12VDC-220VAC pure sine inverter using SG3524 and IR2110

[sinewave123]

Hello everyone, I have found a lot of threads related to my query but it's exhaustive to be in everyone's conversation so I decided to post my own. I am now working for a university project (12 VDC-220VAC pure sine inverter design). I started learning about inverter design just a few months ago so I have a relatively little knowledge about this. A friend of mine is helping so my learning gets accelerated. However, we got into trouble when the MOSFETs (IRF480) are heating up with higher voltage DC input. It only works well with 15 VDC fed into the mosfet fullbridge circuit. Can you share any suggestion?

MOSFET fullbridge is driven by IR2110 and PWM signal is generated by SG3524. How can I make it work with higher voltage DC? Should I use DC/DC converter to convert 12VDC to 350VDC before feeding it to the mosfet fullbridge? Or just step up the lower voltage output sine (from 15VDC input) into higher sine voltage using a transformer coupled after the LC filter so the mosfet fullbridge is saved from high DC voltage input?

 

[betwixt]

I don't think you can use an SG3524 to generate a PWM sine wave, are you using it only for generating the 350V DC or are you trying to use it to control the AC output?

Brian.

 

[sinewave123]

im using it to generate the pwm signal input to the mosfet drivers. why not sg3524?

 

[betwixt]

The SG3524 is really intended to be a push-pull driver and the PWM duty cycle controlled to stabilize the output voltage. For a normal sine wave inverter you can use push-pull drive but you need the DC to be twice the peak AC voltage, in other words about 625V. The method used in most designs is to use a H-bridge so the voltage can be halved and the polarity alternated to produce both positive and negative half cycles. I'm not sure the SG3524 will allow 0% to 100% duty cycle adjustment to produce your 50/60Hz output either. Its a good choice for producing the DC ahead of a H-bridge but I think the bridge driving signals need a more precise control method.

Brian.

 

[sinewave123]

The SG3524 is really intended to be a push-pull driver and the PWM duty cycle controlled to stabilize the output voltage. For a normal sine wave inverter you can use push-pull drive but you need the DC to be twice the peak AC voltage, in other words about 625V. The method used in most designs is to use a H-bridge so the voltage can be halved and the polarity alternated to produce both positive and negative half cycles. I'm not sure the SG3524 will allow 0% to 100% duty cycle adjustment to produce your 50/60Hz output either. Its a good choice for producing the DC ahead of a H-bridge but I think the bridge driving signals need a more precise control method.

Brian.

So you mean SG3524 is not a good tool for sending signals to control the MOSFETs? Can you tell me how can I have a good controlling method using inexpensive materials and simple circuitry?

 

[betwixt]

All PWM inverters are relatively complicated and most use a microprocessor to generate the timing signals. If you look around on this forum there are several threads about sine wave inverters running at the moment, some with schematics. All PWM sine converters start with a DC supply equal to peak AC voltage so you can still use the SG3524 to drive the DC supply generator. Its only the output H-bridge that needs special drive signals.

Brian.

 

[sinewave123]

All PWM inverters are relatively complicated and most use a microprocessor to generate the timing signals. If you look around on this forum there are several threads about sine wave inverters running at the moment, some with schematics. All PWM sine converters start with a DC supply equal to peak AC voltage so you can still use the SG3524 to drive the DC supply generator. Its only the output H-bridge that needs special drive signals.

Brian.

I want to restrict my design from using microprocessors. I had an encounter with someone who constructed an inverter without the use of those, so I am thinking it would be possible for me to do it. My present design's basic principle is this:

sine oscillator ---> SG3524 (it has internal oscillator of carrier wave, i guess) ---> PWM signal generated from SG3524 will be separated into complementary outputs using Xor gates (CD4070) ---> complementary outputs into the mosfet Hbridge + ir2110 ---> LC filter

i do not have my design of a DC/DC converter yet. But the original plan is to feed the Hbridge with high DC voltage from the DC/DC converter to produce 220 VAC. It turns out that the mosfets heat up with higher voltage DC input, so I am thinking now of just converting the low voltage AC output into higher voltage using high freq transformer.

This is an idea from a newbie (it's me) so I want to ask for more insight to improve my design with the least changes.

 

[betwixt]

I think there several flaws in your design but don't let me stop you experimenting.
1. Check the SG3524 will accept a sine input without distorting it, and allow you to modulate the PWM mark/space ratio between 0 and 100%. Anything less will waste power. You can try this in a small test on the bench without using high voltages, just power it up and feed a control voltage in from a potentiometer. Monitor the outputs and check you get a linear and wide enough change in pulse width.
2. The SG3524 already produces complimentary outputs, the two drive signals it produces are complimentary as they leave the IC. You shouldn't have to use the XOR gates.
3. There is a problem with the switching devices in a H-bridge which the SG3524 will not cater for. At the point where the drive signal changes state, there will be a very brief period when one arm of the bridge turns on but the other device on the same side has not yet turned off, this is partly due to stored charges in the FET, partly from the output filter and partly from the load currents. In an optimal system, a brief period when both drive signals are turned off exists, this ensures there is never a current path directly across the DC supply which would cause damage. The duration of the dead-band ideally, is controlled according to the pulse width and you can't do that without some calculating power.

