Lecture 9: Packaging & I/O
Part III: Systems Integration — SCE Futures
Table of Contents¶
Part 1: Packaging¶
- Chip Carriers & Mounting
- Thermal Contact Methods
- Wire Bonding at Cryogenic Temperatures
- Flip-Chip & MCM Integration
Part 2: Electrical I/O¶
Part 3: Advanced I/O¶
Learning Objectives¶
By the end of this lecture, you will be able to:
- Select appropriate chip carriers and mounting methods for cryogenic applications
- Understand wire bonding techniques and materials for cryo environments
- Design flip-chip and MCM assemblies for superconducting circuits
- Choose appropriate wiring materials balancing thermal load vs signal quality
- Implement proper thermal anchoring at each temperature stage
- Design filtering and shielding strategies for noise-sensitive measurements
- Understand optical I/O options for thermal isolation
# Setup
import matplotlib.pyplot as plt
import matplotlib.patches as patches
import numpy as np
COLORS = {
'primary': '#2196F3',
'secondary': '#FF9800',
'success': '#4CAF50',
'danger': '#f44336',
'dark': '#1a1a2e',
'light': '#f5f5f5',
'purple': '#9C27B0',
'cyan': '#00BCD4',
'cold': '#00BCD4',
'warm': '#FF5722',
}
plt.rcParams['figure.facecolor'] = 'white'
plt.rcParams['axes.facecolor'] = 'white'
plt.rcParams['font.size'] = 11
print("Setup complete.")
Setup complete.
1. Chip Carriers & Mounting¶
Chip Carrier Options¶
| Type | Material | Pin Count | Thermal | Best For |
|---|---|---|---|---|
| Ceramic DIP | Al₂O₃ | 24-68 | Moderate | Prototyping |
| Ceramic LCC | Al₂O₃ | 20-84 | Moderate | Standard testing |
| CPGA | Al₂O₃ | 84-400+ | Good | High pin count |
| Custom PCB | Rogers/Duroid | Any | Varies | RF applications |
| Cu block | OFHC Cu | Custom | Excellent | High thermal load |
| Direct mount | N/A | N/A | Best | Maximum cooling |
Material Considerations¶
Ceramics (Al₂O₃, AlN):
- Good thermal conductivity at cryo (AlN better than Al₂O₃)
- CTE mismatch with Si (~7 vs 2.6 ppm/K)
- Electrical insulation
Metals (Cu, Au-plated Cu):
- Excellent thermal conductivity
- Requires insulating layer for electrical isolation
- Larger CTE mismatch with Si
PCB (Rogers, Duroid):
- Good for RF routing
- Lower thermal conductivity
- CTE can be tailored
Differential Thermal Contraction¶
Everything shrinks on cooldown, but not equally!
| Material | ΔL/L (300K→4K) |
|---|---|
| Silicon | 0.022% |
| Copper | 0.33% |
| Aluminum | 0.41% |
| Al₂O₃ | 0.07% |
| G10/FR4 | 0.2-0.4% |
A 10mm chip on a Cu carrier: Cu shrinks 33μm more than Si → stress!
2. Thermal Contact Methods¶
Getting heat out of the chip requires excellent thermal contact at every interface.
Thermal Interface Materials¶
| Material | R_th (typical) | Permanent? | Notes |
|---|---|---|---|
| GE/IMI 7031 varnish | 1-10 K/W | Semi | Standard, thin layer |
| Apiezon N grease | 2-20 K/W | No | Crystallizes at cryo |
| Cry-Con grease | 1-10 K/W | No | Stays soft at cryo |
| Indium foil | 0.5-2 K/W | Semi | Conforms to surfaces |
| Solder | <0.5 K/W | Yes | Best, but permanent |
| Epoxy (Stycast) | 5-50 K/W | Yes | Structural + thermal |
Application Tips¶
GE Varnish:
- Apply thin layer, let dry partially
- Press parts together firmly
- Can be removed with solvent
Indium Foil:
- Use 0.1-0.5mm thickness
- Soft metal conforms to imperfections
- Can cold-weld under pressure
Thermal Grease:
- Use sparingly (excess increases R_th)
- Apiezon N: cheap, but crystallizes
- Cry-Con: expensive, better cryo performance
Mounting Hardware¶
- Screws: Brass or stainless (not magnetic!)
