Wednesday, April 17, 2013

Welding a thin thermocouple with cylindrical junction under a microscope

Thermocouple is a standard to measure temperatures and is widely used in industry and academia. Thermocouples are made by welding two wires with different materials at one end. The temperature of the welded junction is obtained by measuring the voltage between the two open ends. The physical principle of thermocouple is based on thermoelectric effect (Seebeck effect) where the temperature difference can be converted to electricity. The temperature and voltage conversion are tabulated into standard tables for various types of thermocouples made with different materials. These tables (or sometimes polynomial fitted equations) are often used to refer temperature from a measured voltage.

Thermocouples only perform point measurement of the junction temperature. The junction of the most commercially available thermocouples are mostly sphere-like but the geometry is not in a well-defined shape. The junction is also normally bigger than its adjoining wires.

This could introduce two problems: 

1. When thermocouple is used to measure fluid temperature such as hot gas stream, there is more uncertainty in calculating heat transfer when the junction does not have a well-defined geometry; 

2. The junction may have different temperature of its adjoining wires due to different radiation loss.

The merits of making a thermocouple with cylindrical junction is to solve the two problems to improve the accuracy of temperature measurement.

The following video shows how I welded the thermocouple under a microscope using a micro-torch.





The video is also available on youtube:
http://www.youtube.com/watch?v=HUriLTnkCQw



Tuesday, April 2, 2013

Spectral response of Nikon DSLRs (D90 and D300s)

Relative spectral response of two Nikon DSLR cameras were measured and shown in the post.

Background on spectral response:
Spectral response of a detector describes the efficiency of transforming photons to digital signals (i.e. CCD counts). It is related to quantum efficiency QE by the following equation

SR(lambda) = Const*QE(lambda)./lambda

where SR stands for spectral response, QE stands for quantum efficiency and lambda is the wavelength. 

Bayer pattern filter array employed in DSLR cameras
For a DSLR camera, a Bayer pattern filter array with RGB channels are normally employed to yield a vivid color image. The spectral response of each channel was measured and shown below.


Setup:
A stable light source is used to provide the illumination. A calibrated spectrometer is used to disperse the light and provide a known spectrum on the ground glass. The spectrum is then imaged by the camera. The ratio of the real spectrum (on the ground glass) and the imaged spectrum (by the camera) is a direct measure of the relative spectral response.


Setup to measure the spectral response of a DSLR camera

Images are taken in raw format (.NEF for Nikon) to preserve quantitative information. The subsequent image decoding and processing is performed by an open source software OMA that runs on Mac OS.


Results:

The spectral response of two Nikon DSLR cameras are shown here. Despite that D300s is a more advanced model than D90, the two employ the same CMOS sensor. The measured spectral responses of the two are quite similar. The Y-axis shows the relative signal response.


Spectral response of Nikon D90

Spectral response of Nikon D300s

Thursday, April 12, 2012

DIY Canon 400d infrared conversion

Unlike the eye, sensors based on silicon (including CCDs and CMOS sensors) have sensitivities extending into the near-infrared. Such sensors may extend to 1000 nm. There are a few DSLRs are capable of sensing infra red light, e.g. Fujifilm FinePix S3 Pro UVIR. However most of the digital cameras and video cameras are usually equipped with IR-blocking filters to prevent unnatural-looking images.

Objective: I am going to replace the built-in filters in front of the image sensor (CCD or CMOS) with an IR filter.

Such conversion will NOT yield special functions like thermographic imaging which wavelength range is about 9000 nm - 14000 nm. But it does pick up some light which can not be sensed by human eyes in the short wave infrared region.

Tools: 1. a right size screw driver; 2. a tweezer; 3. some lens paper; 4. some ethanol; 5. a dust blower; 6. solder kit (for certain models including 400D, 350D and earlier versions)

Here is my 5 years old Canon 400D for IR conversion.

