[SOLVED] Pure sinewave inverter with toroidal transformer

[schmitt trigger]

I always thought that a low frequency transformer (EI or toroidal) should never "see" such a high frequencies (5 - 20 kHz). Beside, in every LF specifications I saw the maximum working frequency of 400 Hz or so.

If I put the toroidal choke on the output, all the high frequencies will pass the transformer.

The problem is core losses, which increase with frequency. These are offset somewhat with the decrease of flux density with increasing frequency, but nevertheless ferromagnetic materials still have a wide hysteresis curve and there is a frequency limit on which your core will be really cooking.

I'm with you (although I've not made a detailed analysis and thus could be incorrect) that the choke would serve you better on the primary.

 

[Orson Cart]

Yes putting the choke in series on the LV side will give lower losses in the Tx, as the switching delta V applied is less, hence the delta I (switching) in the Tx pri (LV) is less, hence the flux ripple at sw freq is less, hence the core heating is less.
Due to sizing issues most people put the filter comp's on the HV side and live with the slight increase in losses in the Tx. You don't have to follow the herd of course.

 

[red_alert]

I don't mind using a oversized/pricey toroidal choke in primary (LV) circuit if that's the best option.

I'm using this inverter for personal use so there's no economical constraints. I want it to be very reliable and efficient as I'm living offgrid (I'm just using solar/wind energy).

BTW.. does anyone have any experience with IRS21864 (4 Amp MOSFET driver)? I'd like to use it instead of IRS2110 (used in my current design). I want it to drive 6 x IRFP4668 (actually, 12 of them.. 6 on each high side / low side H-bridge switches).

 

[engmunir]

Can u provide the complete schematic of pure sine wave inverter. i want to make it for my home use.

 

[red_alert]

I don't have an actual schematic, it's just the IRS2110 datasheet (high/low MOSFET driver), a couple of MOSFETs and a MCU. Anyway, I'm in the process of redesigning the whole thing so I could post the final results when everything it's working properly.

 

[engmunir]

Dear Sir
thanks for your reply. when complete then post here.
thanks

 

[zeeshan2029]

When switching from main A.C. to inverter using soft start of about 0.5Sec to 1 Sec, dont you think that there will be another problem that switching delay can cause loads like computers to power-off.

 

[red_alert]

My inverter it's an off-grid one. That's it, there's no mains backup connection so there's no switchover either. It's supposed to start once and run indefinitely.

Btw, these days I'll finish the upgrade so I'm going to test various soft start software algorithms (slowly rising the output voltage, starting at the maximum voltage point, short pulses to demagnetize the transformer core).

 

[zeeshan2029]

Sir i have similar problem like yours instead there is also grid involved. Also i have used toroidal transformer of 6kva which has huge inrush current. For huge inrush current i have designed softstart algorithm. which is working fine. but the real problem is that when there is transition from ac main to inverter mode, the softstart takes about 500mS to reach maximum amplitude and load like computer restarts!
can anyone help me??

 

[red_alert]

Have you tried to shorten up the startup algorithm (to only a few waveform periods)? I was thinking of 0.5 - 1 second, too, but I don't have any limitation as I don't have to switch to mains.

Anyway, I have a small UPS for the sensitive equipments (PC, NAS server) so I'm not worried about the inverter start-up time.

Seems like a shorter algorithm is the only option for you (assuming you don't like the UPS alternative).

By the way, can you post a short description of the algorithm you're using? I'd really apreciate it, thanks!

 

[FvM]

I believe practical alternatives for faster saturation free startup have been suggested above in this thread.

 

[zeeshan2029]

My algorithm is very much simple. i am just increasing the amplitude after every 50hz complete cycle by changing the SPWM array till it reaches max required amplitude. i have tried to reduce the startup time by exponentially increasing the amplitude but same inrush current problem occur.

 

[red_alert]

So yes, a voltage ramp is probably the most simple solution. Alternatively, a fast saturation free start could use one halfwave of 50% magnitude and switch to 100% after the first zero crossing.

That was @FvM suggested solution early in this thread so you might try it, as it only takes two wavform periods. You could further minimize the inrush current by starting the PWM pulses at 90 degree (the sinewave peak voltage).

 

[zeeshan2029]

Thanks sir, I will try this solution and will let you know the results.

 

[bitsfromadish]

Hi Red_Alert

I was wondering how the inverter is going as I'm about to start working on making my own inverter.

 

[Warpspeed]

Hi Guys, first post here....
I was searching the internet and stumbled quite by accident upon this rather interesting thread.

This particular problem of feeding high frequency PWM into a toroidal mains transformer has interested me for quite some time.
A couple of years ago I was given a free (dead) commercial 1.5Kw grid tie inverter and reverse engineered it out of curiosity to try to learn the "trick" of exactly how to do this, which all turned out to be a rather interesting project.

This particular PWM output stage used a full bridge fed from about 200 volts dc nominal, PWM'd at 20 Khz into a 2:1 step up toroidal transformer to generate a 240 volt at 50 Hz sinewave output.

There was a 2.7mH choke in series with the primary, and a 5.0uF capacitor directly across the primary of the transformer.
Tests revealed the 5.0uF capacitor and transformer primary resonated at about very roughly 85Hz, with nothing connected to the transformer secondary.

This is all rather clever once it is realized how it's all supposed to work.

