This note comments on the noise performance of the IF system proposed
by Rick Raffanti, as documented in DesignLog103. We find that the
system is subject to amplifier compression and that some design changes
are required to both stay away from amplifier compression while
obtaining HEMT- and ADC-limited noise performance
We want the readout system noise to be limited by the HEMT. The second
largest contributor should be the ADC given that we are pushing the
limits of currently available ADCs. The IF system should be designed to
always be a negligible contributor. This is no problem if one is not
worried about amplifier compression. But we are worried about amplifier
compression because it creates harmonics that can result in an
exponential increase in intermodulation products. Avoiding amplifier
compression requires keeping input and output power levels low at any
given amplifier, which then increases the contribution of amplifier
noise. So one must be careful to simultaneously avoid amplifier
compression and avoid noise degradation by the IF system.
After study the IF board design again to optimize the SNR in front of
ADC, we also find some problem I did not figure out before and some way
to further improve the noise performance of the whole receiver chain.
See DesignLog103 for the IF system schematics with component
descriptors and part numbers.
The problem is:
• In the current design, IF Amp2 (HMC313) after var.
attenuator contribute major amount of noise to the final noise
• I thought -9dBm input power for the current
baseband amplifier BB Amp 1 (AD8353) is OK, but by looking at the data
sheet again, the power compression has a turning point around -10 dBm.
We have to avoid that for our application.
I make another design which try to do the following to solve the problems and improve noise temperature:
• I move the IF Amp 2 before var. attenuator and also
change the part for IF amp 2 from HMC313 to HMC311 to achieve better
noise figure; after doing this, I have to add additional attenuation
before IF amp 2 to make sure the power reaching it doesn’t too high: I
change fix att 1 from GAT-3 to GAT-9 , and also add an extra 6 dB fix
attenuator (GAT-6) between SW1 and new position of IF amp2.
• To solve baseband amplifier problem, I search for
another part AD5530, it has higher max input power level than the one
we are using right now; but AD5530 has less gain. So I delete the fix
attenuator between Demod mixer and BB amp. and change BB amp 1 from
AD8353 to AD5530.
• One more thing I changed is reduce the fix attenuator between 2 LPF from 3 dB to 1dB.
Overall, what have been changed are:
• Change IF amp 2 from HMC313 to HMC 311;
• move the IF amp 2 position from after var. attenuator to before var. attenuator.
• Change fixed atten 1 from 3 dB to 9 dB
• Add an extra 6 dB attenuator between SW1 and IF amp 2 ‘s new position
• Delete the attenuator between demod mixer and BB amp 1
• Change BB amp 1 from AD8353 to AD 5530
• Change the fix atten 3 (between 2 LPFs) from 3 dB to 1dB
the excel spreadsheet for the calculation of both current and suggested design here:
summarized version of current design and suggested new design
Since there is component change here. Once the change is confirmed, I
will update the new components purchase order for this updated list !