sram combo potentially ... and have a thermal imager 320x240 as I wanted... huh... Thanks in advance!
 

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #11 on: April 25, 2018, 01:27:56 pm »
Great to hear You're willing to make progress  :-+
I will write down my observations into a working doc and will be in touch.
Piotr.K

 
The following users thanked this post: jone546

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #12 on: April 26, 2018, 11:51:42 am »
My FLIR ISC0601B sensor work.

My expirience is based on partial disassembly of Autoliv nightvision camera for BMW/AUDI cars.

I did not work on the sensor separately, I decided to keep power (several), grounds and clock signal connections
with original autoliv PCB intact.

My ISC pin numbering is 'folded' - like DB9 standard, i.e. 1,2,345..10-u-turn-11, 12, 13..20
Pin 1 has square pad in autoliv PCB.

My observations were: (sensor pin-wise)

1,3,5,9,11,13,15,17,19 are GNDs
10 is sensor housing
7 power 2,56V
8 power 8,9V
12 power 3,3V
14 power 9,0V

4 Clk [75MHz]
6 FPGA 2,5V logic, sensor CE, pulled-up
16 FPGA 2,5V logic, Frame request on falling edge
18 FPGA 2,5V logic, Image Enable when HI, empty frame when kept LOW
2 2,5V logic output serial data channel A
20 2,5 logic output serial data channel B

I've made an interconnection wire (20-lead) with male/ female connectors for PCB and sensor.
Then added 10 pin IDC with 7 wires to:

- split CLK signal from pin 4
- split a GND connection
- rerouted connections of sensor pins# 6,16,18,2&20. These are no longer connected to original camera PCB.

To receive a frame from the sensor:

- turn on the original camera power (12V to car chassis connector). Camera feeds Power(s) to sensor pins. It also generates CLK signal continually.
- keep sensor pin 6 pulled-up to 2,5V (can be sourced from sensor pin 8 via 1kOhm resistor)
- pullup pin 18 (2,5V logic!)
- send positive pulse (2,5V logic) to pin 16. 20-100 milisecond long, apparently it's time does not change actual 'exposure' time
- few milliseconds later output stream will appear on pins 2 & 20 (odd and even image lines).
- if You want another frame - just pulse 16 again.

Single frame data output (2,5V logic) is about 16ms long, when CLK=75MHz.
Seems CLK can go way lower - like 8MHz to ease timing.
Data is 14bit big-endian, each image line has a header of 1010101010... pattern (1555h, 1554h)

For now I don't know which output channel is odd/even line#.
As I remember - first 2 line data is a dummy or some non-image data.
To see all dummy and header parts in output stream(s) - ground pin 18 then pulse pin 16.
Then sensor spits out a kind of 'black' frame - with zeroed image data, keeping data headers pattern.

Piotr.K
« Last Edit: April 27, 2018, 06:58:46 am by LesioQ »
 
The following users thanked this post: lukier

Online Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Interfacing ISC0601B bolometer
« Reply #13 on: April 26, 2018, 01:59:43 pm »
For info, the microbolometer has two pixel bias voltages that set its thermal sensitivity. Those will be two of the DC voltages noted above. Each individual microbolometer is different so the bias voltages can be slightly different between otherwise identical cameras.

Never operate a microbolometer without its correct clocking signals present...... they tend not to like such a situation and suffer damage !
I was surprised to hear this but it came direct from a thermal camera designer so he should know about such things. In the case of this specific  microbolometer, it appears that a master clock is used to create the three other clock signals so hopefully the ROIC also protects the microbolometer if the correct clock is not present.

Fraser
« Last Edit: April 26, 2018, 02:03:05 pm by Fraser »
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline LesioQTopic starter

  • Regular Contributor
  • *
  • Posts: 66
  • Country: pl
  • Every king should be naked.
Re: Interfacing ISC0601B bolometer
« Reply #14 on: April 26, 2018, 03:44:33 pm »
Interesting. May it be a kind of VCG Voltage Controlled Gain  in this case ?
A thing to evaluate, then.
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 13148
  • Country: gb
Re: Interfacing ISC0601B bolometer
« Reply #15 on: April 26, 2018, 04:00:04 pm »
The bias voltages are set values for differing thermal sensitivity ranges. This is why thermal cameras often have more than one Range to cover a large span of temperatures. Each Range has its optimum bias voltages for the pixels. FLIR also use a 'High Gain' And 'Low Gain' mode in their cameras.

Fraser
If I have helped you please consider a donation : https://gofund.me/c86b0a2c
 

Offline ir.ukrm

  • Contributor
  • Posts: 48
  • Country: ua
Re: Interfacing ISC0601B bolometer
« Reply #16 on: July 26, 2019, 05:01:55 pm »
I think the one who hacks this camera will get the Nobel Prize.
 

Offline jone546

  • Newbie
  • Posts: 1
  • Country: il
Re: Interfacing ISC0601B bolometer
« Reply #17 on: December 25, 2019, 11:28:10 am »
hello
can you give how are you  Interfacing ISC0601B bolometer in your project
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf