Second version of the PWM dimmer and test results

In this article I will give updates on the PWM dimmer I explained in a previous article. I will explain the changes that have been made and present a new schematic. I will also quantitatively compare the performance of the PWM dimmer to the performance of a phase control dimmer.

New schematic and explanation

The new schematic is shown below:

Link to simulation (1Ω resistor and 500pf capacitor are only for the simulation, they are not part of the circuit)

I have made several changes to the design:

Gate drive circuit

The gate drive transformer was replaced with a handmade one made from a toroidal ferrite core and 9 turns of a single twisted pair from a CAT5 cable. An additional circuit was also inserted between the 555 timer and the gate drive transformer to amplify the current on the output. This circuit consists of an NPN and a PNP transistor, with both bases connected through separate resistors to the output of the 555 timer. The capacitors between the output and the bases help the transistors turn on and off quickly by compensating for the parasitic base-emitter capacitance of the transistors.

Filter circuit

To improve the efficiency and make the MOSFET run cooler when the dimmer is used with a fan, a filter circuit was added to the output of the dimmer. This circuit consists of an inductor, two capacitors, and a common mode choke. The inductor and larger capacitor act as a filter on the output of the dimmer circuit. The smaller capacitor acts as a snubber to absorb the energy in the inductor when the MOSFET turns off. However, this circuit does not work very well as the inductor can get hot enough to boil water during operation, and was measured to be over 150°C with an infrared thermometer.
A better design would have an second copy of the MOSFET/rectifier/gate drive circuit between the output and neutral connected to another winding on the GDT, but with the opposite polarity. This would cause the MOSFETS to switch alternately, meaning there is always a path for current to flow in or out of the output, but never a path for current to flow from live to neutral directly.
It is also possible that the core of the inductor is saturating, which causes it to heat up as it dissipates energy rather than storing it in its magnetic field. If this is happening, I would need to replace the filter inductors with appropriately sized ones.

Data collection

The current waveform, noise level, and rotational speed of the fan were measured, both with the PWM dimmer and the phase control dimmer. This would allow me to compare the performance of the two dimmers in a variety of metrics. Both dimmers were tested using a fan and an incandescent bulb. For both dimmers, the current was measured using a home-made current transformer consisting of a transformer from a power supply with a single turn of wire as the primary and the mains-voltage winding with a 316 ohm resistor soldered across it as the secondary.

Measurements

For each dimmer and each increment, I measured the following variables:

1. The RMS voltage on the current transformer. (fan and bulb)
2. The RMS voltage of the principal component of the output of the current transformer (fan and bulb). I used the FFT function of my scope to view the current waveform in the frequency domain and find the amplitude of the 60Hz componet of the current waveform.
3. The overall noise level (in dB) produced by the fan, measured at a distance of around 8 inches. (fan)
4. The frequency of the sound produced. This is directly proportional to the rotational speed of the fan. (fan)

The power factor is a measure of how much of the current consumed by the device is actually converted into useful work (closer to 1 = better power factor). It is also a measure of distorted the current waveform: if the waveform looks more like a sine wave, the power factor is higher. Power factor can be computed by dividing the RMS voltage of only the principal component of the current waveform by the total RMS voltage of the current waveform.

Phase control dimmer

I connected channel 2 of my oscilloscope to the output of the phase control dimmer and channel 1 to the output of the current transformer. With the dimmer set to maximum, I measured the following waveform on the output:

I aligned the waveform so the zero-crossing of the input voltage waveform was aligned with the left side of the screen. The turn-on edge was around 2 grid ticks, or 800μs from the zero crossing point. I made and recorded the measurements. Then, I adjusted the dimmer so the turn-on edge was 3 grid ticks from the left side, recorded the measurements, and repeated until the dimmer was at the lower limit of its range.

PWM dimmer

To get a stable and accurate measurement of the duty cycle of the PWM signal, I connected a filter consiting of a resistor and capacitor to the output of the 555 timer. Using the oscilloscope, I measured the maximum voltage of the output to be 10.5 volts. Thus, to convert the measured voltage on the capacitor to a duty cycle, one must divide the voltage by 10.5 volts. Measurements were made in 0.2 volt increments, starting at 2.9 volts and ending at 8.5 volts.

Analysis

I recorded all of the values collected in a spreadsheet and created various plots.

Noise level

The following plot has the measured frequency on the x-axis and the noise level of the fan on the y-axis. The blue line and red line are the phase control and PWM dimmer respectively. As the phase control dimmer produces more noise when running at the same speed, the PWM dimmer is quieter.

Efficiency of the dimmers

The following plot has the measured frequency on the x-axis and the RMS principal current on the y-axis. The principal current is directly proportional to the amount of real power drawn by the device. The blue line and red line are the phase control and PWM dimmer respectively. For low frequencies, the PWM dimmer draws more current than the phase control dimmer, which is likely explained by the heat generated by the inductor (>150°C) during operation. However, for a large interval of duty cycles (43-58%), the frequency is constant, but the current is not, resulting in a vertical line near 126Hz.

Power factor (separate graphs)

The following plot has the phase control dimmer setting on the x-axis and the power factor on the y-axis. It is obvious that the power factor decreases significantly below 1 (ideal) for lower settings of the dimmer.

The following plot has the PWM dimmer input duty cycle on the x-axis and the power factor on the y-axis. One can clearly see that the power factor is much closer to 1 over the entire range of the dimmer.

Power factor (fan, combined graph)

The following plot has the measured frequency on the x-axis and the power factor on the y-axis. The blue line and red line are the phase control and PWM dimmer respectively. Again, one can see that the power factor of the PWM dimmer is higher at the same frequency.

Data

Use this link to download the data.