If you are too lazy to hook up the PTT from the rig to a (homemade) brick RF amplifier and don't want to make a T/R sequencer, the above schematic diagram of RF VOX or actually a COR (Carrier Operated Relay) may be useful, though it uses "obsolete" components such as mechanical relays and bipolar transistors. The 6 pF RF-pickup capacitor's value may be reduced if used on higher bands such as 144 or 432 MHz.
As you know, the relays, amps and neighbourhood don't like hot-switching. Some of the high gain amplifiers may not be unconditionally stable (break in to oscillation when unterminated) and start to oscillate fiercely the millisecond the input or output is left with no load. Such termination during T/R-switching is difficult to arrange, but if one switches-off the amplifier's DC power (or in some cases just the bias), it can't oscillate (and blow up). The second poles in the input and output relays are used to make sure both amplifier input and output have been connected before DC is switched ON by a separate relay (or a FET switch). When transmission ends the DC feed-relay is released immediately with no extra delay (except for the obvious delay the RF VOX needs for SSB/CW operation), while the 10 to 100 uF electrolytic capacitor holds the RF input and RF output relays still on transmit position releasing them maybe some 200 milliseconds (you can hear it) later to bypass the amp for receiving. The value of the capacitor depends on the relays, the amount of delay desired and other components.
Use common sense when selecting the relays (must have 2 poles or aux contact, at least one of them!) and make sure the 2N2905s can supply their coils without being damaged due to exceeding their maximum allowed collector current. The DC relay should not be excessively slow or you may have to add the delay to make sure things work in proper sequence. Do the first testst with amplifier DC power from the relay unconnected and wil reduced RF drive, untill you know the VOX and the relays work as planned, or damage to the power amplifier may result! If you are afraid the relay wiring might pick up stray RF, use shielded cables and add some small (1 to 10 nF) ceramic RF bypass capacitors where ever needed. Better be safe than sorry.
The directivity achieved by trimmer resistor as terminator, was about 17 dB (at around 85 ohms). Insertion loss was negligible, 0.3 dB on 2 m (but closer to 1 dB on 70 cm). The circuit works through VHF and lower UHF, but it can not easily be calibrated for more than a single ham band. Power levels much above 50 W would call for shorter lines and wide gaps to reduce coupling and for higher bands the lines should be shorter. The RF connector's center pins are connected to the strip-line by short brass sheet jumpers. HP 5082-2800 Schottky diodes were used.
This circuit seems simple, looks good enough to dummy loads, but problems may arise when the antenna ports are replaced with real atennas, which have non-50 ohm random impedance outside their operating bands; HF antenna's impedance on 50 MHz and vice versa. Still, when properly tuned, SWRs only slighly change with real antennas connected. The biggest attenuation occurs on 28 MHz band, but is tolerable. No more switching antennas or working with wrong antenna with those HF&6m radios. Power handling capacity depends on when the trimmer caps flash over, may even take 100 W but not more. Besides, with linear amplifiers, this diplexer is connected between the rig and linear amps input connector(s).
Matched pair of 2SC2290MP was used instead of MRF454 due to price and availability. The amplifier worked when built with Amidon ferrites and 2SC2290 transistors per Granberg's application note, on all bands up to about 10 MHz, but higher up, the initial Pout was low - just 20 W on 28 MHz. To cure this the output toroidal transformer's (T3) primary parallel Mica capacitor bundle needed reduction, both 470 pF Micas were removed and replaced by one around 330 pF for best power on 28 MHz. Also slight changes in the T2 feedback loop's parallel caps increased the output power. As a last resort, the amp aimed to be driven by Icom IC-703 10 W radio, called for a 3 dB attenuator to be inserted at amp input. If you look carefully, you see the AN762 has holes and PCB traces for the pi-input attenuator, but as there would have been too little space for the resistors, I added a patch on the PCB, so the RF drive from the input T/R relay makes a longer route and the original attenuator space remained vacant. The attenuator was eventually bypassed with 820 pF capacitor to increase drive power on higher bands. The whole unit's bypass losses (also RX) were 0.5 dB on 14 MHz and 1 dB on 28 MHz.. The cabling and how you do it, affects, but also the ordinary PCB- type relays induce some losses.
