Review, teardown, and testing of RSP-320-24 Mean Well power supply

The RSP-320-24 is a 24-volt power supply with a maximum current of 13.4 amps. The supported mains voltage is from 100 to 240 volts without an additional switch. The supply measures 8.5 × 4.5 × 1¼ inches (215 × 115 × 30 millimeters), made on a printed circuit board fixed to the base's case. The top cover is perforated at the back near the connection terminals and on the front, where the cooling fan is installed. The fan starts spinning even if there's no electrical load. As the load increases, the fan speeds up, following the load current value. The fan sucks in the air and pushes it through the internal case volume to the perforated holes, including those on the side walls.

The input and output circuits are connected to a standard screw terminal block (1), from right to left: 3 terminals for the input line, neutral and ground wires, and 3 in parallel for common and +24V output. The input voltage from the terminals goes to the fuse (2), then to the pulse limiter (varistor), followed by the RF interference filter (3), and finally to the diode bridge (5). Next comes the active PFC, controlled by the PFC+PWM controller FAN4800 (4). Indeed, with a 234-volt AC power input, we get a rectified voltage on the storage capacitor (8) of 377 volts, approximately 47 volts more than without PFC Boost. The small voltage reserve is confusing since the capacitor installed is rated for 180 uF and 400 volts. All that's left is to rely on Nichicon's quality control.

The power part of the PFC is made of two parallel MOSFETs, IPP60R280P6 (7) and on an ultrafast diode 8A 600V STTH8S06D (6). The temperature sensor (11) is mounted above the PFC elements. The output voltage from the PFC is supplied to the two-transistor forward converter; the transistors are IPP60R280P6 (9) and are controlled by the same FAN4800 controller. The transformer (10) converter voltage is rectified and supplied to the LC filter. The output rectifier comprises eight diodes connected in two parallel groups (12). Total output capacitance: 2 pieces of 1000uF, 35V, designed for operating temperatures up to 220°F (105°C) (13). The output high-current circuits are reinforced with tinned copper busbars.

The control signal from the high-voltage side to the low-voltage side is transmitted through transistor optocouplers (there are two of them in the photo above the transformer hidden under a blob of the compound). One optocoupler is the primary regulation channel, and the second forms a backup channel for overvoltage protection, OVP.

The block diagram in the datasheet shows the "Active Inrush Current Limiting" node. Still, we could not find components on the board that could perform such a role. Still, the inrush current limitation element is present, marked as RTH1 on the board, and installed near the boost inductor PFC; most likely, this is an ordinary NTC.

The high-voltage part of the board, starting with the capacitor (8) and ending with the transformer leads (10), is coated on the high-voltage side with a protective composite, presumably epoxy-based, which further increases electrical safety.

There is additional insulation and a thin sheet of fiberglass between the aluminum case and the board (solder side).

The overall build quality is good.

Test conditions

Most tests use metering circuit #1 (see appendices) at 80°F (27°C), 70% relative humidity, and 29.8 inHg pressure. The measurements were performed without preheating the power supply with a short-term load unless mentioned otherwise.

The following values were used to determine the load level:

Output voltage under a constant load

The high stability of the output voltage should be noted.

Power-on parameters

Powering on at 100% load

The power supply is turned off at least 5 minutes before the test, with a 100% load connected. The oscillogram of switching to a 100% load is shown below (channel 1 is the output voltage, and channel 2 is the current consumption from the grid):

The picture shows three distinguishable phases of the power-on process:

1. The pulse of the input current charging the input capacitors when connected to the grid has an amplitude of about 2 A and a duration north of the mains voltage period.
2. Waiting for the power supply control circuit to start for about 300 ms.
3. (Output Voltage Rise Time) Starting the converter, increasing the output voltage, and entering the operating mode takes 8 ms.

(Turn On Delay Time) The entire process of entering the operating mode from the moment of powering on is 315 ms.

(Output Voltage Overshoot) The switching process is aperiodic; there is no overshoot.

Powering on at 0% load

The power supply is turned off at least 5 minutes before the test, with a 100% load connected. Then, the load is disconnected, and the power supply is switched on. The oscillogram of switching to a 0% load is shown below:

The picture shows three distinguishable phases of the power-on process:

1. The pulse of the input current charging the input capacitors when connected to the grid has an amplitude of about 2.2 A and a duration slightly longer than one mains period.
2. Waiting for the power supply control circuit to start for about 300 ms.
3. (Output Voltage Rise Time) Starting the converter, increasing the output voltage, and entering the operating mode takes 7 ms.

