I have a 50Hz, 230 Volt 700 VA PWM inverter using a SG3524, and the final stage is basically as shown in Fig 1. (There are actually 4 pairs of MoSFET’s, the voltage feedback control has not been shown)


I would like to modify this as a Sine Wave Inverter using a EG8010 + IR2110 module, details here http://www.lz2gl.com/data/power-inverter-3kw/eg8010_datasheet_en.pdf



Instead of a Full Bridge Configuration as suggested in 6.6 of the datasheet, I intend to use the existing setup (Fig 1) using 1LO (Pin 3) and 2LO (Pin 8) of the module for driving the MoSFET’s.



Besides Comments and Suggestions from other members, I have a couple of questions:



1) Where is the Snubber / Filter capacitor to be connected? Some prefer across the 230 Volt output, others across the MoSFET’s Drain - Drain Terminals.

2) In the proposed setup, the transformer will handle 23 KHz Rectangular wave instead of 50 Hz Rectangular. Would not the Transformer losses go up considerably?



Thanks.

 

700 watts at 50 hertz and 23 kilohertz requires COMPLETELY different transformers. What is your end goal with this change?
Are you still driving a motor?

 

No, it's not a specific motor drive. Just for home electrification.

 

1) Where is the Snubber / Filter capacitor to be connected? Some prefer across the 230 Volt output, others across the MoSFET’s Drain - Drain Terminals.

2) In the proposed setup, the transformer will handle 23 KHz Rectangular wave instead of 50 Hz Rectangular. Would not the Transformer losses go up considerably?

The capacitor should be across the 230 volt side if you are intending for it to be for waveform smoothing.

As for the 23 KHz PWM the switching devices are all that will be seeing it being the 50 Hz iron core transformer will totally ignore such a high frequency and just see the averaged out 50 HZ sine wave its replicating.

 

It doesn't ignore the 23khz, it takes the high freqs and using "pure magic" turns them into heat.
The losses will be larger but only you know if that is acceptable.

 

It doesn't ignore the 23khz, it takes the high freqs and using "pure magic" turns them into heat.

Not really. The inductive impedance of a common laminated iron core transformer at that frequency is very high so what loses one may have are typically negligible.

It's why a common power transformer can't be used as an audio impedance matching unit and give any decent high frequency passthrough. To be honest that's why most larger VA sizes don't even work on 400 HZ systems. Even at 400 Hz the impedance is too high to pass through much power.

 

...The inductive impedance of a common laminated iron core transformer at that frequency is very high so what loses one may have are typically negligible....

By inductive impedance I assume you mean from the leakage inductance which limits the high frequency response.
A transformer high frequency response is largely determined by its leakage inductance in series with the winding capacitance and load impedance.

Thus for a high frequency PWM signal modulated at 60Hz, the leakage inductance along with a filter capacitor at the transformer output will remove the high frequencies, leaving the 60 fundamental.

A low frequency transformer becomes lossy at higher frequencies because of the greater eddy-current losses which is proportional to the square of the frequency, which is another reason a 60Hz transformer does not work well at 400Hz.

 

Thank you all for the information.
As I understand, the Eddy Current losses are high at 23 KHz for this transformer and will result in Heating up the transformer Core.
Since I'm not planning an Induction Heater, I'll need to now look into designing a Ferrite Core Transformer.

 

By inductive impedance I assume you mean from the leakage inductance which limits the high frequency response.
A transformer high frequency response is largely determined by its leakage inductance in series with the winding capacitance and load impedance.

Thus for a high frequency PWM signal modulated at 60Hz, the leakage inductance along with a filter capacitor at the transformer output will remove the high frequencies, leaving the 60 fundamental.

A low frequency transformer becomes lossy at higher frequencies because of the greater eddy-current losses which is proportional to the square of the frequency, which is another reason a 60Hz transformer does not work well at 400Hz.

I've worked with countless inverter devices that used higher PWM carrier frequencies that drove common low frequency iron core loads that never had problems. Most any common VFD unit has a multi KHz to 10's of KHz PWM frequencies yet the iron core induction motors they drive have no such issues with overheating from it.

All practical hands on experience I have ever had has not shown any degree of HF inductive heating effects to any degree I ever had any reason to be concerned.

 

Thank you all for the information.
As I understand, the Eddy Current losses are high at 23 KHz for this transformer and will result in Heating up the transformer Core.
Since I'm not planning an Induction Heater, I'll need to now look into designing a Ferrite Core Transformer.

