Project MOSFET H-Bridge to generate single phase signal

[Kai Xiang]

Hi all,
I am currently working on a project which is able to generate three phase signal, using a microcontroller Teensy++2.0 and MOSFET circuits. The MOSFET circuit will be able to convert/smoothen the square waves with variable duty cycles generated by the microcontroller into reasonable sine waves. To begin, I started with a single phase.

NI MultiSim is used as the simulation tool, and for that purpose opamp together with sine wave and triangular wave generator replaced the microcontroller to generate the required square waves for testing purposes.

Objective: To get a reasonable looking sine wave
After weeks of trying and troubleshooting, I am still unable to get the required output (reasonable looking sine wave).

My circuit design is shown below: (Left shows the function gen used) (Mid is the circuit) (right is the output that I generated)


I would appreciate your advice and guidance.
Please help.

Thank you so much.
KX

 

[Harald Kapp]

Try a low pass filter between the output of the H-bridge and the input of the oscilloscope.

 

[kellys_eye]

The grounding around the top opto-coupler seems wrong to me.

 

[Kai Xiang]

Hi Harald and Kellys,

Thank you for the advice. I tried to modify the circuit accordingly, the signal is now much cleaner however still not a reasonable sine wave yet. Below is the revised circuit, appreciate your help to advise.



Thank you so much.
Best Regards,
KX

 

[Harald Kapp]

The high side drive of your circuit is less than ideal. Consider Q1:
With U1 active, the gate is pulled to 0V, turning Q1 off.
With U1 inactive, the gate is pulled up to 9 V by R1. Since the threshold voltage of Q1 is in the range 2V ... 4 V, the source of Q1 will reach max. 5 V ... 7 V. It cannot reach 9 V, for that you need a gate voltage > 13 V (worst case Vgsth = 4 V).
Or use P-channel MOSFETs in the upper legs of the H-bridge.

Also the pull-up by R1 creates a comparatively slow edge: The gate-source capacitance of Q1 is ~ 1300 pF. Since the gates of Q1 and Q4 are in parallel, the total capacitance seen by R1 is 2.6 nF. Together with R1 these capacitances create a low pass filter with a time constant of 1.8 µs.

C4 is not well placed. It will alternately be charged and discharged by Q3 and Q4.This creates a lot of wasted power in these transistors. Remove C4.
When I mentioned a low pass I meant an RC filter from the right side of R6 to the input of the oscilloscope.

 

[Terry01]

HOLY SMOKE HARALD!

That's all....just HSH!

How can you know all of that off the top of your head? That's encyclopedic!
Tip my hat to that one for sure......I wonder if I'll ever know that much. No laughing,I'm serious.

 

[Kai Xiang]

Hi Harald, thank you so much for the help.
There are a number of points which I don't really understand and is rather confused. I tried to interpret as much and came out with a revised circuit using P-channel MOSFETs in the upper legs of H-Bridge.

I would really appreciate your guidance and support.
Thank you so much.

 

[Harald Kapp]

The P-Channel MOSFET needs an inverted gate source voltage.

When the output of the opamp is at low voltage, U1 will be active, pulling the gate of Q1 low. Q1 will be on.The same voltaeg is connected to the gate of Q4, which will be off.
U2 will be inactive, pulling the gate of Q2 high which turns on Q" and turns off Q3.
You have created a short circuit between VDD and GND via Q! and Q2 which are both on.

Similarly a short circuit will be created between Q3 abd Q4 when the output of the opamp is at high voltage.

Swap the gate connections of Q2 and Q4. That way either Q1/Q4 or Q2/Q3 will be on in pairs.

Read here and here how an H-bridge works.

There's also the question of why you use photocouplers. Since you're operating all parts from the same VCC/GND, the isolation by the photocouplers is not necessary.

THe low pass filter at the output of the opamp is also detrimental. R5, C2, C28 will round your pulses. This in turn will slow the turn-on and turn-off of the MOSFETs. Even with correct gate connections (see above) this will increase the power dissipation due to switching and current spikes from VDD to GND when both transistors in a leg (e.g. Q1, Q2) are partly on at the same time.
On the contrary, a good H-bridge driver would not simply invert the gate drive (as your circuit does) but would insert a brief dead-time, a period where none of the MOSFETs is on to minimize this effect. See the above links.

