Humidity control update

The new closet approach works very well. I topped up the humidifier on Thursday, and on Monday afternoon, there was plenty of water left in the humidifier. One contributing factor is the relatively warm weather over the weekend. If this trend continues, we will likely only need to top up the water twice per week, until the summer months.

Starcheck uniformity experiment

Back in January, we mounted Starcheck on a Velmex linear stage, again, to measure the response uniformity of all the detectors in a rigorous way.  By moving the profiler 9mm per step and taking a 300MU profile each step, we tie 22 detectors directly to the central chamber.  By measuring profilers at 3mm and 6mm offset, we managed to tie remaining two groups of 22 detectors to the central chamber, in an indirect way.  Note that the starcheck central chamber has a 6mm gap between it and its neighbours, rather than the 3mm spacing for other chambers.  This was a cause of grief in analysis, as we can't simply multiply the detector number by 3 to get the chamber position.  

Looking at the data, what we see is that all three groups follow the same shape, suggesting a systematic bias.  For now, I speculate that the bias is due to beam quality modification by the wedge.  Since we already know that the Starcheck central chamber show strong energy dependence, it is not surprising that we observe beam quality effect due to beam softening by the wedge.  Note that the detector at -78mm is an obvious outlier, showing a 2% deviation from the central chamber.  Given the 1%, 2SD uncertainty quoted by the manufacturer, this is possible.  Further characterization may be needed before any conclusion about this detector can be made.  The three colors represent the three groups of detectors, offset from each group by 3mm.  The red dots represent detectors tied directly to the central chamber, while the other two colors represent detectors tied through the 3mm and 6mm measurements.

New data on Jan 23rd Vickers experiment

A significant discrepancy between Al chamber measurement obtained by Claudiu and Al film measurement obtained in October was observed. Adding Monte-carlo data into the mix, I found an unusually large discrepancy between MC and chamber measurement. If we add Jan 23rd data of Aluminum into the mix, which we know is 2 sheets of aluminum(#32 and #33), the profile is very close to Monte-Carlo simulation. The question now is, what is the foil used in October's Vickers experiment?

February 07 update:  The foil used in October's Vickers experiment is the same as Jan 23rd.  The results from October seem to be 0.5% narrower than Jan 23rd film measurement, which can be due to different distance between exit window and film.  Also note that the green circle just shows the excellent circular symmetry of electron beam, allowing easy contouring and circular fitting in Matlab.  I'll try to submit a Vicker X-ray contour soon to compare.

Starcheck detector response

With the Friday Morning Talk out of the way, we can now focus on other stuff, such as Starcheck response characterization.  Previously we compared non-rotated and rotated Starcheck profiles to get a feeling of uniformity.  Now we put it in a more rigorous test, by scanning a dose wedge with Starcheck.  Mounting starcheck on a Velmex linear stage, we move the profiler 9mm between successive scans.  The result is 23 scans, which can be used to link detectors on a line to central detector, as they are measuring the same dose.  The data points are 3 detectors apart, except for the central detector, which is 6mm away from its neighbours.  This exception actually caused me fair bit of grief in trying to figure out what's wrong with the parsing code.  After adjusting the code, we have a comparison of 22 detectors with the central detector.  Note that a systematic trend can be observed, with the edge detectors reporting lower dose than central detectors.  This may be attributed to the fact that for edge detectors, when they are at the beam axis, the central detector is at the edge of field.  Since Starcheck uses only the central detector for absolute dose measurement, a change in beam quality as we move across the wedge profile would cause a change in charge to dose conversion coefficient.  With that in mind, we note that detectors in the central 10cm are within 1%, which is within the manufacturer's specification.

EBT film monitoring

After much procrastination, I finally got to the monitoring of EBT films. A TinyTag logger, the one used in the room next to Malcolm's office, was put inside the cardboard box of film cutter, along with two sheets of film. The two films were from a good batch, not the most recent batch that contains two boxes. The box was put into a

Gold foil and A16 measurements-finally some agreement.

