PWM 5 - 15V/35A

Thread Starter

riko

Joined Sep 7, 2014
11
Hello everyone.
I have to create PWM that will supply fuel cell.
Here is the concept: TRANSFORMER - RECTIFIER - PWM - LOAD
I can handle the transformer, rectifier and MCU that will produce the pulses, but I have troubles with the power driver/module that will amplify the pulses. For example: what transistors should I use, should I use inductance at the output, should I use reverse diode etc.? The following ling shows my idea which is probably wrong, but the enclosed part of the schematic is what I am not familiar with.
https://drive.google.com/file/d/0B_EYtdS7J5eKcXI3dm96N3pvQVk/view?usp=sharing
This schematic is only a concept, but I need something steady that will withstand 15V/35A for a long time. Therefore I ask you for help. I need help only for the powerful part of the schematic. The transformer, rectifier and the MCU is my concern.
If you need more info, dont hesitate to ask.
Any help would be much appreciated!
 

Papabravo

Joined Feb 24, 2006
21,159
Hello everyone.
I have to create PWM that will supply fuel cell.
Here is the concept: TRANSFORMER - RECTIFIER - PWM - LOAD
I can handle the transformer, rectifier and MCU that will produce the pulses, but I have troubles with the power driver/module that will amplify the pulses. For example: what transistors should I use, should I use inductance at the output, should I use reverse diode etc.? The following ling shows my idea which is probably wrong, but the enclosed part of the schematic is what I am not familiar with.
https://drive.google.com/file/d/0B_EYtdS7J5eKcXI3dm96N3pvQVk/view?usp=sharing
This schematic is only a concept, but I need something steady that will withstand 15V/35A for a long time. Therefore I ask you for help. I need help only for the powerful part of the schematic. The transformer, rectifier and the MCU is my concern.
If you need more info, dont hesitate to ask.
Any help would be much appreciated!
What comes out of the bridge rectifier is an AC waveform with about 21.2 Volt peaks. A capacitor of some description would be required to make this into a DC voltage with some ripple. PWM with an AC voltage does not make much sense after all. From your requirement that the PWM output needs to be from 5 to 15 volts I would suggest a pre-regulator that produces such a DC output. Once we have a stable DC output we can then apply a the PWM. Your preliminary drawing shows two MOSFETS in parallel. Was this your intention? You need to ensure that component variations do not allow one of them to pass more current than the other. You also need to ensure that your driver turns the device on and off quickly. If it spends too much time in the linear region the MOSFETS will get very hot. I've tried several times to build discrete drivers for MOSFETS and always been disappointed in the results. A driver IC that can bootstrap a higher gate voltage is something I've always wanted to try but never got around to.

Good Luck.

Can you supply a link to the MOSFET datasheet. I've had no luck with IRF4095 or IRF4005 and I can't read your schematic.
 
Last edited:

ronv

Joined Nov 12, 2008
3,770
What is your power supply for the micro? The reason I ask is that it may be easier (cheaper) to not filter the big power supply and then use an NFET to switch the ground side of the load.
 

Thread Starter

riko

Joined Sep 7, 2014
11
Thank you for the prompt reply.
The specification of the load (fuel cell) requires the supply voltage not to be filtered by capacitor, it works better with rectified not filtered voltage. So the schematic I post shows that there will be only the positive waves after the bridge rectifier. I admit my explanation was not exact. The MCU should detect the zero crossing and based that signal, will realize pulse with different duty (up to 10ms for 100% duty). The pulses should drive the transistors so that they will conduct only part of sine wave depending on the duty being set. Thus I plan to regulate the output power. It acts as dimmer, but transistors are used instead of thyristor.
The micro will ne supplied voltage regulator 7805 with 5V. Pretty reliable filter is required because of the glitches produced by the heavy load.
 

Alec_t

Joined Sep 17, 2013
14,280
Depending on the type of fuel cell and its thermal capacity, could you simply omit occasional cycles to control average power? That would allow switching to be done at the zero-current points and so avoid the need to switch heavy currents, which in turn would mean fewer glitches and less EMI generation.
 

