Its is a 3KVA transformer. i wound it myself. i made it IN:7V OUT:240V

My challenge now is when higher load ( not exceeding the rating of the transformer ) is placed on the inverter, the wave becomes distorted and the voltage largely attenuates .what can i do ?

 

Hi,

AC converters are not the easiest thing to build if you havent done DC converters before.

There are a lot of things you would have to look into, such as saturation of the core and unequal loading vs output impedance.

If your transformer works ok for the voltages you are inputting (sounds questionable really) then your output impedance may be a little higher than you expect and so if you regulate on a cycle by cycle basis you'll always get distortion when you connect a load that is not linear. The only way to get around that impedance issue is to move to sub cycle regulation.
Sub cycle regulation means you regulate the output every switch period rather than every AC cycle or half cycle. That allows non linear loads to work better because then the feedback is countering the loading effect over the entire wave rather than over the average value.

So if your output impedance is not low enough you have to move to sub cycle regulation, and so you'd have to implement that in code. The code would be similar to what you do now, but it would work over a much shorter period. In short, it works like a DC to DC converter where the DC output is just one small part of the AC sine wave, and then when the amplitude changes the pseudo DC voltage changes and so the regulation point changes. Looking over the whole half cycle, it looks like you have a DC to DC converter that keeps changing the voltage regulation point.

It could just be that your output drags the voltage down too low, and when the feedback adjusts the input to the transformer, it presents the primary with a voltage that is too high for the core vs primary winding turns and thus saturates the core in some places in the cycle. If the input is really designed for 7vac and you have to input 12vac to get the right output with load, you could very well be saturating the core. If that is not an issue, then it is probably the output impedance vs non linear loading issue, unless of course it is something much simpler that you overlooked.

If you could post a scope shot of your output when you connect the offending load, we could take a look and see which of these it is, or maybe something else even.

 

This is the image of the output wave form when the inverter is been loaded with my soldering iron.
When loaded with my PC charger, the output waveform and voltage keeps rising and dropping.
This is serious.

 

I also noticed, when i changed one of the mosfet, the output waveform tends to lean more to the right .
Can the mosfet also contribute to this challenge ?

 

I also noticed, when i changed one of the mosfet, the output waveform tends to lean more to the right .
Can the mosfet also contribute to this challenge ?

Yes, if you don't have a proper and adequate gate driver circuit and or a adequately sized switching device. Under driven gate and or overworked mosfet.

The typical rule I have used in sizing switching devices is 2x over maximum working voltage and 4x minimum on average anticipated switching currents.

So if you are using s basic push/pull switching circuit that has to work with a typical automotive type 12 volt power source that could realistically see 15+ volts the bare minimum switching device voltage rating I would go with 60 volts.
After that if you are pushing some 3 KVA into a 7 volt primary the realistic peak currents involved could easily be over 500 amps resulting in you need somewhere near 2000 amps of peak switching capacity per leg of the Push Pull power switching circuit to make it reliable. You can get by with less but reliability at high loads plus peak overload capacity diminishes rapidly.

 

Thanks. I used an H-bridge design. with two MOSFET on each legs. a combination of irf3205 and irf740. Are you suggesting i should increase the number of mosfet on each legs ?.

My driver circuit is good. i have used the same driver circuit to design a square wave inverter, and it worked well.

Although, the MOSFET on the high side of the bridge remains cool while the MOSFETs on the lower side of the H- bridge gets HOT easily, even when just a small table lamp is loaded on the inverter.

I used a 4.7 ohms resistor on each gate of the MOSFETs

What are your recommendations ?

 

Thanks. I used an H-bridge design. with two MOSFET on each legs. a combination of irf3205 and irf740. Are you suggesting i should increase the number of mosfet on each legs ?.

Are they operating in parallel to each other as in having a 55 volt 110 amp switching device working side by side with a 400 volt 10 amp device on a power circuit that you intend to handle something like 3000 watts or more?

Although, the MOSFET on the high side of the bridge remains cool while the MOSFETs on the lower side of the H- bridge gets HOT easily, even when just a small table lamp is loaded on the inverter.

Well obviously if they are heating up with such a small load on them your driver circuit clearly isn't doing its job properly.

What math and reasoning did you use for the design?

Mine says that even if using theoretical ideals 3000 watts off of a 12 volt source would be something like 250 amps minimum current draw going through each leg of the H bridge when it's active but when considering you are pushing it into a 7 volt rated primary the PWM will be nearly doubling that number at minimum.

So, what do you think about using a mismatched pair of switching devices with a at best 120 amps and grossly mismatched on state resistance values will do when loaded up to anywhere near that current level?

 

The mosfets are connected in parallel.
so, what is your advice , what should I do ?

 

Also, I want to ask,since am using a full bridge configuration for the H-Bridge, should the Modulated SPWM be applied to the high side or low side of the H-bridge ?

 

Hello again,

Wow, nasty waveform. Does not look like saturation though. Looks like maybe the more narrow pulses nearer to the zero crossings do not get through, which would point to the gate driver.

You switch both low and high sides at the same time. That generates a bipolar output. You also ensure there is enough dead time.

If you got it to work with a square wave, then go back to that configuration and then gradually add more pulses to make a pseudo sine. That might tell you what went wrong. From the square wave go to three pulses, then maybe five, etc., per half cycle.

Also, you do need to make sure the mosfets can turn on and off fast enough. That requires a good gate driver.
As you move from square wave to pseudo sine, you are adding pulses within the same time frame so that means the mosfet switching cycle gets smaller. That means it must be able to turn on and off quickly because it has to generate more pulses per second. This is very important. It may appear to work fine at 50Hz (square wave) but once you start adding pulses the frequency goes up. Just three pulses per half cycle and we are up to 300Hz, and the higher we go the better the gate driver has to be. A gate driver that can handle 1 amp peaks is probably the minimum for this kind of project.

Is your test soldering iron made for AC only or for either AC or DC?
If only for AC, it may have significant inductance which could cause distortion.
You should probably compare that to a totally resistive load of the same power level.

 

Also, I want to ask,since am using a full bridge configuration for the H-Bridge, should the Modulated SPWM be applied to the high side or low side of the H-bridge ?

There are two camps on that design issue.

Myself I learned H bridge PWM technique on old industrial applications and a lot of them used the low side device only concept for the PWM work so that's my preference.

My understanding of the concept was that by doing so the top devices don't have to deal with dissipating as much energy losses being they are not operating at the higher PWM base frequency. That and the driver circuits for their high side operation could be made lower powered and simpler too because of it.

Now to take the concept even further if a circuit such as a low voltage source based power inverter like you are building was being constructed to further reduce forward voltage drop issue typically a simple two sided push/pull primary power circuit is preferable as well over a H-bridge design. Half the switching devices are needed an thusly half the overall complexity and losses are avoided as well which when dealing with high currents that's a substantial reduction in both component count and energy loss avoidance.

.1 volts at 500 amps is 50 watts so if the top half of the H bridge is not used that's 50 watts of energy not lost to heat that will need to be dealt with.

Also when it comes to DIY power inverter construction unless you're building it for the learning experience as a hobby project there is no way you will ever design and build one for less than you can buy on or even modify a factory built cheapo block wave type to have a SPWM output.

Especially so now that mass produced SPWM high voltage H-bridge driver boards can be had for under $5.

https://www.aliexpress.com/w/wholesale-spwm-driver-board.html

and manual for it.

http://www.egmicro.com/download/EGS002_manual_en.pdf