- Spring washers: Maintain pressure through contraction
- Clamps: For samples that need to be changed frequently
3. Wire Bonding at Cryogenic Temperatures¶
Wire bonding connects chip pads to package/carrier pads.
Wire Materials¶
| Material | Diameter | Resistance | Cryo Performance | Notes |
|---|---|---|---|---|
| Al (1% Si) | 25 μm | Moderate | Good | Standard, reliable |
| Al (pure) | 25-50 μm | Lower | Excellent | Softer, harder to bond |
| Au | 18-25 μm | Low | Good | Expensive |
Bonding Techniques¶
Wedge Bonding (Ultrasonic):
- Standard for Al wire
- Works with Nb pads (with proper metallization)
- First bond: pressed and ultrasonically welded
- Second bond: similar, then wire broken
Ball Bonding (Thermosonic):
- Standard for Au wire
- Requires heated stage (~150°C)
- Ball formed by spark, pressed to pad
Pad Metallization for Bonding¶
Nb pads alone are difficult to bond to. Common solutions:
| Stack | Bondability | Notes |
|---|---|---|
| Nb only | Poor | Oxide layer prevents bonding |
| Nb/Au | Good | Au cap, but watch purple plague |
| Nb/Al | Good | Al cap, matches Al wire |
| Nb/Ti/Au | Better | Ti adhesion layer |
Cryo-Specific Considerations¶
Thermal cycling stress: Wires flex with each cooldown
- Keep loop height reasonable (not too flat)
- Avoid sharp bends
- Consider redundant bonds for critical signals
Bond strength: Actually improves at cryo (materials harden)
Testing: Pull-test samples before committing to cooldown
# Visualize: Wire bond geometry
fig, axes = plt.subplots(1, 2, figsize=(12, 5))
# Left: Good wire bond loop
ax = axes[0]
# Chip
chip = patches.Rectangle((0.5, 0.5), 2, 0.3, facecolor=COLORS['primary'], edgecolor='black', linewidth=2)
ax.add_patch(chip)
ax.text(1.5, 0.65, 'Chip', ha='center', va='center', fontsize=10, color='white')
# Package
pkg = patches.Rectangle((3, 0.3), 2, 0.5, facecolor=COLORS['light'], edgecolor='black', linewidth=2)
ax.add_patch(pkg)
ax.text(4, 0.55, 'Package', ha='center', va='center', fontsize=10)
# Bond pads
ax.add_patch(patches.Rectangle((2.2, 0.75), 0.25, 0.1, facecolor='gold', edgecolor='black'))
ax.add_patch(patches.Rectangle((3.1, 0.65), 0.25, 0.15, facecolor='gold', edgecolor='black'))
# Good wire bond - gentle loop
t = np.linspace(0, 1, 50)
x = 2.35 + t * 0.85
y = 0.85 + 0.4 * np.sin(np.pi * t) # Gentle arc
ax.plot(x, y, 'b-', linewidth=2)
ax.plot([2.35], [0.85], 'ko', markersize=6) # First bond
ax.plot([3.2], [0.8], 'ko', markersize=6) # Second bond
# Annotations
ax.annotate('Loop height\n~100-200μm', xy=(2.7, 1.2), xytext=(2.7, 1.6),
fontsize=9, ha='center', arrowprops=dict(arrowstyle='->', color='black'))
ax.annotate('Bond 1', xy=(2.35, 0.85), xytext=(1.8, 1.0), fontsize=9, ha='center')
ax.annotate('Bond 2', xy=(3.2, 0.8), xytext=(3.7, 1.0), fontsize=9, ha='center')
ax.set_xlim(0, 5.5)
ax.set_ylim(0, 2)
ax.set_aspect('equal')
ax.axis('off')
ax.set_title('Good: Gentle Loop', fontsize=12, fontweight='bold', color=COLORS['success'])
# Right: Bad wire bond examples
ax = axes[1]
# Chip and package
chip = patches.Rectangle((0.5, 0.5), 2, 0.3, facecolor=COLORS['primary'], edgecolor='black', linewidth=2)
ax.add_patch(chip)
