Development of infrared focal plane arrays for space - Astro Pas ...

TABLE II. Clock settings.Clocks Active V Inactive VFIG. 4. Schematic showing the output from one pixel vs time and thecorrelated triple sampling SRP mode. S, R, and P refer to the signal, reset,and pedestal levels. In this mode, four levels P1, R1, S2, and R2 aresampled and the combination (S2R2)(P1R1) is recorded, where P1and R1 refer to the pedestal and reset levels in the beginning frame and S2and R2 refer to the signal and reset levels in the end frame. However, since256 unit cells in each row of CRC-744 are reset together, it is only possibleto perform true correlated triple sampling on the first four unit cells in eachrow.unit cell itself see Fig. 3. The configuration was from a fourtransistor architecture as in the CRC-644 and CRC-463multiplexers to a three transistor architecture as in theSBRC Aladdin 10241024 readout, SBRC-084. With thisdesign, the unit cells are reset by row. This minimizes thetime necessary to reset the whole array about 25615 s,which is advantageous for ‘‘Fowler’’ or sampling up theramp SUR nondestructive multiple sampling. 3,8 However,it is not possible to perform correlated double or triple samplingsee Fig. 4 with this architecture in the same way aswith the CRC-463. A new program was written to clock theCRC-744 multiplexers muxes in Fowler sampling mode.However, as discussed in Sec. III, we used the CRC-463clock program to make noise measurements in correlatedtriple sampling mode. A further modification was the inclusionof a clamp circuit V ggcl and V ddcl in Fig. 3. This reducesthe clock feed-through to the detector integrating node.III. EVALUATION OF MUX LOT SPLITSFour CRC-744 muxes were tested, one from each of thelow temperature optimized splits Lawrence 2 m process,Lawrence 3 m process, standard cryo-CMOS process, andTRW 2 m process. The tested muxes are listed in Table I.There was a fifth split, optimized for operation at T30 K,which has not been tested. rst 5.5 3.0 syncS 7.0 1.0 1S 7.0 1.0 2S 7.0 1.0 syncF 4.0 1.0 1F 4.0 1.0 2F 4.0 1.0 ggcl 5.0 1.0The first activity was to determine the clock patterns,clock voltages, and bias voltages to employ for best performance.The initial voltages and currents utilized came fromthe engineering design see Tables II and Tables III. Therewere two areas of special concern. First, it was necessary todetermine whether to use the current source per frame P1, orthe current source per column P38 see Fig. 3, to supplycurrent to the unit cell amplifiers and second, it was necessaryto determine how large the current should be. The designspecifies each P38, which uses the current from V gg1 asthe reference current and provides 1 A; and P1, which usesthe current from V gg2 as the reference current and provides 1or possibly 10 A. In turn, the current from V gg1 should be4 A and the current from V gg2 should be 4 or possibly 40A. A 400 k resistor was used with V gg1 anda40kresistor was used with V gg2 . Initial operation of the first muxat room temperature indicated better uniformity from row torow using P1 rather than P38. However later, when operatingat T6 K, we observed a significant anomaly, namelythat the signal level in the correlated triple sampling modewas 0.3 V higher than the pedestal level, with no illuminationof the mux. These two levels should be approximatelyequal. It was found that turning V gg2 off 0 Vand V gg1 on(3 V) cured this problem, and also gave a much moreuniform Reset level. We concluded that the current sourceper column P38 gave superior performance. All the testsreported here used this current source. In addition, V dduc ,V ddout , and rst were also changed from 3.0, 1.0, and2.0/4.5 V to 3.5, 1.2, and 3.0/5.5 V, respectively,to improve dynamic range. Later in the FPA tests,when the gate protection Zener diode was replaced by a 1M resistor, V gg1 was adjusted to give a reasonable waveTABLE I. Test summary of CRC-744 bare muxes.Lawrence 2 mLawrence 3 mCryo-Std.CMOSTRW 2 mWafer No. 17 20 08 14R-P charge dump V0.28 0.32 0.32 0.30output referredTime constant s2.4 2.4 2.4 2.4rising, 5%–70%Time constant s0.7 0.6 0.7 0.6falling, 70%–5%dc gain 0.827 0.873 0.863 0.827Dynamic range of V rstuc V 3.85 to 2.75 3.75 to 2.75 3.95 to 2.95 3.9 to 2.9Current of V gg1 A ••• 4.2 4.0 •••Current V gg2 A 0 0 0 03568 Rev. Sci. Instrum., Vol. 68, No. 9, September 1997 IR focal plane arraysDownloaded 21 Jun 2012 to 128.151.144.191. Redistribution subject to AIP license or copyright; see http://rsi.aip.org/about/rights_and_permissions