Modifying Square Wave Inverters to Sine Wave Equivalents
Basically the working waveform of the modified sine wave power inverter circuit presented here has been carved by positioning many discrete square waves in succession. This structure replicates almost an actual sine wave and provides an efficiency that may probably exceed 90%.
Every electronic enthusiast will know how difficult it is to design an efficient sine wave inverter circuit. A sine wave being an exponentially varying waveform is difficult to optimize. Also, unless a PWM technique is employed, the transistors may dissipate too much of power in the form of heat, wasting precious battery power. A modified sine wave inverter is relatively easier to design and the cost involved in building one is also much cheaper.
The present circuit design incorporates a modified version that’s quite unique in its concept and may be understood even by an electronic novice. The idea has been exclusively designed by me, but has not been practically tested. Although it looks impeccable, I would definitely like to hear if anybody feels the design may not be practically successful.
Before going through the circuit details, it would be important to asses a typical modified sine wave structure derived from multiple square waves.
Modifying Sine Wave Using Discrete Square Waves
Referring to the figure alongside (click to enlarge) we can see an interesting design of a single modified sine wave cycle made by chopping a few square waves. Here, each positive and negative half cycle contain 3 discrete individual narrow square waves, each block is separated by a notch, the center two “pillars" are identical but are twice in magnitude than the extreme ones.
The average value of this special arrangement of discrete square waves effectively imitates a sinusoidal wave. This configuration is as good as a pure sine AC waveform and thus will be suitable to operate almost all appliances safely (my assumption).
In fact the present design is much more efficient than the usual circuits shown on many websites. From this circuit it’s possible to get an efficiency of almost 90%, because here the output devices are either turned fully ON or fully off.
Let’s assess how the circuit actually functions.
You will require the following parts for building this modified sine wave inverter:
R1, R2, R3 = 10K, 1/4 Watt, CFR 5% (Approximate Value),
R4-----R7 = 1K, 1Watt,
D1----D4 = 1N4148,
D5----D16 = 1N4007,
D17, D18 = 1N5408,
T1, T3, T5, T7 = TIP122,
T2, T4, T6, T8 = BDY29,
IC1, 2 = 4017,
General Purpose Board, Heatsink, Connecting Wires etc.
Transformer = As per requirement, (Please modify the base resistors accordingly).
Battery = As per the transformer used.
To create the particular waveform explained in the previous page for this modified sine wave power inverter, first of all obviously we need a square wave generator.
Primarily, we need to break each square wave into pieces so that they are equally spaced with notches and also to make them grow in size gradually from start to the center and reduce as it ends.
Ideally we may think of a PWM IC for this, but the condition is also very effectively met using a couple of Johnson decade counters. The circuit may be understood as follows:
We know that the IC 4017 produces a continuously shifting sequential logic high pulse across its 10 outputs in the order: 3-2-4-7-10-5-6-9-11.
Here, two 4017 ICs are cascaded to provide the above sequence in 20 continuous steps. D1---D4 are appropriately configured to make both the ICs conduct in tandem.
If you inspect the outputs of each IC carefully, you will find that a few of the alternate pin-out sequences are cleverly skipped, while the remaining are joined together through diodes.
The diode junction goes to the base of the power transistors. These transistors respond and conduct exactly according to the switched pin-out sequences and remain shut-off intermediately due to the skipped pin-outs, producing pulses exactly as shown in the diagram.
The above DC pulses are forced through the windings of the transformer where it is stepped up to the required primary voltage level.
The frequency of the AC will particularly depend on the input clock pulses applied to the ICs.
Since the outputs shift in response to every rising edge of the input clock, one complete sequence from the start of IC1 (pin #2) to the end of the IC2 (pin #9) constitutes a single AC pulse. However, since one complete sequence should happen 50 times (for 50Hz) per second, implies that the input clock should have a frequency of 50 × 20 = 1000 Hz or 1 KHz or 1.2 KHz to get 60 Hz (for 120 volt Outputs).
The circuit incorporates a simple oscillator circuit using IC 4049, however other standard oscillator configurations may also be tried. For example a simple oscillator circuit using IC 4060 as discussed HERE may be employed for driving the above circuit.
One avid reader Ali Rexhepi did some very interesting practical investigations with my circuit of this modified sine wave power inverter. The following pictures and data reveal the rigorous effort and analysis done by him to improve the results.
The following e-mail was received by me from Ali Rexhepi:
Subject: wave forms
Date: Sun, 21 Nov 2010 17:40:23 -0500
Swagatam here are those screen shots.There are four pics, two of the wave forms PWM at 59.760 Hz are at diode junctions IC1 & IC2. The third one is of the square wave clock input at 1.067 KHz .The difference between the upper and lower pulses of the clock input is 51 ms. I was actually able to improve on the oscillator circuit by trial and error. The one you posted had difference in the upper and lower pulses of 134 ms. I used 150 Ohm & 5.66 KOhms for the two resistors and C1 used 0.1 uF Cap.The last one is of the complete circuit on a proto board, don't mind my lousy breadboarding techniques, lol! Thanks for all your help, this was an ingenious design. I am going to use power Mosfets for the power circuit, is that OK. They are SUP90N07 from Vishay continuous drain current of 90 amps each, parallel 4 per side. Could you please recommend what the Gate resistors I should use instead of the 1 K ohm resistors ?
Sincerely Ali Rexhepi Barrie,Ontario, Canada 46 years young, father of two of Kids 15 years Licenced Auto Mechanic Passion for hobby electronics
My answer to the above question:
Thanks a lot, these pictures are amazing and your breadboarding is much better than mine. I'll post all of these screen shots in the article itself along with all the given information... it may take some time to appear. Calculating MOSFETs can be very complex and honestly, I don't have much practical experience with MOSFETs, so won't be able to help in this regard, but I'm sure when you could do such complicated and comprehensive analysis with the circuit, then off-course finding about MOSFETs would be just as easy for you.
For your reference I would suggest you visit the following discussion links where your question has been elaborately discussed:
Yahoo Answers: How to calculate the current needed to drive a MOSFET
Edaboard Forums: About MOSFET gate resistance
Thanks and Best Regards,
The Final Modified Sine Wave Inverter Design
After some discussions with me, the proposed modified sine wave design was finally built and verified by Ali Rex. Here's what he wrote to me:
Hi Swagatam, here's the revised schematics for our circuit. This schematic begins from the Two Diode banks 1 and 2 of our circuit.
R4,5,6,7 are 1K Ohms resistors 1/4 W , R8,9,10,11 are 21K Ohms resistors 1/4 W, and Mosfets Q1,2,3,4 are IRFZ44 , the Transformer is a Center Tap Transformer 12-0-12 on the Primary side.
This setup seems to work with four Mosfets run in 2 banks of 2 mosfets in parallel.
Output Waveform of the Above Design
More To Explore