Published in DUBUS 4/2011
Since the publication of my article presenting EXTRA-2 144 MHz Contest Preamplifier REV5 version , a total number of 150 preamps have been produced in 3 months, so the time has come to summarize the experiences and consider the development possibilities.
First of all I would like to emphasize that the original circuit met the demands of contesters. The primary design requirement was the strong signal performance (IMD characteristics) and further important requirements were good selectivity unconditional stability. However, the desire for a low noise figure was pushed into the background, so in this article I examine the possibilities of decreasing the noise figure while keeping the excellent stability and selectivity of the preamp.
Many thanks to the hams using my preamp for giving their feedback, and several suggestions which I tried to take into consideration during the further development of this product. Special thanks to DC8RI, IZ4BEH, DK5EW and HA1YA. This article describes only the 144 MHz version, but under the influence of the many words of encouragement, I have started the development of the 6 m and 70 cm versions using the same ATF-51389 extra high dynamic range FET.
Frequency range: .. 144-146 MHz
Input/Output impedance: 50 Ω
Gain: . 14 - 20 dB (adjustable)
Noise Figure .. <0,6 dB (typ. 0,5)
S11: .. -9 dB @ 144MHz
1 dB Bandwidth: .. 2,5 MHz
3dB Bandwidth: .. 4,5 MHz
IP3 . >22 dBm
Broadcast rejection: ..>63 dB @ 100 MHz
Rejection @ 432 MHz: >50 dB
Moderate sensitivity to ESD
Supply voltage: ... +12 15V/152mA
Powered through antenna feed line, or separately
+40 dBm IP3 GaAs FET (ATF-53189)
Connectors: ...... BNC or N
Dimensions (Box): .. 74 x 37 x 30 mm
Modifications of version REV6
- Adjustable output attenuator to select an appropriate level of gain
- Applying a special source degeneration method (unconditional stability)
- Modified input matching circuit (lower noise figure, applicable also for EME)
- Improved drain and gate decoupling circuits (higher OIP3)
- Modified gate bias circuit (more precise adjustment of ID current, better thermal stability)
- Built-in Schottky diode protection for the input and output (necessary for real-life operation)
- Improving the operation of the protection relay (special PTT control is not needed)
Design and simulation
By omitting parallel feedback in the original circuit, the noise figure will decrease by approximately 0.2 dB and the amplification will increase by 2-3 dB. This increased gain could cause overload of the following RX in some cases, and would only be acceptable if the attenuation of the RX cable is high, or if it is possible to put an attenuator between the preamp and the transverter. To meet the needs of a wider range of users, a 100 Ohm potentiometer has been included as part of the output attenuator, so the amplification can be easily adjusted over a range of about 6 dB. Table I shows the average values of the measurements of 20 completed preamplifiers.
|PMIN (0 ohm)||PMAX (100 ohm)|
Noise figure can be decreased to as low as 0.3 dB by modifying the input matching circuit, but series inductive feedback in the source is then needed to avoid instability. This is not easy because the FET in its SOT-89 package needs to be cooled efficiently. The special layout developed for the inductive source degeneration can be seen in the simplified circuit diagram of Figure 1. A short tapered transmission line connects the source tab of the FET package to a section of broader transmission line which provides both the inductive feedback and good heat conduction to the large number of through-board grounding vias. While working with the test circuit I found that the optimal width of the broader transmission line is 200 mil (0.200 inches, 5.08 mm). At this value the cooling and the inductance are both appropriate. Then I examined the S11 parameter and the K stability factor with a high frequency circuit simulator for various lengths of line.
Figure 2 shows the S11 results when the source inductance is decreased from a line length P = 300 mil down to P = 0 (tapered section only) and also for the case of zero source feedback. The corresponding variations in the K stability factor over a very wide frequency range are seen in Figure 3. In order to have suitable input match at 144 MHz, I first chose P = 300 mil, which seemed to be the most preferable value. The simulated noise figure increases slightly (0.05 dB), the K stability factor is excellent and the gain decreases by approximately 2 dB. Ut in the prototype of this layout, the relatively long series inductor caused too much positive feedback between the output and the input, leading to oscillation. Based on that practical experience, I chose P = 110 mil for the further simulations. No more oscillation was observed and the wideband K stability factor is still adequate.
The high K value allows further optimization of the input matching circuit to achieve a lower noise figure. At this point a compromise is needed between s good board layout, acceptable selectivity and a low noise figure. I examined all the possible input circuit configurations, and found that the original configuration still met the mentioned requirements the best, so the further optimizations concentrated on that input circuit.
