FEBiasOptimizationNotes

Notes, discussion, and planning for improving FEIC cartridge bias optimization routines.

Next steps:

  • Band 3 bias optimization is fairly automated except for manual updating of INI file and manual stepping of the LO. Automate these steps.
  • Band 7 defluxing should always be followed by a full SIS heater cycle (30 seconds of pulsing the heater followed by ~1 minute for mixers to cool back down.)
  • Bands 7, 8, 9 (6?) should routinely have a full demag/deflux cycle whenever powered-on. Prompt the user "perform standard mixer demagnetization / deflux now?"
Band 7 optimization procedure:
  1. Demag/deflux and then find 2nd minimum on Bessel function of Isis vs. Imagnet. Repeat a few times to get consistent results.
    1. Pavel asserts that there is still a software or other problem causing non-repeatability here.
    2. Once optimum is found, saving the configuration should be automatic, not a user operation.
  2. For now, we use the Vsis vs. freqLO curves provided with the cartridge, rather than attempting to reoptimize these ourselves.
    1. This approach produced very good noise temperature results for band 7 SN 16 in FE SN 02.
    2. Pavel is following up on what to do to optimize this parameter going forward.
    3. The supplied bias curves use a fixed Vsis for most LO frequencies, and then stay 0.2 mV away from the Josephson resonance at upper frequencies. In some cases it is necessary for the Vsis to be set below the Josephson resonance at the top LO frequencies.
  3. With the magnet set, the LO tuned and locked, and the LNA and SIS biased as specified in the cartridge-provided tables:
    1. Sweep each polarization LO PA drain voltage first viewing the cold load and then the ambient load. Select the operating point with the best noise temperature for both sidebands (max sum of SB1 and SB2 Y-factor.)
    2. Record the LO PA drain voltage settings and the SIS current readings.
    3. Repeat for all LO frequencies, in 4 GHz steps.
Band 7 process improvements:
  • If the maximum Y-factor vs. LO power occurrs at the same setting as the maximum IF output power, then the automated search code in use for band 3 could be repurposed for this phase (3.1 above) of the band 7 optimization process.
  • In the original procedure, the SIS voltage is set to a fixed value and the LO PA drain voltage is servoed in order to get the Isis to about 28 uA. This could be done using the existing "Adjust PA" routine.
  • In the original procedure, the SIS voltage is then swept, looking for the best sum of Y-factor within a range defined by the photon step and avoiding the Josephson resonance. If the optimum here corresponds to the maximum IF output power within the allowed range (as it does for band 9) then some of the automatd search code in use for band 3 could be repurposed for this phase of the band 7 procedure.
Band 7 measurements:
  • 2-D magnitude plot of IF power vs. LO frequency and SIS voltage.
  • 2-D magnitude plot of Y-factor vs. LO frequency and SIS voltage.
  • Plots of IF Power or Y-factor vs. LO power and SIS voltage?
  • Does best Y-factor vs. LO Power always occur at where the IF total power is maximum?
Band 9 optimization procedure:
  1. Demag/deflux then find 2nd minimum of "Isupercurrent" bessel function of Isis vs. Imagnet. Repeat a few times to get consistent results.
    1. Once optimum is found, saving the configuration should be automatic, not a user operation.
  2. With the magnet set, the LO tuned and locked, Vsis to 2.2 mV, the LNAs set according to the cartridge provided tables:
    1. Find the LO PA drain voltage giving the best noise performance (max Y-factor)
    2. Sweep Vsis to find the maximum IF output power, keeping away from the Josephson resonance region.
    3. Record the LO PA drain voltage settings, SIS voltgage settings, and the SIS current readings.
    4. Repeat for all LO frequencies, in 4 GHz steps.
Band 9 process improvements:
  • If the maximum Y-factor vs. LO power occurrs at the same setting as the maximum IF output power, then the automated search code in use for band 3 could be repurposed for this phase (2.1 above) of the band 9 optimization process.
  • According to Pavel, the best magnet current setting should be optimized even further than simply finding the 2nd minimum of the Isupercurrent Bessel function:
    • Use the 2nd minimum as a starting point and plan to sweep a small range from below to above that setting.
    • At each magnet current setting, measure IF power vs. Vsis with the LO power fixed at the optimum value found in 2.1 above.
    • Select the magnet current setting where the size of the Josephson resonance effect is smallest.
    • We should be looking at the Josephson peaks in Power vs. SIS voltage, and make sure that max-min of the two peaks is within limit (need to establish a threshold - which can be function of the cartridge, at least band 9 (different SIS mixers technologies). If not - perform de-flux;
Band 9 measurements:
  • 2-D magnitude plot of IF power vs SIS voltage and fine-stepping magnet current.
  • Does best Y-factor vs. LO Power always occur at where the IF total power is maximum?