IC-703 provides PTT output, which goes low during transmit. The necessary T/R relay drivers were built on separate PCB not shown here. That PCB also includes the SWR protection system's final end, from the sensitivity trimmer onwards. SCR triggers the T/R system when the SWR is too high and switches the T/R relays on bypass and +Vcc feed relay open. The reflected power sensor with a toroid, is better placed at the T3 secondary before the LPF switching relay RL3, if the LPF is selected manually and can be on wrong band by mistake. There is no extra space reserved for the SWR detector, but I used a small 2- sided PCB over the RL3 relay. The RF from T3 secondary gets looped via the detector and back to the amp PCB via short jumper. If the LPF switching is automatic, then the SWR detector can be installed on the coax that runs from T/R output relay to ANT connector. When the SWR protection (SCR) triggers, you need to switch off momentarily the amp from the ON/OFF switch to reset, or switch DC power off and on.
The LPFs uses 500 V Mica capacitors and Amidon toroid coils on both 5- element low pass pi-filters. The coils are made with typical toroid materials, the T50-0 may run little warm, so I would afterwards opt for the next larger toroid ring. The LPFs used here, were selected by the antenna in use; a 20-15-10 m tribander. This way and by using 2- pole relays, the design only has two LPF switching relays instead of 4 or 10. If a need arises to work on low HF bands, such as 7, 3.5 or 1.9 MHz, LPFs for those can be built on separate enclosures on antenna feeder cables. That way the LPFs are always "selected automatically" right, provided you switched on the right antenna and if you did not, the SWR protection will trigger. This scheme works, unless you have a 5- band trap-antenna, or something that is fed with just a single feeder on all bands.
Note: Ferrites I used in AN762: T1: Amidon BN 61-202, T2: Amidon
T 68-6 2pcs, T3: Amidon FT 50-61 14pcs, L3 and L4: Amidon T 68-6 one each,
L1 and L2: VK-200 RFCs, Q3: MJE3055 (TO-127), D1: BD135. (C5 and C6 capacitance
changed for best output on 28 MHz)
Use 1 mm copper wire or rivets on PCB feed-throughs and see that you get them all there. You may add some on the input and output ends of the PCB here and there, to make the ground planes work. Use adequate heat sink for proper cooling and preferably a copper bar plate as heat spreader under the transistors. Pay attention the PCB's stand-off's holes and those of the 2SC2290 flanges are positioned precisely. Test the bias circuit without the RF transistors first - you should get from about 0.7 volts up to little over 1 volt. When the unit is fully wired and PCB is secured on the heat sink, use some thermal conducting paste and bolt the 2SC2290s on, do not over tighten and use as thin layer of thermal paste as necessary. Last: solder the 2SC2290 leads on the PCB - avoid mechanical strain on the RF transistors. With current limited PSU, adjust the bias current to 200 mA and use reuced RF drive power when doing the initial tests and tweaking the amp. The unit I built, draws 21 A on low bands with 160 W output and little less on 14 MHz, but only about 11 A on 28 MHz where the achieved output in the end was about 90W.
Check the PCBs upper and lower side's patterns and holes matches - these
images are edited from a hand-made PCB's photos - scale right and verify!
The AN762 application note can be downloaded elsewhere on the net with detailed description of the design and component values.
Please observe the ferrite material availability and the alternate Amidon ferrites used here.
The VLA 150's built-in preamplifier had severe oscillation problems, it oscillated on it's working frequency 50 MHz and at parasitic around 2 GHz. The both oscillations occurred for separate reasons and could be initiated at different conditions. The 50 MHz oscillation may be due to bad PCB layout with RF- feedback or just too narrow band matching making it prone to oscillate. The 2 GHz parasitic is likely from PCB layout and construction around the MMIC, lack of vias, lack of use of SMD components. etc. Using a microwave component on low-VHF still needs the designer to take care of the circuit properties up to several GHz, so the active component works as planned. To get around these oscillation issues, the MMIC's output needed broad band loading to damp 50 MHz oscillation. Biasing was not ideal neither. It also needed some gain reduction on higher frequencies to stop the parasitic oscillations. The tricks tried and proved successful, were to replace the L1 choke inductor at MMIC's Vcc feed with 330 ohm resistor and as the voltage reduction resistor was originally 1200 ohms for some odd reason, got changed to 330 ohms also, so the device was now optimally biased as the MMIC's manufacturer Mini Circuits suggest in their MAR-6+ data sheet. The device has 22 dB gain, here just 13 dB now observed (impossible to measure before the mods. because of the oscillations!) but it's better so, less receiver overloading with strong signals. The noise figure of this MMIC device is typically 3 dB but as 50 MHz has strong galactic noise, this is still satisfactory value. The preamp will not improve S/N ratio, if you receiver has low-noise FET stage at input, but as it makes the S-meter to move up more easily with weaker signals, it could thus be helpful when peaking antenna heading. Bypass loss of VLA 150 linear amplifier was 0.25 dB. With 10 W drive and 14 Vdc, the power amplifier's carrier output was 125 W, much as advertised..