Fig. 3: Stability factor versus source inductors
Fig 4: S11 versus C1
Firstly, I examined the variation of S11 with the series input capacitor C1 and found 15 pF to be optimal (Figure 4). With that value fixed, I then optimized the C2 capacitor (Figure 5) and chose the green curve for C2 = 22 pF. The rectangular spiral inductor then needs to be re-optimized, and the simulator allows the number of turns (N) to be varied in ¼-turn steps (Figure 6). I chose the green curve for 3.75 turns as optimum. Attentive readers may notice that I did not choose the values of L1, C1 or C2 that gave the best possible input match, for reasons that will be revealed in the next series of simulations.
Fig 5: S11 versus C2
Fig 6: S11 versus number of inductor turns, N
The input circuit of the preamp needs to be optimized not only for the good impedance match, but also to give a lower noise figure. Figure 7 shows the variation of the noise figure with C1, and 15 pF proves to be an acceptable compromise between the impedance match and the NF. This can be seen from a comparison between Figure 4 and Figure 7.
C2 also needs to be optimized by the same criteria. Figure 8 shows the variation of NF with C2, and by comparing Figure 8 with Figure 5 I chose C2 = 22 pF. Finally, the number of turns of L1 needs to be re-optimized on the basis of Figure 9 and Figure 6. N = 3.75 remains the optimal value.
The simulated noise figure of the preamp using all of these optimized component values can be seen in Figure 10. The website of the manufacturer does not supply NFMIN data under 500 MHz for NF, so I had to apply linear interpolation during the simulation. This can cause a small difference compared to the built circuit. Matching to such high impedance requires very high-Q components to minimize circuit losses. NF = 0.45 dB can be achieved assuming component Q values of QC = 300 and QL = 160. To achieve the assumed value of QC = 300, C1 and C2 will need to be low-loss microwave capacitors. For L1, QL = 160 is achieved by a spiral inductor optimized to maximal QL on the popular FR4 (TanD = 0.02). This solution also provides easy reproducibility. The box slightly mistunes the spiral inductor, so C2 consists of a fixed capacitor of 22 pF with a small trimmer capacitor (CT) to provide accurate tuning.
Figure 11 shows the K stability factor of the EXTRA-2 REV6 144 MHz contest preamplifier optimized according to this article, between DC and 4 GHz. S-parameters of the preamp without output filter can be seen in Figure 12 (this simulation assumes an output attenuator of -10 dB). Figure 13 shows the characteristics of the output filter. The input circuit also has some selectivity; it attenuates -18 dB on 50MHz compared to 145 MHz, which may be useful for multi-band contests. The S-parameters of the preamplifier with the output filter and the -10 dB attenuator can be found in Figure 14.
Fig. 11: Stability Factor
Fig 12: S parameters without BP filter (Simulation)
The above results were finally achieved after more than 100 optimization processes, having found the best compromise between input matching and NF in each process. The noise figure can be further reduced to even 0.26 dB, but this makes no sense because the noise of the galaxy is much higher than this at 144 MHz. For this reason, the design gives more emphasis to input matching than to minimal NF.
I was not able to simulate the IP3 characteristics because the nonlinear circuit model of the ATF-53189 has only be published by AVAGO Technologies for the companys own (expensive!) ADS simulator . Despite this, I modified the gate and the drain decoupling circuits to increase the IP3 value. In the gate circuit C3 = 100 pF provides effective RF decoupling in the GHz range, C4 = 1 nF in VHF band and C5 = 100 nF for frequencies below a few MHz. In the drain circuit C6, C8 and C9 are responsible for the same functions. The closer together the two IMD test frequencies F1 and F2 are, the lower frequency of the 2nd- order the IM product (F2 F1) will be. Therefore, as input test frequencies F1 and F2 come closer together, higher capacitance is needed to achieve best possible bypassing of this low frequency product.
I also modified R1 to 22 Ω, being the optimal value between the noise figure and damping the low-frequency resonance of the gate bias choke L1, and slightly modified the setup of the quiescent current resulting in more precise adjustments and even better temperature stability. The Schottky diodes D1-D5 (VISHAY Semiconductor BAS70-04) are for the protection of the FET. They have no significant influence on NF or IP3 but are well proven (and valuable!) in real-life operation.
At the end of all these detailed design considerations, the final circuit diagram for the EXTRA-2 REV6 144 MHz contest preamplifier can be seen in Figure 15.
The layout of the PCB can be found in Figure16. Dimensions are 72 x 34,5 mm which fits a standard tinplate box. Figure 17 shows a panel of milled PCBs and Figures 18 and 19 show the finished PCB with its milled notches for the coaxial connectors and the corners of the tinplate box.
Fig 16: The PCB
If the PCB has been purchased already assembled, skip to the next paragraph. The bare PCB as removed from the milled panel (Figure 17) must be cleaned up with a fine file where necessary. Solder the components on the PCB using solder paste and a hot air soldering instrument. For ESD protection it is very important that the FET must be the last component to be placed on the PCB.