 

[Kai Xiang]

Hi Harald,

I have read through and have revised the circuit.
There are some parts which I couldn't understand in the links given above.

The circuit is still unable to generate the required signal sine wave and is getting further.
I would really appreciate if you could help me out.

I based this project on an example found online (Please find attached)

Thank you so much.
KX.

 

[Harald Kapp]

There are a number of points which I don't really understand and is rather confused. I tried to interpret as much and came out with a revised circuit using P-channel MOSFETs in the upper legs of H-Bridge.

No, you don't. Your circuit still shows IRF840 for all 4 MOSFETs.

You also missed this part:
QUOTE="Harald Kapp, post: 1743291, member: 22009"]Swap the gate connections of Q2 and Q4. That way either Q1/Q4 or Q2/Q3 will be on in pairs.[/QUOTE]
As it is, with th eoutput of the opamp high, U1 is off, U2 is on.
U1 off provides a positive gate voltage to Q1 and Q4. but Q1 being an NMOS will not turn on completely, cf. post #5. And if Q1 were a PMOS, as I suggested, it would be off.

Maybe this video helps you more than my verbal explanations?

 

[Kai Xiang]

Hi Harald,

Can the IRF840 be used as P-channel by switching the drain and the source?
If not, do you have any idea what P-channel MOSFET is equivalent to the N-channel IRF840?

I have swapped the gate connections of Q2 and Q4. Am I doing it correctly?

Thank you so much for the advice and guidance and I appreciate it so much.
KX.

 

[Harald Kapp]

Can the IRF840 be used as P-channel by switching the drain and the source?

No, these are different technologies.

do you have any idea what P-channel MOSFET is equivalent to the N-channel IRF840?

Sorry, I don't. But you don't need a 500 V MOSFET here. USe any P-channel MOSFET with a Drain-source voltega higher than VDS (e.g. use 20 V) and a low Rdson (<100mΩ)

I have swapped the gate connections of Q2 and Q4. Am I doing it correctly?

Looks good if you use P-channel MOSFETs for Q1 and Q3.

 

[Kai Xiang]

Hi Harald,

Changing Q1 and Q3 into P-Channel MOSFET, the output is as follows. Have I done it correctly?





Not that much of a sine wave yet, I would really appreciate your help and guidance.

Thank you so much.
Regards,
KX.

 

[Harald Kapp]

Have I done it correctly?

Unfortunately not. the source of the PMOS needs to go to VDD.

 

[Kai Xiang]

Hi Harald,

I have connected the source of PMOS to VDD, but the output seems to be the same.



Please advice me.
Thank you.
KX

 

[Harald Kapp]

remove C2, connect R5 directly to Q3/Q4. put an RC low pass filter between h-bridge and scope:

Adjust the filter frequency to ~1/10 PWM frequency.

 

[Kai Xiang]

Hi Harald,
Am I doing it correctly?
Can you explain the ~1/10 PWM frequency?

Thank you.
KX

 

[Harald Kapp]

Can you explain the ~1/10 PWM frequency?

If, for example, your pwm frequency is 10 kHz, set the low pass filter corner frequency to 1 kHz.

Am I doing it correctly?

No. Why did you connect the gate of Q4 to the gate of Q2? Q3'S gaet needs to be connected to Q4's gate. THsi was correct in post #11, but then you changed it.

 

[Kai Xiang]

Hi Harald,

Following your guidance, the waveform is now so much nicer!
Is it normal for the signal to start high and slowly reduces to the sine wave?

I tried setting the low pass filter to 1.5kHz, but was not successful, the frequency fluctuates and reduces to 1kHz eventually. My output from the op-amp is 15kHz.

\



Thank you so much!
KX

 

[Harald Kapp]

C2 is still in the wrong place. See my post #16 for the correct position. M1 is equivalent to Q4 in your circuit, M2 is equivalent to Q3.

tried setting the low pass filter to 1.5kHz, but was not successful, the frequency fluctuates and reduces to 1kHz eventually

???
The frequency response of an RC lowpass filter is determined by the values of R and C and does not fluctuate and settle.