Can someone explain how a pure sine wave inverter makes a pure sine wave?

Thread Starter

LMF5000

Joined Oct 25, 2017
130
I've spent a few days searching with google, but all I can dig up are poorly written articles about the difference between modified sine wave (MSW) and pure sine wave (PSW) inverters, without going into the actual electronics.

Basically, I'd like to know how PSW inverters actually generate a sine wave at their output.

I assume that the conversion starts the same way as for a modified sine wave inverter - i.e. battery voltage (12V) is chopped into a ~30kHz square wave and fed to a transformer to boost it to >300VAC (@ 30kHz), then it's rectified and fed to the output stage.

It's my understanding that at this stage a MSW inverter uses transistors to turn the ~340VDC into a 50Hz, 240V DC (RMS) quasi-square wave that sort of approximates a sine wave (aka a "modified sine wave") - by simply switching on and off the transistors at the right intervals.

But how does a pure sine wave inverter do it? Does it use PWM of the transistors and a capacitor across the input? Or some kind of magnetic component to smooth out the waveform?

Could you have a signal-generating IC inside the inverter that makes a perfect sine wave as a reference, then use a comparator and voltage divider to compare inverter's actual output with the sine wave reference and adjust PWM to maintain a perfect sine wave regardless of load?

Could you add electronics to a MSW inverter, or something to the output, to make it PSW?
 

AnalogKid

Joined Aug 1, 2013
10,986
I have not torn into a modern sine wave inverter, but my guess is that your explanation adds a couple of steps they skip. If you are going to chop the battery voltage (12 V, 24 V, whatever) into a transformer, why not chop it into the output transformer? The 12 V powers a sinewave oscillator modulating a PWM stage, driving the transformer. With a 28.3:1 turns ratio, 12 V peak > 340 V peak. Of course, a real one is a bit more complex that that, but them's the basics.

ak
 

Alec_t

Joined Sep 17, 2013
14,280
Does it use PWM of the transistors and a capacitor across the input? Or some kind of magnetic component to smooth out the waveform?
Essentially, yes, but it's the output to the load which is smoothed. Often the load is a motor and the motor's inductance does the smoothing.
 

Thread Starter

LMF5000

Joined Oct 25, 2017
130
I have not torn into a modern sine wave inverter, but my guess is that your explanation adds a couple of steps they skip. If you are going to chop the battery voltage (12 V, 24 V, whatever) into a transformer, why not chop it into the output transformer? The 12 V powers a sinewave oscillator modulating a PWM stage, driving the transformer. With a 28.3:1 turns ratio, 12 V peak > 340 V peak. Of course, a real one is a bit more complex that that, but them's the basics.

ak
Because the output is at 50Hz. A transformer that can handle 1500W at 50Hz will weigh 8-12kg and be the size of a car battery. The way I described is the way it's done in MSW inverters - since the transformer operates at 30,000Hz, it weighs a few dozen grams and is about the size of a bottle cap.
 

Thread Starter

LMF5000

Joined Oct 25, 2017
130
Essentially, yes, but it's the output to the load which is smoothed. Often the load is a motor and the motor's inductance does the smoothing.
So they just add PWM to the transistor output? If it's that simple why does that make a PSW inverter 2-4x the price of a MSW inverter? And if the load provides the smoothing how do they get such nice sine waves on oscilloscopes when they run them with no load?
 

Thread Starter

LMF5000

Joined Oct 25, 2017
130
Thanks for that - I actually stumbled on that datasheet by accident earlier today, and it did help me gain a good understanding of what's involved, but they recommend a transformer at the output stage but none of the PSW inverters I've seen disassembled have had a large enough transformer (compared to their advertised power) to be using that topology.

Eventually I saw this video which includes a teardown and a block diagram - that's when it really "clicked" for me -
 

MrAl

Joined Jun 17, 2014
11,389
I've spent a few days searching with google, but all I can dig up are poorly written articles about the difference between modified sine wave (MSW) and pure sine wave (PSW) inverters, without going into the actual electronics.

Basically, I'd like to know how PSW inverters actually generate a sine wave at their output.

I assume that the conversion starts the same way as for a modified sine wave inverter - i.e. battery voltage (12V) is chopped into a ~30kHz square wave and fed to a transformer to boost it to >300VAC (@ 30kHz), then it's rectified and fed to the output stage.

It's my understanding that at this stage a MSW inverter uses transistors to turn the ~340VDC into a 50Hz, 240V DC (RMS) quasi-square wave that sort of approximates a sine wave (aka a "modified sine wave") - by simply switching on and off the transistors at the right intervals.

But how does a pure sine wave inverter do it? Does it use PWM of the transistors and a capacitor across the input? Or some kind of magnetic component to smooth out the waveform?

Could you have a signal-generating IC inside the inverter that makes a perfect sine wave as a reference, then use a comparator and voltage divider to compare inverter's actual output with the sine wave reference and adjust PWM to maintain a perfect sine wave regardless of load?

Could you add electronics to a MSW inverter, or something to the output, to make it PSW?

Hello,

There are a number of ways to generate a sine PWM and that is the start of most modern converters today. in the 1980's i worked on designs of low THD sine converters on a regular basis.
I will describe a few of the techniques.

