MRF300AN 27MHz VSWR Protection and Recovery Strategy We are developing a 300W, 27MHz CW RF generator for medical aesthetic applications using the MRF300AN. The load (handpiece) can momentarily open-circuit, and we need a robust protection and recovery strategy. We are not looking for an open-load safe power level (we know the 65:1 VSWR ruggedness is pulsed). We are trying to optimize the system-level protection architecture and recovery timing to prevent CW overstress. What we have implemented so far: A three-toroid directional coupler samples forward and reflected power. The detected DC levels feed a comparator with hysteresis; when reflected power exceeds the threshold, a latch is set and a shunt FET rapidly pulls the MRF300AN gate bias toward zero. The MCU can reset the latch, and the MCU also has control over the small-signal RF input level. In other words, we have the basic detect, compare, latch, and gate-bias pull-down chain in place, and the MCU can reset the latch and adjust drive level. Our concern and the specific gaps we need help with: The current recovery sequence is essentially MCU resets the latch, gate bias returns, and drive is restored. We recognize that this is too symmetric: protection should be fast, but recovery should be deliberate. We are not sure whether our overall architecture is the right one for this application, and we are not sure how to design the recovery timing so that (1) the PA is not re-activated while the open-load condition is still present, and (2) the first re-entry attempt does not itself create a secondary overstress event. Q1: Architecture Is our comparator-latch-gate-pull-down approach the right primary protection for 27 MHz CW, or is there a better topology for this power class and application? We have seen references to several alternatives, and we would appreciate NXP's application-level judgment on which direction is most appropriate for MRF300AN at 27 MHz in a CW medical generator: - PIN-diode-based shunt or switch protection at the PA output, to present a controlled impedance or isolate the PA during fault conditions. - Circulator/isolator with a terminated load, as commonly used in MRI and other medical RF chains, accepting the cost and volume penalty. - Layered protection, where a very fast hardware path handles the sudden open-circuit event and a slower path (MCU or ALC) manages foldback and recovery. - Keeping the comparator + latch + gate-bias pull-down as the primary protection and adding a TVS clamp as a last-line voltage clamp. For MRF300AN specifically at 27 MHz CW, is there a reason to prefer or avoid any of these? We are particularly interested in whether a PIN-diode output switch would meaningfully reduce the stress on the device compared to gate-bias pull-down alone, or whether it introduces its own loss and reliability trade-offs at this frequency and power level. Q2: Recovery timing We understand that blindly resetting the latch after a fault can cause the PA to chatter at the protection threshold and repeatedly inject stress. We would like NXP's guidance on the holdoff and re-entry scheme: - Should the system enforce a fixed holdoff time after a fault before any re-entry attempt is allowed? - During re-entry, is it acceptable to use a short low-power probe (e.g., a brief low-drive burst) to check whether the load has returned to a safe VSWR, and then ramp the drive slowly rather than stepping back to full power? NXP community guidance already recommends gradual power-up rather than abrupt full drive during initial bring-up; we are trying to extend that principle to fault recovery. - Is there an NXP-recommended ramp profile or re-entry power ceiling for MRF300AN in CW operation, or should this be treated purely as a system-level state-machine design independent of the device? Q3: TVS role In the NXP 27 MHz reference design discussion, a 110 V TVS from drain to ground was discussed, with the guidance that the standoff voltage should sit above the maximum normal drain voltage. In our system, we already have active reflected-power protection. Should the TVS be treated as a last-line clamp for the brief interval before the active protection responds, or is it expected to carry more of the protection burden? At 27 MHz CW, do the junction capacitance and response speed of a TVS introduce any concern for normal RF waveform fidelity? What would be most helpful: Any application-level guidance, a recommended architecture, a pointer to an existing NXP application note on VSWR protection and recovery for LDMOS PAs, or even a brief statement of how NXP's own reference designs handle the fault-holdoff-re-entry sequence, would directly inform our design. Thank you. Re: MRF300AN 27MHz VSWR Protection and Recovery Strategy Hello @fengyuqi ,
Thank you for the detailed description. Your comparator, hardware latch and active gate-bias pull-down form a reasonable primary protection path. For a 300W CW application, we would recommend using this as part of a layered architecture rather than relying on a single mechanism.
Where possible, the fast hardware path should inhibit the RF drive and pull down the gate bias simultaneously. The MCU should not be part of the immediate shutdown loop; it should manage the subsequent holdoff, diagnostics, controlled re-entry, and repeated-fault lockout.
A PIN-diode switch could reduce stress only if it transfers the PA output to a known and adequately rated impedance. Simply disconnecting the handpiece may present another open-circuit condition. At this power level, insertion loss, diode heating, voltage rating, switching transients and the impedance presented during switching must all be considered. A circulator with a rated termination provides stronger load isolation, but with significant size, cost and loss trade-offs.
For recovery, a fixed minimum holdoff followed by a low-power probe and controlled drive ramp is a reasonable system-level approach. Both forward and reflected power should be monitored during the probe because the reflected-to-forward ratio may become inaccurate near the detector noise floor. Any new fault should immediately return the system to the latched state, and repeated unsuccessful attempts should result in a longer backoff or permanent lockout.
NXP does not specify a dedicated holdoff time, ramp profile, or re-entry power ceiling for the MRF300AN. These values depend on the protection-chain response, stored energy in the matching network, detector dynamic range, bias settling, thermal conditions and the expected load behavior. They must therefore be determined and validated at the system level, including testing at multiple mismatch phase angles.
The drain TVS should be considered a last-line transient clamp for the interval before the active shutdown becomes effective. It should not be expected to absorb continuous reflected RF power. Its standoff voltage, clamping behavior, pulse-energy capability, junction capacitance and package/layout parasitics must all be included in the RF and transient analysis, since the device can affect the 27 MHz output matching.
The published MRF300AN 27 MHz reference design does not define a qualified fault-holdoff-re-entry sequence, so the complete protection architecture remains an application-specific design responsibility.
Best regards,
Pavel
記事全体を表示