I would try experimenting with it before writing things off. My books say that even if you used the windings from the transformer as an air core inductor the inductive impedance (high turns ratio) would be too high to pass much power to begin with at 23KHZ.
I suspect that if you could test your transformers input winding set it would have a inductance of several hundred uH which at a 12 volt input would only allow a few watts of power to pass at that frequency at best meaning it will largely ignore the 23 KHz PWM frequency and just let the low frequency go through.

 

As I understand, the Eddy Current losses are high at 23 KHz for this transformer and will result in Heating up the transformer Core.

That's only if the transformer actually passes the 23kHz from the primary to the secondary through the core magnetics.
A standard power transformer will have sufficient primary leakage (series) inductance that will prevent significant 23kHz from passing.
You could add a small series inductor at the primary input to further suppress the 23kHz signal but I don't think that will be necessary.

I'll need to now look into designing a Ferrite Core Transformer.

That's not what you want.
Since you don't want to pass the 23kHz, only the 50Hz modulation, you don't want (or need) a high frequency transformer.
Also you still need a high magnetizing inductance to avoid saturation of the core by the 50Hz modulation, and that would require a huge ferrite core.

 

That's only if the transformer actually passes the 23kHz from the primary to the secondary through the core magnetics.
A standard power transformer will have sufficient primary leakage (series) inductance that will prevent significant 23kHz from passing.
You could add a small series inductor at the primary input to further suppress the 23kHz signal but I don't think that will be necessary.

That's basically how I see it.

In the early years of inverter power supply based welders and plasma cutters a few brands used laminated iron core transformers rather than ferrite cores due to at the time large ferrite core manufacturing was very limited and rather pricey.

The problem with getting a laminated iron core to work in the multi 10's of KHz range was almost a black magic art in itself. Typically they used non interleaved E and I core laminations with a huge air gap (~1/8" - 1/4" on a 6" x 6" core) between the E and the I sections to get them to run and even then they were rather fussy about the air gap to core size to operating frequency ballance.

Too high of operating frequency (~+ 500 - 1000 Hz would put them over the edge of their frequency response and the output would lose its I/O power/current/voltage ratio.
or to wide or narrow of air gap and they would lose output because the core inductive impedance would go too low or high.

Or if the unit got wet and the iron core laminations started to surface rust too much that too would change the magnetic properties enough to push them out of tune.

Then add any a combination of any of that and just wield stuff would happen where the unit would work at one output power/current or voltage and not another.

I only ever worked on one that had a core meltdown due to inductive heating though and I suspect that it only did that due to excessive high power use and abuse in a very dirty environment where it likely built up so much fine metal dust in the HF transformer assy that it started to melt down then took everything around it with it.

 

.......Surely the leakage inductance is high (relatively) and hence presents a high impedance at the primary and surely that limits the power that can be passed ?

That's only true if you are trying to pass the high frequency through the transformer, which here you are not.
In this case you actually want a high leakage inductance.

You have to understand the signal characteristics. It's a high frequency PWM signal that is modulated with a low (50-60Hz) signal.
It is the average value of this modulated signal (the 50-60Hz) that is of interest and that is obtained simply by running the PWM signal through a low-pass (LC) filter.
In this case the low-pass filter is provided by the transformer primary leakage inductance and stray capacitance.
What remains then is the 50-60Hz signal going through the transformer, with little of the carrier frequency.

So you can see that you don't need (or want) the high frequency signal going through the transformer, only the low frequency modulation, thus a standard, low-frequency mains transformer is fine for such an application.

 

.......Actually I can see no advantage in using said iron, I am pushing 2Kw through the top ferrite in my avitar yet its only 60mm square by 40mm thick. I guess there is an element of simplicity ......

You are still missing the point.
The purpose of the PWM is to efficiently generate a 50-60Hz sinewave from a DC supply.
You can push all the high frequency signals through a transformer you want but that really won't help us here.
You need that iron to pass the 50-60Hz without core saturation.
If you didn't have the large transformer, then you would need a large inductor to recover the 50-60Hz, so nothing would really be gained.

 

that ferrite transformer carries 50Khz modulated by 50hZ, think class D audio amplifier

That uses the speaker inductance and limited mechanical frequency response to filter the PWM signal.
You'd need to add an inductance at the output of the high frequency transformer for a 50-60Hz inverter.
So the tradeoff is between one larger 60Hz transformer, or one small, high-frequency transformer and a high current inductor and capacitor filter.

 

Thank you all for this informative discussion.

Suppose I go with the existing Power stage, but with the Gate Drive chopped at 23 KHz, like this,


and compare the No Load losses with that at present. That should give an idea on the transformer losses, which I guess, will be in the form of heating of the core.

Maybe the 23 KHz astable will require some sort of a sync with the 3524 clock?