2012/02/18 Duan
2012/03/28 Duan Test of microphonics with cable from Crystek.

Abstract

From the device to the readout electronics, there are 3 part of losses:

•    cable loss at inside cryostat, assume 3 dB loss
•    warm run of cables will be around 4.6 dB total, from the cryostat feedthrough to the power splitters, broken down as follows:

◦    60cm to cryostat bulkhead plate = 1 dB 1m from cryostat bulkhead plate to telescope bulkhead plate, inc bulkhead connector = 1.8 dB 
1m from cryostat bulkhead to splitter, inc bulkhead connector = 1.8 dB

•    Splitter loss 3dB
We did taken into account of item 2 before, In this log we discuss the effect of 0-6 dB cable loss (item 2) for the whole readout system.


Method we used

First of all, all the calculation is based on this excel file: MUSIC_receiver_chian_SNR_calculation_18feb2012.xls

In this excel file, the value we care about are:

•    the HEMT noise temp : B26
•    the cable loss (item 2) : D22
•    digital attenuator on IF board : L22
•    total power at ADC inpout (max input power is 6 dBm): R30 (we change digital attenuator to try to fully use the ADC)
•    the SNR degradation: S56 (how much SNR we loss from the electronics and cables etc)

Results

•    Here is a brief summary that assume there are additional 0-6 dB loss from cable connect between dewar and IF board: MUSIC_cableloss_effect.pdf
1.    In the case of 4.6 dB cable (from dewar output to the IF board) loss, we willl loss SNR around 3.69 dB compare to the SNR at device. (loss inside the dewar, and splitter etc have already taken into account). In the case of there is 0 dB cable loss, we will loss around 2.29 dB SNR. (This is due to the fact that 3dB splitter loss and 3 dB cable loss inside the dewar. In the original design phase, we did not take splitter into account, at that time, we can achieve less than 1.5 dB degradation in SNR.) 
Here attached a excel file to calculate the above numbers. 4.6 dB loss is at D21 and D22 (for the case of 1 or 192 resonators), and SNR degradation is at S55 and S56. 
The large loss (total 10.6 dB) between device and IF board is the reason for out high SNR degradation( this amount of loss will make us really hard to have SNR less than 2dB, because if we take a look from row 56, SNR already reduce 1.7 dB before it reach the electronic system ).

•    To help with the situation, we can either increase the amplification or reduce the attenuation at early stages on the IF board.

1.    For the case of reduce attenuation, we can change the fixed attenuator G21 and G22 from 9 dB to 3 dB (change 2nd fixed attenuator J21and J22 won't help much), and tune the variable attenuator L21 and L22 on IF board to 9.3 dB, this way we will reduce SNR degradation from 3.69 to 3.10(also attached a excel for this).

    MUSIC_receiver_chian_SNR_calculation_18feb2012.xls MUSIC_receiver_chian_SNR_calculation_18feb2012_modified.xls

2.    Another options we have is to use a external amplifier before our splitter:

3.    Due to the limitation of the input power of 3rd LNA on the IF board (L20) , we can not have amplification greater than 11dB, otherwise our L20 amplifier will work in its non-linear region. Then the target is to find a LNA that works from 2-4.5GHz, SMA connectors, with Gain less than 11dB and better Noise figure. I find one from Miteq which has 9dB gain and noise figure of 2. (still trying to find ones that has better NF) By putting this into the calculation, we get

    MUSIC_receiver_chian_SNR_calculation_18feb2012_additional_amp.xls

    for 2 K SIGe/HEMT, we loss SNR 2.58 for 5 K SiGe/HEMT, we loss SNR 1.25
    if we can find 9dB Gain, NF 1, we will get
    for 2 K SIGe/HEMT, we loss SNR 2.37 for 5 K SiGe/HEMT, we loss SNR 1.14

•    The best option we have is to use low loss cables: Sunil find someone that at 3 GHz, coax cable, the cable loss is 0.3dB/meter (0.8dB/m at 18 GHz). and they suggested their SMA will loss 0.05 dB each one. so for 1 meter two SMA cable, total loss is 0.4 dB. The dielectric is celloflon. (they do have other connector as well).

1.    We might be able to do better than 2.6 dB, but if we assume with these cables, the total loss is 2.6 dB, we will have: Cable loss reduce from 4.6 dB to 2.6 dB:

2.    for 2 K SIGe/HEMT, we loss SNR 2.97 for 5 K SiGe/HEMT, we loss SNR 1.47

•    (28 Mar, 2012) We ordered 1 meter cable, we have already tested the loss with network analyzer, the loss is 0.31 dB for 1 meter @3GHz (I would expect 2 meter long cable loss to be 0.5-0.6 dB).

1.    Then I connect the cable from RF_out and RF_in of the SDR electronics, and try to test microphonics etc by move the cable by hand for 60 seconds (usually wait 10 seconds and move the cable for 10 seconds and repeat).

    cable_stability.pdf (page1: Cable_noMove; page2: Cable_moveByhand; page3: zoom_of_page2; page4: zoom_of_page2)

Conclusions
A summary of all different situation