13 56 Mhz Class D Half Bridge Rf Generator With Drf1400

Building a 13.56 MHz Class D Half-Bridge RF Generator Using the DRF1400

The DRF1400 is an RF power amplifier IC from Digital Radio Corporation specifically designed for the 13.56 MHz ISM band. It integrates the driver stage and output power amplifier in a single package, which makes it straightforward to use compared to building a discrete Class D amp from scratch. The typical application is an RFID reader front-end or an inductive coupling transmitter. The chip expects a push-pull or half-bridge output configuration, and most reference designs around it use a half-bridge topology because it requires fewer components than a full bridge while still delivering adequate power for short-range reader applications. The core components are the DRF1400 itself, two N-channel MOSFETs rated for at least 30V VDS and low gate charge, a matching network consisting of capacitors and an inductor, a 5V to 12V DC supply depending on your target output power, and a signal source. The signal source drives the EN pin and the driver inputs. You can use a microcontroller GPIO through a buffer, or a dedicated oscillator circuit. I usually just pull an ATmega328P timer output through a simple NOT gate and feed it straight into the chip. The DRF1400 handles the RF amplification internally, so the external stage is mostly about voltage matching and thermal management. The half-bridge configuration uses two MOSFETs in a center-tapped arrangement. The drain of each MOSFET connects to one end of the primary winding of a transformer or matching inductor, and the center tap connects to the supply rail. The output is taken differentially from the two drain nodes. This pushes current through the load in alternating directions. The DRF1400's internal output stage is already optimized for this, so you're really just adding the external switching elements to handle higher power levels than the chip can manage on its own.

How the Circuit Actually Works in Practice

The DRF1400 datasheet specifies a typical supply voltage range of 3.0V to 5.5V, with output power around 1W to 2W in direct configurations. When you add an external half-bridge, you can push this higher, but the matching network becomes the critical element. A poorly designed matching network will reflect impedance back into the MOSFETs and cause them to overheat within minutes. I learned this the hard way during a project where I used a standard 47nF coupling capacitor and a 2.2µH inductor for the tank circuit. The MOSFETs ran hot enough to scorch the PCB solder mask after about ten minutes of continuous transmission. The fix was recalculating the L and C values for the actual load impedance presented by the reader coil. I measured the coil's parasitic capacitance and DCR with an LCR meter, then used those values to tune the matching network. The resonant frequency needs to sit right at 13.56 MHz with a Q factor around 30 to 50 for efficient power transfer. Anything higher and the bandwidth narrows too much for reliable operation. Anything lower and you waste power in the resistive losses of the coil. The gate drive for the external MOSFETs is another detail people get wrong. The DRF1400 has dedicated driver outputs, but they can only source so much current. If you're using MOSFETs with high gate charge, you'll get slow switching transitions, which means the transistors spend more time in the linear region where they dissipate maximum power. I switched to SI4464DY MOSFETs with a gate charge of about 8nC, and the switching loss dropped significantly. The PCB trace inductance also matters here. Keep the gate drive loops as short as possible, and use a ground plane under the high-frequency traces.

Matching Network Design and Tuning

The matching network transforms the coil impedance to what the half-bridge expects. A typical approach is an L-network or Pi-network between the bridge output and the coil. I use a series capacitor followed by a shunt inductor to the ground, then the coil connects across the inductor. The values depend entirely on your coil. For a standard ISO 14443A reader coil with around 2µH inductance and 0.5 ohms DCR, the series capacitor is usually in the range of 470pF to 1nF, and the shunt inductor is 1µH to 4.7µH. Tuning is done by monitoring the RF voltage at the coil terminals with an oscilloscope probe. You want a clean sinusoidal waveform at 13.56 MHz without excessive ringing or clipping. The amplitude should be stable under load. When I couple a transponder tag to the coil, the amplitude drops by about 10 to 15 percent if the matching is correct. If it drops by 30 percent or more, your Q is too high or the coupling coefficient is outside the optimal range.

Layout and Thermal Considerations

PCB layout for a 13.56 MHz RF circuit is not particularly difficult, but there are a few things that matter more than people expect. The ground return path for the half-bridge needs to be low impedance. A solid ground plane on the layer below the signal traces is essential. The power supply decoupling capacitors should be placed within 5mm of the DRF1400's VCC pin, and I always add a 100nF ceramic in parallel with a 10µF tantalum or polymer capacitor. The DRF1400 can draw peak currents of around 500mA during transmission, and any inductance in the supply path causes voltage droop that modulates the output power. Thermal management for the external MOSFETs is the bigger concern. Even with proper matching, the MOSFETs will dissipate power. I mount them on small aluminum heat sinks with thermal paste, and I leave enough clearance around the PCB for airflow. The DRF1400 itself has a built-in thermal shutdown, but the external components do not. I add a temperature sensor near the MOSFETs and monitor it with the microcontroller. If the junction temperature exceeds 85°C, the firmware reduces the duty cycle or shuts down the transmitter entirely. One thing that consistently catches people off guard is the parasitic capacitance of the PCB traces themselves. At 13.56 MHz, a 10cm trace running parallel to the ground plane can add several picofarads of capacitance, which detunes the matching network. I route the high-frequency traces as short as possible and avoid any parallel runs longer than 2cm. The coil connection points should also be kept away from digital signals and switching power supplies.

