An RF preamplifier can reduce noise on weak signals. A low-noise preamp can ensure that external noise, not tuner sensitivity, limits reception. A preamp can breathe new life into an older tuner that lacks an RF amplifier, has inadequate IF limiting, or is misaligned. Mounted at the antenna, a preamp can overcome the loss of a long feedline, which otherwise directly degrades tuner sensitivity. If the RF signal feeds multiple tuners, each signal split degrades tuner sensitivity at least 3 dB. Placing a preamp ahead of the power divider can mitigate this loss.
Gain, noise figure, third-order intercept, and return loss characterize a preamp. Gain is signal and external noise amplification. Noise figure quantifies the noise a preamp adds. Third-order intercept relates desired and spurious signal levels to measure distortion. Return loss indicates how closely the input or output impedance matches 75Ω. Lower noise figure, higher third-order intercept, and higher return loss are always better. Higher gain is better, but too much can overload a tuner front-end, which is likely to distort before the preamp does.
A low-noise preamp with 5 dB gain can improve sensitivity with low risk of tuner overload. Unless large losses follow the preamp, diminishing returns set in above 15 dB gain and overload risk becomes significant. Preamp gain ranges from several dB for distribution amplifiers to 30+ dB for mast-mount TV preamps.
Preamp noise figures range from about 0.5 to 5 dB, with low values for optimized designs using modern devices and high values for older designs. When given, the noise figure for a consumer preamp usually is 1 to 3 dB.
Tuner noise figures range from perhaps 3 to 9 dB, or even higher for older designs. As a rough approximation, noise figure plus mismatch loss equals monophonic 50 dB quieting sensitivity minus 10.5 dBf. I derived this figure by measuring six tuners using full IF and audio bandwidth. They yielded differences of 10.0, 10.4, 10.4, 10.5, 10.5, and 11.3 dB.
I built this RF preamp from junkbox parts. I could not find a recommended source impedance for lowest noise figure. Instead, I went for a good match since input mismatch loss directly degrades noise figure. I measured the input return loss as > 18 dB from 88 to 108 MHz. The corresponding mismatch loss is < 0.07 dB.
The untuned input inductor, 25 turns on a T-25-7 core, provides a DC ground. As long as its reactance is high, its value is noncritical. You can eliminate it entirely if the antenna is at DC ground. The output inductor is 15 turns on a T-37-10 core tapped 3 turns from the cold end. It resonates with the J309 drain/gate capacitance. I adjusted the turns spacing until the gain was about equal at the band edges. Output return loss is poor, but this lowers the midband gain peak and has almost no effect on noise figure. It does complicate measurement since the output impedance is not near 75Ω. To avoid inconsistencies, I measured everything without reconnecting the preamp.
The favorable input return loss is due to the J309 transconductance, which is about 13 mS. Inverting that yields an input impedance close to 75Ω. With additional parts you can adjust the drain current and transconductance, but I didn't bother. The J309 IDSS spec is 12–30 mA. My circuit drew 17 mA.
At midband the third-order output intercept was 27 dBm. I measured the following gain and noise figure values:
Freq Gain NF 88.0 12.6 3.0 96.9 14.8 3.1 107.7 12.2 3.8
I wasn't sure I'd keep this circuit so I didn't cut the transistor leads. It's best to make them as short as possible to minimize stray inductance.
To avoid the output coil tap, reduce the output coupling capacitance. 12 turns on a T-37-10 core yielded about 15 dB gain midband and 11 dB at the band edges.
I built an even simpler preamp to mount at the antenna. The output inductor forms an L-network with the J309 drain/gate capacitance. I used 16 turns on a T-37-10 core and spread them to adjust resonance. Gain was about 16 dB midband and 12 dB at the band edges. The input coil is a Nytronics Wee-ductor. I found the Scientific Atlanta DA-PI power inserter at a garage sale. 220Ω for R1 yielded 13 V at the drain from a wall wart that delivered about 18 V with no load. Current was 20 mA.
Don't duplicate these designs. They are intended to inspire you to throw something together from whatever is in your junkbox. Unless your tuner's noise figure is unusually low, a homebrew preamp using an older transistor should improve it. To use a modern, low-noise device, follow the manufacturer's datasheet.
Amazon and eBay offer a variety of low-noise amplifier modules at low cost. Gain and noise figure measurements for several are here. Modules with an SPF5189Z require 3–5 volts at 60–90 mA and should have 25–27 dB gain and 0.5–0.6 dB noise figure at 100 MHz. An output attenuator can tame the gain and match 75Ω. Use SMA-to-F adapters or replace each SMA with an F. To better match 75Ω and help protect the sensitive pHEMT against electrostatic discharge, add a 340 nH inductor (6t #22, ⅜″ ID, ¼″ L) across RFin. To install the LNA at the antenna, put it in a waterproof enclosure, jumper the output capacitor or add 1 µH (11t #22, ⅜″ ID, ⅜″ L) between VCC and RFout, and use a power inserter followed by a variable attenuator at the tuner.
R1 R2 R3 Gain NF 82 75 240 18 0.6 68 110 160 15 0.6 62 160 120 12 0.7
This table shows the performance of an LNA plus fixed output attenuator when the LNA alone has 27 dB gain and 0.5 dB noise figure. Resistance is in ohms, gain and NF in dB. Use this calculator to design a different attenuator.
At garage sales I find older cable TV amplifiers like this one. They work fine as an indoor preamp for less sensitive FM tuners. This particular amplifier had 13.7 dB gain in the FM band. The third-order output intercept was 29 dBm and saturated output was 40 mW. Return loss was 14.5 dB input and 12 dB output. The noise figure was 4.8 dB. This is not a low value, but the amplifier still improved the 50 dB quieting sensitivity of a Technics ST-9030 from 18.1 to 15.4 dBf.