pkg = patches.Rectangle((3, 0.3), 2, 0.5, facecolor=COLORS['light'], edgecolor='black', linewidth=2)
ax.add_patch(pkg)
ax.add_patch(patches.Rectangle((2.2, 0.75), 0.25, 0.1, facecolor='gold', edgecolor='black'))
ax.add_patch(patches.Rectangle((3.1, 0.65), 0.25, 0.15, facecolor='gold', edgecolor='black'))
# Bad: Too flat
x_flat = np.linspace(2.35, 3.2, 50)
y_flat = 0.85 + 0.05 * np.sin(np.pi * (x_flat - 2.35) / 0.85)
ax.plot(x_flat, y_flat, 'r-', linewidth=2, label='Too flat (stress)')
# Bad: Sharp bend
ax.plot([2.35, 2.35, 3.2, 3.2], [0.85, 1.4, 1.4, 0.8], 'r--', linewidth=2, label='Sharp bends (fatigue)')
ax.set_xlim(0, 5.5)
ax.set_ylim(0, 2)
ax.set_aspect('equal')
ax.axis('off')
ax.set_title('Bad: Stress Concentrations', fontsize=12, fontweight='bold', color=COLORS['danger'])
ax.legend(loc='upper right', fontsize=9)
plt.tight_layout()
plt.show()
print("Wire bond failures at cryo are usually from:")
print(" 1. Thermal cycling stress (too flat = wire can't flex)")
print(" 2. Sharp bends (fatigue crack initiation)")
print(" 3. Poor initial bond (didn't stick in first place)")
Wire bond failures at cryo are usually from: 1. Thermal cycling stress (too flat = wire can't flex) 2. Sharp bends (fatigue crack initiation) 3. Poor initial bond (didn't stick in first place)
4. Flip-Chip & MCM Integration¶
Flip-Chip Bonding¶
Chip mounted face-down with solder/metal bumps connecting directly to substrate.
Advantages over wire bonding:
- Shorter interconnects → lower inductance
- Higher I/O density (area array vs perimeter)
- Better thermal path through bumps
- More robust mechanically
Bump Materials for Cryo:
| Material | Bump Height | Cryo Performance | Notes |
|---|---|---|---|
| Indium | 5-20 μm | Excellent | Soft, forgiving, SC below 3.4K |
| SnPb (legacy) | 50-100 μm | Good | Traditional, Pb restrictions |
| Au stud | 20-50 μm | Good | For fine pitch, thermocompression |
| Cu pillar | 30-100 μm | Moderate | High current capacity |
Indium bumps are preferred for SCE because:
- Indium is superconducting (Tc = 3.4K)
- Soft metal accommodates CTE mismatch
- Low bonding temperature (room temp possible)
Multi-Chip Modules (MCM)¶
Multiple chips on a common substrate:
flowchart TB
subgraph mcm[MCM Substrate — Si or Ceramic]
direction TB
subgraph chips[" "]
direction LR
C1[SCE
Logic]
C2[SCE
Memory]
C3[SCE
ADC]
end
IC[═══ Superconducting Interconnects ═══]
C1 --- IC
C2 --- IC
C3 --- IC
end
style C1 fill:#00BCD4,stroke:#00838F
style C2 fill:#00BCD4,stroke:#00838F
style C3 fill:#00BCD4,stroke:#00838F
style IC fill:#FFD54F,stroke:#F57C00
MCM Benefits:
- Mix technologies (Nb logic + NbN detectors, etc.)
- Known-good-die assembly (better yield than monolithic)
- Shorter die-to-die interconnects than package-to-package
MCM Challenges:
- Substrate must support superconducting traces
- Alignment tolerance for fine-pitch bumps (<5μm)
- Testing/rework after integration
Note on cryo-CMOS: While sometimes proposed for hybrid integration, cryo-CMOS typically dissipates 100s of mW — often negating the efficiency gains of SCE. Generally not recommended unless absolutely necessary.
5. DC & Low-Frequency Wiring¶
For bias currents, control signals, and slow monitoring.
