On 09/16/2020 07:56 PM, Minutolo, Lorenzo wrote:
Yes, that’s basically what we are doing. And we enjoy using the Ettus
products because of the versatility: we can do VNA scans to find the
resonances and then use a comb of tones to read them out. The basic
readout method which uses a single fixed sinewave per detector and is
good for very small signals as the resonance shift slightly from the
original position as the inductor gets excited. When the photon/phonon
flux increases, or you have detectors with higher responsivity
(frequency shift per aW), the change in the S21 on the original tone
frequency becomes very small and the SNR gets too low. One solution
I’m experimenting with is to use narrowband noise for the readout of
each detector (a paper should come out next season) but this technique
is not very scalable because of the stress induced on the cold low
noise amplifier by all the signals (the power level used for readout
is not arbitrary). The other technique is “tone tracking”: the
sinewave used to read a detector follows the shift of the resonance
itself: in this context the round loop latency determines the maximum
bandwidth at which you can use the detector.
Please let me know if you have in mind a point in the libraries where
I should start implementing this, it would make the USRP the ultimate
KIDs readout. Colleagues at SLAC, Fermilab, JPL, INFN and Caltech are
already using the GPU system I designed for lab testing but for field
deployment I have to step up this game.
Best,
Lorenzo
Ordinarily, customers of Ettus hardware who need a combination of
low-latency and high bandwidth will write code in the FPGA--that's
what it's for.
I understand that you have special requirements. Probably the place to
start is in "transport" uhd/host/lib/transport/*udp* but apart from
that I have nothing specific to add. Perhaps others on the list might.
How subtle is the phase shift? If you sent a continuous tone to a
sensor, the output would be phase-shifted by photon flux--would that drive
a PM demodulator adequately with enough SNR to be useful? Maybe you
already do that. I'm trying to understand where the low-latency
part comes from.
_______________________________________________
USRP-users mailing list
[email protected]
http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com
