# Cloudlifter CL-1 circuit research

Researched 17 October 2026. Scope: the specific Cloud Microphones CL-1 topology, following John's clarification. The earlier two-BJT alternative is excluded from the build.

## What the evidence supports

Cloud Microphones describes the CL-1 as a phantom-powered Class-A discrete JFET booster, with up to 25 dB gain and nominal 3 kΩ input impedance. Its microphone input does not supply phantom power. These are the manufacturer's product specifications, not measurements of this DIY project. [Cloud product page](https://cloudmicrophones.com/products/cloudlifter-cl-1)

The company's patent describes paired JFET cascode branches and the use of the receiving preamp's external phantom feed resistors as both power feeds and part of the gain load. Patent examples are not a production BOM. Figure 11 variants have different component values. [US9167327B1, especially Figures 11A–11D and their description](https://patents.google.com/patent/US9167327B1/en)

MostlyBuilds directly documented a CL-1 teardown, then built a custom PCB using the traced circuit. This is the practical construction source for this educational reconstruction; it is independent of Cloud and is not an official factory schematic. [Author's teardown and build](https://mostlybuilds.com/microphone-activator-series/) · [KiCad repository](https://github.com/MostlyBuilds/microphone-activator)

The reference checkout is pinned to `a41f27b773d3bbb2653ae5907460e5cf38217a87`. The course's board, pads, tracks, filled ground pour, placements, and net table are extracted from that PCB, rather than invented for an attractive render. The board outline is **27.305 × 19.050 mm**. The enclosure and connector bodies are photo-informed teaching geometry; they are not machining drawings of a CL-1 or Neutrik housing. See [the photo/iteration record](REALISM.md) and [the simulation record](SIMULATION.md) for the 17 October realism and numeric-study update.

## Circuit and package details

The build has **two dual-JFET SOIC packages**, containing four transistor sections in total. Q1 is the input pair; Q2 provides the cascode devices. A cascode is not two independent voltage-gain stages whose gains can simply be multiplied. The input gates receive the balanced microphone signal; the output drains connect toward the receiving preamp and its phantom supply. The local source resistors establish bias and degeneration. Actual gain depends on the devices, operating point, phantom source, and receiving input load.

The input has a 3 kΩ differential resistor and two 33 kΩ resistors to circuit ground. Their DC differential load is approximately `3000 || 66000 = 2869.6 Ω`; the two 1.8 nF shunt capacitors also affect impedance at high frequency. The capacitors are RF shunts, **not series phantom-blocking capacitors**. There are no input electrolytics in this reconstruction.

The SOIC pin functions are **1 S1, 2 D1, 3 substrate, 4 G1, 5 S2, 6 D2, 7 substrate, 8 G2**. Pins 3 and 7 float in the reference PCB. The TO-71 version has a different package mapping and cannot be placed on this footprint. Use the exact ordered-package datasheet when building. [Linear Systems LSK389 application note](https://www.linearsystems.com/_files/ugd/7e8069_5339d0e5c21c409abccc4701d82d1976.pdf) · [Manufacturer datasheet hosted by NAC](https://cms.nacsemi.com/content/AuthDatasheets/LISIS00420-1.pdf)

## A source discrepancy that matters

The schematic's instantiated Q1/Q2 values are **LSK389C**, consistent with the author's written parts list. The PCB value strings and reused symbol metadata still say **LSK389B**. C and B indicate different IDSS grades, not cosmetic labels. This course displays **LSK389C**, retains both source strings in `board-source.json`, and explicitly warns against automatically buying the B grade from the stale PCB field. This is a documented source conflict; it has not been resolved by inspecting a physical CL-1 or testing alternative grades.

R3 uses a capacitor-named 1206 footprint in the source PCB. It remains a **33 kΩ resistor**. A library footprint's name does not decide the component's electrical function.