Wire Material Selection¶
| Material | Thermal Load | Resistance | Best For |
|---|---|---|---|
| Manganin | Very low | 43 μΩ·cm | DC bias, low current |
| Constantan | Very low | 49 μΩ·cm | Thermocouples |
| Phosphor bronze | Low | 10 μΩ·cm | General purpose |
| NbTi | Near zero (SC) | 0 (below Tc) | High-current, best thermal |
| Copper | High | 1.7 μΩ·cm | Short runs at 4K only |
Typical Configurations¶
DC bias (e.g., SFQ bias):
- Manganin or phosphor bronze twisted pairs
- 32-36 AWG typical
- Filter at room temp and possibly at 4K
Voltage sensing:
- 4-wire (Kelvin) sensing to eliminate lead resistance
- Use same material for all wires in a group
Temperature sensors:
- Constantan or phosphor bronze
- Shielded twisted pairs for low noise
Thermal Anchoring¶
Critical: Thermalize wires at each temperature stage!
flowchart TB
RT[Room Temp
Connector] --> A40
subgraph A40[40K Anchor]
W40[Wrap 5-10 turns
varnish or grease]
end
A40 --> A4
subgraph A4[4K Anchor]
W4[Same treatment]
end
A4 --> DEV[Device]
style RT fill:#FF5722,color:#fff
style A40 fill:#FFECB3
style A4 fill:#00BCD4,color:#fff
style DEV fill:#4CAF50,color:#fff
Without proper anchoring, heat conducts straight to 4K!
6. RF & High-Speed Signaling¶
SCE circuits can operate at 10+ GHz, requiring careful transmission line design.
Coaxial Cable Options¶
| Type | Material | Attn (dB/m @ 10GHz) | Thermal Load | Use |
|---|---|---|---|---|
| SS semi-rigid | Stainless | 3-5 | Low | 300K↔40K |
| CuNi semi-rigid | CuNi | 2-3 | Moderate | 300K↔40K |
| Cu semi-rigid | Copper | 0.5-1 | High | 4K only |
| NbTi SC coax | Superconducting | ~0 | Very low | Ultimate |
| Flexible (SS) | Stainless | 4-8 | Low | Where flex needed |
Thermal Break Strategy¶
Use attenuating (lossy) cables to reduce heat load:
300K ──[SS semi-rigid, 30cm]── 40K ──[SS semi-rigid, 15cm]── 4K ──[Cu, 5cm]── Chip
The SS sections provide thermal isolation; Cu section at 4K preserves signal.
Connector Options¶
| Connector | Freq Range | Size | Notes |
|---|---|---|---|
| SMA | DC-18 GHz | Large | Standard, robust |
| SMP/SMPM | DC-40 GHz | Small | Push-on, space-efficient |
| 2.92mm (K) | DC-40 GHz | Medium | High performance |
| 1.85mm (V) | DC-67 GHz | Small | Very high frequency |
Cryo considerations:
- Connectors may need re-torque after thermal cycling
- SMP push-on connectors are convenient for sample changes
- Hermetic feedthroughs required at vacuum boundary
7. Filtering & Amplification¶
Why Filter?¶
Noise from room temperature electronics can swamp sensitive cryo measurements:
- Thermal noise: ~4 kT bandwidth
- EMI/RFI pickup
- Digital switching noise
Filter Types¶
| Type | Cutoff | Attenuation | Location | Notes |
|---|---|---|---|---|
| RC (lumped) | 1-100 MHz | 20-40 dB | Any | Simple, cheap |
| LC (lumped) | Tunable | 40-60 dB | Any | Sharper rolloff |
| Powder (metal powder) | 1-10 GHz | 60-100 dB | Cryo | Excellent HF |
| Eccosorb | >1 GHz | 40-60 dB | Cryo | Absorptive |
| Copper powder | 100 MHz-10 GHz | 80-120 dB | Cryo | Best attenuation |
Filter Placement¶
flowchart TB
RT[Room Temp Electronics] --> F1
subgraph F1[RT Filter]
RF1[Blocks EMI from
entering cryostat]
end
F1 --> VF[─── Vacuum Feedthrough ───]
VF --> A40[40K Anchor]
A40 --> F2
subgraph F2[4K Filter]
RF2[Blocks thermal
radiation in wires]
end
F2 --> DEV[Device]
style RT fill:#FF5722,color:#fff
style F1 fill:#FFF9C4
style F2 fill:#BBDEFB
style DEV fill:#4CAF50,color:#fff
Cryogenic Amplifiers¶
For weak signals, amplify at 4K before thermal noise accumulates:
| Type | Noise Temp | Gain | Power | Notes |
|---|---|---|---|---|
| HEMT | 2-10 K | 25-40 dB | 5-15 mW | Workhorse, commercial |
| SiGe HBT | 10-30 K | 20-30 dB | 1-5 mW | Lower power |
| JPA | ~quantum limit | 20 dB | μW | Narrowband, complex |
| TWPA | ~quantum limit | 20 dB | μW | Broadband, cutting-edge |
HEMT amplifiers (Low Noise Factory, Cosmic Microwave) are the standard choice.
