Theory

Noise Sources in Vacuum Tubes

Shot noise, thermal noise, flicker noise, and partition noise — understanding each mechanism

Every electronic device generates noise. In vacuum tubes, four fundamental mechanisms contribute. Understanding each one lets you minimize the total noise through proper tube selection, operating point choice, and circuit topology.

Shot Noise
i_n = √(2qI_p × BW)
Random arrival of electrons at the plate. Proportional to √I_p. Dominant in low-current stages.
Thermal Noise
e_n = √(4kTR × BW)
Johnson noise from resistors. The plate load and grid-leak resistor are primary contributors.
Flicker (1/f) Noise
S(f) ∝ 1/f^α (α ≈ 0.8-1.2)
Surface contamination and cathode irregularities. Dominant below 100Hz-1kHz. Oxide cathodes are noisier than pure tungsten.
Partition Noise
Only in pentodes/tetrodes
Random division of electron stream between plate and screen. Adds 3-6dB of noise vs triode. This is why low-noise preamps use triodes.
Equivalent input noise: e_n ≈ √(4kT × 2.5/gm) for triodes
Calculator

Noise Figure Calculator

Compute noise figure and S/N ratio for your preamp stage

gm1.6mA/V
R_source600Ω
R_grid1.0k
BW20kHz
e_n tube719.4nV
e_n source445.8nV
Total noise0.85μV
NF5.6dB
S/N (1V in)121dB
Visualization

Noise Spectral Density

Reference

Low-Noise Tubes Ranked

Noise figure depends primarily on gm — higher gm means lower noise

Tubegm (mA/V)NFApplicationNotes
5842 (417A)111.2dBMicrophone preamps, lab instrumentsThe gold standard for low noise
E88CC (6922)5.02.5dBPhono preamps, headphone ampsBest dual-triode NF. Frame-grid.
12AX7 (ECC83)1.64-6dBGuitar preamps, general audioNoisy — high rp, low gm
12AT7 (ECC81)5.52.8dBPhase splitters, driversMuch quieter than 12AX7
6SJ71.658-10dBMicrophone preamps (old)Partition noise limits NF
EF86 (6267)1.93-4dBLow-noise pentode preampsExceptional for a pentode
6S45P45< 1dBUltra-low-noise preampsRussian military. Extreme gm.
12AU7 (ECC82)2.25dBCathode followers, buffersModerate noise, good for followers
Practice

Building Low-Noise Tube Circuits

1. Tube Selection: gm is King

The equivalent input noise voltage is inversely proportional to √gm. The 12AX7 (gm = 1.6mA/V) has 3× worse noise than the E88CC (gm = 5mA/V). For the lowest noise, use frame-grid triodes like the 5842/417A (gm = 11mA/V) or the Russian 6S45P (gm = 45mA/V). The tradeoff: high-gm tubes have low rp and low mu, so they need careful impedance matching.

2. Operating Point: More Current = Less Noise

Running a tube at higher plate current increases gm, reducing noise. For a 12AX7, increasing Ip from 0.5mA to 1.2mA improves noise by ~4dB. But there's a limit — excessive current causes the tube to leave its linear region and increases distortion. The sweet spot for noise is typically 60-80% of the maximum rated plate dissipation.

3. Heater Supply: DC is Essential

AC heater hum couples into the signal via cathode-heater capacitance and leakage. For serious low-noise work: (1) use DC heater supply with proper filtering (< 1mV ripple), (2) elevate the heater reference to +50-70V DC to reverse-bias the cathode-heater junction, (3) use a separate heater winding for the input tube. These measures can improve hum rejection by 40-60dB.

Quiz de synthèse

Testez vos connaissances

Question 1 / 5

Which tube parameter most directly determines the equivalent input noise voltage?

References

  1. Morgan Jones, Valve Amplifiers, 4th ed., Newnes, 2012. ISBN 978-0080966403Modern engineering treatment of tube audio design.