This time around, with the beam structure resolved, we can finally begin to compare A16 measurements with film measurements. By applying 1% expansion correction to film, centering the A16 profile, and correcting the A16 profile for monitoring chamber effect and chamber size effect, we obtained a perfect match. Of course, the drawback of such approach is that we could be attempting to obtain a match by applying corrections. The monitoring chamber correction applied to A16 profiles is derived from window only profile, and may not be applicable. The size correction is also derived from window only profile, but we have more reasons to use that correction. Given the 1mm size of film pixel, we can approximate the film as a point detector, and the A16's relatively large detector size requires a -0.6% correction in X-axis. With all these corrections applied, and A16 centered, the gold profile matches perfectly, except for film non-uniformity issues near the peak.

New Vickers Electron work, gold foil #1

New experiment performed on Gold foil was scanned today. Gold foil #1, 30 micron thick, was film 27B. The curves look symmetric, without major features. A look at lateral and longitudinal profiles suggest that no difference between the two profiles can be detected beyond the uniformity level of the film. In this experiment, we performed a repeat film measurement without moving the gold foil, thus allowing the beam to hit the same spot of the foil. The results of the repeat confirmed the idea that differences between the two profiles are purely due to film non-uniformity.

January 22, 2009 -- Finally, the day of measurements

After 3 days of strugling with various equipment, we managed to collect window only, Gold, and Aluminum data. 5 film irradiation was done, and Window only, Aluminum, and gold scanning was collected. Preliminary results showed good agreement with previous scanning data, other than start-up effects seen at the peak.

January 22, 2009 Vickers scanning-- scattering issues?

9:47 AM: After struggling with the electrometer #29 for a bit, we managed to acquire a window only profile using version 1.5 of software. The results show noisy peak at 2% level, as previously shown. A careful inspection of monitoring data however, revealed an up to 10% bump at the center. Another inspection of scanning chamber reading versus position showed a noisy top up to 10%, but correlation between the two chambers seemed to cancel this effect, giving a noisy level at the peak region of only 2%.

10:22AM update: The cause is feebback effect of Joakin chamber. Since the beam is blocked by the A16 chamber, the Joakin chamber show lower reading, which cause Vickers to increase beam current. Now the noise is better.

January 21st Vickers experiment

8:36 AM: Continuing from yesterday. The Helium bag is holding up well. We'll work on the electrometers first.

8:41AM: Electrometer #30 voltage can't be set with software. Set it manually to -300.

9:10 AM: Electrometer #30 and #27 let software set its range, but not voltage. The software communication actually set voltage back to zero. Reverting to Labview program version 1.5 doesn't make a difference.

9:24AM: It seem like a software problem. Any electrometer put on the first box of the electrometer page gets set correctly, and the second box gets its voltage set to zero. If we leave it as channel 3, the manually set voltage is not erased. We recall that the Labview source code was edited for Laval group visitors, but version 1.5 has a modification date of 2007. Stay tuned, investigation is ongoing.

9:33 AM: The -300V set by software seem to be wiped out in subsequent measurements in a scan, at least for software version 1.5a. We found out that besides Malcolm, only Fred seemed to have experience with Labview. We contacted him, and hope that he will find time to give us a hand today.

3:30PM: After Jugo's talk, we figured out what's wrong: the "monitor chamber" tab need to be set to "field chamber". The default is "transmission chamber". This is the root of our problem. After that, we took out Elekta chamber, Jan was having a discussion with Carl while Claudiu and I took the Elekta chamber out.

Experimental setup:
10s integration time.
York monitoring chamber at 2.5m, for Vickers control

NE2581 as field monitoring chamber

A16 as scanning chamber

Film:

12 second beam time

York monitoring chamber at 2.5m

no other chamber used

7mm from bag metal edge to film

film not centered on dot.

January 20th, 2009 Vickers experiment.