If you are prepared to make the jump to a microprocessor, there are many with built in PWM H-Bridge drivers which produce all the signals for you. All you have to do is hook up the IR2110 and MOSFETs and tell it what frequency and modulation rate to use.

Brian.

 

[sinewave123]

Wow, I can see you're sharing much. Thanks a lot. Would you mind if I send my schematic diagrams to your e-mail? You might want to criticize and improve it. I cannot share it here because the schema is not my intellectual property. It is someone else's. So I am sharing it only to those I can rely on for help. By the way, have you done an inverter before?

 

[sinewave123]

By the way, I am doing it for school project and I have to finish it very soon or else I'd be disqualified for graduation. I only have three weeks to finish everything but I am willing to do it day and night. What I am saying here is, this is a very important project for me. It's like the rest of my life depends on it. And I would be forever grateful to all those whoa are willingly lending their help

 

[betwixt]

You can contact me through my web site, it is listed in my profile. I has disabled direct mailing from here due to the number of people asking me for help when it would be better to ask publicly here. It would be better if you moved this thread to the "Power Electronics" thread though, you will find people with more expertize than me in that part of the forum. Quote the page address of this thread in the first message so people can refer back to what has been said already.

Brian.

 

[eee_psc]

i would like also to suggest using optocouplers or optoisolator with IRF5210 MOSFET..

 

[sinewave123]

Quoted from:

https://www.edaboard.com/threads/202321/#post853640


Hello everyone, I have found a lot of threads related to my query but it's exhaustive to be in everyone's conversation so I decided to post my own. I am now working for a university project (12 VDC-220VAC pure sine inverter design). I started learning about inverter design just a few months ago so I have a relatively little knowledge about this. A friend of mine is helping so my learning gets accelerated. However, we got into trouble when the MOSFETs (IRF480) are heating up with higher voltage DC input. It only works well with 15 VDC fed into the mosfet fullbridge circuit. Can you share any suggestion?

MOSFET fullbridge is driven by IR2110 and PWM signal is generated by SG3524. How can I make it work with higher voltage DC? Should I use DC/DC converter to convert 12VDC to 350VDC before feeding it to the mosfet fullbridge? Or just step up the lower voltage output sine (from 15VDC input) into higher sine voltage using a transformer coupled after the LC filter so the mosfet fullbridge is saved from high DC voltage input?

 

[sinewave123]

You can contact me through my web site, it is listed in my profile. I has disabled direct mailing from here due to the number of people asking me for help when it would be better to ask publicly here. It would be better if you moved this thread to the "Power Electronics" thread though, you will find people with more expertize than me in that part of the forum. Quote the page address of this thread in the first message so people can refer back to what has been said already.

Brian.

I have already posted this thread to the Power Electronics section. But I cant find a way to send you the schematic diagrams without showing it to everybody else. So I decide to show it here. This circuit below is suppose to be connected to a MOSFET fullbridge via IR2110. But as it turned out, it doesn't work as intended, so please tell me how should I make it work.

 

[betwixt]

Well for a start, you are generating two identical drive signals to the IR2110s.
You need FOUR different drive signals, one to each arm of the H-Bridge and each with it's own dead band generator. The high voltage DC (which you do not show) goes to the top of the 'H', ground goes to the bottom and the AC output is taken between the centers of the two arms. I still have doubts about 100% modulation on the SG3524 but I've never tried it so I could be wrong.

Brian.

 

[sinewave123]

As far as I know, those two signals are complementary. So, strictly speaking they should not be identical. Please correct me if I am wrong. By the way, I get the basic concepts of my design from this link:

https://www.google.com.ph/url?sa=t&...sg=AFQjCNEZKpN0Au0ipm_kcr6T4GzoW6tc1w&cad=rja

Can you tell me anything about it?

 

[betwixt]

I assume you mean the schematic on page 48. If you look closely, there are four drive signals to the IR2110 ICs which in turn connect to the FETs in the H-bridge.
Your schematic shows only one signal which is duplicated and sent to both 'HI' and 'LO' .

The page 48 schematic also shows the classic way of generating PWM with a triangle wave and comparators, this is what happens inside the SG3524 as well but I'm not sure it works over the full 100%:0 to 0:100% range of pulse widths as the device is not intended to be used that way. You will have to experiment to check that the amplitude on pin 9 is sufficient and the on/off ratio varies linearly according to pin 9's voltage.

As I said, I'm not an expert in this field, I'm going from general knowledge of electronics and many years of experience.

Brian.

 

[Genomerics]

Off on a rant..

One thing I fail to understand is why this 'question' is being presented as an end of course/final year project and nothing seems to have moved on beyond using something like a, not, modern day equivalent of a TL494...

