MOSFET burns - 220V inverter for induction motor speed control using Arduino.

shortbus

Joined Sep 30, 2009
9,330
is it possible to make a VFD with this breadboard?
i want to try on it then, later after I found my way then migrate to PCB board, but I would like to know about this.
What breadboard are you talking about? If you mean the PCB in my link many have done it. It's in the manual for that PCB. Things like this shouldn't be done on a breadboard. Breadboards are for low voltage low amperage things.
 

Thread Starter

Claudez

Joined Mar 24, 2019
28
What breadboard are you talking about? If you mean the PCB in my link many have done it. It's in the manual for that PCB. Things like this shouldn't be done on a breadboard. Breadboards are for low voltage low amperage things.
I mean solderbreadboard, perfboard.
 

Thread Starter

Claudez

Joined Mar 24, 2019
28
Ooh, i saw that board. Additional from my circuit after comparison is voltage, temperature and current feedbacks sensors.
But looking for more and i will migrate to pcb later.
 

Thread Starter

Claudez

Joined Mar 24, 2019
28
Why would you even think those things aren't needed or relevant to what you want to build?
I didn’t think they are not needed, they are safety features. I kinda point out the weaknesses of my circuit.
Though i expected at-least for the circuit to work, but maybe due to these factors my design fails.
I’m considering buying that board, but i saw it doesn’t work with higher DC source.
 

Irving

Joined Jan 30, 2016
2,562
Your design has limitations. It works OK into a small load with a power-factor of 1 ie a resistive load.. Here is a simple simulation. I couldn't find a working Spice model for the IRF830 so I've used something very similar and also included the intrinsic diodes, here modelled by actual diodes of similar characteristics, The stuff inside the dotted line to the left is a representation of the driver circuit, generating the necessary gate voltages and currents for PWM to the lower MOSFETs and phase on/off to the upper MOSFETs as is common practice; you never PWM the upper devices. These signals are shown in the upper two traces. The lower two traces show the voltage across and current through a 600ohm resistive load designated R_motor (R1 and R2 are there to separate the d1,d3 nets from the r,r1 net for later load types). As can be seen, the voltage and current in the load is a faithful representation of the PWM signal on both positive and negative 1/2 cycles (see also the close-up of the central zero crossing). Because the load is resistive there is no integration of the PWM, but the heating effect on the load will be near equivalent to that of a real AC mains feed. It is 99W against the theoretical 80W (220^2/600) of a real AC feed because the PWM is slightly overdriven.

1638706191508.png
1638706977812.png

Now lets add in the inductance of the fan motor - the one I have here is a similar loading to yours and measures 475mH...
As expected the inductor serves to integrate the current so we see the expected sine wave with a small phase shift from the voltage, hence the discontinuity. The phase shift represents a power factor of about 0.9, typical for a motor of this type. Also shown is the power in the motor, around 72W, down from the resistive case since the reactance of the inductor comes into play. Here the lower MOSFETs are dissipating around 90mW each over the mains cycle, but individual switching transients are close to 600W during the PWM period (see the close up) - its hard to be sure but its possible this is just outside the safe operating area for the IRF830.

1638710571725.png1638711747353.png

Coming, finally, to your filter - this produces some very odd effects because it is not symmetrical. And as seen above, its not needed.

1638735979224.png
 

Irving

Joined Jan 30, 2016
2,562
Does this board work with 310VDC or i would need to step up the output voltage via transformer?
Let's be clear, that board is just a driver, it generates the necessary signals to drive the power output stages but it's not an inverter. If the problem you have is in your output stages i doubt it's going to help. It will probably take you just as long to make it work as you have taken already but with added complexity. However the data sheet shows how the output stages can be organised including the correct arrangement of the output filter and you should look at that.

Coming back to your board, do you have an oscilloscope or some way to look at the driver signals? I'm not convinced you are generating them correctly.

To be honest, if controlling the speed of a fan motor is your end goal then this is the wrong approach. The near ubiquitous Triac dimmer is a much better solution.
 

Irving

Joined Jan 30, 2016
2,562
Simulating your circuit with a revised filter layout as per that inverter driver data sheet and playing with the values a little shows that the power factor can be 'tuned out' and a much better waveform created. This gets rid of the discontinuity and improves the power factor to around 0.95, bringing down the peak dissipation in the MOSFETs to <400W. Try it...