By resonating the primary at about roughly 1.5 times the sinewave operating frequency, direct high amplitude resonance at 50Hz is avoided.
The tendency to resonant energy buildup is damped by causing each half cycle to be out of phase with the next half cycle of the 50Hz exciting frequency.
That way you avoid massive destructive resonant buildup of voltage with zero load on the inverter output.

But its still close enough to 50Hz resonance for the Xl of the transformer primary to cancel out much of the Xc of the 5.0uF shunting high frequency filter capacitor.

So the whole system passes 50 Hz fairly efficiently without massive no load losses due to the rather large 5.0uF shunting capacitor.
Note that this is for a 200 volt dc system.

If running at lower voltage, you may need a lot more than 5uF to resonate the primary at a suitable frequency, as it will have fewer turns and a lot less inductance.

You are going to need a very good capacitor, one that has a very high ripple current rating, and a very low self inductance if it is going to both work properly and survive.

At the 20 Khz PWM frequency, the 5uF capacitor and 2.7mH choke act as a two pole low pass filter, as in a buck regulator, so little of the 20 Khz remains as ripple across the primary of the transformer.
Fr works out at around 1.37 Khz, so its about 14.6 times lower than the 20 Khz switching frequency. Almost four octaves at 12db per octave is pretty good attenuation.

The 2.7mH choke is an iron cored device, made from two tape wound C cores, and has to handle at least 15 Amps of low frequency current without core saturation.

The whole thing was very carefully thought out by someone, who knew exactly what they were doing. And it works very well with a completely clean looking output sinewave.
This seems to be the way to do it, although it still needs to all be properly engineered and thought through to suit the dc voltage and power level of your output transformer. Its far easier to do effectively and get good results at a higher dc operating voltages and lower currents.

Cheers, Tony.

 

[Orson Cart]

When you load the Tx much of the effect of the 5uF cap goes away and the res freq goes way up above 85Hz...
2.7mH at 15A is a fair sized choke...!

 

[Warpspeed]

Reactance of the 5uF capacitor at 20Khz is about 8 ohms.
At full load (1.5Kw in this case) the primary will have about 120 volts at 12.5A or a load impedance of about 9.6 ohms, hopefully mostly resistive.

Both quite low, and will shunt most of the 20Khz, as the choke impedance will still be up somewhere around 340 ohms, assuming minimal core saturation effect.
The high frequency attenuation will remain very good.

Yes, any reactive power factor load on the inverter will move the 85Hz resonance around, and also heavily damp it out resistively.
But that resonance is only really needed at zero load, to prevent the output voltage from any uncontrolled resonantly peaking.

Under any reasonable load, the inverter output should be very well controlled.

It's all pretty complicated, but the design as stated above works very well in an existing commercial product.
I take no credit for it personally, its just reverse engineering and attempting to understanding something that is already proven to work.

Trying to understand all this may give some useful insight into designing your own PWM inverter project.

 

[babatundeawe]

Hi all,

Must really say thank you to all that have contributed immensely to this thread. thanks for your wealth of knowledge that you have shared. it has really sent me back to the drawing board to review a lot in my current design for a 4KW inverter using spwm @16Khz carrier frequency but i will be using the EI core as that is what is widely available to me here.

Please i would like to ask some few questions.

1. I am also planning to implement a soft start algorithm to my software. but i m kind of confuse here as the way i am thinking, two methods had been enumerated here, (Starting at 90 degrees or sort and or starting the dutycycle at a lower value from the maximum duty cycle say the 50% and increasing gradually until the nominal voltage ) this is my understanding and i hope this i what is been said here.

2. Since i will be using the EI core (which is what is available to me). i want to couple the low voltage side of the Xformer directly to the bridge and hoping the leakage inductance of the low voltage side will aid in filtering and place a capacitor in parallel to the secondary output terminal and also along with some EMI filtering. I hope this method is ok. though i have seen a lot of this design during my research and have seen some of them work. but for that they might not be they are actual good design or do i need to add the extra inductor and capacitor at the primary for filtering as well.

NB: will be inverting from 24vDC(has a battery bank of 24v already) to 230VAC 50Hz. so my trasnformer spec is 250VAC to 20VAC. hope this is ok too

pls what do you suggest.

thanks all

Tunde

 

[Warpspeed]

If you always start the PWM cycle off from a zero crossing, soft start comes for free.
Even with no external load, flux doubling in the transformer can give the inverter quite a significant loading at start up if the the transformer core saturates.
If you also incorporate a very fast current limit, the two should cope very well together in most transient situations.

A standard EI transformer certainly works, but magnetizing current, and zero load inverter current will be higher than with a silicon steel toroid.

Your suggestion of placing a capacitor directly across the secondary is fairly common practice in many commercial designs that use EI transformers.
It will need to resonate with the secondary transformer inductance "near" 50Hz but not close enough to set off a huge troublesome 50Hz resonance at zero load.

One thing to watch out for will be high amplitude current spikes in the primary, if your transformer leakage inductance is not as great as you expect.
Its trying to switch at 16Khz straight into a large shunt capacitor across the secondary.

It may need some help from slight additional external series inductance with the primary. But you are not going to know until the beast is up and running.
This acts as a turn on snubber, and slows the rate of current rise, only a very few uH can work wonders ! It can dramatically reduce turn on switching losses.

Cheers, Tony.