Not that I have advanced much further and obviously it can be done with a PIC processor. Excuse me whilst I laugh in that general direction.

For your delectation I present...



Based on one of these..

https://focus.ti.com/lit/ds/symlink/uc3827-1.pdf

blah here

**broken link removed**

Ignoring the blah then, with trepidation, the results according to LTSpice are..



So my harmonics are 80dB down.. That qualifies as Hi-Fi. Not that I would expect to achieve that in reality and I am sure it would burp or break elsewhere plus you can see QBODGE hanging about the place.

I might be sure it will not 'work' but I would be tempted to 'hit it'

I met this 'student' problem some time back in 1996 and they are still running it and there still seems to be the same lack of education that results in 'enforced' efforts at rubbish solutions.

Not that I would wish to claim mine is any better or workable.

Not that I would wish to claim mine is any better or workable.

I could still run the numbers.

That still leaves the OP out on a post because quite frankly I am stupefied that the 'Lecturer' in charge of this project offered it, again after so many years, in the first place. I don't get it.. How did we advance from TL494 to SG3524 and so rapidly.

I know nothing but SMPS, in its horrible glory and the associated Loop Compensation is conceptually HARD. Obviously if you have a PIC then you just change the bit you divide something by and hope things don't go BANG whilst you were looking elsewhere. In the mean time it would seem your PIC still has to deal with the loop compensation.

Woot

https://www.wpi.edu/Pubs/E-project/Available/E-project-042507-092653/unrestricted/MQP_D_1_2.pdf

Our project failed because we were not taught stuff and did not have stuff and could not buy stuff so it was stuffed. Please follow our example.

Genome.

 

[sinewave123]

Off on a rant..

One thing I fail to understand is why this 'question' is being presented as an end of course/final year project and nothing seems to have moved on beyond using something like a, not, modern day equivalent of a TL494...

Not that I have advanced much further and obviously it can be done with a PIC processor. Excuse me whilst I laugh in that general direction.

For your delectation I present...



Based on one of these..

https://focus.ti.com/lit/ds/symlink/uc3827-1.pdf

blah here

**broken link removed**

Ignoring the blah then, with trepidation, the results according to LTSpice are..



So my harmonics are 80dB down.. That qualifies as Hi-Fi. Not that I would expect to achieve that in reality and I am sure it would burp or break elsewhere plus you can see QBODGE hanging about the place.

I might be sure it will not 'work' but I would be tempted to 'hit it'

I met this 'student' problem some time back in 1996 and they are still running it and there still seems to be the same lack of education that results in 'enforced' efforts at rubbish solutions.

Not that I would wish to claim mine is any better or workable.



I could still run the numbers.

That still leaves the OP out on a post because quite frankly I am stupefied that the 'Lecturer' in charge of this project offered it, again after so many years, in the first place. I don't get it.. How did we advance from TL494 to SG3524 and so rapidly.

I know nothing but SMPS, in its horrible glory and the associated Loop Compensation is conceptually HARD. Obviously if you have a PIC then you just change the bit you divide something by and hope things don't go BANG whilst you were looking elsewhere. In the mean time it would seem your PIC still has to deal with the loop compensation.

Woot

https://www.wpi.edu/Pubs/E-project/Available/E-project-042507-092653/unrestricted/MQP_D_1_2.pdf

Our project failed because we were not taught stuff and did not have stuff and could not buy stuff so it was stuffed. Please follow our example.

Genome.

Thanks for your post. Well, first thing to say, I am from Philippines so I have a somewhat limited English and I am not sure if I understood everything you said correctly. Anyway, I can see that you're suggesting for me to use a PIC for the project. Am I right? Well, due to series of failures with my circuit without the PIC, I thought about welcoming the idea of using them. But what would I want to know here is: How much would it cost? And will it really make my circuit simpler and my success more probable? Can you show me a schema of a working inverter (maybe something you have done before) using PIC? How about microcontrollers? I heard Arduino microntrollers are available in my place and they are capable of generating PWM signals. I want to know how much can it offer.

 

[betwixt]

I would suggest you visit the Microchip web site, most PIC devices have PWM generators in them as standard. I think my first choice would be the PIC18F1320 because it is very low cost and already has a four-phase PWM generator built in to it. This means you can produce the drive signal to all the FETs (using IR2110 to shift the voltage levels) directly by setting some values in PIC registers. It will allow you to produce any on/off ratio between 0% and 100%, all you have to do is produce the sine wave in software or use a look-up table of sine levels then send them to the PWM register in time to modulate at the right frequency. The PIC also has several ADC channels which you can use to stabilize the output or monitor the load.

The data sheet and free assembler are at Microchip Technology Inc. is a Leading Provider of Microcontroller and Analog Semiconductors, providing low-risk product development, lower total system cost and faster time to market for thousands of diverse customer applications worldwide. where you will also find trial versions of 'C' compilers which should be adequate to complete this task without having to buy a full version.

Incidentally, I am using a similar PIC to generate speech using PWM so low frequency power AC signals should be no problem!

Brian.