[SOLVED] Pure sinewave inverter with toroidal transformer

If you approach the design as a Class E Woofer Amplifier with a constant Sinusoidal source and go from 95 to 98% efficiency with rising power levels in kW using super fast recovery diodes.

Warpspeed said:

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.

Click to expand...

Are you saying that starting the inverter in such a way that the first output of a 50/60 Hz waveform starts at a zero crossing is automatically a soft start? Why is this? What would be an example of non-soft startup?

The Electrician said:

Are you saying that starting the inverter in such a way that the first output of a 50/60 Hz waveform starts at a zero crossing is automatically a soft start? Why is this? What would be an example of non-soft startup?

Click to expand...

A non soft start would be hitting 100% duty cycle hard in one direction.
Basically full supply rail dc (in one direction) to the ac output.

At the zero crossing, either both switching devices will be off, or both switching devices will be at 50% duty cycle at the switching frequency. The PWM output filter will then average that to zero output.

I fear, it's still unclear what you are suggesting. A waveform would help.

Ac waveforms pass through zero each half cycle.
If you start off at a zero crossing (at initial power up) the output ramps up over the next ninety degrees to the peak.

Surely you do not need to see a picture of a sine wave to understand this ?

Surely you do not need to see a picture of a sine wave to understand this ?

Click to expand...

Not particularly. I didn't believe that your verbose previous post just suggests to switch on the inverter at the voltage zero crossing, which is the perfect way to drive the transformer into saturation. The question has been extensively discussed at the start of this thread, I won't repeat the details. To avoid saturation, the voltage must be either switched on at maximum (90° phase), or ramped.

Not all PWM inverters use an output transformer.
But your point is correct and well taken.

Warpspeed said:

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.

Click to expand...

Please can you suggest means of arresting this kind of trouble " Core saturation

Warpspeed said:

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.

Click to expand...

please can you suggest a range of leakage inductance that is suitable for this kind of application and also if i have to include a series inductor at the primary low voltage side of the xformer, at the 16Khz switching frequency and what, any idea of value of the inductor?

thanks

tunde

- - - Updated - - -

FvM said:

Not particularly. I didn't believe that your verbose previous post just suggests to switch on the inverter at the voltage zero crossing, which is the perfect way to drive the transformer into saturation. The question has been extensively discussed at the start of this thread, I won't repeat the details. To avoid saturation, the voltage must be either switched on at maximum (90° phase), or ramped.

Click to expand...

FvM, i guess your response here provides answer to my first question, of method of softstart, which is by switching at 90 degrees phase or ramping the pwm over time.

If it is so. I will use the latter as i dont have the slightest idea of how to switch at 90 degrees phase.(would love to learn that if any can help).

thanks for your help all the way.

babatundeawe said:

please can you suggest a range of leakage inductance that is suitable for this kind of application and also if i have to include a series inductor at the primary low voltage side of the xformer, at the 16Khz switching frequency and what, any idea of value of the inductor?.

Click to expand...

Zero transformer leakage inductance would cause the switching devices to turn on into a virtual dead short circuit, because of the relatively large filter capacitance connected directly across the secondary.
That secondary shunt capacitance will be reflected back into the primary, magnified by the turns ratio
So its going to be particularly high for a 12v dc system.

Any transformer leakage inductance will be detrimental to full power transfer at 50 Hz.

But you definitely need some inductance in series with the switching device to limit the rate of di/dt rise at turn on.
How much depends on the power level, system voltage, and your own sympathy for the switching devices.
Just build it, and test it, and observe the current waveform in the switching devices.
If it works to your satisfaction, fine, that is good.

If you see dangerously high current spikes, and the switching devices are almost on fire, something will need to be done.
Just be aware of the potential problem, and gradually increase the applied voltage at very first turn on and observe the current waveform.

I know of several cases where replacing an iron EI transformer with a silicon toroid actually increased the zero load inverter idling current.
This was the exact opposite of what was expected.
Adding a few microhenries of external inductance reduced the inverter idling current by about three quarters, without introducing any bad side effects.
So its well worth a bit of practical experimentation.

As the original poster, I could share my recent experience.

The soft start procedure for the 5kW toroidal transformer turned to be the easiest part of this project.

Taking account of @FvM and @BradtheRad suggestions, I've just designed a simple software procedure to start generating the sPWM signal at 90 degree, with a significant initial attenuation.

I must say that I'm using a sine wave look-up table with 50 rows, corresponding to a sine wave voltage attenuation range of 45-95% (1% stepping).

So, if you have 128 sine wave interpolations during every half wave, you have to set the look-up table index to 64 (90 degree) and the attenuation to 50% or something.

In my software routine, I check the output voltage (230V) during each sine wave zero crossing (I read an average value, anyway) so for the very next sine wave, the attenuation is changed accordingly.

If the battery is in a good state of charge, the attenuation stays in 70-75% range so it takes 20-25 half waves from the startup to normal operation (25 * 10ms = 250ms).

With the above parameters, there's almost NO inrush current from that big 5kW toroidal transformer.

To further tune up the soft-start procedure, the inverter is turning off (due an alarm or a manual/remote action) when the sine wave is at -90 degree thus the remanent core magnetization it's opposed to the one developed at startup (+90 degree).


Anyway, I have some problems with a noisy DC-link (due to big switching spikes). I've already increased the gate resistors to reduce the switching speed, I put some RCD snubbers and big TVS diodes across every switch (and DC-link) and now I'm designing a simple fly-back converter to generate four isolated power supplies for every MOSFET driver (for now, I'm using bootstrap technique for the high-side drivers).

My problem seems to be the (slow) MOSFET built-in freewheeling diodes. The Trr is about 200ns and during this period very high voltage spikes are developed. Anyway, I couldn't find any suitable external Schottky diodes in a 100-150A range thus I have to play with snubbers and such.

- - - Updated - - -

@Warpspeed:

Welcome to edaboard! You must be the same guru member from thebackshed/energymatters boards, right?

I've read a lot about your hard work (you and the other cool Aussie guys: Oztules, Andrew, Orson Cart..). I've just seen Oztules complaining on fieldlines board about the lack of understanding of how that mighty series inductance manage to reduce the idle current almost ten times!

Maybe (with the help of edaboard experts) we could find some good explanations on how to actually dimension that inductance for a specific situation.

I forgot to mention that I was also using a 47uH toroid inductor and a 0.47uF capacitor across the transformer output. Due to those annoying spikes, I also decreased the PWM frequency to 6.4kHz.

red_alert said:

As the original poster, I could share my recent experience.

The soft start procedure for the 5kW toroidal transformer turned to be the easiest part of this project.

Taking account of @FvM and @BradtheRad suggestions, I've just designed a simple software procedure to start generating the sPWM signal at 90 degree, with a significant initial attenuation.

I must say that I'm using a sine wave look-up table with 50 rows, corresponding to a sine wave voltage attenuation range of 45-95% (1% stepping).

So, if you have 128 sine wave interpolations during every half wave, you have to set the look-up table index to 64 (90 degree) and the attenuation to 50% or something.

In my software routine, I check the output voltage (230V) during each sine wave zero crossing (I read an average value, anyway) so for the very next sine wave, the attenuation is changed accordingly.

If the battery is in a good state of charge, the attenuation stays in 70-75% range so it takes 20-25 half waves from the startup to normal operation (25 * 10ms = 250ms).

With the above parameters, there's almost NO inrush current from that big 5kW toroidal transformer.

To further tune up the soft-start procedure, the inverter is turning off (due an alarm or a manual/remote action) when the sine wave is at -90 degree thus the remanent core magnetization it's opposed to the one developed at startup (+90 degree).
.

Click to expand...

Thanks for your response, but please what do you mean by attenuation. from what i can deduce you have an array of 50 different dutycyles from 45% to 95%, then each table depending on the output voltage. am i correct?

if its like this i will be so happy because i am currently implementing it already but i actually meant to use for output voltage regulation but i never knew it can be used for soft start

And also how many samples of voltage do you take for your output voltage, do you take within a cycle or you just let it happen.

My plan is to sample the output voltage and current at 5Khz, i am going to use a timer to initiate an ADC and get the value in an ADC interrupt store the values for reference.



Thanks for your effort and definitely will not take your thread.

@Warpspeed thanks again.

i shall put all your advice into careful consideration and implement wherever possible. i will also try and see i could get a good toroidal xformer or make use of my EI which is readilly available here. thanks so much

tunde

@babatundeawe:

You're welcome, no problem at all with hijacking this thread!

Yes, I was talking about using look-up table rows for output voltage regulation. During normal operation, a 50% attenuation might be necessary when the battery is full and the load is very small. Anyway, you could force this higher attenuation at start-up to further reduce the inrush current.

For now, I'm using 64 samples per half wave (and a 6.4kHz PWM frequency) but at the beginning I was using 100 samples/10kHz. Anyway, the output waveform looks almost perfect in both situation.

Like I mentioned in my previous message, I check the output voltage level during each sine wave zero crossing point and I change the attenuation accordingly (for the next half wave) only if it's necessary.