Notes from original telecon...

Software support to Henry:

  • Needs to be using the FEControl2 driver level software, not the Front End Cartridge model
  • Wants "direct access to parameters"
  • Discussed this with him on the phone Tuesday 12/7. He is aware of that. Also reinforced that new interfaces and routines can be provided.
  • Check SW version Henry is using.

Magnet current optimization, mainly band 7:

  • Enough time between setting current and measuring SIS current?
  • There are no delays in current software, apart from CAN and log file writing delays.
  • Try adding delay
  • Try timestamping data
  • Try removing file write from the inner loop
  • In operations will want to sweep a narrow range to find the minimum, not the full control range.
  • Email from Pavel 12/9: "1. Whereas it might be essential to de-flux between the measurements indeed, also in case of Band 7 (Band 9 is confirmed), I'm still not convinced that just de-fluxing is the only cause for the non-repeatability between the measurements with different ranges. In case the flux is trapped - it's there, and the position of the minimum shall not depend on the starting current of that test, and obviously not on the end setting. We have done many tests, and in random order, and the minimum position *is* a function of the scanning range."

SIS voltage optimization:

  • Confirmed that the SIS voltage calibration was not in force in Pavel & Henry's original experiment. Email from Pavel 12/9: "2. Henry has preliminary confirmed that the voltage correction bit was set to "0" in our test - he will double check and re-do the measurements."
  • Try scan from -V to +V to check for gain errors. Josephson resonance should be offset from nominal by same amount, not symmetrical
  • Band 7 strategy is to use their Vbias vs. LO curves and re-optimize LO PA drive for max Y-factor.
  • Spot-check IF power curves that we are not operating on the steep slope edge of the Josephson regoin at critical LO freqs.
  • Need to increas margin away from Josephson resonance region? Pesently it is 0.2 mV.
  • No evidence of this in FE 02 / B7 16. Got good results with cart mfrs. Vbias vs. LO curves.

Band 9 magnet optimization:

  • Ideal strategy according to Pavel would be to optimize Imag not necessarily to 2nd minumum but to minimize magnitude of Josephson resonance peaks.
  • This would imply a two-dimensional search around the initially found 2nd minimum: IF power vs. Isis x Imag
  • Should be fairly easy to add code take this data to the existing software framework. Plotting it a little harder. Searching it for optimum perhaps harder still.
Topic attachments
I Attachment Action Size Date Who Comment
FEBiasOptimizationNotes__ALMA__NRAO-Public.txttxt FEBiasOptimizationNotes__ALMA__NRAO-Public.txt manage 2 K 2010-12-15 - 14:32 MorganMcLeod Pavel's email message of 12/5. I have fixed the errors and made the clarifications he points out on this page.
band_7_optimization-followup.pdfpdf band_7_optimization-followup.pdf manage 483 K 2010-12-09 - 11:55 MorganMcLeod Repeated Pavel's experiments on FE-02 / Band 7-16
band_7_optimization.docdoc band_7_optimization.doc manage 417 K 2010-12-09 - 11:52 MorganMcLeod Pavel's original 12/2 measurements for band 7 magnet current and bias voltage
Topic revision: r4 - 2010-12-15, MorganMcLeod
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback