COMMENTS
to
FIRST HET IF SYSTEM

Victor Belitsky
Chalmers University of Technology
S-412 96 Gothenburg, Sweden

Considerations About IF Frequency Choice

The main goal for HET instrument is to have state of art sensitivity of all mixers. It is obvious that the resulting system noise temperature depends on IF chain noise temperature, the mixer noise temperature and the RF conversion loss. With submm wave SIS mixer conversion loss optimistically 6 dB we will have IF system noise contribution to the system noise 4 times of IF noise temperature. Correspondingly we must optimize the IF noise temperature keeping in mind SIS or HEB mixer specific features, HEMT transistors potential to have the best noise temperature, cryogenic constraints.

In the provisional IF system design (please see Figure 1 for block-diagram) we have the IF frequency 102 GHz with total 4 GHz bandwidth.

The high center IF frequency (for SIS mixers typical IF center frequency is 1.5 … 4.5 GHz with 1.. 1.2 GHz band) have the follow advantages:

With high RF frequency it is easier to provide SSB operating (USB is separated from LSB by 2xIF center frequency). NB. HET band is from about 500 GHz up to 1.2 THz continuos coverage plus possible THz channel at about 1.8 THz.
With high IF center frequency the LO band can be reduced down to 2xIF center frequency times of LO source number.
By keeping IF center frequency high even for 4 GHz wide IF band we have DIF/IF_center=0.4 that simplifies the IF amplifier design (relatively narrow-band IF amplifier).
Still there is a hope that InP transistors will solve the problem of power dissipation so that such amplifier can be "integrated" with SIS mixer. InP HEMT transistors are demonstrated to be better for frequencies above 10 GHz.

Point 1 is listed here just for keeping general approach, while after long discussions within HIFI group during 1996 DSB operation was chosen for HET instrument. Point 2 is related with available LO power at higher frequencies. It is, perhaps, the most serious recent constrain implying use of diplexers for LO signal injection for high frequency end of the HET band. Furthermore, by using several switchable LO sources per one mixer band one can gain up to 2xIF center frequency per each source in LO frequency coverage.

However, besides the advantages there are also drawbacks of using the high IF center frequency.

The state-of-art noise temperature of microwave amplifiers at cryogenic temperatures can be estimated as 1 K/GHz according to [2] (I have a few references with noise temperatures confirming that at least below 12 GHz). That gives for 42 GHz IF frequency T_IF4 4 K while for 10-12 GHz T_IF10 10 K (see for instance CTH design for limb sounder IF amplifier) which is more than 3 dB difference.

Looking at provisional IF chain block diagram one can find the coaxial cable connecting the mixer and 1^st IF amplifier is a critical component. Indeed, it is much better to get rid of any long cables (with losses!!) in between a mixer and IF amp. However, due to cryogenic constrain we need to isolate thermally the SIS mixers from external heat flow. Total power dissipation at SIS mixer cold plate is about 5 mW only according PDD. Hence, we need to use pure stainless steel cable (or other low thermal conductivity material) and, perhaps, increase the length of the cable to keep thermal flow through it at the certain desirable level. The calculation I made for thermal flow shows that with 30 cm cable the thermal flow is as great as 0.7 mW (in the Table 1 it was used data from reference 1 see also Figure 1) for 10 IF amplifiers we have in the HET assuming that IF amps are at 9 K ambient temperature. Low heat conductivity of the cable material implies higher RF resistive loss (even in spite of the silver covered inner surface of the shield and rod of the cable, like UT 85 SS-SS). The cable will increase the resulting noise temperature of the IF amplifier. Again, the loss in the cable is frequency dependent, so typically the loss per meter of the cable for 10 GHz is 2.2 dB higher than for 4 GHz. You can see the resulting noise temperatures due to the cable loss in the Table 1.

Table 1

Table of comparison of the HET IF system components at 102 GHz and 42 GHz

Component Main characteristic Value at 10 GHz Value at 4 GHz

Cold IF amplifier Noise Temperature 10 K 4 K

Cryogenic IF cable
UT 85 SS-SS Loss per 1m 5.1 dB/m 2.9 dB/m

Cryogenic IF cable
UT 85 SS-SS Heat flow mW over 1 cm
with temperature differ.
4.6 mWcm 4.2-23 K
2.2 mWcm 2.5-9 K 4.6 mWcm 4.2-23 K
2.2 mWcm 2.5-9 K

IF noise temperature calculated based on above 14.2 K 4.9 K

HET IF SYSTEM (Provisional, suggested in PDD, thin Red Book)

Another issue that limits IF band of an SIS mixer is capacitance of SIS mixer. Figure 2-4 display the SIS junction topology and equivalent circuitry for the SIS mixer at RF and IF. The main conclusion from that is that the capacitance that is a sum of the SIS junction capacitance and the capacitance of RF integrated tuning circuitry. The latter is result of drastic difference of RF and IF, so that the RF tuning microstrip circuitry must be considered as lump capacitor at IF. Without special precautions the capacitance together with 50 impedance of the coaxial cable connecting IF amplifier creates classical filter structure with corresponding IF coupling loss (Figure 5).

Typical solution used to solve this problem like to reduce the cable impedance (by employing transformers) is not useful with SIS mixers. This will lead to immediate increasing of the mixer conversion loss L, which is in proportion to the ratio of ~R_RF/R_IF, where R_RF is the mixer RF impedance (constant for given mixer and RF frequency) and R_IF is the load at IF mixer port. The lower R_IF the higher the ratio and the more mixer loss.

The same comment can be applied for the series inductance compensation presented at Figures 6-9. Q-factor of the compensation circuitry has to be to kept at reasonable level (low to avoid critical dependence on the circuitry parameters but enough to get resonance). That with decreasing of the compensation inductance L at high frequency for given circuitry parameters lead to reducing the impedance connected to the inductance from IF amplifier (again by using transformer). Figure 9 shows results of modeling for such a compensation. Note that this circuitry becomes very sensitive to the varying of C_j, C_comp and L that complicate the mixer design. We still lose of about 5 dB all together due to coupling circuitry and decrease of R_IF (see paragraph above).

SIS junction: topology

Figure 2. An SIS junction is planar mm-scale size structure that is alike planar capacitor. The figure shows typical dimensions of cross-section of Nb-AlO_x-Nb SIS junction. Thickness of the top Nb electrode and SiO/SiO₂ dielectric layer can vary depending on an integrated tuning circuitry.

SIS junction: RF equivalent circuit

Figure 3. RF equivalent circuit for SIS junction mixer. Integrated tuning circuitry which is the only solution to provide RF coupling in some frequency band is schematically shown by inductance L. In practice the tuning circuitry is a set of planar transmission lines providing resonating out the SIS junction capacitance, Cj, and transforming of the SIS junction RF impedance to match the latter to Signal & LO guide system (waveguide mount or planar antenna).

SIS junction: IF equivalent circuit

Figure 4. IF equivalent circuit for SIS junction mixer. Integrated tuning circuitry that is the only solution to provide RF coupling in some RF frequency band is schematically shown now by capacitance C_comp due to F_RF>>F_FI.

Possible improvement of SIS mixer-IF coupling

Reducing the R_IF or the circuitry IF capacitance

Figure5. Solid line presents SIS junction IF coupling to 50 cable for Cj+Ccomp=0.7 pF; o-line shows IF coupling with impedance of 16.7 in stead of 50 . While square-line shows possible improvement in the case of reducing SIS mixer capacitance down to 0.35 pF.

Series inductive IF compensation

Figure 6. Principle of IF series inductance compensation. For the given above parameters it was modeled the circuit above for different values of the inductance L. The choice of the inductance was predefined by capacitance C_IF=C_j+C_comp and desirable frequency of IF (4 and 10 GHz center). The quarter-wave transformer was used to reduce the impedance of 50 to keep Q-factor at reasonably low level (to avoid critical dependence on the circuitry parameters) and to allow the circuit to work towards higher frequency while L is reducing.

Figure 7. Principle of IF series inductance compensation: the real part if impedance substituted to the SIS IF port.

Figure 8. Principle of IF series inductance compensation: the imaginary part if impedance substituted to the SIS IF port.

Figure 9. Principle of IF series inductance compensation: the SIS mixer -- IF coupling (including loss due to drop of the real part of impedance (see Figure 7)). The magenta dash-dot trace shows changes in the IF coupling for 20% change of SIS mixer capacitance, confirming that such a compensation is very sensitive to the parameter values and might be tricky to use.

