Embed
Email

mce

Document Sample

Shared by: gegeshandong
Categories
Tags
Stats
views:
1
posted:
12/30/2011
language:
pages:
20
Multi-Channel Electronics

SCUBA2 MCE

• 4 readout cards (RC)

– each reads 8 output columns through 14-bit 50MHz ADCs



• 1 address card (AC)= SQ1_bias

– addresses the [41] rows at add ≤ 850kHz, set by L/R

– frame ≤ add /41≈ 20kHz



• 1 clock card (CC)

– master card: interprets commands and synchronizes all the

cards. Drives fibre to PC.



• 3 bias cards (BC)

– BC 1: Squid Series Array Feedback (x32) + TES bias(x1)

– BC 2: Squid 2 feedback (x32)

– BC 3: SQ2 bias(x32) + TES heaters(x1)

SPIDER MCE

• SPIDER (as well as SPUD and BICEPII) have

up to 33 rows and 16 columns for each array





• Repackaged MCE:

– 2 RC’s

– Reorganized BC’s duties





• Additional features:

– More DET_BIAS (already in place)

– SQ2_fb fast switching (to achieve a better lock and to

remove summing coil pickup switching off all SQ1_bias and

adjusting SQ2_fb. However with fixed values. Or we could

effectively connect the dark SQ sq1_fb with the sq2_fb)

Cards in SPIDER MCE

• 3 BC’s

• BC1  0/16 SSA_fb, 0/16 SQ2_BIAS, DET_BIAS_ORG

• BC2  0/16 (possibly fast switching) SQ2_FB

• BC3  0/16 DET_BIAS

• In the current configuration slow SQ2_FB and 16 DET_BIAS

slightly different from DET_BIAS_ORG





• 2 RC’s

• RC1  0/7 SQ1_FB, 0/7 SSA_BIAS

• RC2  8/15 SQ1_FB, 8/15 SSA_BIAS





• 1 AC, 1 CC

Subrack in SPIDER MCE









T Felton design

Backplane in SPIDER MCE









SQ1_fb: it was 450A, now 150 A because of the

increase in the mutual inductance of MUX06. Is it

fine?

DET_BIAS: the maximum we can achieve now is

4mA. It can be increased. Should it?

PSU in SPIDER MCE

• Power supply:



– We currently use an ACDCU with a switching

power supply



– For SPIDER:

• we may use batteries





– 119W for SCUBA2. For SPIDER: 16 columns

 73W (23W each RC)

Cooling SPIDER MCE

to the

• Cooling: gondola

frame,

used as

– We currently heatsink

use fans and

filtered air with

positive pressure



– SPIDER MCE will

be radiatively

cooled using

circulating

fluid with a pump

T Felton design

SEU in SPIDER MCE

• Cosmic rays on FPGA and SEU tolerance: 1 per month at

JCMT  ~1 SEU / board / LDB flight. We are already in

contact with the TRIUMF particle accelerator for realistic

neutron fluency test (in fact, test flight would not reveal the

real effect).



• Altera Stratix III FPGA’s offer additional on chip monitoring

and testing features wrt Stratix I. We would not use it for the

test flight. Maybe for SPIDER.



• PC’s in pressure vessel (HD). Two with crosscomunication.



• The syncbox could be there as well.

Schedule for SPIDER MCE

• For the new repackaged MCE with no cards redesign:

– Finalize the design  1-2 weeks

– Backplane  6-8 weeks

– Chassis  our machine shop, 6 weeks

– New FPGA? Probably not for the test flight. Maybe for

SPIDER

– Cooling system  4-5 weeks

– New filter box and flex (rigid-flex, 1 connector less, and no

ground plane)  6-8 weeks

– Test  6-8 weeks



– SEU test  ?