Attach the 2 coaxial connectors to the drilled box using the upper screws only. Then fit the PCB into the box from the bottom side and solder the pins of the connectors. Solder the bottom layer of the PCB to the sides of the box at the points shown on Figure 18. A Weller PTC-8 soldering iron is recommended for this step. Now the lower screws of the coax connectors can be fitted. The box has been designed so that BNC, N-female or N-male connectors can be used. SMA connectors can only be fixed by soldering, because the holes are not at the right place.
Only two further wires need to be soldered. The bottom side jumper wire can be seen on Figure 18, and finally connect the feedthrough capacitor C21 to the pad labeled +12V with a short Teflon insulated wire.
Adjust the current limit of a 13.5 V power supply to 250 mA and connect the preamplifier to it. The original factory setting of the potentiometer P1 will require very little change to achieve the necessary ID = 135 mA value. In this case 135 mV must be measured between TP1 and TP2 points. The total current consumption of the preamplifier should be approximately 152 mA. The power supply voltage can be varied between +10 and 15 V, without influencing the operation.
After setting the DC conditions, potentiometer P2 must be set to maximum gain (turn fully clockwise). Tune the BP filter L3-L4, using a spectrum analyzer and a tracking generator, until the maximum of the selectivity curve is at 145 MHz. This requires 2 turns inward from the original factory settings of both cores.
Please note that the pre-assembled PCB (Figure 19) has been pre-tuned on the assumption that the board will be used inside the specified metal box. The box will move the preamp tuning upward by 0.5 MHz while the gain will rise by 0.8 dB, and the unboxed boards are pre-tuned to allow for this later change. Many hams like to use the unboxed PCB version of the preamp for modernizing their own transverters or outdated preamplifiers, so some further tuning may be necessary to allow for any difference in the screening environment. Finally, in absence of a calibrated noise source, the CT input trimmer capacitor must be adjusted to the middle of its range. This setting gives the lowest noise figure, and based on measurements, the gain decreases by only 02 dB compared to GMAX. The preamp does not require any further adjustment, except for the gain potentiometer P2 to a level that will maximize the dynamic range of your particular receiving system.
Part List in PDF
The ATF-53189 provides the best performance at frequencies above 500 MHz, which is why it is very difficult on 144 MHz to achieve both the best input match and the minimum NF together. In the case of minimal noise figure (0.3 dB) the input return loss is very unfavorable, as S11 is only -2 dB and the input selectivity is virtually zero. The input return loss of the REV6 version designed with the necessary compromises can be seen in Figure 20. The measurement of S22 in Figure 21 shows the good the good impedance matching of the output filter. The maximal and minimal gain curves can be seen in Figure 22. The 1 dB bandwidth of the preamp is shown in Figure 23 and the 3 dB bandwidth in Figure 24. These figures show the excellent selectivity and suppression of broadcast frequencies. This is provided not only by the output BP filter, but also the input matching circuit contributes to it. Figure 25 shows the transfer function of the EXTRA-2 REV6 preamplifier between 20 and 500 MHz. The measurement results of the S parameters coincide with the simulation curves shown in Figure 14. The Third Order Intercept measurement of the preamp can be seen in Figure 26 (at maximum gain).
Fig. 20: Input Return Loss
Fig 21: Output Return Loss
The measurement of the noise figure is shown in Figure 27. This figure shows the measurement results of one of the noisiest examples tested, including the loss of the input relay K1 (approximately 0.15 dB). Figure 28 shows a possible wiring mode of the EXTRA series preamplifiers in the antenna relay box.
I have tested approximately 30 pieces of the EXTRA-2 REV6 preamplifier and the results are nearly the same. The circuit is stable and easily reproducible. About 20 pieces of the preamp have been tested by contest stations in IARU Region I VHF Contest, and the experiences have proved the measurement results and the simulation. It can be stated that the high-level behavior and the selectivity of the preamp is outstanding in real-life situations, and the noise figure is certainly low enough.
It was these positive experiences with the 144 MHz version that encouraged me also to develop the 50 MHz and 432 MHz versions of the preamp. Figure 29 shows the EXTRA-6 50 MHz contest preamp, and Figure 30 the EXTRA-70 for 432 MHz. On this latter frequency it is possible to achieve both an optimum impedance match into the FET at the same time as a very low noise figure of 0.38 dB.
Fig 29: The EXTRA-6 50 MHz contest preamp
I provide continuous information about EXTRA contest preamplifiers on my web site (www.HA8ET.hu) to everybody who may be interested.
I would like to say many thanks again to the Hams who sent numerous supporting e-mails, and for sharing their practical experiences from using the EXTRA-2 preamp.
1. Dipl. Ing. Gyula Nagy, HA8ET: EXTRA-2 144 MHz Contest Preamplifier. DUBUS 4/2010 p.78
2. ATF-53189 Datasheet. AVAGO Technologies (www.avagotech.com)
© HA8ET 2012