IRTF Optical Alignment

After a long night of observing, Matt, Tommy, John and Andreas have concluded that we are losing a great deal of light at the slit.  This may be attributable to adjustments made inside EXES to compensate for astigmatism in our oBox powered mirrors, or possibly to a misalignment between the telescope optical axis and that of the instrument.

As a test, we'll start by placing our laser fiber at the expected telescope first focus, and looking at our beam as it traverses our foreoptics.  Placing both the end of the fiber and the center of our beam-defining iris at the design height of 4.162" above the optics plate, and centering both on the centerline of the optics plate, we se that the beam hits near the center of the ellipsoid, but is steered nearly off the hyperboloid and enters the window.  Here is the beam at the hyperboloid:

Admittedly hard to see, but the edge of the red laser beam is essentially coincident with the edge of the hyperboloid.  Because it is off-center on the hyperboloid, it enters the instrument window off center:

Again, kind of hard to see, but the spot is entirely in the upper left quadrant of the entrance window.  Suspect that this is not a good arrangement.

At his point we turned the laser/iris assembly around so that it shines toward the fold mirror.  It was *not* hitting it on center, be we decided to adjust the beam for level travel along the optics plate as we had for the Z-beam optics, and attribute the angle to manufacturing tolerances in the bracket.  We placed tissue paper across the fold mirror to define an edge that coincides with 4.162" above the optics plate:

The line between the tissue paper and the hand-held card is a 4.162" off the optics plate, and it pretty well splits the beam, suggesting that the laser beam is traveling parallel to the plate at the right height.  Whether this beam its the mirror at the same spot as the telescope central ray is the big question.  We should be able to translate and/or tilt the fold mirror to send the sky beam back along the right direction.  We'll need actual star light to verify this.

Array Linearity (low flux)

Looking at a front surface mirror just outside the entrance window (7µm filter, hi-res mode, narrow slit w/decker), we collected a series of frames at varying integration times and biases.  Picking a 32x32 box in a bright area of one order and converting to electrons based on 35 e per DN, we have the following result:

In the plot above we see good linearity up to ~25,000 electrons for high bias, which is about what I'd expect to be a quarter-full well.  To map the onset of debias/non-linearity, we'll need more light than we're getting off the mirror through the 7µm filter.

If we fit a line to the data above at each bias , we get a curve for detector response as a function of applied bias in perfectly arbitrary units:

This curve doesn't show the characteristic 'S' shape of an IBC detector response, but investigating that will have to wait until after the IRTF run.  We're seeing very linear detector performance up to ~25k electrons, and we'll need to repeat the test with more light to characterize the debias curve.

Verify Gains and Conversion Factors

Reference to the transfer function data in the previous post in comparison to the display data numbers logged as part of the same measurement (cf. Vol 3, p.22) confirms the eBox preamplifier 'Gain 0' measurement made in May 2010 (Vol2, p55).  Taking the May measurements of the four available eBox gain settings, we can estimate conversion factors from DN to electrons based on a 33fF assumed input capacitance and the measured SF gain of 0.72:

eBox Gain Selection:

'0' :  Preamp gain = 2.61, 9.1 µV/electron @ ADC input, 35.9 e/DN
'1' :  Preamp gain = 3.21, 11.2 µV/electron @ADC input, 29.2 e/DN
'2' :  Preamp gain = 5.07, 17.7 µV/electron @ ADC input, 15.6 e/DN

'3' :  Preamp gain = 8.87, 31.0 µV/electron @ADC input, 10.6 e/DN

Operating Point @ IRTF

The table at left shows the starting values for biases and clocks for our IRTF run.  These may need to be revised, but seem to work well.  With these values, we can map the source-follower transfer function to deermine the linear range of the device, shown in the plot above.

The measured gain is linear for VDDUC values between -2.2V and -3.8V.  Setting VDDUC more negative than -3.8V results in glow along the row mux and an unacceptable increase in device temperature.  VDDUC values more positive than -2.2V rail the output source follower.  The gain itself, at 0.72, is lower than that measured for the SB226 (cf. Lab book Vol 1, p97), but there is no obvious reason to doubt the data.  The gain is higher than the ~ 0.57 value measure warm for the bare mux (Vol. 3, p.16), and with a wider linear range.

Assuming an input capacitance of 33fF, this gain would result in a conversion factor of 3.5µV/electron at the device output.  We can leave a little headroom to account for spatial nonuniformity by choosing a value of -2.25V for VDDUC/VRST.  This gives a source-follower-limited range of ~320k electrons, probably enough to cope with the highest detector bias we can maintain.

EXES pumping and LN2 filling before IRTF run

Here are the notes from Lars and IRTF about EXES pump down and cooling at the IRTF.....

12-17-10 @ 11:30: pump pressure 7.2e-5 Torr. Valve off dewar. Shut off pump.

@15:00: pump manifold 7e-2 T.

12-22-10 @ 11:00: pump manifold 2.5 T.

12-29-10 @ 9:45: cracked dewar valve (with pump on). pressure rise to 2e-3 T.

@13:00: 9e-4 T.

@15:30: 5e-4 T.

12-30-10 @04:20: 8e-5 T

@15:00: valve closed. 5e-5 T

1-2-11: ~19:00: valve opened. P 4e-4 T.

1-3-11 @10:30: pump at 5e-5 T. Start filling outer can with LN2. Pressure dropped. Valve closed.

@10:45: outer can full. switch to inner can.

@11:30: inner can full. switch to outer can (still full).

@12:00: topped off before lunch (still had some liquid) - NOTE: not certain which can.....

@14:00: inner can empty, outer still fuming. topped off both cans. 15 min on inner.

@14:25: storage dewar empty

@14:50: new dewar. restarting N2 can fill

@15:30: last top off, both cans.

@21:15: outer still fuming. inner empty. fill inner. push in screwdrivers. top/middle echelle may not have clicked in, but rest moved in nicely.

@21:30: inner full

@21:40: start data logger to CD11.txt (1 second intervals changed soon to 1 minute)

@21:45: start filling outer can

Temps: A=113.3; C2 = 74.129; D4=73.907; D2=107.45

@22:00 outer full.

@22:15 Temps: 111.21; 73.85; 73.74; 104.17.

@23:15 top off inner. took 5 minutes. outer had low enough boiloff to leave. try to push in screwdrivers, but all pretty solid.

@23:30 head down.

updated 1-4-11@12:20: both cans still full with low boil-off. plan to fill before leaving or when they run out (will not fill inner since we will be pumping soon). temperatures available in cd11.txt

updated 1-4-11@15:00: both cans still have liquid. filled outer can (5 minutes). put LN2 into inner can for 2-3 minutes (since we'll be pumping tomorrow, didn't want it full). Changed temp log rate to 5 minutes.

1-5-11@10:30 - rotated EXES to 3 clicks short of vertical so that LN2 runs out of cans (necks down). Start pumping on inner can. Using backing pump on turbo. Have gauge on pump and installed our pressure gauge on neck tube port. Have balloon on one-way valve to check for leaking. Have plastic in-line valve connecting two lengths of blue hose. in-line valve is cracked right now. getting ~7 torr at pump and ~450 and dropping on our gauge. temperature coming down. going to let sit for a bit before opening in-line more. filled outer can.