– New boards  new year

Modulation schemes



– TES bias



– SQ1 bias



– SQ1 fb (switching)



– SQ1 fb (ramping)

TES bias

– The TES bias could be square (or sine) wave

modulated at ~1kHz. We have the firmware for s.w.

that can run as fast as half the framerate

– However, if the bias crosses zero for more than few

hundreds s, than the TES’ may not recover





Superconducting branch



Normal branch

Superconducting

transition









Acquired on the ACT CCam camera

TES bias

– In any TES bias modulation scheme we need a

firmware that has to run two separate servo loops.

Synchronization is needed. Major firmware change

for the read-out and the demodulation.

– It can be tested in software since we can already

acquire fast data at 2.2kHz (33 rows, 32GB/h), 26kHz

(1row). We’d need bolometers to do tests.









Acquired on the ACT MBAC

TES bias

– In any TES bias modulation scheme we need a

firmware that has to run two separate servo loops.

Synchronization is needed. Major firmware change

for the read-out and the demodulation.

– It can be tested in software since we can already

acquire fast data at 2.2kHz (33 rows, 32GB/h), 26kHz

(1row). We’d need bolometers to do tests.

SQ1 bias modulation

– We visit the SQ1 at up to 20kHz with RS on bias.

We could select 2 different on bias’ and switch at

sub-harmonics of 20kHz.

– The firmware demodulation (i.e. remove off from

on) would be straightforward since it would be a

modification of the coadding we already implement.









Acquired on the ACT CCam camera

SQ1 bias modulation

– We visit the SQ1 at up to 20kHz with RS on bias.

We could select 2 different on bias’ and switch at

sub-harmonics of 20kHz.

– We could invert the polarity of the SQ1_bias

(although now it is not bipolar) and keep the same

inverted polarity lock point.









Acquired on the ACT CCam camera

SQ1 fb modulation (switch)

– The SQ1_fb could be switched so that the PID

feedback loop works in opposite SQ1 slope

– A firmware should be created that enables opposite

locking. That could be done with I-term=I·(-1)#frames

except that it would take time

– However it should require a fine tuning and











Acquired on the ACT CCam camera

SQ1 fb modulation (switch)

– The SQ1_fb could be switched so that the PID

feedback loop works in opposite SQ1 slope

– A firmware should be created that enables opposite

locking. That could be done with I-term=I·(-1)#frames

except that it would take time to relock

– However it should require a fine tuning and

any asymmetry of the SQ1 V-phi doesn’t help









Acquired on the ACT MBAC

camera

SQ1 fb modulation (switch)

– The SQ1_fb could be switched so that the PID

feedback loop works in opposite SQ1 slope

– A firmware should be created that enables opposite

locking. That could be done with I-term=I·(-1)#frames

except that it would take time to relock

– However it should require a fine tuning and

any asymmetry of the SQ1 V-phi doesn’t help

a combination of

SQ1_bias and SQ1_fb

modulation so that there

are 4 different locking

points and/or we keep

the same lock point as

K.Irwing suggested.

SQ1 fb modulation (ramp)

– We throw away the PID feedback loop and we

continuously ramp the SQ1_fb. This creates an effective

modulation of the signal coming form the SQ1 V-phi

curves

– We already have a data mode that ramps the SQ1 fb. In

the 14bits DAC each step can be as high as 10kHz

– We look forward to test in software on a system with TES









Acquired on the ACT CCam camera



Related docs
Other docs by gegeshandong
A_E_KY-4PSE30WUSeries-Rev1012A
Views: 0  |  Downloads: 0
688_xls
Views: 0  |  Downloads: 0
2-1 辫措康
Views: 0  |  Downloads: 0
VINPR Lit Order Form New Jan 09
Views: 0  |  Downloads: 0
WRECKED - Torino Film Festival
Views: 0  |  Downloads: 0
project2btestcases
Views: 0  |  Downloads: 0
Fund Account transfer form9.2011
Views: 0  |  Downloads: 0
By registering with docstoc.com you agree to our
privacy policy

You are almost ready to download!

You are almost ready to download!