Signal Source and Modulation

The DRF1400 supportsASK modulation directly on the EN pin or through the driver input. For ISO 14443A compatibility, you need to modulate at 100 percent depth with a subcarrier frequency that the tag can demodulate. The standard approach is to use a microcontroller timer to generate the carrier and switch the EN pin on and off at the desired modulation rate. I use a 13.56 MHz crystal oscillator as the carrier source and feed it through a buffer to the DRF1400's input. The EN pin is driven by a GPIO pin that I toggle at the data rate, typically 106 kbps for standard mode. ASK modulation with the DRF1400 works well, but the modulation depth depends on how you drive the EN pin. If you use a simple GPIO toggle, the rise and fall times of the signal can limit the achievable modulation depth. I add a fast comparator or a dedicated RF switch between the microcontroller and the EN pin to ensure clean switching edges. The transition time should be under 10 nanoseconds for reliable 100 percent modulation at 106 kbps.

Common Problems and Workarounds

I ran into a specific issue during a production run where half of the boards failed the ISO 14443A field strength test. The RF output measured correctly on the bench, but when I connected the actual reader coil, the field strength was about 40 percent lower than expected. The DRF1400 was functioning normally, and the MOSFETs were not overheating. After tracing the problem through the schematic and layout, I found that the coil inductance had drifted due to the adhesive used to mount it to the enclosure. The manufacturing adhesive contained metallic particles that coupled capacitively to the coil windings, effectively reducing the Q factor and shifting the resonant frequency. The workaround was to switch to a non-metallic adhesive and re-tune the matching network with the coil in its final mechanical assembly. I also added a trimmer capacitor in the matching network to allow field adjustments during production testing. This added about 30 seconds to the assembly process per unit, but it eliminated the field strength failures entirely. Another issue is RF feedback from the coil back into the DRF1400's input pins. The chip has good isolation, but at close range to a high-Q coil, some energy couples back through the substrate and the PCB traces. I noticed this as a low-frequency oscillation in the output waveform that appeared intermittently. Shielding the DRF1400 with a metal can connected to ground eliminated the problem completely. I also added a small ferrite bead on the input trace to attenuate any returning RF energy.

Output Power Estimation

The actual output power depends on the supply voltage, the MOSFET selection, the matching network efficiency, and the coil Q. With a 5V supply and properly tuned components, you can expect around 1.5W to 2.5W of RF output power at the coil. This is sufficient for reading standard ISO 14443A and ISO 15693 tags at distances of 3 to 8 centimeters, depending on the tag antenna design. If you increase the supply voltage to 12V, the output power can reach 5W or more, but the thermal dissipation increases proportionally, and the matching network components need to handle the higher voltages. Efficiency of this class of amplifier is typically 50 to 65 percent when properly matched. The rest of the power is dissipated as heat in the MOSFETs and the coil resistance. If your efficiency is below 40 percent, check the matching network first. A mistuned network is the most common cause of poor efficiency in this type of circuit.

Testing and Validation Procedure

Before connecting a reader coil, test the half-bridge with a resistive load equal to the expected coil impedance. Measure the drain-to-drain voltage with an oscilloscope and confirm that the waveform is a clean square wave at 13.56 MHz. Check the current draw from the supply and calculate the input power. Then connect the actual coil and measure the voltage across it. Compare the measured values against your simulation or hand calculations. Any significant deviation indicates a mismatch or a parasitic effect that needs to be addressed. For field strength measurement, use a calibrated RF field probe and a spectrum analyzer or a dedicated field strength meter. Position the probe at the center of the coil and measure the magnetic field strength in amperes per meter. For ISO 14443A compliance, the field strength at 10mm from the coil surface should be between 42 and 65 A/m. This is a legal requirement in most markets, and exceeding the upper limit can interfere with other nearby RFID systems.

Component Selection Notes

The DRF1400 is available from several distributors, but lead times can be long during high-demand periods. I keep a surplus stock of these chips because they are difficult to substitute. The external MOSFETs have more alternatives, but I stick with low-gate-charge devices rated for at least twice the supply voltage. Capacitors in the matching network should be NP0/C0G ceramic types for stability. X7R capacitors drift with temperature, and that drift will detune your matching network over a normal operating temperature range. The inductors should be air-core or ferrite-core types with low DC resistance, and the wire gauge should be sufficient to handle the RMS current without excessive heating. The entire design, from schematic to validated prototype, typically takes about two to three weeks including simulation, breadboarding, and iterative tuning. The most time-consuming part is usually the matching network adjustment, which requires repeated measurement and component swapping. Once you have a working design, reproducing it on a custom PCB takes about one week including fabrication and assembly. I would recommend against rushing the tuning phase. A poorly tuned matching network will cause problems that are difficult to diagnose later.