How can I filter out an inverter for less noises

[gaetan8888]

Hello

A true-sinewave inverter cost too much for me. So I have a Xantrex XPower 1000 inverter.

This 1000 watt inverter do have a modified sinewave output.

Is it possible to filter the output voltage to bring it closer to a sine wave signal and have much less noise ?

And I did buy mosfets and diodes to double them in parallel on the inverter board, I will use a bigger heat-sink, so it may go a bit more than the 1000 watt, but I can not find any number on the two transformers.

Anybody would know what type of transformers it can be ?

I include photos of the insides of my Xantrex inverter, with some parts numbers on the first photo.

BTW, Xantrex do not reply to any parts request or technical question.

Thank

Bye

Gaetan

[star882]

Just add a LC filter to the output.

[gaetan8888]

Hello

Any LC filters I've seen in the web was not for a very variable load like for an inverter, the pole frequency cut was varying a lot with the load, and I would like to filter as close as possible to 60hz.

Any web page to suggest ?

Thank

Bye

Gaetan

[star882]

If all you want is a simple solution, just put some inductors in series with the lines. They'll smooth out the edges.
A more advanced solution involves downsizing the high voltage capacitors and modulating the duty cycle of the DC/DC converter (effectively soft starting every edge), but that will decrease the capacity of the inverter and is also tricky to implement correctly.

[Tahmid]

Hi Gaetan,
Just adding an L-C filter will not do. To make sine wave, you must modulate the PWM signal to the full bridge in the KHz range and then use a low-pass L-C filter to convert it to sine wave.
Present 50Hz modified sine wave can not be used for creating sine wave.
We can see there are 2 transformers. These are probably E type cores (EE/ETD/ECR). Could you please let us know the length and width of each transformer as well as how many pins each has. Then, it will be easier to sort out what type of cores they are.
Thanks.

[zigzagflux]

It would be next to impossible to use a passive output filter to generate a reasonably clean (~10% THD) sine wave from a stepped approximation.

Problem is the rms value of the stepped wave is 120V, and may or may not be regulated, really doesn't matter.

However, the FFT of this wave contains the fundamental at a much lower value than 120V. So even if you are successful with a lot of inductance and capacitance to clean up the wave, you'll end up with maybe 80V rms. I suppose you could then step this up with an autotransformer or something, but you're really not helping the cause from a cost and simplicity standpoint.

I've been trying various methods to generate a clean (0.1%) sine wave to power the filaments of my DHT's for a few months now, and in the end, I think the cheapest method is a sine wave oscillator (pick your circuit) followed by a class-D amplifier/transformer.

Last resort would be to run this inverter output into a ferroresonant transformer. Those babies work with square wave input, and produce a 3-5% THD output, regulated to boot. Harmonic spectrum is weighed toward the 3rd, 5th, 7th, 9th, and 11th, and not much over that. You might be able to locate one on ebay for reasonable cost. New they're a couple grand at 1000VA.

[jackinnj]

Decades ago we used McIntosh tube amps driven by an HP200 signal generator to derive ultra-clean power for the stuff I worked on as a lab-rat. Temperatures were near absolute zero, lots of dewars with liquid nitrogen, hydrogen and vacuum pumps. Noise couldn't be tolerated.

You will get some bandwidth limiting using an isolation transformer on the output of the inverter. Every hole in that inverter is letting out EMI, RFI so bypass all of the power leads into the inverter from the battery.

I would chuck the inverter. Get a couple hundred watt Class-D amplifier, a 60Hz source (you can use an XO in the form of a watch oscillator), divide down and use an 8th order eliptical filter from Maxim to generate a very stable, low THD sine wave. A transformer on the output of the amplifier will get you the a.c. voltage you need and you can put a feed back loop in to control the output .

[star882]

You can downsize the high voltage capacitors, then modulate the duty cycle of the DC/DC converter on every transition of the H bridge. That would reduce peak currents and harmonics. Increase the duty cycle of the H bridge to compensate for the reduced RMS voltage. Maybe also try rapidly modulating the frequency of the H bridge to spread the spectrum.

Note that it is mainly motors that care much about waveform and frequency. Switching power supplies work well on lightly filtered square waves or (for some supplies) DC. Linear power supplies work well with filtered square waves at higher than usual frequencies (to prevent saturation).

Those wanting to PWM an actual sine wave would be best off with DDS (Direct Digital Synthesis) implemented with a microcontroller. It should be easy to modify the inverter, although the H bridge drivers might have to be replaced to get the switching fast enough.

[golam]

Star882 is right. DDS with Micro controller is the easiest way. But not easier for those who don't know the modalities of Micro controller.

One can use pwm ic to convert low battery voltage to high voltage dc (like 350v dc for converting 220v sine wave) and then convert this high voltage dc to sine wave with Micro controller (I use Pic Micro ) and mosfet drivers through Full Bridge Topology and after wards passing through low pass filter. One can use hardware modalities of Micro controller or can use complete software route and in both cases, they require Lookup Table consists of Sine routine.
With Thanks.

[star882]

A DDS is simply a counter cycling through the values in a ROM with the output magnitude compared to another, much faster counter. It is possible to implement it in discrete logic but a microcontroller is much simpler.