(Turn On Delay Time) The entire process of entering the operating mode from the moment of powering on is 320 ms.

(Output Voltage Overshoot) The switching process is aperiodic; there is no overshoot.

Power-off parameters

The power supply was turned off at 100% load, and the input voltage was nominal at the moment of powering off. The oscillogram of the shutdown process is shown below:

The picture shows two phases of the shutdown process:

1. (Shutdown Hold-Up Time) The power supply continues to operate due to the input capacitors holding charge until the voltage across them drops to a certain critical level, at which maintaining the output voltage at the nominal level becomes impossible. The phase takes 14 ms.
2. (Output Voltage Fall Time) Reduction of the output voltage, stopping voltage conversion, and accelerating the voltage drop takes 18 ms.

(Output Voltage Undershoot) The shutdown process is aperiodic; there is no overshoot.

Right before shutdown, the current waveform at 100% load is close to sinusoidal with an amplitude of 4.22 A.

Ripple output voltage

100% load

At 100% load, the low-frequency ripple is approximately 3 mV.

At 100% load, the ripple at the converter frequency is approximately 50 mVp-p, and the noise is 120 mVp-p.

75% load

At 75% load, the low-frequency ripple is approximately 3 mV.

At 75% load, the ripple at the converter frequency is approximately 40 mVp-p, and the noise is 120 mVp-p.

50% load

At a 50% load, the low-frequency ripple is approximately 3 mV.

At 50% load, the ripple at the converter frequency is approximately 30 mVp-p, and the noise is 70 mVp-p.

10% load

At 10% load, the low-frequency ripple is approximately 2 mV.

At a 10% load, the ripple at the converter frequency is approximately 20 mVp-p, and the noise is 90 mVp-p.

0% load

No-load current consumption measured with a multimeter: 53.5 mA.

(Power Consumption) The first assumption of excessive standby power draw of more than 6.5 watts is wrong since the current in this mode is predominantly reactive. Indeed, the input filter in the circuit contains two capacitors with a combined capacitance of 1.5 μF. Measuring the exact active power consumption at a 0% load with a basic set of instruments (oscilloscope, multimeter, etc.) is impossible.

At 10% load, the low-frequency ripple is approximately 2 mV.

At 10% load, ripples at the converter frequency are masked by the 90 mVp-p noise.

Dynamic characteristics

TA mode with periodic switching between 50% and 100% load was used to evaluate the dynamic characteristics. The process oscillogram is shown below:

It is evident that the supply’s response to loading changes is close to aperiodic, and there is no overshoot, which indicates a good stability margin. The magnitude of the response to load changes is just 60 mV.

Overload protection

The claimed protection type is "hiccup mode, recovers automatically after fault condition is removed." This was confirmed during testing. When a short circuit occurs, the power supply periodically tries to turn back on and, if the overload is still present, turns off again until the next attempt. This operating mode reduces energy losses and heating during overload. Still, it does not allow the parallel connection of multiple power supplies with a common output.

The output current for the overload protection to kick in is 17 A.

Input circuit safety assessment

(Input discharge) Safety assessment is based on the discharge time constant of the input circuits when disconnected from the grid; the value is 0.26 s. This means that when operating on a 120 V input voltage, the time required to discharge the input circuits to safe values (<42 V) will be 0.41 s:

Important: The result is valid for this particular power supply unit; it was obtained for testing purposes and should not be taken as a safety guarantee.

The leakage current at the ground pin is less than 10 µA.

Thermal conditions

When operating with no load connected, no component overheating had been noticed. Thermograms were captured at three power levels: 80, 90, and 100%, fully assembled and with the lid removed. Thermal images show that the most loaded element of the block is the input diode bridge, and its heating seriously stands out against the background of all the other components.

Unfortunately, already at 80% load, the diode bridge heats up to an unacceptable level of 259°F (126°C), which is dangerous for long-term operation.

80% load

90% load

100% load

Conclusions

Overall, the RSP-320-24 is well-built: this power supply has good dynamic characteristics, low noise, and ripple, good accuracy in maintaining the output voltage, and is well put together.
The load should be limited to 70–80% of the nominal for long-term operation.

Important: The results are valid for this particular power supply unit; they were obtained for testing purposes and should not be used to evaluate all the units of the same type.