Output harmonics at 100 W, preamplifier response (G=13 dB), bypass frequency response.
Further comments: I would like to see with VLA 150: 1 m longer power cords, maybe some ferrite bead too, 2 more fixing screws on the base lid plate and foot pads to place under the amplifier.
A directional coupler, constructed of double-sided FR-4 PCB material
with strip-lines, is inserted in the amplifier's output coaxial line after
LPF filter. The coupler's directivity is peaked with the termination resistor
at the other end of the coupler strip-line
by running RF power through the coupler to a 50 ohm dummy load and varying the termination resistor's value to find minimum DC voltage output at BAT83 Schottky-diode's cathode (or at 50 k SWR sensitivity trimmer's hot end). The coupler
delivers DC-voltage proportional for reflected power, which is fed to a DC-coupled amplifier. This 3-stage transistor adjustable-gain amplifier simply pulls down the IRF510 power FET's bias voltage (at regulator's input), thus reducing the RF output significantly. The gate bias voltage regulator circuit shown here, has slightly different component values and some added parts from original OZ1PIF design. A 5.6 V zener is added to absorb ESD spikes, though a spike charge directly to FET gate will not be reduced (due to 33 k resistors) and should be avoided when handling the amplifier..
A high-SWR indicator LED with it's own BC547 driver provides visual
warning of antenna mismatch. The built unit reduces power (P.A. fed with
+28 Vdc) to 30 W when SWR is 1:2.3, but the reduction level may be altered
with Rx, the pull-down
series resistor. Control loop's gain created 130 Hz oscillation to bias voltage (effectively AM modualtion) which was solved with adjustable series R-C feedback in the DC amplifier (BC557). At SWR 1:2.3 the circuit shows about 2 ms delay, before the bias (and RF power output) has gone low. The mis-match conditions for adjusting the protection circuit, is easily created by connecting the RF power amplifier to a 50 ohm dummy load with a 1 m long RG-59 cable (1/4-wl. 75 ohm). But do not run full power now.
Set 50 k trimmer to max. sensitivity and 22 k damping trimmer to maximum resistance. Use scope or AM receiver to observe is there is oscillation on bias voltage when the RF is keyed on and SWR protection activates. Adjust 22 k damping trimmer to kill oscillation
and 50 k sensitivity to fold-back output (when the SWR is 1:2 witht the RG-59 test cable). Adjust/change Rx resistor to desired RF ouput level when SWR is bad, the lower the power output, the safer. Re-adjust 22 k and 50 k trimmers, they may interact.
Change to 50 ohm cable (now good SWR) and see the cicuit does not pull down the regulator input now. Few have 100 W- class mis-match load tester box, so this setting method is rather simplified. If you feel the protection kicks-in too eagerly at fair SWR values, reduce sensitivity some with the 50 k trimmer.
No warranty is given this SWR protection circuit saves the FETs on all cases of mismatch - still better than nothing. Tested at 120 W output and 28.7 Vdc with 3 diffent lenght open-ended coax - no damage. If an amplifier is ran at reduced voltage (ie. 28 V) and at 50% output power (125 W per module), the currents and voltages will be low enough not to cause FET damage. The SWR protection is not meant to be active under normal operation - if the LED lights up, fix the antenna's mismatch fault as soon as possible.
This circuit could be used for other FET RF amplifiers with similar bias voltage and possibly adopted for commercial RF pellet amplifiers with "mute-input pad".
