1. Single Pixel Mixer Tests July 2008 (version 2, posted 11 Aug 2008)
Measurement goals:
Rn of device G-22(+)
RF tests of G-22(+)
MMIC LNA oscillation measurements
Along the way:
Taming the VDI LO (new to this setup)
Numerous power failures, cold head failures, etc...
LO = 336 GHz I-V and T-P sweep
Trec~ 70K DSB at 2.05 mV

2. Rn of device G-22(+)
A previous device, I-22(0), showed a high Rn of 31 ohms (24 ohms targeted). Does this mean that the RnA product for the wafer is off (i.e. low), or is this device simply very undercut?
New device G-22(+) shows a low Rn of ~18 ohms, so RnA is probably fine; I-22(0) was undercut. [Does that sound right, Jacob?]
A sample no-LO sweep of device G-22(+)
shows a low Rn of 17.9 ohms.

3. RF performance across the band
waveguide atten.
VDI x2x2x3 multipliers
Performance in the GHz range is consistent with expectations, with DSB noise temperatures in the 70-80K range with no real effort.
Beyond this range, initial RF performance was simply awful (>500K DSB). Amplitude noise from the synthesizer-driven LO chain (new to this setup) is the expected culprit based on follow-up data.
Instead of attenuating the RF going into the x2x2x3 multipliers to get the right mixer current, we decided to run the LO at full bore and physically attenuate the LO output to the RX. This improved the amplitude stability.
Spacek
doubler
and RF
amp.
D60 bias box
'New' VDI GHz LO chain poses challenges at band edges.

4. Example: 319 GHz RF performance as a function of LO configuration
All waveguide attenuation on 30 GHz input to LO multipliers, poor stability, Trec ~ 270K
(Vmon on WR8x2 doubler ~ 1 V).
Less waveguide attenuation, more attenuation on LO output, better stability, Trec ~ 120K
(Vmon on WR8x2 doubler ~ 2 V).
Least waveguide attenuation, most attenuation on LO output, Trec ~ 110K
(Vmon on WR8x2 doubler ~ 4 V).

5. At the other end of the band: behavior at 370 GHz
Although driving the LO hard (and attenuating the output radiation) improves stability at most frequencies, at the top end of the frequency range, this results in irregular pumping of the junction. (?!) The total IF power is 1/10th of that detected at 360 GHz.
Reducing the RF drive to the multipliers regains the normally-moded IV curve, but the RF performance suffers. The total IF power is 4 times that measured at 360 GHz.

6. 336 GHz
Summary Chart of RF performance
358 GHz
PSAT current x100 (A)
IF power x100 (uW)
Trec(K)
365 GHz
LO freq (GHz)

7. RF input to VDI LO
There is variation in how hard the 1st LO amplifier stage is driven – see PSAT currents in previous plot. We should be at a constant 3.6 A for best stability control. Are we mis- feeding the LO?
Looks mostly fine, actually...
Total power measurement of -2 dBm signal from HP freq synthesizer after length of RG-174 coax, 3 dB pad and ~23 dB Miteq amplifier. Operating at the nominal maximum input to VDI RF amp is +20 dBm made little difference, except at ~365 GHz, where Trec improved ~15K over the previous plots.
Total power measurement of -6 dBm signal from HP freq synthesizer after length of RG-174 coax and 3 dB pad.

8. MMIC LNA & SIS interaction
Vd = 0.6V
Single pixel cryostat sees a similar oscillation problem as does the 1x8 mixer block of Supercam. Appears to not be a DC grounding problem.
Onset of oscillation and its effect on the SIS junction's IV-curve depends on the MMIC output termination.
open output oscillates at < 4mW
50 ohm load oscillates at 9 mW
connected to room temp amps oscillates at 6 mW, 8 mW with a 3 dB attenuator. (So there are standing waves on output...)
Small evidence for sensitivity to the input match; oscillation first seen at 2.8 mV (SIS bias) before propogating elsewhere.
No sensitivity to B-field.
Vd = 0.75V
Vd = 0.9V

9. Vd=0.65V
Vd=0.70V
Vd=0.75V
Spectrum Analysis of MMIC output Measured up to 22 GHz Strong oscillation readily measured in the 10 GHz band. The contribution of additional sources above 22 GHz is unlikely but not directly measured. No oscillation is seen when the mixer is warmed up.
GHz
Vd=0.83V
GHz