Number 16 on your Feedbackcard ATV Trans a Microwave Oven! Low-cost high-power microwave operation has arrived. by David Pacholok KASBYI netic field having a secondary effect on this. I First, I modified the magnetron cavi- ty to couple RF to a transmission line instead of to the oven compartment. I removed the interior radomelsplatter cover, field stirrer blades, and mag- netron output matching section. Next, I shorted the waveguide open end with a plate (Photo A) and installed an E-field probe to couple the RF to an N-connec- tor output jack. (Photo B shows the details of the construction of the E-field probe.) Magnetron current, voltage, and fre- quency were measured and plotted in- dependently to quantify performance in he majority of the amateur spec- this modified cavity. In power output T trum allocation lies above 1300 vs. cathode current measurements, for MHz, yet when you scan those bands, a power out range of 50-400 Watts, and you rarely hear anything but band a cathode current of 50-250 mA, I noise. Hams have let these regions lay found a very linear relationship. See fallow because of the idea that mi- Photo A. Waveguide shorting plate, to prevent the mi- Figure 1 for the frequency vs. current crowave equipment is complex, expen- crowave RFffom entering the cooking chamber, and to curve. This data suggests that: sive, or just unavailable. reflect this energy back to an E-fieldprobe. 1. The 2M189A magnetron is a cur- To be sure, there are concepts unique I rent-operated device. The anode-to- to microwave design, but they are nit neces- cathode voltage changes only about 1 per- sarily harder to grasp than those in lower cent, with a 2: 1 change in cathode current Ik. frequency RF design. And, as microwave 2. Power output is a linear function of Ik. applications find a larger place in society, as 3. Output frequency is a non-linear (but with ovens, and satellite TV, affordability monotonic) function of Ik, with increased and availability of surplus microwave equip- current causing an operating frequency in- ment constantly increases. crease. The average frequency "pushing" coefficient is about 0.1 MHzJm.4, with a use- Project Features ful frequency swing of about 20 MHz. The goal for this project was to provide an inexpensive, relatively simple high power What Mode To Use? microwave transmitter using a microwave The above conclusions ruled out AM oven as the foundation. This project meets the double-sideband video, because of the large following goals: incidental FM that would result. On the other *Low cost-less than $200. hand, an FM deviation of 2 MHz would cause *High power output-250 Watts minimum. Figure I. Graph showing frequency versus Ik incidental AM of only 15-20 percent, so I *Parts readily available from consumer elec- for the magnetron. This shows that output investigated wideband FM video trans- tronic supply houses. ffequency is (non-linearly) related to current mission. *Emission type compatible with standard to the magnetron. To check compatibility with existing TV low-cost BIW television receivers. receivers, I used an FM video-modulated sig- *Frequency of emission in the 2390-2450 *The basic transmitter scheme is adaptable to nal generator as a signal source for an MDS MHz amateur band, compatible with Multi- other emission modes, such as narrowband downconverter and a 5-inch monochrome re- point Distribution System (MDS) TV down- FM, with phase-lock circuitry described ceiver. I got a fair quality picture with the converters. (Historically, these downcon- below. television adjusted for IF slope detection, and verters have been misused to "pirate" with sync and vertical lock achieved at devia- television movie distribution at 2156 and Modification Description tions of 700 kHz to 3 .O MHz. The best picture 2162 MHz. They have been widely sold A microwave oven magnetron is a self- quality occurred at 2.2 MHz deviation. through magazine advertisements and elec- contained, crossed-field power oscillator. tronic flea markets, so there are tens of thou- Built-in cavities primarily determine oscilla- Modulator Circuit Description sands of them in existence.) tion frequency, with anode voltage and mag- The modulator serves two purposes. First, 54 73AmateurRadio July, 1989 it is a high-voltage current source with high ed-grid voltage amplifier with a current gain A floating screen supply of about + 100 open-loop gain, setting the magnetron cur- of unity, with enough voltage capability to volts is provided, with R28 included to limit rent to a known value, and establishing a drive the magnetron. However, to an input screen dissipation. The floating supply al- frequency and power output. See Figure 2. voltage at U1, a transconductance amplifier lows only plate