50 hz, 4 kw square wave inverter filter design.

Thread Starter

COBBERS

Joined Oct 20, 2022
4
I have purchased a 4 kVA switchmode 12 vdc to 230 vac inverter which has a modified sine wave output (not seen the waveform yet) and I need to build a filter to reduce the interference and harmonic content. The inverter will be used to power the household including a few 1hp induction water pumps so the load will be variable. I am a retired rf design engineer but I don't know much about heavy current power filters. I live in rural Africa and will probably wind the inductors myself with air cores. I only have simple electronic tools, dvm and oscilloscope. I see on the forums that an Ott filter is used, I am not familiar with this type, or why it is recommended. Can anybody help me with a simple design.

many thanks,

Brian Cobern
 

Audioguru again

Joined Oct 21, 2019
5,420
I have made air-core speaker crossover filters at 2kHz/100W but your switchmode frequency is unknown and is 40 times the power. The resistance of your inductor will probably throw away much of the 4kW making heat.
 

Thread Starter

COBBERS

Joined Oct 20, 2022
4
Thanks Audioguru for your reply, I am aware of copper losses and ohmic losses are manageable with coil design and wire gauge.
 

Ian0

Joined Aug 7, 2020
6,715
The component values for a 4kW (13Ω) load calculate as 40uF, 186mH, 81uF, 41mH.
186mH @ 16A is not an inductor that can feasibly be implemented as air-cored.
Using a silicon-iron core it's going to weigh at least 10kg

Filter coefficients came from p.249 the 1967 GE SCR Manual
 

Hymie

Joined Mar 30, 2018
1,103
I recently purchased a 12V to 230V, 50Hz inverter – attached is a no load trace of the output waveform with a 500V/div vertical scale and 5ms/div time scale.

I would think the total harmonic distortion to be >50%, adding a filter to get the waveform anywhere near a sinewave would be near impossible. That said, I powered a small desk fan with the unit and it worked OK. I would suggest you connect the inverter to the water pumps you propose to power with it and see if it works.

Bear in mind that if the motors don’t start rotating, they will effectively be presenting a locked rotor (heavy) load to the inverter (best have a suitably rated fuse between the inverter and motor, just in case).

It might be worth adding a suitably rated MOV between Live & Neutral in an attempt to reduce the output voltage spikes (which are >1kV on my inverter under no load).
 

Attachments

Audioguru again

Joined Oct 21, 2019
5,420
The waveform from your inverter will not have the spikes with a load.
The modified sinewave has very few steps and appears to be completely wrong.:
1) It is narrow at 0V and wide at the voltage peaks which is wrong.
2) It should have the same pulse widths at 0V as at the peaks.
Look here:
 

Attachments

Ian0

Joined Aug 7, 2020
6,715
The waveform from your inverter will not have the spikes with a load.
The modified sinewave has very few steps and appears to be completely wrong.:
1) It is narrow at 0V and wide at the voltage peaks which is wrong.
2) It should have the same pulse widths at 0V as at the peaks.
Look here:
I've seen this circuit implemented in commercial products with a fixed step-up ratio from the battery voltage, to give a high-voltage bus voltage with varies with the battery voltage, then the rms voltage of the output is "regulated" by varying the mark-space ratio. Obviously it does nothing for the third-harmonic cancellation.
 

Ian0

Joined Aug 7, 2020
6,715
A question for my own interest,
if its a 4 Kw squar eish wave,
filtered off to a sine ish wave,

how much power is lost in the filter ?
Hypothetically, none at all as it is a LC filter. Practically - it depends on how much you want to invest in the inductors and how much space you can used.
Having looked at the Ott filter, adding it after an inverter with hf DC to DC section, would cost more, weigh more and be larger than using PWM and a 50Hz isolation transformer, where the transformer leakage inductance forms the L in an LC output filter.
 

Thread Starter

COBBERS

Joined Oct 20, 2022
4
The component values for a 4kW (13Ω) load calculate as 40uF, 186mH, 81uF, 41mH.
186mH @ 16A is not an inductor that can feasibly be implemented as air-cored.
Using a silicon-iron core it's going to weigh at least 10kg

Filter coefficients came from p.249 the 1967 GE SCR Manual
Thanks Hymie, my water pumps have electronic soft starters and its these that I am concerned about damaging. MOV's are a good idea and no electronic loads such as incandescent lights should be OK, but what about LED lights, they have electronics to reduce the voltage etc. would they be at risk?
 

crutschow

Joined Mar 14, 2008
31,129
I have purchased a 4 kVA switchmode 12 vdc to 230 vac inverter which has a modified sine wave output (not seen the waveform yet) and I need to build a filter to reduce the interference and harmonic content.
Once we see the waveform, we can give you a better idea about what you need (if anything) to filter the wave.
 

Ian0

Joined Aug 7, 2020
6,715
but what about LED lights, they have electronics to reduce the voltage etc. would they be at risk?
The first thing in the circuit is a rectifier followed by a smoothing capacitor, unless there is active power factor correction.
If it's the rectifier-capacitor version (<25W) then the capacitor will deal with most of the spikes, but the inverter might not like the capacitive load, because it will generate a big current spike at the start of the pulse.
If it's the active pfc version (>25W), then the first component is a filter, followed by a rectifier followed by an inductor which is part of a boost converter. That should deal with the spikes nicely, but how well it will behave when presented with a waveform that does't remotely resemble a sinewave is anybody's guess. My guess is that the least sophisticated PFC circuits will handle it better.
 
Top