pulse width modulated inverter

[seven_segment]

Pulse width modulation

I'm breadboarding a sine wave inverter using variable PWM and adjusting the mark/space ratio. I have a PIC micro runing @ 20Mhz with a quarter wave sine lookup table with 10 values per quadrant. I'm trying to achieve an output frequency of 50Hz, PWM input to the transformer is currently at 2Khz. The transformer is a standard 50Hz chassis mount, I had hoped that it would filter the input and give a relatively clean sine wave, but that doesn't seem to be the case.

The driver circuit consists of two sets of BC108/2N3055 darlington pairs. The transformer has dual secondary coils, the centre tap is wired to positive and the outer taps are grounded at set intervals during each half cycle using timing values from the lookup table. Note that only one transistor is involved in switching during each half cycle, the other one is left in the off state.

The transformer is rated at 20VA. From the primary I'm driving another smaller (3VA) transformer to give (along with a resistor voltage divider) a suitable voltage to input to my sound card so I can record the waveform. I have attached an image showing the output them I'm getting from this thing. It vaguely resembles a sine wave, although wasn't what I was hoping for. Can anyone tell me where I'm going wrong?

Cheers,

7 segment

 

[seven_segment]

Pulse width modulation

Anyone?

 

[flatulent]

Re: Pulse width modulation

Could you post the part of the circuit that this is a measurement of.

Have you measured the base drive and collector waveforms? Another possibility is that the darlington transistors have slow response time.

 

[jorgito]

Pulse width modulation

Hello,
Could you please show us your complete power stage schematics?
Have you used freewheeling diodes?
When you turn off the active transistor, the current flowing in the transformer inductance must be continuous.
Best regards

 

[seven_segment]

Re: Pulse width modulation

No problem, I've attached a schematic to this post. As you can see, it's pretty simple, bear in mind this is only a prototype.

Thanks again,

7 segment

 

[jorgito]

Pulse width modulation

Hello,
You don't have load.
Please try loading the first transformer.
While any transistor ON, you are storing energy in the transformer, but have to dump it somewhere before the next cycle.

 

[seven_segment]

Pulse width modulation

Unfortunately that's something I've already tried, using a 15w bulb as the load I'm still getting the same problems

 

[electronrancher]

Pulse width modulation

i'd like to make a suggestion that a pic is somewhat weak for driving a power bipolar. Can you use your scope to take a photo on the bjt side of the 1k? Or better yet, break the 1k into 2 500Ohm, and scope between them to get current sense info as well.

maybe you want to make a cheap buffer out of hex inverters to drive your power bjts - wire all six inverters in parallel, then use this module to drive the 1k (which should by the way be lowered to 100Ohm). These will probably get hot, so you may need to replace them from time to time if they die.

Also, commercial power supply driver IC's can be obtained, which give you the equivalent of a 1-amp inverter you can control with just a logic signal - that may be an idea for rev 2!

Last, please put freewheeling diodes in as jorgito said. A diode from each collector to the center 12v tap should help. (stripe to +12v supply) This diode is reverse biased when the bjt is on, but drains off the inductive kickback that occurrs when the bjt turns off - that may be the reason you get so much negative "fuzz" on the positive half-cycle, I bet you have a lot of kickback with no diodes. give it a try.

 

[seven_segment]

Pulse width modulation

OK thanks for the suggestions everyone, I'll try out some of your suggestions and get back to you on how it went.

I'm not sure that the problem is caused by driving from the PIC without a buffer, each output can soirce/sink about 20mA on the model I'm using. Taking in to account the combined gain of both transistors, this should be more than enough current. However, I will scope it as you suggest.

 

[jorgito]

Pulse width modulation

Maybe you could also reduce switching times placing a capacitor in parallel with the base resistors.

 

[seven_segment]

Pulse width modulation

I haven't tried your capacitor suggestion yet, but I have tried adding freewheeling diodes and I've also checked the waveform at the base of Q1 and Q3. The waveform there looks as it should, and adding the diodes didn't help, it just severely reduced the amplitude at the primary of L1

I've had a quick check on the maximum switching frequency of the transistors I'm using, and as far as I can tell I am within the specified limits. I'm getting close to giving up on this one, looks like I might end up going down the modified sine wave route.