current (magnetron current) U2, a 7805 5-volt regulator, establishes a is formed, with transconductance given by: to be included in the control loop. Additional reference voltage adjusted by R5 and R6. components with functions are: This voltage is applied to the non-inverting .R3, R14, and R15, which prevent parasitic input of high-speed op amp U1, which drives oscillation in U1 and V1. source follower Q1. The output of Ql, plus *R12 and R13, which aid current sharing in R9 and R7, provide negative feedback to U1 Bandwidth of this amplifier must be suffi- Vla and Vlb. in the ratio 5.7: 1. At equilibrium, Ql's drain/ cient for the modulator's second purpose- source current produces a voltage drop across video modulation. This must be 4.5 MHz, if *D3, which protects Q1 in the case of V1 Rl1 that equals 5.7 times Ul 's non-inverting you want to include the audio subcarrier. arc-over. voltage. Frequency response measured with a current *Conventional power supply rectifiers, fil- Temporarily ignoring screen grid current, probe in the plate leads of V1 was down 4 dB ters and bleeders. plate current equals cathode current in V1 at 4.5 MHz. Adding C6 (1200 pF) provides a (a,b combined). Since Vl 's cathode current pole for this frequency, flattening the re- WaveguideICavity Operation I equals Ql's drain current, VD rises or falls sponse to beyond 6 MHz. C1 and R8 serve to The waveguide circuit is deceptively sim- until the V1 grid 1-to-cathode bias causes couple an external 4.5 MHz subcarrier gener- ple: The oven's TElo waveguide feed (from Ip = Ik = ID = IS. V1 is therefore a ground- ator to the modulator. tube to cavity) is shorted with a copper plate. Figure 2. Transmitter schematic. 56 73Arnateur Radio July, 1989 Figure 3. NBFMphase-lock system schematic. 58 73AmateurRadio July, 1989 (See Photo A). This is analogous to of RG-174 that comes off pin 6 of a coaxial or microstrip short, the LF357 IC. Attach this to pin 2 where wavefronts are reflected of U1 in the transmitter circuit back with a 180 degree phase in- (Figure 2). Before doing this, how- version. At a quarter guide wave- ever, be sure to remove the 4.5 length from the short: MHz audio subcarrier at R8, and A the video input. Ag'd(~l~c)2- 1 You have now converted the mi- crowave oven transmitter to use where Ac = 2X guide broadwall with NBFM (t-5 kHz) voice mode! dimension. Now adjust the magnetron cavity The reflection is in phase with probe length and R6 until the mag- the incident wave from the mag- netron locks up at all times during netron, and an E-field probe (see the magnetron anode warmup (5-7 Photo B) is inserted at this voltage minutes). \ maximum. Ordinarily, maximum power transfer occurs when this Transmitter Improvements probe is A14 in length. Deliberately for NBFM shortening the probe introduces a .Bypass D5 and D6-Dl7 with reactive mismatch at the magnetron output wire this circuit into the transmitter unit. 0.0005 to 0.001 pF 3kV minimumcaps. This port. After an unknown number of degrees Refer to Section A on the schematic-the reduces "hum bars" in the picture and low- rotation within the feed structure (Matsushita overtone VXO circuit. The entire unit should level audio buzz in the NBFM mode. would not provide tube data), this causes the be temperature controlled at 70°C by "crys- *Isolate the metal case of the 0.74 pF (on TI) magnetron to be pulled lower in frequency by tal ovens," or something similar. The oscil- capacitor from ground with plastic blocks, some 25 MHz from its design frequency, lator drifts at around 100 Hz per degree, nylon screws, or other means. This will also ensuring legal amateur band operation. causing about 1.6 kHz per degree for the reduce hum bars and buzz. frequency out drift. Stability is traded off for *Using insulated standoffs, isolate T1 lami- Floating Operation simplicity in this design. nations, and frame from ground. This will One important feature of this conversion is Refer to the 151.85 MHz crystal in the further reduce hum bars and buzz, and will the modification of the high voltage power VXO circuit. Choose this crystal after you result in better insulation in TI after mods. supply for floating operation. The original build the oven video transmitter and measure *Disconnect the magnetron filament power transformer had one end of the sec- the stable operating frequency range using feedthrough from ground! Otherwise you ondary grounded to the frame. I lifted this end one UBP585 and a 600 MHz counter. won't get full video bandwidth, and the and attached it to a high-voltage lead wire. Now refer to the crystal oscillator tank NBFM mode PLL filter won't work (no This modification eliminates the need to float coils, to the upper right of the crystal on the phase margin).