Presentation on theme: "ELT AO WFS Technology Validator e2v KO Feedback Mark Downing Instrumentation division ESO 11 Jun 20101ELT AO WFS TVP KO Feedback."— Presentation transcript:
ELT AO WFS Technology Validator e2v KO Feedback Mark Downing Instrumentation division ESO 11 Jun 20101ELT AO WFS TVP KO Feedback
Has e2v considered to add an extra stage of pipelining whereby the pixel processing is done in parallel to the ramp ADC (rather than sequential). This way the full 34us (30kHz) could be used to noise band limit (filter) the signal. This could dramatically reduce the read noise below current values. This may require the proposed voltage amplifier stage to do a CDS (signal – pixel reset) to pre-cancel the kTC noise before further processing. 11 Jun 2010ELT AO WFS TVP KO Feedback2 Pixel Rate pix/sec 1 M10 M 100 k 10 k e2v Read noise vs speed Decrease noise BW
If these new lower noise source followers are NOT available for the TVP but become available and are proposed for subsequent phases, is it a simple matter of using them or does it require another phase of optimization. Same question asked a little differently: how relevant will be the source follower optimization of the TVP in respect of using these new low Vt transistors. 11 Jun 2010ELT AO WFS TVP KO Feedback3
What limits how much gain can be applied; –Linear voltage range - 120uV 1.44/e * 4000e * x3 gain = 1.44V ? –Other? 11 Jun 2010ELT AO WFS TVP KO Feedback4 Will all these improvements be addressed in the TVP?
I could not find any mention of this investigation of dark signal. Is this still planned? 11 Jun 2010ELT AO WFS TVP KO Feedback5
Can this circuit please be update to include the voltage preamplifier? Will an additional mode be added to sample before/after the preamplifier? What impact will this have on the test plan. 11 Jun 2010ELT AO WFS TVP KO Feedback6
Has e2v considered to implement Correlated Double Sampling (either analog clamp and sample or digital CDS by oversampling then adding) off chip for mode 1. This way the pixel (noise and lag) could be fully characterized independent of the comparator and ramp; including investigating benefits of reducing noise bandwidth and longer pixel processing time. 11 Jun 2010ELT AO WFS TVP KO Feedback7 This is especially important if the comparator noise happens to dominate the total noise and thus mask the noise difference between pixel variants.
Why not save the images? Does it not make sense to save some of the raw images so that if required they can be re-analyzed later? 11 Jun 2010ELT AO WFS TVP KO Feedback8
Can mode 3 please be explained more fully and why it is the mode of choice to measure pixel lag? Where is “mode 3 input”? 11 Jun 2010ELT AO WFS TVP KO Feedback9
Note that the NGSD may be used for both LGS and NSG applications thus requires regions of programmable gain even though Req21 is “greyed out”. 11 Jun 2010ELT AO WFS TVP KO Feedback10 What are the issues?
11 Jun 2010ELT AO WFS TVP KO Feedback11 There is no direct mention of building in testability in the NGSD or LGSD design documents. What plans are for building in testability?
The proposed photodiode has an asymmetry in X and Y thus PSF will be different in X and Y. Is this a concern? Should ESO ask e2v to make the photodiode square rather than rectangular? 11 Jun 2010ELT AO WFS TVP KO Feedback12
Can a discussion of QE please be added to this document or to the NGSD design document. Note that it is unlikely that using mirrors will be an acceptable solution; degrades PSF and centroid accuracy. Note that 10um epi (without mirror) will meet specs but will go no where near meeting the goals. This limits the extension of use of the device to NGS applications. What silicon resistivities are proposed (available)? How applicable (what needs to be changed/modified) will the TVP results in std silicon be to devices made in high resistivity silicon? 11 Jun 2010ELT AO WFS TVP KO Feedback13
Ease of mounting (e.g. how to attach heat exchanger, how to mount the detector to a PCB, how to clamp it….) and use of the package is important to ESO and thus it is important to add a section to this document that considers how to mount the package inside the camera (holistic approach). This should include ease of aligning (and keep it aligned) the detector once mounted inside the camera to an optical beam. 11 Jun 2010ELT AO WFS TVP KO Feedback14
Why a discrepancy between ELT WFS requirements and e2v package requirements? 11 Jun 2010ELT AO WFS TVP KO Feedback15
We should provide early feedback on these proposals; action item on ESO to study these options and provide advise. Initial response is that both do not seem satisfactory. 11 Jun 2010ELT AO WFS TVP KO Feedback16
Yes, agree that ESO has to provide a reasonable thermal control system but the thermal elements (Peltier and temperature sensor), characteristics of the package and detector thermal sensitivity should not impose unreasonable constraints on this control system; e.g. temperature stability down to ±0.1K is reasonable where as ±1mK is not. 11 Jun 2010ELT AO WFS TVP KO Feedback17
If necessary, this type of solution can be discussed further. 11 Jun 2010ELT AO WFS TVP KO Feedback18
ESO (controller and RTC teams) should provide some feedback on this design choice. 11 Jun 2010ELT AO WFS TVP KO Feedback19 ESO agrees.
The ESO controller team should provide feedback on this scheme. Note for image data transmission, the receiver end will be the controller and a commercial IC (FPGA) thus the scheme should be dictated more by what is available (industry standards) and reliable than constrained by “large circuit overhead”. 11 Jun 2010ELT AO WFS TVP KO Feedback20
AI: ESO to confirm that proposed read out of reference pixels is acceptable. 11 Jun 2010ELT AO WFS TVP KO Feedback21 Why only cover 8 columns of pixels and why not more; e.g.16?
20 rows of pixels was suggested but not all pixels need to be black pixels. A mix of say 16 black pixels, 2 transition pixels and 2 dummy pixels is quite acceptable. The overhead is to read out an extra subaperture (2.4%). 11 Jun 2010ELT AO WFS TVP KO Feedback22 Nothing beats reading real black pixels to get the most accurate measure of offsets and offset variations with time, temperature or temporal noise. Taking calibration frames (biases that can be subtracted to remove offsets and flat fields that calibrate out FPN) is standard but may only be taken once per night. Having reference pixels per row is basically good insurance.
Agree that this is a safe way to proceed to guarantee that the NGSD works while at the same time investigate more speculative enhancements. 11 Jun 2010ELT AO WFS TVP KO Feedback23 Agree that this is important to start discussing and deciding at an early stage. On the ESO side, the controller and RTC teams will need to be involved. The KO referred here is NGSD KO or TVP KO.
AI: ESO to provide an indication of minimum data reliability that will not impact the final application. 11 Jun 2010ELT AO WFS TVP KO Feedback24 Will the NGSD/LGSD have some built in testability (digital pattern generator) that can be used to check/confirm data transmission reliability.
How realistic is this option? Cost? Maturity of the technology at TowerJazz? Taking into account the total number, pin reduction (340 to 250) is not all that significant. Are there other significant advantage? Power dissipation or ease of interface? 11 Jun 2010ELT AO WFS TVP KO Feedback25
Interesting concept worth considering especially for applications that do not use the full array. 11 Jun 2010ELT AO WFS TVP KO Feedback26
With only one review involving RIXs, it is difficult to see how 2.5 months can be saved during the TVP by streamlining the review process. The final TVP review is important and should not be shortened, but of course work could continue in parallel on the next phase. Note that the TVP design is planned to be presented at a bimonthly progress meeting. 11 Jun 2010ELT AO WFS TVP KO Feedback27