Sources & accuracy
The Instrument is a stylised dilution refrigerator, drawn to show how Rahul Nainala builds software. It is not a working design and the site does not simulate qubits. The physical numbers are typical published values for one well-documented build — the ETH Zürich Bluefors XLD400 measured by Krinner et al. — cross-checked against the manufacturer. Corrections welcome: open an issue.
Stage by stage
| Stage | Typical temperature | Cooling power | Gap to plate above | Drive attenuation (illustrative) | Sources |
|---|---|---|---|---|---|
| 01 · Top flange | ~295 K | — | — | 0 dB (total 0 dB) | Bluefors, Krinner et al. |
| 02 · 50K flange | ~35–40 K | 30 W at 45 K | 200 mm | 0 dB (total 0 dB) | Krinner et al., Bluefors |
| 03 · 4K flange | ~2.85–3 K | 1.5 W at 4.2 K | 290 mm | −10 dB (total −10 dB) | Krinner et al., Bluefors, RF Essentials |
| 04 · Still | ~0.88 K | 40 mW at 1.2 K | 250 mm | 0 dB (total −10 dB) | Krinner et al., Bluefors |
| 05 · Cold plate | ~82–100 mK | 200 µW at 140 mK | 170 mm | −20 dB (total −30 dB) | Krinner et al., Bluefors |
| 06 · Mixing chamber | ~6–10 mK | 19 µW at 20 mK | 140 mm | −30 dB (total −60 dB) | Krinner et al., Bluefors, RF Essentials |
Nominal flange names are not temperatures: the “50K” flange typically sits at 35–40 K and the “4K” flange near 3 K. Per-stage attenuation follows Krinner et al.; other published schemes split it differently but also total about 60 dB. Thermal-photon numbers are deliberately not quoted.
Numbers used in the copy
- Cooling power, top stage vs. mixing chamber
- 30 W → 19 µW, about 1.6 million times less — Ratio computed from the two published values. [Krinner et al.]
- Total drive-line attenuation
- ≈60 dB (a millionfold reduction in power) — Per-stage split is illustrative; published schemes differ but total about the same. [Krinner et al., RF Essentials]
- Base temperature
- 6 mK measured; 7 mK (Bluefors); 10–15 mK (IBM) [Krinner et al., Bluefors, Popular Science]
- Compared with the afterglow of the Big Bang
- ~300× colder (2.725 K ÷ 6–10 mK ≈ 270–450×) — Computed. The site avoids “colder than space”. [NASA WMAP, Krinner et al.]
- Nominal vs. actual flange temperatures
- “50K” ≈ 35–40 K, “4K” ≈ 2.85–3 K, still ≈ 0.88 K, cold plate ≈ 82 mK [Krinner et al., Bluefors]
- Plate spacing, top to bottom
- 200 / 290 / 250 / 170 / 140 mm [Krinner et al.]
- Cooldown time
- about 24 h (XLD400) to about 48 h (IBM) [Université de Sherbrooke, Popular Science]
- Pulse-tube “heartbeat”
- ≈1.4 Hz [arXiv:1603.03146 — pulse-tube cryocooler vibration measurements]
- Why the plates are gold
- Gold-plated OFHC copper resists corrosion; no nickel underlayer, because nickel is magnetic [Hubs, IQM]
- Lines per qubit
- ≈3.8 (76 lines for 20 qubits) [IQM]
- Amplification on the way back up
- TWPA/JPA at the mixing chamber, HEMT at 4 K, then room temperature [RF Essentials, Krinner et al.]
What is invented
- The model is procedural and stylised: proportions follow published plate spacings, but part counts, positions and the chip layout are design choices.
- The chip is not a real device. Its 45° square lattice, meanders and bond pads are drawn to read as a superconducting chip, nothing more.
- The telemetry on the cold-plate stage is a simulated feed generated in your browser.
- The finale's orbital is sampled from the real hydrogen 3d(z²) distribution; the qubit is a single Bloch vector with ideal gates. A playful nod to measurement, not a simulation of hardware.
References
- Krinner et al., “Engineering cryogenic setups for 100-qubit scale superconducting circuit systems”, EPJ Quantum Technology (2019) — Measured stage temperatures, cooling powers, plate spacings and attenuation for a Bluefors XLD400.
- Bluefors, “Components of the dilution refrigerator measurement system” — Nominal flange names vs. actual temperatures (“50K” ≈ 40 K, “4K” ≈ 3 K); still, cold plate, mixing chamber.
- RF Essentials, “What is the required attenuation at each temperature stage of a dilution refrigerator?” — An alternative per-stage attenuation scheme; also ≈60 dB in total. Amplification on the way back up.
- IQM, “Technology and performance benchmarks of IQM's 20-qubit quantum computer” (arXiv:2408.12433) — Square qubit lattice; 76 lines for 20 qubits; gold-plated copper hardware.
- Hubs, “Creating hardware for quantum engineers” — Gold over OFHC copper, without a nickel underlayer because nickel is magnetic.
- arXiv:1603.03146 — pulse-tube cryocooler vibration measurements — The ≈1.4 Hz pulse-tube cycle.
- Université de Sherbrooke, high-capacity dilution refrigerator (XLD400) — Cooldown of about 24 hours.
- Popular Science, “In photos: journey to the center of a quantum computer” — Cooldown of about 48 hours on an IBM system; 10–15 mK base.
- NASA WMAP, cosmic microwave background — 2.725 K is the standard CMB temperature (linked for reference; not re-checked on this page).