KFPA LO Modifications (from single pixel results) Critical Design Review January 30, 2009 G. Anderson

KFPA LO Critical Design Review, January 30, Concerns Several issues were raised during single-pixel KFPA prototype testing with regard to the LO system: Why does the output power fall off rapidly at the band edges? Can we increase the frequency range of the 1st LO output? Should the output power specification be modified? Should additional spectral purity requirements be added to the specification?

KFPA LO Critical Design Review, January 30, Original Requirements Two local oscillator sources (HP/Agilent synthesizers that generate up to 20 GHz) are available in the GBT receiver room. The original specification called for the following to be supplied to each of 7 Downconverter modules: –LO1: 25.7 to 32.4 GHz At 25.7 GHz, -2 dBm (+/- 5 dBm) At 32.4 GHz, -4 dBm (+/-7 dBm) A gain slope of -0.3 dB/GHz at the mixer input would give nominal levels. –LO2: dBm (+/- 5 dBm)

KFPA LO Critical Design Review, January 30, No spectral purity specification was imposed. However, a self-imposed specification of 30 dB rejection of the third harmonic of the 1st LO input signal (as provided by the generator) seemed reasonable and was used as a design guide. The following are slides from the Conceptual Design Review giving the schematic of an LO Distribution Module. One module will supply LO power to all 7- pixels. (one output is not used and is terminated) Original Requirements

KFPA LO Critical Design Review, January 30, LO Distribution Module (High Frequency Side) 3dB Input: GHz +7 dBm (from LO Driver Module) Hittite HMC578 (Active Multiplier) Output +17 dBm for +4 dBm drive BPF X2 3dB Output (each): -1 dBm Loss = 1.5 dB MAX

KFPA LO Critical Design Review, January 30, LO Distribution Module (Low Frequency Side) 3dB 8.9 GHz -1.5 dBm (from LO Driver Module) Hittite HMC451 Gain = 22 dB Output IP3 = +32 dBm NF = 7 dB Each Output: +4 dBm dBm

KFPA LO Critical Design Review, January 30, Single Pixel Testing Single-pixel tests were performed using a test fixture consisting of the multiplier and bandpass filter shown in the schematic (LO Distribution Module (High Frequency Side)) above. Tests showed the following problems: The LO multiplier output dropped off sharply outside the original frequency range. During test, the LO drive desired for the input of the downconverter was significantly higher than called for by the LO1 specification. A spur was detected at the noise floor during a one minute integration. Its frequency indicated that one of the mixing inputs that created it was the LO1 input to the x2 multiplier. This was verified during test when an improvised high pass filter (waveguide) was added to the x2 output and the spur disappeared.

KFPA LO Critical Design Review, January 30, Corrective Action The multiplier output filter was originally designed around the initial specification. The original bandwidth was kept as narrow as possible to improve the 3rd harmonic rejection and to give some margin to allow for manufacturing tolerances.

KFPA LO Critical Design Review, January 30, New Filter

KFPA LO Critical Design Review, January 30, Specification Change –LO1: 24.8 to 34.3 GHz At 24.8* GHz, -5 to +3 dBm At 34.3* GHz, -3 to +4 dBm Specified differently at upper and lower band edges to allow a positive gain slope. This will largely counteract the negative gain slope from cables. –LO2: dBm (+/- 5 dBm)

KFPA LO Critical Design Review, January 30, The wider LO1 frequency range will allow operation over the sky frequency range of 18.0 to 27.5 GHz. The minimum power levels to be applied to the Down- Converter Module were also increased.

KFPA LO Critical Design Review, January 30, High Pass Filter The original plan was to add a high-pass filter to the multiplier output. –The spurious output that drove this addition was at the noise floor for a 1 minute integration. –Discussion with T. Minter and J. Lockman revealed that the maximum integration time one could ever expect was 1000 hours……worst case. –The reduction to the noise floor would be proportional to the square root of the ratio of integration time…..yielding a reduction in the noise floor of 23.9 dB.

KFPA LO Critical Design Review, January 30, Bandpass Filter Instead of adding a high pass….why not cascade another bandpass filter? –Still reduces the feedthrough of any x2 fundamental. –Adds additional rejection to any x3 multiplier output. Why not increase the number of poles of the original filter? –We already have greater than 50 dB rejection of the x2 input. –Approaches ultimate rejection of the filter where stray coupling dominates.

KFPA LO Critical Design Review, January 30, New Amplifier The addition of the high-pass filter and the change in the output power specification forced the addition of an amplifier to the multiplier output. The large attenuators placed at the input and output serve two functions: –to lower the amplifier input power –to mask the input and output return loss of the amplifier (output return loss is 6 dB).

KFPA LO Critical Design Review, January 30, Final Version LO Distribution Module (High Frequency Side) 3dB Input: GHz +7 dBm (from LO Driver Module) Hittite HMC578 (Active Multiplier) Output +17 dBm for +4 dBm drive BPF X2 3dB Output (each): 1 dBm* Loss = 2 dB MAX 3dB BPF 10 dB Hittite HMC283 Amplifier +21 dB gain 3 dB Loss = 2.0 dB MAX