Thread Starter

riko

Joined Sep 7, 2014
11
Thank you Alec_t. What you suggest is very interesting idea, but the customer wants to be able to fine tune the power. If I get the idea correct, it wouldnt be possible to perform fine tune by cutting occasional cycles. Having omited one cycle per 50cycles not good idea.
 

crutschow

Joined Mar 14, 2008
34,285
If you want to efficiently control the current with PWM then you will need a series inductor and a free-wheel diode in conjunction with the load. Otherwise you will get high current spikes through the fuel cell (or would spikes be acceptable?)
 

Thread Starter

riko

Joined Sep 7, 2014
11
Thank you for the valuable advice!
Dear rony - http://store.comet.bg/en/Catalogue/Product/6336/ The input voltage is allowed to reach 35V. I am planning to supply the 7805 from the power transformer which secondary voltage will be approximately 15VAC.
Dear crutshow, how to calculate the value of the inductor needed, more specifically the inductance [u\mH]?
The same transformer will supply the fuel cell.
Taking in account that the load will reach current of 35A may be more, what transistors should I use? I know that two or more transistors in parallel is not good idea, since the current will flow (mostly) trough the one of them. I need to develop schematics that will cause minimum heat at the transistors. I will drive the output nearly at zero voltage, but the transistors off-state might be everywhere. If I switch off the transistor at the peak of the sine wave, that would produce glitches, right? The transition from state on to state off, would cause heating of the transistors. This is the (transition)time, I want to decrease at minimum. That's why I need your help, to show me some tested techniques that will make my device work steady.
The following diagram will shed some light on what I want to achieve: https://drive.google.com/file/d/0B_EYtdS7J5eKbEpuMVVUOWRldTg/view?usp=sharing
 
Last edited:

ronv

Joined Nov 12, 2008
3,770
OK, so you will need another regulator so the gate voltage does not go to high.
Here is the same circuit with the waveforms like your picture.
You do know that you need to filter the voltage going to the 7805? You could make the input to the 7805 from the output of the 7809 in the schematic. It might help with noise.
The fet I posted in the schematic will just need a small heat sink - maybe 6 or 10C per watt.
 

Attachments

crutschow

Joined Mar 14, 2008
34,285
...................
Dear crutshow, how to calculate the value of the inductor needed, more specifically the inductance [u\mH]?
The same transformer will supply the fuel cell.
Taking in account that the load will reach current of 35A may be more, what transistors should I use? I know that two or more transistors in parallel is not good idea, since the current will flow (mostly) trough the one of them. I need to develop schematics that will cause minimum heat at the transistors. I will drive the output nearly at zero voltage, but the transistors off-state might be everywhere. If I switch off the transistor at the peak of the sine wave, that would produce glitches, right? The transition from state on to state off, would cause heating of the transistors. This is the (transition)time, I want to decrease at minimum. That's why I need your help, to show me some tested techniques that will make my device work steady.
The following diagram will shed some light on what I want to achieve: https://drive.google.com/file/d/0B_EYtdS7J5eKbEpuMVVUOWRldTg/view?usp=sharing
The inductance depends upon the switching frequency frequency and the allowable current ripple. The formula is L = V (dt/di) where V is the difference between the input (switch) voltage and output (load) inductor voltage, dt is the switching period, and di is the allowed ripple current. The inductor needs to be able to carry at least 40A without saturating which likely makes for a reasonably large inductor.

You can use any MOSFET rated for at least 30V and 50A. The lower its ON-resistance the lower the losses when it is conducting. Any glitches should be shunted away by the free-wheel diode. The switching losses should be low if the MOSFET gate is driven by a high current driver that can rapidly charge and discharge the large gate capacitance of a MOSFET.
 

ronv

Joined Nov 12, 2008
3,770
Carl,
I don't think he is making a power supply, just a low frequency PWM so he doesn't boil over his cell.
 

crutschow

Joined Mar 14, 2008
34,285
Carl,
I don't think he is making a power supply, just a low frequency PWM so he doesn't boil over his cell.
I understand that. But the concern is does he want good efficiency and can the cell tolerate high peak currents. If he wants good efficiency and/or low peak currents then he likely need to add an inductor. If not, he doesn't.
 

ronv

Joined Nov 12, 2008
3,770
It's cleaver turning it on at the zero crossing, so he doesn't need to charge the plates with a fast rise time. Was that your concern?
 
Top