I traced out this circuit. The emitter voltage is −12 V and total current is 15 mA. Rated power consumption is 2 W. The feedback flattens the gain and lowers distortion, but it degrades the noise figure. Breaking the feedback loop improved the noise figure 1.4 dB and increased gain 5 dB, but it degraded return loss to 6 dB input and 9 dB output. Adding 68 pF with 1½″ leads from the emitter to ground restored return loss to 13 dB input and 14 dB output. The modified preamp improved the 50 dB quieting sensitivity of an Onkyo T-4150 from 17.5 to 14 dBf.
A similar amplifier with four outputs had 7.0 dB gain. The circuit was the same as that above minus the input diodes and with tiny ferrite transformers to divide the output four ways. In this unit the transistor was a 2SC3777. The third-order output intercept was 21 dBm. The power division degrades the noise figure just 0.2 dB.
This amplifier has adjustable gain. In the FM band I measured 18.5 to 29.5 dB. This is too much gain for all but the most bulletproof of tuners in a benign signal environment. Reducing the gain to minimum degraded the noise figure so much that sensitivity for the amplifier plus tuner dropped below that of the tuner alone.
The Kitz Technologies KT-501 gain is adjustable to 17.5 dB. The noise figure, measured with an HP 8970A, is 0.85 dB and increases to about 1 dB as gain is reduced. Input and output return losses are > 10 dB. The third-order output intercept is 32 dBm. The unit has input electrostatic discharge protection and no FM trap. The mast-mount model is KT-501-COAX. Kitz is a small outfit that has been making amplifiers since 2000.
The PBD CX-208 has 16 dB gain, 1 dB noise figure, 4G LTE filter, no FM trap, and can be mast-mounted.
The RCA TVPRAMP12E has 16 dB gain, < 2 dB noise figure, switchable FM trap, and can be mast-mounted.
Adding variable attenuation after a high-gain preamp provides a flexible system. You can dial back the gain until any tuner overload ceases. Output attenuation degrades the effective preamp noise figure surprisingly little. For example, if you lower the gain of a 25 dB preamp with a 2 dB noise figure by 15 dB, the noise figure increases 0.3 dB. The preamp plus attenuator is equivalent to a preamp with 10 dB gain and 2.3 dB noise figure.
You can locate the preamp at the antenna to overcome feedline loss and place the attenuator at the tuner for easy adjustment. A simple 100Ω potentiometer should work fine. Minimum attenuation is 2.8 dB. Output return loss is 11.3 dB minimum. The output impedance won't misload a tuner front-end by varying wildly from 75Ω. However, input return loss is 11.6 dB maximum and drops toward 0 as attenuation increases. Today virtually all preamps are unconditionally stable, but an odd load impedance might cause an older, marginally stable preamp to oscillate. Interchange the attenuator ports if this occurs.
To improve return loss, add fixed resistors. Minimum return loss is 10.6 dB input and 13.9 dB output. Attenuation varies from 4.2 to 19.8 dB.
Wiper Pot Pot + Fixed % Atten RLin RLout Atten RLin RLout 100 2.8 11.3 11.3 4.2 35.1 15.8 90 3.7 11.6 14.4 5.1 43.2 20.3 80 4.7 11.5 18.3 6.0 49.5 27.9 70 5.8 11.0 24.1 6.9 43.2 40.9 60 6.9 10.1 39.0 7.9 35.1 25.7 50 8.2 9.0 29.8 9.0 28.8 21.0 40 9.8 7.6 23.2 10.2 24.0 18.3 30 11.8 6.0 20.1 11.7 20.0 16.5 20 14.7 4.2 18.3 13.5 16.5 15.3 10 19.9 2.2 17.3 15.9 13.5 14.4 0 ∞ 0.0 16.9 19.8 10.6 13.9
This table gives attenuation, input return loss, and output return loss in dB as a function of wiper setting for a linear pot taper. Preamp and tuner impedances are 75Ω and stray reactances are ignored.
A Radio Shack attenuator I found at a garage sale provided 1.5 to 20.5 dB of attenuation. I was happy with it until I measured the return losses: 10–25 dB at ANT, but 1–17 dB at TV. Maximum DC resistance at TV was 900Ω, which will unload a 75Ω tuner input circuit. Swap the connections to avoid this.
System noise is likely to depend mostly on external RF noise. In the FM broadcast band, sky noise and manmade noise are the most significant external sources.
Sky noise level depends on solar activity, sun position, and galactic center position. I derived these noise figure values for 98 MHz from noise temperature curves in Thomas A. Milligan's Modern Antenna Design, 2nd ed.:
Maximum 18 dB Average 11 Minimum 4
Manmade noise varies so much with time and location that any attempt to account for it is highly speculative. I calculated these noise figure values for 98 MHz using equations in recommendation ITU-R P.372-14:
City 22 dB Residential 17 Rural 12 Quiet rural 0
This program calculates system noise figure. It orders components by signal path. Two coax entries let you locate the preamp anywhere in the feedline. System noise figure does not depend on the physical order of the output attenuator, coax #2, or power divider, whose loss is typically 3.3 dB/split (e.g., 9.9 dB for eight outputs). See the antenna models and tuner writeups for typical mismatch loss and noise figure values. Ferrite balun loss is typically 0.75 dB. RG-6 loss is about 2 dB/100′. Preamp benefit reference: all three preamp entries 0.
Experiment with preamp noise figure and gain. Try various values for sky noise and manmade noise, both of which can vary over time. If you believe your external noise is generally low, you may find a preamp worthwhile. But if you expect it to be more often high than low, you may decide that a preamp offers little benefit.