I've been messing with AC regenerators. My problem is HF noise, mostly the intermodulation harmonics produced from the PWM inverter. Commercial "true sine wave" inverters (and two-stage UPS units) use an H-bridge to "chop" DC into AC. They all use the same control strategy. A sine wave oscillator feeds a voltage-to-PWM modulator, which then controls the H-bridge MOSFETs. If the output voltage is too high, the PWM modulator throttles back some (reduces pulse width). If the output voltage is too high, the PWM modulator add some, all while the oscillator is delivering the fundamental 60Hz sine wave. The resulting output isn't actually a "true" sine wave, but an approximation consisting of hundreds of tiny pulse-width modulated square-wave pieces. Each piece, being a square wave, has its own frequency and harmonic contribution to the overall output. Additionally, whenever a MOSFET turns on or off, there is a rising edge and falling edge, containing harmonics in the MHz range (commutation noise).
I've built some of these. My latest version uses a PFC (power factor correction) rectifier that chops the 120VAC line voltage into 300VDC. The HVDC (high voltage DC) is "pi" filtered with capacitors and choke. I used a demo kit from Texas Instruments for the PFC 300VDC rectifier, and some big chokes from Hammond. So the HVDC is "clean" (as measured with o-scope).
The 300VDC can be inverted to 120VAC using an H-bridge without a transformer. My first revision used an H-bridge and 11kHz PWM in closed-loop control mode. It had objectionable HF content, due to (1) the PWM harmonics, and (2) switching noise.
I put some AC filters after the H-bridge to absorb some of the switching noise. I also added some common-mode filters to capture the switching noise, which was in the high MHz range.
The second version operated open-loop, which means the output voltage was unregulated, and would fluctuate if the load changed. This was OK for me, because my loads are essentially constant, even my power amplifiers (Pass Zen, SET, Class A). The switching noise remained.
I'm still working on the third version. I've improved the snubber networks on the power electronics, and the MHz commutation noise is much better. However, the PWM harmonics are still a problem. I'm switching to a PIC microcontroller-based "Magic Sinewave" (google it). This technique dramatically reduces the harmonics produced by the individual PWM square wave pieces. My first try will be in open-loop mode (no voltage regulation), but as previously discussed, this doesn't present a problem for me.
One feature of the commercial high-end (expensive) AC Regenerators are "alternative" waveforms. I've looked into increasing the AC frequency of the output. This results in more peaks and more often, which reduces the power supply capacitor peak currents and voltage sag between peaks. A sneaky trick is to blend multiple sine waves of different frequencies to optimize the shape of the waveform peaks. Care must be taken so that the waveform peak doesn't exceed the 50Hz transformer saturation. The result would be broad, big peaks that are closer spaced together, but would pass through conventional 50Hz transformers.
I've built some of these. My latest version uses a PFC (power factor correction) rectifier that chops the 120VAC line voltage into 300VDC. The HVDC (high voltage DC) is "pi" filtered with capacitors and choke. I used a demo kit from Texas Instruments for the PFC 300VDC rectifier, and some big chokes from Hammond. So the HVDC is "clean" (as measured with o-scope).
The 300VDC can be inverted to 120VAC using an H-bridge without a transformer. My first revision used an H-bridge and 11kHz PWM in closed-loop control mode. It had objectionable HF content, due to (1) the PWM harmonics, and (2) switching noise.
I put some AC filters after the H-bridge to absorb some of the switching noise. I also added some common-mode filters to capture the switching noise, which was in the high MHz range.
The second version operated open-loop, which means the output voltage was unregulated, and would fluctuate if the load changed. This was OK for me, because my loads are essentially constant, even my power amplifiers (Pass Zen, SET, Class A). The switching noise remained.
I'm still working on the third version. I've improved the snubber networks on the power electronics, and the MHz commutation noise is much better. However, the PWM harmonics are still a problem. I'm switching to a PIC microcontroller-based "Magic Sinewave" (google it). This technique dramatically reduces the harmonics produced by the individual PWM square wave pieces. My first try will be in open-loop mode (no voltage regulation), but as previously discussed, this doesn't present a problem for me.
One feature of the commercial high-end (expensive) AC Regenerators are "alternative" waveforms. I've looked into increasing the AC frequency of the output. This results in more peaks and more often, which reduces the power supply capacitor peak currents and voltage sag between peaks. A sneaky trick is to blend multiple sine waves of different frequencies to optimize the shape of the waveform peaks. Care must be taken so that the waveform peak doesn't exceed the 50Hz transformer saturation. The result would be broad, big peaks that are closer spaced together, but would pass through conventional 50Hz transformers.
You guys do things the hard way, a device called a power conditioner is what you need, these were common in the 70s and 80s they are a transformer and capacitor network resonant at mains frequency with a Q of 5 or so so, they filter out mains distortion and also stabilise supply voltage. I threw one out last year. You should be able to pick one up for near scrap price if they haven't all been scrapped already.
If you had bothered to listen to what this device does to audio you would have thrown it many years ago. But thanks for the tip anyway
I could be a ground loop or some common mode voltage appearing on the conditioner output. or even stray magnetic field from the conditioner.
My static power conditioner gave a nice clean sine wave as opposed to the mains which had quite bit of 3rd harmonic distortion but it made no difference to the amplifier. I would be far more wary of PWM supply conditioners, high frequency switching noise can be quite insidious not to mention when they fail they might be difficult to repair just like UPS's which are becoming throw away devices.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.