You'll need to unscrew about 20-30 small screws, and unsolder the metal plate on the main circuit board. The camera is delicate and fragile, make sure your touches are gentle, especially while disconnecting the ribbon cables and dealing with the CCD sensor. All the process should be at least performed in a clean environment with minimal dust present.

The brief procedures are as below.

Camera Dissemble

Unscrew the bottom and side screws until you can open the back.
Detach the big Ribbon cable and take the back (LCD display and control panel) away
Unsolder the metal plate in order to access the bottom ribbon cable. (Be very careful and do not let your solder  gun touch the cable or any components on the circuit board)
Flip the main board over and you'll see the back of the CMOS assembly
Unscrew the necessary screws. Flip the CMOS assembly over. In front of the CMOS sensor, there are actually a series of optical component, including the IR blocking filters that I want to replace.

Here is a closer look of the CMOS sensor assembly.
In front of the imaging sensor, there are 1. Low-pass filter 1 that separates image in horizontal direction; 2. dichroic mirror and glass filter to pass only visible light from 400 nm to 700 nm; 3. Phaser layer (a quarter wave plate) to converts linear polarized light into circular; 4. Low-pass filter 2 that separates image in vertical direction.
Notes: The piezoelectric element is used for dust cleaning; the Low-pass filter 2 is attached on CMOS sensor as a protective window, I'll just leave it there.


                                                                            CMOS sensor assembly layout

The CMOS sensor is dissembled from the assembly, leaving only the filters in front of it.
CMOS without filter
Back of CMOS


The Low pass filter 1 and dichroic mirror (the two are attached together) can be taken off. This part will not be put back in the re-assembling process.
Leaving only the glass filter on the supporting frame
The sides of the glass filter is attached on the frame by tape
Cut the tape
Apply some force and push the glass filter off from the frame. The sides are glued on the frame, but one can easily push it off. (put a lens paper between the finger and filter)
I now get rid of the IR blocking filters. You can screw everything back, that will give you a full spectrum camera. However since here I'd like to make an IR camera. I'll put my customized IR filter in.


Choice of IR filters:
There are many choices of IR filters in the market. Instead of purchasing a $200+ glass IR filter, here I considered some cheap ways to make a usable filter.

1. Developed Kodak negative film
I learned from the web that the developed negative film can be used as a IR filter. I bought the film from Walmart for ~$3, exposed it under sun light for 3-5 seconds and have it developed in Costco for free. My transmission measurement of the developed film is shown as blue curve in the figure below. As you can see the developed film still passes some light at 400 nm and 600 nm.

2. LEE Filters 3x3" Infra Red #87 Polyester Filter
I bought it from B&H for ~$14, the LEE filter is much better at rejecting visible light as can be seen from the red curve below.

Given the measured spectral transmission curves, LEE filter seems to be a much better choice than the exposed film.


In order to minimize the problem of focusing, (i.e., matching the refractive index of the IR filter to the original filter) I glued the sides of the IR LEE filter to a clean microscope slide. (I broke the slide to make it similar size with the IR blocking glass filter.) That gives me more or less the same optical path compared with the original filter.


A piece of micro slide
Infrared LEE filter
The LEE filter is glued on the glass slide
Put the customized IR filter back into the frame. Put everything back in the reverse order of the dissembling process. The two things should not be put back are the dichroic mirror (and Low pass filter 1 of course since they are bonded together) and the green glass filter.


Tuesday, March 6, 2012

The spectral output of computer LCD

I examined two computers that I have, one is a imac and the other is a macbook. The two screens were set to the same Gamma settings. Blank white Microsoft word page were open on both computers. A calibrated compact CCD spectrometer was used to scan the "white" color from two displays. The scanned spectral outputs were shown in the figure below, with the X-axis to be wavelength in nm and Y-axis with arbitrary unit.


It is interesting to find out the two displays have dramatically different spectral outputs. The imac has a few peaks at specific wavelengths (e.g. at 550 nm and 615 nm), while the macbook is broader.