8. Grounding & Shielding¶
Ground Loop Prevention¶
Ground loops cause low-frequency noise and can completely mask small signals.
Star Ground Topology: All grounds connect at ONE point (typically cryostat chassis), no loops.
Coax Shield Grounding¶
Single-end grounding: Ground the coax shield at the warm end (feedthrough) only. Do NOT connect the shield to ground at the cold end.
Why? A shield grounded at both ends creates a loop between 300K and 4K. Temperature-dependent thermoelectric voltages and conducted noise will flow through this loop.
What about getting ground to the chip?
The coax center conductor carries the signal to the PCB. The chip's ground reference comes from:
- PCB ground plane — a continuous copper layer on the interposer/PCB
- Single chassis connection — PCB ground connects to cryostat chassis at ONE point
- Multiple wirebonds — chip ground pads connect to PCB ground plane via MANY wirebonds
This is not a contradiction:
- Coax shields: single-end ground (prevents warm-cold loops)
- Chip wirebonds: multiple ground bonds (low inductance for RF return)
- PCB to chassis: single point (star ground topology)
RF ground return: For high-frequency signals, you need low-inductance ground return paths. Use multiple ground wirebonds distributed around the signal bonds. These all connect to the same ground plane — this is NOT a ground loop, it's a low-inductance return path.
Two-Domain Ground Isolation¶
For µV-level measurements (AQFP readout), even star grounding may be insufficient. The cryostat body couples to building ground through the compressor/cold head chain.
Two-domain architecture separates:
- Domain 1 (Earth): Building mains → compressor → cold head → cryostat body
- Domain 2 (Isolated): Isolation transformer → readout electronics → dedicated ground wires → 4K PCB
The domains are electrically isolated by:
- Alumina disk at 4K (thermally conductive, electrically insulating) between cold finger and measurement PCB
- Isolation transformer feeding all readout electronics
This keeps compressor motor noise and 50/60 Hz building ground noise in Domain 1, while measurements reference the quiet Domain 2 ground.
See Lecture 11 (Testing) and the ground_isolation_memo technical note for detailed implementation.
Common Grounding Mistakes¶
| Mistake | Symptom | Fix |
|---|---|---|
| Shield grounded both ends | 50/60 Hz noise, ground loop | Ground at warm end ONLY |
| Single ground wirebond to chip | Poor RF return, ringing | Multiple ground bonds |
| PCB ground at multiple chassis points | Ground loop | Single star point |
| Digital and analog share ground | Switching noise | Separate grounds, join at star |
| No chassis ground | EMI pickup | Ground cryostat chassis |
Shielding¶
Cable shielding:
- Twisted pairs: Reject magnetic pickup
- Braided shield: Good flexibility, moderate shielding
- Solid shield (semi-rigid): Best shielding, no flexibility
Magnetic shielding at 4K:
- Finemet: High permeability, no annealing needed
- Cryoperm: Similar performance
- Lead (Pb): Superconducting shield below 7.2K, perfect diamagnetic shielding
RF shielding:
- Copper or aluminum enclosures
- All seams must make good contact
- Filtered feedthroughs for DC lines
9. Optical I/O¶
Optical fiber offers unique advantages for cryogenic I/O:
Advantages¶
- Zero electrical heat conduction: Only photon energy
- Extremely high bandwidth: 100+ Gbps possible
- Complete galvanic isolation: No ground loops
- EMI immunity: No pickup in fiber
Fiber Types¶
| Type | Core | Bandwidth | Cryo Behavior |
|---|---|---|---|