The first, probably the oldest, is to generate a higher frequency triangle wave and a sine wave at the signal level. Now whereever the triangle intersects the sine, that is where the transistor will switch on and then off later. That gives you the basic PWM you need to get a sine.
This PWM goes to a transistor bridge that does the switching of the DC buss. The output is then filtered with whatever they want to use like an inductor and capacitor, or no filtering.

Another way is to store the sine table into a microcontroller chip. The microcontroller chip then outputs the sine pattern which is used to switch the transistor bridge transistors. This is a modern approach.
The sine table values are used to switch the transistors on and off in order to generate the right pattern.
The feedback varies all the pulse widths so that the output can be regulated.
Some form of filtering may be used on the output as well.
The main idea in the pattern is to use the sine values to indicate how wide the pulse should be at that particular phase angle, relative to the pulse width at the peak of the sine wave. So if the width was 1.000ms at the peak then at the angle of 45 degrees the width would be 1.000ms times sin(45 deg) equals 0.707ms.

Now in more advanced units like the ones we used to manufacture, you take into consideration the DC current that ends up in the primary of the transformer because of transistor CE voltage differences. The pattern is adjusted to adopt to this and thus the DC current goes very very low.
Also, sometimes the plant is considered along with the design of the PWM pattern to take into account various non idealities of the output filter and possible transformer characteristics. The PWM is adjusted to account for these effects. Sometimes the feedback is used to accomplish this rather than a hard coded design, and that really helps to reduce THD because even if there is unequal loading across the 360 degree sine wave output the feedback will adjust the PWM widths to help keep the THD low.

Lastly, there are sensors placed at key points to detect abuse of the system such as a current sensor in the output circuit and sometimes current sensors in each leg of the transistor bridge. Of course input over and under voltage detection is also added.
 
Last edited:

AnalogKid

Joined Aug 1, 2013
10,986
none of the PSW inverters I've seen disassembled have had a large enough transformer (compared to their advertised power) to be using that topology.
What makes you think that?

Also, my vision of a bottle cap is 1.1" dia. and 0.25" thick. What I see in the video are two transformers each larger than a golf ball. I suspect they weigh more than two ounces.

ak
 
Last edited:
Do some searching on Class D audio amplifiers. That may shed some light. AS i understand it, you have basically a triagular wave and a sine wave fed to a comparitor that created a series of pulses that fit within the sine wave envelope. The pulses are at a high frequency. Those are "power filtered" at 60 Hz. the filter has to be handle the VA requirements.

The H-bridge stuff should come out of the wash, but that's another concept. It's how you get dual polarity from a single supply. It's a common topolgy for driving motors.

Quasi sine wave just has a digitally synthesized pulse widths. You might probably do the same with a fully synthesized PWM sine wave so you don't need the analog class-D analogy. The key is the filter.
 

Thread Starter

LMF5000

Joined Oct 25, 2017
130
Not if it is a switching power supply, which is what I described.

ak
You did not describe a switching power supply, your exact words were:

I have not torn into a modern sine wave inverter, but my guess is that your explanation adds a couple of steps they skip. If you are going to chop the battery voltage (12 V, 24 V, whatever) into a transformer, why not chop it into the output transformer? The 12 V powers a sinewave oscillator modulating a PWM stage, driving the transformer. With a 28.3:1 turns ratio, 12 V peak > 340 V peak. Of course, a real one is a bit more complex that that, but them's the basics.

ak
Which would require the transformer to operate at the output frequency, namely 50Hz -> meaning a beast of a transformer.

On the topic of switching power supplies, you should be aware that they work a little like the input stages of inverters - an SPS rectifies the input to 330VDC, chops it at circa 20kHz, feeds it into a small step-down transformer (small because it operates at the high frequency of the chopper), then rectifies and filters that - and there's your output.

I'd rather not derail the thread by discussing whether the transformer is the size of a golf ball or bottle cap - that was just an analogy. The point is that transformer size at constant power is inversely proportional to switching frequency.
 
Last edited:

Thread Starter

LMF5000

Joined Oct 25, 2017
130
Hello,

There are a number of ways to generate a sine PWM and that is the start of most modern converters today. in the 1980's i worked on designs of low THD sine converters on a regular basis.
I will describe a few of the techniques.
Thanks for that very interesting post! So if I understand you correctly, the sine table is a way of numerically storing a sine wave in a microcontroller or ASIC? I.e. just a list of the PWM duty cycles to output at different times? So for example, if the output frequency is 50Hz, the length of one sine wave (360°) is 1/50 = 0.02 seconds. So at, say t=0.007 seconds the controller output would be sin (0.007/0.02 * 360) = sin (126°) = 0.809 so the base PWM duty cycle to the output stage would be 80.9% at that moment. Is that about right (ignoring any trimming due to feedback)?

Your last paragraph was very interesting, in fact that's what I was about to ask. As a mechanical engineer I usually prefer using negative feedback to run things closed-loop rather than trying to model everything precisely and have the system run open-loop. Once the product is out in the wild, you never know what unusual loads the customer is going to throw at it, so it's great that you build some degree self-correction into the system.
 

AnalogKid

Joined Aug 1, 2013
10,986
a <signal> modulating a PWM stage
is the heart of most lower-cost switching power supplies.

I thought you knew what PWM stands for, and how it applies to the context of this thread. My error.

With a pulse-width modulated squarewave signal driving the transformer primary (through saturated switching power transistors), the transformer never "sees" the 50 Hz component of the modulated waveform.

ak
 
Top