Hot Electron Bolometer mixer (HEB), the new element for heterodyne mixer technology differs from SIS mixer by inherently limited the upper IF frequency. In the case of using really low-noise IF amplifier the IF upper frequency of HEB mixer can be, perhaps, extended up to 56 GHz with drop of IF signal towards the high frequency of the IF band , however, with the same system noise temperature. It is clear that using of HEB mixer in the HET will require another IF amplifier than provisional for SIS with 102 GHz IF band.

SUMMARY

Table 2

Table of comparison of the HET IF system components at 102 GHz and 42 GHz

Component Main characteristic Value at 10 GHz Value at 4 GHz

SIS junction mixer IF coupling^†
without compensation/with L-series compensation
-15.5 dB/-5 dB
-5 dB/-1.5dB

HEB mixer IF coupling maximum IF cutoff frequency 5..6 GHz

IF noise temperature calculated on based above IF chain (the cable output) 14.2 K 4.9 K

IF contribution into the system noise assuming 6 dB SIS mixer RF conversion loss only 56.8 K 19.6 K

IF contribution into the system noise assuming 6 dB SIS mixer RF conversion loss as well as IF coupling (with series L-compensation) 179 K 27.5 K

The SIS mixer IF chain with 102 GHz band is incompatible with HEB mixer technology. We have to consider that issue from technical point of view (we can have two different IF chains) as well from point of view of the project cost.

Possible reducing of the IF center frequency increases importance of HET LO system assessment. Perhaps using of 3 and even 4 LO source for each mixer band (3 LO for the 3^rd and 4 LO for 4^th mixer) can solve the problem. Increase in the quantity of LO sources will be compensated (at least partly) by lower cost of each source. We also can benefit from using passive (beamsplitter) LO injection scheme that will simplify the HET optical architecture. The lower frequency LO sources are going to be more reliable. Then, perhaps, we can reduce redundancy level for low band mixer(s) to compensate complexity of having extra LOs for high frequency end mixers.

Possible solutions

1 SIS mixer development program focused on IF coupling in 8-12 GHz IF band

integrated IF tuning (only for open structure mixers, no space in waveguide mount!).
decreasing the circuit capacitance (decreasing the SIS junction size requires E-beam lithography, submicron area junctions are less stable and much more sensitive to static discharge).

2 Changing the IF frequency to 2-6 GHz band with using balanced IF amplifier [2]. This also requires development work. However, this solutions brings use for both HEB and SIS mixers.

References

"The Heterodyne Instrument Concept for FIRST" by Eri Cohen, Neal Erikson, Margaret Frerking (editor), Paul Goldsmith, Andrew Harris, Charles Lawrence, MJ Mahoney, Tom Phillips, Jonas Zmuidzinas
Ilcho Angelov, Microwave Technology Dept., Chalmers University. Privet communication.
Micro-Coax Components, Inc. Catalogue.

NB:

The first time these notes were presented during Opto-Mechanical Workshop (devoted to the HET optical architecture) at SRON, Groningen March 13, 14 1997. I found some points in my at SRON presented paper that needed more clear comments and some correction.

Some pictures and numbering have been transferred into HTML format not correctly by MSWord Internet Assistant.

You can get this file in form of MSWord 6 document, contact me.

Victor Belitsky

Radio and Space Science Department

Chalmers University of Technology

S-412 96 Gothenburg SWEDEN

E-mail belitsky@ep.chalmers.se,

tel. +46 31 7721893

FAX +46 31 164513

Table of comparison of the HET IF system components at 102 GHz and 42 GHz
Component	Main characteristic	Value at 10 GHz	Value at 4 GHz
Cold IF amplifier	Noise Temperature	10 K	4 K
Cryogenic IF cable UT 85 SS-SS	Loss per 1m	5.1 dB/m	2.9 dB/m
Cryogenic IF cable UT 85 SS-SS	Heat flow mW over 1 cm with temperature differ.	4.6 mWcm 4.2-23 K 2.2 mWcm 2.5-9 K	4.6 mWcm 4.2-23 K 2.2 mWcm 2.5-9 K
IF noise temperature calculated based on above		14.2 K	4.9 K

COMMENTS to FIRST HET IF SYSTEM

Victor Belitsky Chalmers University of Technology S-412 96 Gothenburg, Sweden