| Single-mode (SMF-28) | 9 μm | Highest | Works well |
| Multi-mode (OM4) | 50 μm | Lower | Works well |
| Polarization-maintaining | 9 μm | High | Birefringence shifts |
Cryo-Optical Considerations¶
Fiber contraction: ~0.1% from 300K to 4K
- Leave slack in fiber routing
- Avoid tight bends (becomes brittle)
- Minimum bend radius ~10mm at cryo
Connectors:
- FC/PC, FC/APC work at cryo
- May need realignment after first cooldown
- Use vacuum-compatible feedthroughs
Detectors at 4K:
- InGaAs photodiodes: work but reduced responsivity
- Ge photodiodes: good cryo performance
- SNSPDs: superconducting, for single photons
Applications¶
- Clock distribution: Low-jitter optical clock to 4K
- High-bandwidth data input: Streaming data into cryogenic systems with minimal heat load
10. Flex PCB / Kapton Wiring¶
Flexible printed circuits offer advantages for cryogenic wiring:
Why Flex/Kapton?¶
- Consistent geometry: Controlled impedance, repeatable
- Low mass: Less thermal anchoring needed
- High density: Many traces in small space
- Flexible: Can bend around corners
- Kapton survives cryo: -269°C to +400°C range
Design Considerations¶
| Parameter | Typical Value | Notes |
|---|---|---|
| Trace width | 75-150 μm | Finer for high density |
| Trace spacing | 75-150 μm | Limited by manufacturing |
| Copper thickness | 9-35 μm | Thinner = less heat load |
| Kapton thickness | 25-50 μm | Standard polyimide |
| Impedance | 50Ω typical | For RF traces |
Thermal Performance¶
Heat load per flex cable: $$Q = n_{traces} \cdot \frac{A_{trace}}{L} \int k_{Cu}(T) \, dT$$
Example: 20-trace flex, 9μm Cu, 50mm long, 300K→4K:
- ~50 mW total (compare to 20× individual wires: ~200 mW)
Flex Cable Routing¶
flowchart TB
RT[Room Temp Connector
ZIF / board-to-board]
RT --> F1
subgraph F1[Flex Cable]
S1[Serpentine
for length]
end
F1 --> T40[40K Thermalization]
T40 --> F2
subgraph F2[Flex Cable]
S2[Continues
to 4K]
end
F2 --> DUT[4K Connection
to DUT]
style RT fill:#FF5722,color:#fff
style T40 fill:#FFECB3
style DUT fill:#00BCD4,color:#fff
style F1 fill:#FFF9C4
style F2 fill:#FFF9C4
11. Practical Wiring Guidelines¶
Design Rules Summary¶
- Minimize wire count: Each wire is a thermal leak
- Use appropriate materials: Balance thermal vs signal quality
- Thermalize at every stage: 5-10 wraps with varnish/grease, or thermal clamps
- Route carefully: Separate digital from analog
- Document everything: Label wires, keep wiring diagrams
Material Selection Quick Reference¶
| Application | Material | AWG | Notes |
|---|---|---|---|
| DC bias (<10 mA) | Manganin | 36 | Lowest thermal load |
| DC bias (>10 mA) | NbTi | 32-36 | Zero resistance below Tc |
| Digital signals | Phosphor bronze | 32-36 | Good balance |
| RF (300K→40K) | SS coax | - | Thermal isolation |
| RF (40K→4K) | SS coax | - | Continued isolation |
| RF (at 4K) | Cu coax | - | Low loss |
| Thermometry | Phosphor bronze | 32-36 | Twisted pair, shielded |
Checklist Before Cooldown¶
- All wires thermally anchored at 40K and 4K
- Continuity verified at room temperature
- No shorts between channels
- Shields connected (at one end only)
- Connectors tight and labeled
- Wiring diagram documented
- Spare wires for contingency
# Visualize: Complete wiring scheme for SCE test system
fig, ax = plt.subplots(figsize=(14, 10))
# Temperature stages
stages = [
(8.5, 12, 1, '300 K', COLORS['warm']),
(6, 10, 1.2, '40 K', '#FFECB3'),
(3, 8, 1.5, '4 K', COLORS['cold']),
]
for y, w, h, label, color in stages:
rect = patches.FancyBboxPatch(((14-w)/2, y), w, h, boxstyle="round,pad=0.05",