1638796085687.png
 

Thread Starter

Claudez

Joined Mar 24, 2019
28
Thank you so much for your explanations, actually, I have an oscilloscope and I have checked the output of my gate driver, with the small load, it is fine. The pulses are generated fine, even when I use a small load it is good waveforms.

I read somewhere that the eg8010 is good for low voltages, that's why I asked, but I'll follow your recommendations and see the waveforms too. and learn from it in comparison to my circuit.

I have generated an spwm signal from Arduino that also I can vary the frequency down to 0, but currently, I try on fixed freq first.
I also applied the filter for the small load and the output was a decent sine wave haha.

please refer to the attached post, and my scope can take 300V is it safe to use it in this test? I read that it is possible because the probes can take 1000V, but I'm not confident and sure, haha.

I have never scope the output of filter cap from the inverter because I’m not sure, my scope is 300V rated, i heard because my probes are 1000V rated is ok to measure over 300V, how is this?
This is my probe photo?
 

Irving

Joined Jan 30, 2016
2,562
I wouldn't without an HV probe - I managed to blow up a 300V rated Picoscope when fixing an SMPS a few years back. There are transients there to 1000v+ and though your scope input is rated at 300v (mine says 400vpk) it doesn't take much to punch through the sensitive input circuits.

A simple HV probe is easy to make. We need a 100:1 reduction against the 1Mohm input resistance of the scope, so 99M in series. Since most resistors are rated at 200v typically we need at least 5. I used 9 x 10Mohm + 1 x 9.1Mohm laid out in a zigzag on a piece of stripboard with a banana plug soldered to the end so I could attach various probes. That went into a 1.5cm diameter rigid plastic tube/handle.

A better solution that overcomes the grounding problem (where is ground on a rectified AC input? unless you have an isolating transformer its neither DC- or DC+ its actually in between giving you a 150v DC offset) is to make a differential probe like this:

1638820836570.png
 
How much time did it take to fail? 1) Instantaneous 2) 3 to 10 Seconds 3) about minute OR total random.
Try to run the system at 110 V AC if possible. Increase voltage in steps.
I think 4N35 is too slow for 10khz switching.
MOSFETS must need RC snubbers across the Drain and gate.
 

Thread Starter

Claudez

Joined Mar 24, 2019
28
How much time did it take to fail? 1) Instantaneous 2) 3 to 10 Seconds 3) about minute OR total random.
Try to run the system at 110 V AC if possible. Increase voltage in steps.
I think 4N35 is too slow for 10khz switching.
MOSFETS must need RC snubbers across the Drain and gate.
Once i turn ON the power, the MOSFET explode. And gate driver fails too.
Actually 4n35 is slow, but I archived fast turn OFF but pulling it’s gate to emitter by 10k ohm. And i check the waveform with the oscilloscope, it is pretty well.

Here it is.
 

Attachments

Thread Starter

Claudez

Joined Mar 24, 2019
28
I also think the high side MOSFETS never turn OFF? Or is this the signal from the other side of MOSFETs?

If i connect only half bridge, i can see the MOSFETs are never ON at the same time, high side MOSFET should have signal higher than power source but here when it is supposed to be off signal is low, does it still affect when we have higher source voltage?
 

Attachments

Thread Starter

Claudez

Joined Mar 24, 2019
28
I wouldn't without an HV probe - I managed to blow up a 300V rated Picoscope when fixing an SMPS a few years back. There are transients there to 1000v+ and though your scope input is rated at 300v (mine says 400vpk) it doesn't take much to punch through the sensitive input circuits.

A simple HV probe is easy to make. We need a 100:1 reduction against the 1Mohm input resistance of the scope, so 99M in series. Since most resistors are rated at 200v typically we need at least 5. I used 9 x 10Mohm + 1 x 9.1Mohm laid out in a zigzag on a piece of stripboard with a banana plug soldered to the end so I could attach various probes. That went into a 1.5cm diameter rigid plastic tube/handle.

A better solution that overcomes the grounding problem (where is ground on a rectified AC input? unless you have an isolating transformer its neither DC- or DC+ its actually in between giving you a 150v DC offset) is to make a differential probe like this:

View attachment 254363

Actually, I didn’t understand the reason you have explained about using resistors in such fashion, does is has much to do with handling of higher voltage signal that i want to look on oscilloscope?
Because i want to measure the ripples on the rectifier, bank capacitor and at the h-bridge output so i can see the waveforms.

But since the vdc will be 310V, that’s why i was concerned.
Can you please make it clear? I actually didn’t understand. Thank you.
 
Top