Created: Nov. 2021
[For MMIC-based 435 MHz preamp, go to "Very low noise preamp (0.4 dB NF) for 435 MHz / 70 cm"]
Although VHF/UHF pre-amps based on the BF988 & BF998 (same chip in different packages) have been previously reported by others, there is a dearth of information on its large signal (P1dB & OIP3) and stability characteristics [1, 2]. Moreover, prior arts did not provide a circuit model that can be used for either optimization or adapting to different frequencies. Therefore, the value of this work lies in elucidating the previously unreported performances and in creating a viable simulation model.
The BF988 is a popular dual-gate MOSFET offered by several companies: Philips (NXP), Siemens (Infineon) and Telefunken (Vishay) . The package style is TO-50. In this design, Q1 is biased at Vds = 8V and Ids = 10 mA.
The amplifier is supplied via the coax; i.e. phantom power. The series combination of R5 and FB separates the DC supply from the RF path. R5 drops the 12V supply on coax to 8V for Q1.
The operating frequency is adjusted by either compressing or spreading the coils L1 & L2.
Fig. 1 Circuit
Table 1: part list
Fig. 2: The components are assembled "dead bug" or "Manhattan" style on a single-sided PCB blank. A shield is later installed to minimize coupling between the coils L1 & L2
Fig. 3: Project enclosure using PCB. To improve the shielding effectiveness, copper tape is wrapped around the edges for better contact with the lid
The Siemens-provided BF988 s2p file lacked the noise parameter. So, to enable noise simulation, the BF998 s2p file was used instead. The BF998 has the same electrical characteristics as the BF988, except the former is packaged in a tinier SOT-143. We cursorily assume that the different packages will not significantly affect the results at the target frequency.
For simulating non-linear parameters, Philips' spice model of Bf998 (1993) is used. The spice model is also used for simulating the effect of the second gate's parasitic inductance because the s2p file has no provision for G2.
Fig. 4 Datasheets of BF988 (top) and BF998 (bottom)
Fig. 5: simulation circuit
Fig. 6: The shown L-section network presents a source termination for minimum noise. The 1.2 dB & 1.5 dB constant noise figure circles are shown in relation to the matching trajectory
Fig. 7: Although the BF988 is potentially unstable over 300-500 MHz, it can be stabilized with a resistive shunt load, R1= 390 ohm
The mosfet's G2 has to be returned to ground via the lowest possible inductance in order to prevent out-of-band instability. When a leaded C4 is used as G2's ground return, the combination of the two components' lead length causes a gain peak at 1.5 GHz.
Fig. 8: A previous version used a leaded C4. Unfortunately, the parasitic inductance due to the 5mm distance between C4 and Q1 resulted in a deleterious 1.5 GHz gain peak. To resolve this issue, C4 was replaced with a 0805 chip and soldered at ~ 1mm away
Fig. 9: The second gate G2 is very sensitive to parasitic inductance (Lpst) in its AC ground return (via C4). Simulation shows that a parasitic >= 2nH (blue trace) will create an unwanted gain peak at 1.7 GHz
The prototype initially attempted to minimize L1 & L2 coupling by positioning them at right angle to each other. Despite the coils' perpendicular positions, the prototype failed to achieve unconditional stability inband, as evidenced by mu <1 over 320-450 MHz. Subsequent investigation reveals that the potential instability arises from the coils' non-zero coupling degrading the reverse isolation (S12).
In hindsight, a chip inductor should have been chosen for L2. The chip's smaller size and lower unloaded Q can probably lessen coupling.
Fig. 10: Weak coupling (k) between L1 & L2 significantly degrades reverse isolation (S12) leading to inband instability. Simulation shows that k=0.03, degrades S12 by ~18 dB (pink trace) over the ideal k=0 condition (red trace)
Fig. 11: Installing a shield (25 x 18) between L1 & L2 improves reverse isolation to cure the inband potential instability
FB is measured in a series gap fixture.
Fig. 12: The ferrite bead with 5 turns have a parallel self resonance at ~ 450 MHz and so can minimize loading of the RF path.
Fig. 13: Setup for linear characterization: gain, return loss, reverse isolation & stability factor. The biastee inserts a 12V supply thru the pre-amp's output connector
Fig. 14: Input and output return loss show good agreement between simulated (dash) and measured (solid)
Fig. 15 Simulated (dash) and measured (solid) noise figure (F) curves agree within a fraction of a dB. Although the experimental noise match is off-tuned to ~410 MHz, this can be corrected by widening L1 spacing if desired
Fig. 16: Simulated (dash) and measured (solid) gain agree within 1 dB
Fig. 17: Installing a shield between L1 & L2 improves measured reverse isolation (S12) by ~10 dB
Fig. 18: Replacing the leaded C4 (black trace) with a 0805 chip (blue trace) eliminated the unwanted 1.5 GHz gain peak.
The prototype is unconditionally stable over 50-2000 MHz (fig. 19). Ideally, it should have been evaluated up to 4.5 GHz - the upper limit of the BF988's gain response.
Fig. 19: Adding a shield between L1 & L2 resulted in unconditional stability over 0.05-2.0 GHz (black trace). Without the shield, coupling between inductors leads to potential instability (mu <1) over 320-450 MHz (red trace)
Fig. 20: Simulated (blue dashes) & measured (solid black) 1dB gain compression point (P1dB) agree well; i.e. 10.7 dB vs. 9.3 dB
The third-order intercept point (OIP3) is simulated but not measured. I am currently unable to measure OIP3 because my wideband receiver, an Icom PCR-1000 has broken down. In the past, I used the PCR-1000 as a spectrum analyzer in conjunction with the TalkPCR software.
Fig. 21: Simulated third-order intermodulation showing an output intercept (OIP3) of 19.7 dBm.
[1] VE7BPO, "VHF-FM Supplement — Amplifiers", Oct. 2012 [Online] Available: http://www.qrp.pops.net/fm-sup.asp
[2] D. Dobričić , "Ekonomični niskošumni antenski pojačavači za VHF / UHF", [Online] Available: https://www.academia.edu/44598190/Ekonomični_niskošumni_antenski_pojačavači_za_VHF_i_UHF_opseg