We tried to repeat the Vickers experiment today. The experimental setups were documented in photo. Hopefully we have better luck repeating it next time. The photos for this experiment are here. In the morning, we tried to reproduce the same experimental setup as in October 2008, but we put in the Elekta monitoring chamber by mistake. This extra element in the beam path caused some trouble in fitting all elements into the 119.2 cm space between exit window and chamber. We eventually managed to set up everything, only to found out that our previous data did not have Elekta monitoring chamber in the beam. Then we decided to run with Elekta chamber in first, since all Monte Carlo simulations to date were done with Elekta chamber in. The scanning chamber used was A16, connected through cable #2 to electrometer #29. The monitoring chamber is 2581, connected through cable #1 to electrometer #30. The Elekta chamber was connected through black cable to electrometer #27. Upon running the software, both electrometer #27 and #30 were showing zero bias voltage, and we decided to collect data from the two chambers only, discarding Elekta data. The results showed noisy data for 2581, likely due to the lack of bias. The magnitude of measurement is small, for 10s integration time and a relatively large chamber. Thus we need to fix the electrometers first before we can make further measurements. The data from this noisy scan showed broader distribution than previous measurement, which is expected due to Elekta monitoring chamber's scattering effect. The experiment was window only.

The oxygen sensor was showing overly high readings. Carl put in more electrolyte and it seemed to behave as expected now.

Starcheck OnStage update -- wedge field work.

Here is a surface plot that I didn't really expect. In fact, I don't really know what to expect. The individual lines on the 0-25 axis are the 23 profiles obtained when the Starcheck was moved in the V-W direction, and wedge profiles were measured with different offsets, ranging from 9mm to 99 mm. So this is not a surface plot in any way, but hopefully will provide a reference on what such scans should look like. Each lines has 83 elements, and the units are really just element numbers, as I still haven't figured out how to plot in Matlab with designated axis.

V700 update

The Epson V700 seem pretty noisy today. Seem noisier than what I remembered. I do need to keep in mind that this is a consumer grade product, and it probably can't endure too many repeated scans, as I had done with the darkening study. The noisy seem to come from the motor that moves the lamp. We could have just received a defective product, but it was not too loud when we first bought it. If this gets worse, we may be able to say something about the scanner.

Calibration curve, from various literature

Digitization of calibration curves published in literature threw somewhat of a wrench into the one parameter idea. It seems like EBT calibration curves differ by a huge amount. Since extracting the red channel amounts to integrating over a segment of EBT's transmission spectrum, we expect variations due to different scanner lamps, but data from Epson 1680 and Epson 10000, which are the same class of scanners, show huge differences. Without knowing details of other authors' film protocol, we can only speculate that the variation in film sensitivity is the main reason behind this difference. As usual, more data is needed... Stay tuned for more digitized data. For the record, I use the software "GetData", which is quite convenient, and has 30-day trial period. The full version will cost $30, which is quite reasonable.

Starcheck on stage-linear

On December 11th, 2008, the Starcheck stood on a linear translational stage, with4.4 cm extra built up and 10cm backscattering.  The irradiation performed was 6MV beam wedge field 20x20cm.  21 measurements at 300MU were performed, at 1 cm apart.  Direction of travel was along the gun-target direction, as the purpose of this experiment was not to produce a map of the wedge field, but rather to examine the uniformity of Starcheck chambers.  Relative accuracy of 1% was quoted for Starcheck, but no verification has been done so far.  This will be the first test.  

Fist Post!

Well, what's the point of this blog? OK, let's be clear. This is not a stage for my drama, but rather a record of my daily work in GafChromic EBT film. As the data, both what I collect myself, and what I digitize from the literature, grows, there need to be a way to organize them. For now, I believe that a blog would be a good way for me to practise my writing skills, and provide a searchable record of data that I have collected.

On Januar 06, 2009, I returned to work after 3 week holiday break to visit my family in Vancouver. First order of business is reading the temperature log of TinyTag, which is part of our study on the effect of humidity on sensitivity of film. Three weeks of winter did gave us many days of relatively dry environment, but the humidity never went below 5%. Not having a background in diffusion physics, I can not comment on how our environemnt compares with Rink etc's drying oven, which is essentially zero humidity at 50 degree celsius over 24 hours, compared to our 5% humidity over several days. The graphs showed that humidity recovered during the day, due to relatively warm temperature outside. The next step would be to parse these films to extract the length of time spent in sub-7% environment. Again, this criteria is very arbitrary, as I do not know how the speed of water loss scales with environmental humidity.