The FET amplifier's input match was perfected with Pi- low pass filter just ahead the T65 material input ferrite transformer: 36 pF to ground, in series an 8-turn coil, ID 6 mm, 1 mm Cu wire tightwound, 43 pF to ground. SWR before was 1/1.6, now less than 1/1.2
Amplifier performance data on 50 MHz, after adding the input Pi-section
and changing the gate coupling capacitors to 150 pF from 180 pF and the
input ferrite material to 43 (BN-43-3312), which surisingly makes
it a thad better on 50 and 70 MHz.
Also the SWR was in the 1/1.35... 1/1.5 range 50 to 51 MHz with 43 ferrite material with out the Pi- filter match. The filter needs different capacitors, if the input is built with 43 ferrite, if left as for T65 the SWR is about 1/1.3:
Vsupp. 28.7 V, 5 W in, after LPF 66 W out
Vsupp. 28.7 V, 10 W in, after LPF 122 W out
Vsupp. 28.7 V, 13 W in, after LPF 140 W out
Vsupp. 28.7 V, 16 W in, after LPF 166 W out
Saturated at 255 W (after LPF).
Small-signal testing when biased to A-class, show about 3 dB less gain
on 70 MHz. 0 dB gain at 100 MHz.
Below is an alternate FET amplifier SWR protection circuit, the SWR
bridge is HF version, should work on 50 MHz too, but the circuit is
built for 28 MHz version of the amplifier with 6 FETs.
Use a sensitive 30 or 50 uA panel meter as power/SWR indicator, to avoid
loading of reflected sample DC voltage from SWR bridge,
or SWR protection may not activate, if meter switch is left on REFL position. 100 W to a dummy load, or same drive level - no atenna connected:
Pout was fold back to just 5 to 10 W byt the BS170 FET draining down the amplifier bias voltage.
Tune the filter for minimum loss at 50 MHz by stretching carefully the
coils (or use low power and insert brass or ferrite to see which direction
is better before chancing the coil's
length - the 2.5 mm solid strand enameled wire is rather stiff). The loss and SWR curve point has a dip, you need to make it come to 50 MHz. Antenna analyzer with the
filter connected to a dummy load is good way to peak it.
Use Unelco- type compressed micas or 50 V ATC HQ chips (always two in parallel). Ordinary ceramic capacitors fail with 100 W in a few seconds.
The enclosure used was big enough to have toroid type SWR bridge included. The space was split with inter wall of brass to reduce RF leak between input and output.
The OZ1PIF 250 W 50 MHz amplifier I built, has 4R7 ohms SMD series resistors on FET gates to prevent oscillation, as suggested and tried by OH7LMQ. The source of this, sometimes destructive oscillation, noted by some IRF510 (HF) amplifier builders not using the "gate series resistors", come from resonance of input transformer's leakage inductance (a stray inductance in series), gate circuitry lead inductances and from Feet's gate capacitance. The gate series resistor is just one of the methods to dampen this oscillation, sometimes called also as ringing, described on numerous articles and application notes on the web, but this is not the only method and usually two methods are used to achieve stability. From practical experiences with FET-damages in RF amplifiers, the common cause seems to be neglecting to add the gate series resistor on FET (IRF510) amplifiers. Some designs have them, some do not.
An other method is "gate shunt resistor". This circuit does not fully exist in OZ1PIF design, though there are 220 ohms in series with 33 nF from gates to ground. The 220 ohm resistor may be of too high value to provide good shunting effect. The Polyfet paper's amplifiers use 16 ohm resistors. If series and shunt resistors as used, both dissipate some of the drive RF-power and thus reduce amplifier gain. Some of this must be sacrificed for stability. The Polyfet article also shows series resistors near input transformer's secondary where they are bypassed with capacitors to prevent drive power & gain loss.
See about "Rg" in this paper:
This is a more elaborate approach on FET-power amplifier stability:
1997 MTT Presentation
Stabilizing Mosfet Amplifiers
Polyfet Rf Devices
S. K. Leong
The OZ1PIF amplifier does not use drain-gate feedback and it may not
even be necessary and useful, or a complete drain circuit Ferrite loading.
The L4 choke is there, but DC-feed to output transformer's center tap has
no ferrite (coil), should it be there?
According to the Polyfet paper, this would improve stability under moderate to high SWR conditions and would be very easy to accomplish.
© I. Yrjölä, 1999... 2013