## Parts to verify before ordering

| References | Quantity | Source value | Package / role |
|---|---:|---|---|
| Q1, Q2 | 2 | LSK389C | SOIC-8 dual matched N-channel JFET; see grade conflict above |
| R1 | 1 | 39 Ω | 1206; common source return |
| R2, R5 | 2 | 22 Ω | 1206; branch source resistors |
| R3, R6 | 2 | 33 kΩ | 1206; input gate DC returns |
| R4 | 1 | 3 kΩ | 1206; differential input load |
| C1, C2 | 2 | 1.8 nF | 1206 PEN film, nonpolar; source-matched RF shunts |
| J2 | 1 | Female XLR input | Header pads 1/2/3 map to the numbered XLR contacts |
| J3 | 1 | Male XLR output | Receives phantom power from the interface |
| PCB | 1 | Source layout | 27.305 × 19.050 mm; fabrication thickness must be specified |
| Enclosure, insulation, wire, fixings | As required | Mechanical integration | Use actual connector/enclosure drawings for final fit |

This is a researched BOM, not an order placed or a stock/price quote. Use the author's linked exact parts and their current manufacturer assembly limits when selecting substitutes. The course does not invent capacitor voltage ratings or a generic reflow temperature: those belong to the exact purchased component and solder-paste profile.

## Bench sequence for a real prototype

1. With all cables disconnected and phantom off, identify actual connector contact numbers. XLR pin 1 is screen/ground, pin 2 hot, pin 3 cold. The solder side and mating side are mirrored. [Neutrik wiring instructions](https://www.neutrik.com/media/8061/download/crimp-xlr-cable-connectors.pdf?v=1)
2. Populate the six 1206 resistors, then the two RF capacitors, then both SOICs. Verify values and orientation under magnification after each group. Follow the exact film-capacitor thermal limits; excessive rework can damage them.
3. Connect J2 to the female input and J3 to the male output by contact number. Keep balanced conductors paired and short. Insulate the PCB from the metal enclosure; plan the chassis/screen bond deliberately.
4. Compare every pin/net against `NETS.md`. Check for solder bridges and output shorts. The input contains intentional resistive paths, so "everything must read open" is the wrong continuity rule. A no-power meter reading across semiconductors is not a complete electrical test.
5. First power with **no microphone connected**, monitors muted, and a protected/current-limited P48 test fixture operated by someone comfortable with electronics. The fixture needs the normal matched feed resistors; a naked 48 V bench supply is not equivalent. Record both output-to-pin-1 voltages and derive each leg's current from the measured feed drop. Do not put a meter's current range directly across XLR pins.
6. Check input pins 2 and 3 relative to pin 1 for unexpected DC, and between 2 and 3. If supply-scale voltage reaches the microphone input, a leg is pinned to a rail, or readings are strongly asymmetric, power down and investigate. There are no measured reference bias values for this build yet; guessing a pass window would be misleading.
7. Check for oscillation before attaching valued equipment. An ordinary grounded oscilloscope clip must not be attached to a phantom-carrying signal leg. Use suitable differential, isolated, or DC-blocked measurement arrangements.
8. Apply a small balanced test signal with DC-safe equipment. Measure differential input and output RMS at 1 kHz without clipping; calculate `20 log10(Vout/Vin)`. The 25 dB product specification corresponds to a voltage ratio of about 17.8, but is not a guaranteed DIY result. Record frequency response, headroom and level-matched noise before describing performance.
9. Only after electrical checks, try an inexpensive dynamic microphone, then the intended SM7B. Disconnect and discharge before changing the assembly. The SM7B itself does not require phantom; the booster does. [Shure phantom-power FAQ](https://service.shure.com/articles/en_US/Knowledge/does-the-sm7b-need-phantom-power)

## Evidence boundaries

Completed: primary-source research, extracted PCB data, authored Blender geometry and lesson content, and the local interactive lesson prototype (see `evidence/QA.md` for actual render/browser checks).

Not established: physical assembly, live phantom-power behavior, DC isolation under faults, measured gain/noise/bandwidth, thermal performance, EMC, or manufacturing acceptance. The virtual controls never certify a physical build. Cloudlifter is Cloud Microphones' trademark; this is an independent circuit study, not a Cloud product or licensed factory build guide. Public availability of a design or patent does not itself grant manufacturing rights. The source repository has no general license file in the inspected revision; reference material remains in the local research folder and is linked/attributed in the lesson player.