Thanks once again for all the suggestions, although I still can't figure out what I'm doing wrong.

 

[jorgito]

Pulse width modulation

It was previsible.
When you turn, let's say, Q4 ON, you have 24 V on the Q2's colector due to transformer action. this voltage forward biases the corresponding diode and reduces efective voltage on the primary.
I see only one source for your negative voltage pulses and it is the turn off kickback in the conducting TR. Again, by transformer action, it results in a negative voltage in the other (non conducting) TR.
Diodes in parallel with both transistors may supress this negative voltage sending some energy back to supply.

 

[bctang]

Re: Pulse width modulation

I'm breadboarding a sine wave inverter using variable PWM and adjusting the mark/space ratio. I have a PIC micro runing @ 20Mhz with a quarter wave sine lookup table with 10 values per quadrant. I'm trying to achieve an output frequency of 50Hz, PWM input to the transformer is currently at 2Khz. The transformer is a standard 50Hz chassis mount, I had hoped that it would filter the input and give a relatively clean sine wave, but that doesn't seem to be the case.

The driver circuit consists of two sets of BC108/2N3055 darlington pairs. The transformer has dual secondary coils, the centre tap is wired to positive and the outer taps are grounded at set intervals during each half cycle using timing values from the lookup table. Note that only one transistor is involved in switching during each half cycle, the other one is left in the off state.

The transformer is rated at 20VA. From the primary I'm driving another smaller (3VA) transformer to give (along with a resistor voltage divider) a suitable voltage to input to my sound card so I can record the waveform. I have attached an image showing the output them I'm getting from this thing. It vaguely resembles a sine wave, although wasn't what I was hoping for. Can anyone tell me where I'm going wrong?

Cheers,

7 segment

Hai, seven_segment

Do you known how to write the simple PWM signal using PIC?

Actually, i need the Fully-Wave-Rectifier-Sine-Wave modulated pulses at 50Hz program.
Than, using this pulses to compare the triangle-wave and produce the PWM signal.

I am using PIC16F84A or PIC16F877A chip and i using CCS compiler and C program.

Bellow is showing the practical result:



Thank You........

 

[seven_segment]

Pulse width modulation

Sorry, but it sounds like you're going about this a different way to me. I'm not comparing a triangle and sine wave to create the pulses, I'm simply retrieving the values from a sine lookup table. My project still isn't working yet, but I'll let you know if I succeed.

 

[platonas]

Re: Pulse width modulation

Hi
Try connecting a resistor to the base of Q2 and Q4 that might reduce leakage currents of previous stage (transistor). This resistor should be connected from base to emitter (ground) and must have a relatively low value, eg between 820 Ohms to 2,2K. Usually a 1K or 1,2K should work.
To me it seems that the coplimentrary pair is softly on hence the other cycle has some level.

 

[seven_segment]

Pulse width modulation

Thanks, I will give this a try and see how things go whenever I next get chance to work on the project. I've been pretty busy recently and haven't had chance to check this forum, so sorry if my reply is a little late.

 

[kileen]

Re: Pulse width modulation

I think we are missing the basics here. A higher freq PWM signal has to be passed thru a low pass filter to recover the sine wave info in it. Try it.

 

[probert]

Re: Pulse width modulation

Hi 7segment,

For switch mode applications it is advised to use ferrite core transformers
and fast recovery reverse diodes, or a power switches that has them internally integrated. You can find lot of information at International Rectifier, Intersil or on the others websites about power switchers.

Probert

 

[seven_segment]

Re: Pulse width modulation

Sorry about the late reply guys, I've been away on holiday for a while and haven't been able to keep an eye on the forum.

Kileen, I'm using a standard 50Hz transformer for prototyping, I assumed since this is designed to operate at 50Hz that it would act as a low pass filter and recover the sinewave?

Probert, as mentioned above, I am using a standard 50Hz power transformer in the hope that it will act as a low pass filter. However, the plan is eventually to switch to a ferrite cored transformer and seperate filter, but not until I can get this prototype working

 

[kileen]

Re: Pulse width modulation

My turn to apologise...I too was on a vacation...

Seveseg...Will a inductor alone offer the necessary reactance for the low pass filter ?

I do have doubts and feel you would need a capacitance at the output side..