- - - Updated - - -

Oh, you were asking about how many samples of output voltage I take for the average value. Well, I'm using the ADC in continuous sampling/conversion mode and I set the averaging value to 16 samples or something.

But, like I said, I only read the ADC value once per half wave, during zero cross points.

Btw, I am using an obsolete output voltage feedback control (using a small 230/6V transformer across the output). The low voltage (6v) is further rectified and filtered so there's already an average value at the ADC input. The output regulation is working fine, anyway.

@Warpspeed:

Welcome to edaboard! You must be the same guru member from thebackshed/energymatters boards, right? .

Click to expand...

Yes, that is the same me.

Dimensioning a suitable series inductor is really an experimental iterative process, guided by the results.
Some of the cumulative total inductance that the switching devices see can be just distributed parasitics in the high current wiring.
Its far easier to just test it and adjust according to the results.
When switching perhaps a hundred amps in less than a microsecond, even a few inches of wire length can become significant, especially at low dc voltages.

@red_alert:

Thanks again.

Hmn! i am glad i came across your thread, here i have solved about three to four problems i have been having like (ac voltage measurement using differential opamp amplifier thanks to @BradtheRad for pointing to Faltstad simulator) and also confirm some of my implementations are right.

i am using 160 sample/16khz pwm frequency so that means i start the duty cycle at index 80 till i get to index 159 and then i reverse the bridge start from zero or i when i reverse the bridge i go back to index 80 again.

secondly You said i could force a higher attenaution at start - up to further reduce the inrush current, will this not negate the soft - start procedure, or am i missing something.

Yeah Got to see the update to your post after, hmn! thanks another confirmation for me again.

i had use that old method too before but now i want to use the differential amplifier method, its cheap (in my own understanding). i can provide the circuit if you are interested.
thanks

tunde

secondly You said i could force a higher attenaution at start - up to further reduce the inrush current, will this not negate the soft - start procedure, or am i missing something.

Click to expand...

Nope, you're still generating a sine wave starting at 90 degree (maximum voltage) but with a lower amplitude. Being a PWM signal, it might look the same (by applying the full DC-link voltage across the primary anyway) but that 50% duty-cycle (at 90 degree) it's still smaller than the 95% one (at 90 degree, too) thus you're applying the DC-link across the primary for a shorter period.

Btw, I might have been using the wrong terms before: by 45-95% attenuation I was talking about 45-95% of the maximum duty-cycle (but I guess you get it right).

red_alert said:

Nope, you're still generating a sine wave starting at 90 degree (maximum voltage) but with a lower amplitude. Being a PWM signal, it might look the same (by applying the full DC-link voltage across the primary anyway) but that 50% duty-cycle (at 90 degree) it's still smaller than the 95% one (at 90 degree, too) thus you're applying the DC-link across the primary for a shorter period.

Btw, I might have been using the wrong terms before: by 45-95% attenuation I was talking about 45-95% of the maximum duty-cycle (but I guess you get it right).

Click to expand...

Thanks again, yes i got what you mean about the duty cycle percentage.

and please can you help clarify my doubt too the first question in my last post. that is should i start 90 degrees too for the next half of the cycle or i just start at the beginning?

Thanks.

tunde

You just start at 90 degree then the sine wave is generated normally. You only need to progressively decrease that duty-cycle attenuation after each zero crossing point till it reaches the rated output voltage.

Actually, you just have to let the inverter run into normal operation, as it will self regulate the duty-cycle (attenuation) to keep the output constant.

It happens that I've designed it to change the attenuation (during normal operation) with a step of +/-1% after checking the output voltage at every zero crossing point. Maybe you could increase it by a large factor - I've not tested that.

Most sine wave inverters increase the rms output voltage from zero to 230Vac (for e.g.) in 1 to 5 seconds, this is easily done in software and is great for starting into transformers and motor loads....

red_alert said:

You just start at 90 degree then the sine wave is generated normally. You only need to progressively decrease that duty-cycle attenuation after each zero crossing point till it reaches the rated output voltage.

Actually, you just have to let the inverter run into normal operation, as it will self regulate the duty-cycle (attenuation) to keep the output constant.

It happens that I've designed it to change the attenuation (during normal operation) with a step of +/-1% after checking the output voltage at every zero crossing point. Maybe you could increase it by a large factor - I've not tested that.

Click to expand...

Whao! i am really armed now with a lot of conviction and new ideas, got to put them into practice and see the outcome.

thanks for your time and explanations.

Tunde

Red_Alert

Are you using Unipolar or Bipolar modulation? To my mind if you use unipolar modulation it means you only need snubbers and or high speed diodes on the high speed switching quadrants of the bridge. ??