The MFJ-998RT is an automatic remote antenna tuner rated at 1.5 kW. It is powered by 12VDC on the RF coaxial line. It is the remote version of the MFJ-998. Impedance conversion is done with an L network consisting of a series L and a shunt C. The shunt C can be placed before or after the inductor as required. The inductor consists of a series of binary weighted inductors that are shorted out by relays when not needed. The capacitor similarly consists of parallel connected binary weighted capacitors that are switched in and out of the circuit as needed (the ground side is switched by relays). A sampler combines a sample of the input RF voltage and current to determine the forward and reflected power at the input of the tuner. These samples go to analog inputs of the microcontroller which then adjusts the LC relays as required to minimize the SWR at the tuner input. In addition, an RF sample drives a microcontroller timer input to count the transmit frequency. If this frequency had been previously used, the LC settings previously used are set. If the input SWR is low with these settings, they are used. If the SWR is not low with these memorized settings, the tuner adjusts to find the optimum LC values.
The MFJ-998RT is mounted on a metal panel with a fiberglass weatherproof cover. Below is a photograph of the unit with the fiberglass cover removed.
Note that there is an internal display and switches similar to those on the MFJ-998. These are not used in normal operation. In addition, there is a loudspeaker that sends QRP or QRO if the input power is too high or low for the tuner to change its settings. While the tuner can handle 1.5 kW, the relays cannot "hot switch" this power. Instead, the tuner adjusts at low power. When the user sees the SWR on the transmission line is low as indicated on a meter in the shack, the power can be increased. The photograph below shows a closeup of the display.
On 30 November 2022, the tuner stopped working. This was during operation on 75 meters with about 500 watts. As shown in the photo below, the DC power tap failed. This circuit, not shown in the schematic below since it is in the RT version only, consists of the following parts:
The choke is between the incoming coax, which has RF and 12 VDC on it, to the DC input of the MFJ998. The capacitor and MOV are from the DC input to ground. This forms a low pass filter
to pass the DC while blocking the RF. As can be seen from the photograph, the capacitor vaporized and the choke overheated. It's possible that the choke overheated but was still good.
However, I replaced the capacitor and found that the tuner reported low reflected power but the meter in the shack still reported a high SWR. I then disconnected the choke from the incoming
coax connector and powered the tuner with an external 12 VDC supply. The tuner then worked properly, and the SWR meter in the shack reported a low SWR.
The image above shows the replacement parts received from MFJ. The capacitor is marked "103M 1 KV." The MOV is marked "P33Z1."
The schematic of the bias Tee is below. This shows just the choke and capacitor (not the MOV). The AC voltage source driving the circuit is the voltage for 1.5 kW into 50 ohms. If the
SWR is 2:1, double this voltage could be present.
The image below shows the RF voltage across the capacitor. With an SWR of 2:1, this voltage could be doubled. The graph shows an RF voltage of 2.158 volts at 1.8 MHz.
The image below shows the RF current through the capacitor. At 1.8 MHz, it is 244 mA.
The LTSPICE file for the schematic is here.
The MFJ 4117 Bias Tee in the shack was powered by switched 12 VDC from the SEA245 transceiver. From the SEA 245 power distribution schematic, we see that the +12VSW is fused with a 5A fuse. Should the capacitor or MOV short, the available DC current would exceed the 2A rating of the choke, leading to its overheating. The 998RT manual specifies a DC current of 1.4 A or less. After the repair, the MFJ 4117 Bias Tee will be powered by a current limited power supply.
In April, 2023, the tuner stopped working once again. This was in the middle of a 20 meter RTTY QSO at about 400 watts. Invesigation revealed that R121 had burned up, also melting the case of K1. Since the manual of the 998RT does not give a theory of operation, we have to guess at how the tuner operates. The manual states "When you key your transmitter, MFJ's InstantRecallTM checks its memory to see if you have operated on that frequency before. If so, tuning is instantaneous and you're ready to operate. If not, MFJ's IntelliTuneTM algorithm (based on MFJ's famous SWR Analyzer technology) kicks in. It measures the complex impedance of your antenna. Next it calculates the components needed and instantly snaps them in. Finally, it fine-tunes to minimize SWR, and you're ready to operate--all in a fraction of a second." However, the closest circuit we see in the schematic to an impedance measurement instrument is the SWR bridge formed by T1 (current transformer) and the voltage divider consisting of C169, VC1, and C171. The voltage and current samples are summed to yield forward and reflected power. These can determine SWR, but not the complex impedance. The complex impedance could be anywhere on the circle corresponding to the measured SWR on a Smith chart.
Guessing at how the tuner operates, I suspect that if the frequency is not already in memory, the tuner goes into bypass and measures the SWR of the antenna. It then adds R121 (25 ohms) in series with the input (which is connected directly to the antenna at this point) and determines if the SWR increases or decreases. If the SWR increases, the antenna impedance is higher than 50 ohms. If it decreases, the impedance is lower than 50 ohms. From this determination, it decides whether the shunt capacitor goes in before or after the inductor. K1 is then closed (shorting out R121), and the tuner "hunts" for the optimum component values for minimum input SWR.
It appears that mid-QSO, K1 opened putting high power on R121, burning it up and melting the case of K1. The first image shows the destroyed R121 (several resistors in parallel on a small PC board) and the melted K1
The second image shows the burned parts (top) and new replacement parts (bottom).
The third image shows the tuner with the replacement parts. R121 was mounted with long leads above the PCB so that if it were to fail again, it should not be near enough to K1 to melt it. The R121 assembly has a drop of RTV adhesive between the resistor assembly and K1 to keep it spaced away from K1 and the inductor on the other side.
Finally, disassembly of the tuner is interesting. I removed the shield above the microcontroller and associated circuitry, then the spacers below that and the other nuts holding down the PCB. The SO-239 connectors (including the input reflectometer), antenna insulator, ground wires, were unscrewed from the bottom panel. Then the PCB was lifted off the spacers. The relay pins were bent over on the PCB, making removal of the relay difficult. I finally bought an Hakko FR-301 desoldering tool to get the relay unsoldered.
Burned up R121 and melted K1 found on opening unit.
New and old R121 assembly and K1
New K1 and R121 installed. Leads were left long on R121 so damage to K1 would be minimal should R121 fail again. R121 spaced from K1 with blob of RTV.