facecolor=color, alpha=0.4, edgecolor='black', linewidth=2)
ax.add_patch(rect)
ax.text(7, y + h/2, label, ha='center', va='center', fontsize=11, fontweight='bold')
# DUT
dut = patches.FancyBboxPatch((5.5, 3.3), 3, 0.8, boxstyle="round,pad=0.02",
facecolor=COLORS['secondary'], edgecolor='black', linewidth=2)
ax.add_patch(dut)
ax.text(7, 3.7, 'SCE DUT', ha='center', va='center', fontsize=11, fontweight='bold')
# Wiring paths
wire_configs = [
(2, 'DC Bias\n(Manganin)', COLORS['success'], [9.5, 7.2, 4.5, 3.7]),
(4, 'Digital I/O\n(PhBr TP)', COLORS['primary'], [9.5, 7.2, 4.5, 3.7]),
(6, 'Clock\n(SS coax)', 'gray', [9.5, 7.2, 4.5, 3.7]),
(8, 'RF Out\n(SS→Cu)', COLORS['purple'], [9.5, 7.2, 4.5, 3.7]),
(10, 'Optical\n(SMF)', COLORS['cyan'], [9.5, 7.2, 4.5, 3.7]),
]
for x, label, color, ys in wire_configs:
# Draw wire path
ax.plot([x, x], [ys[0], ys[1]], color=color, linewidth=2.5)
ax.plot([x, x], [ys[1], ys[2]], color=color, linewidth=2.5)
ax.plot([x, x], [ys[2], ys[3]], color=color, linewidth=2.5)
# Thermal anchors at 40K and 4K
for y_anchor in [7.2, 4.5]:
circle = plt.Circle((x, y_anchor), 0.15, facecolor='gold', edgecolor='black', linewidth=1, zorder=5)
ax.add_patch(circle)
# Labels
ax.text(x, 10, label, ha='center', va='bottom', fontsize=9)
# HEMT amplifier
hemt = patches.Rectangle((7.7, 4.2), 0.6, 0.4, facecolor=COLORS['purple'], edgecolor='black', linewidth=1.5)
ax.add_patch(hemt)
ax.text(8, 4.4, 'HEMT', ha='center', va='center', fontsize=7, color='white')
# Filter symbols
ax.text(2, 9.2, 'F', fontsize=10, ha='center', bbox=dict(boxstyle='round', facecolor='white', edgecolor='black'))
ax.text(4, 9.2, 'F', fontsize=10, ha='center', bbox=dict(boxstyle='round', facecolor='white', edgecolor='black'))
ax.text(2, 4.8, 'F', fontsize=10, ha='center', bbox=dict(boxstyle='round', facecolor='white', edgecolor='black'))
# Legend
ax.text(12.5, 8, 'Legend:', fontsize=10, fontweight='bold')
ax.text(12.5, 7.5, '● Thermal anchor', fontsize=9)
ax.text(12.5, 7.0, 'F Filter', fontsize=9)
ax.text(12.5, 6.5, 'TP = Twisted pair', fontsize=9)
ax.set_xlim(0, 14)
ax.set_ylim(2.5, 10.5)
ax.axis('off')
ax.set_title('Complete SCE Test System Wiring Scheme', fontsize=14, fontweight='bold')
plt.tight_layout()
plt.show()
12. Summary¶
Key Concepts¶
Packaging:
- CTE mismatch is the enemy: Si shrinks less than Cu/ceramics
- Thermal contact: varnish, indium, grease — minimize interface resistance
- Wire bonding works at cryo with Al wire and proper loop geometry
- Flip-chip with indium bumps is ideal for high-density SCE
DC Wiring:
- Manganin for lowest thermal load (DC bias)
- NbTi for zero-resistance high-current lines
- Phosphor bronze for general purpose
- Thermalize at EVERY temperature stage
RF Wiring:
- SS coax for thermal isolation (300K↔40K↔4K)
- Cu coax only at 4K (low loss)
- HEMT amplifiers at 4K for sensitive readout
Grounding & Shielding:
- Star ground topology, one ground point
- Shield grounded at ONE end only
- Finemet or lead for magnetic shielding at 4K
Common Mistakes to Avoid¶
- Cu wire from 300K to 4K (massive heat load)
- No thermal anchoring (heat goes straight to 4K)
- Ground loops (50/60 Hz noise)
- Shields grounded both ends (ground loop)
- Wires too short (strain on cooldown)
Quick Reference¶
| Application | Material | Key Specs |
|---|---|---|
| DC bias | Manganin | 36 AWG, twisted pair |
| High current | NbTi | SC below 9K |
| RF thermal break | SS coax | 3-5 dB/m @ 10 GHz |
| RF at 4K | Cu coax | <1 dB/m @ 10 GHz |
| Flex wiring | Kapton/Cu | 9-18 μm Cu |
| Optical | SMF-28 | Works at cryo |