Dear friends,

I'm trying to design a step down buck converter. I have couple of issues here to discuss.
Let me give you a brief idea on what I'm working on:
- The input voltage is 15 to 30 DC volts, and the output is 5 volts.
- The output current is 5 Amps.
I can figure the duty cycle, I ripple, and all that stuff EXCEPT! I was trying to find inductance, but couldn't figure out how to determine frequency!
I wanna know what determines frequency and how to calculate it in order to find the value of components (inductance and capacitance).

another issue is the variable input. If I have a variable input then the duty cycle will be an interval of numbers!

Please I need head ups on finding frequency

Thanks

 

Hi,

Selecting the operating frequency is somewhat arbitrary. In a perfect world it wouldn’t matter what frequency you picked, but in the real world it matters.

Start by considering the specifications of the control chip and the switching device (transistor or MOSFET) that you want to use. The datasheets will give you an idea of how fast they can switch, therefore giving you an upper limit of oscillating frequency. If the control chip is made for SMPS applications then I like to use the oscillator frequency recommended, or used in an application note as a starting point for my design.

Generally, higher frequencies allow you to use, smaller, less expensive components, and/or easier to make inductors. Since your current requirement is 5 Amps, the inductor will need to have a low winding resistance. Therefore, the smaller the inductance value is, the easier the inductor is to make. This means the frequency will have to be higher to get the required duty cycle for max load.

Sometimes I’ll have an inductor on hand that I’d like to use so I re-work the formula to get the frequency of the oscillator. If that frequency is suitable for the other components then I can proceed with the design.

The controller will automatically adjust the duty-cycle to compensate for varying load and input voltage. However, be careful not to be too demanding with input Voltage range and output load change to a point where the duty-cycle can’t change enough to compensate. A minimum load may be required.

What parts are you using? Post a schematic.

Regards,
Ifixit

 

Thank you. This is very helpful.
This is my circuit. I'll be using those components in the schematic

I will probably buy an inductor instead of using one. This kind of circuit can probably be built on a breadboard, and the components shouldn't be big!

I can use minimum input voltage and calculate everything else based on it (15V). However, I'm afraid if I do so and try to increase input voltage, something might burn or the output voltage might change!

Plus, I want to know what's the relationship between frequency and the duty cycle, how can I use frequency to obtain the calculated duty cycle


Thanks again

 

http://www.national.com/an/AN/AN-556.pdf

 

You won't be able to build a 5A switching supply on a breadboard; you'll have too many parasitic problems, and the current is too high.

You'll need to use a circuit board; if nothing else build it "dead bug" style.

 

You won't be able to build a 5A switching supply on a breadboard; you'll have too many parasitic problems, and the current is too high.

You'll need to use a circuit board; if nothing else build it "dead bug" style.

my converter will be a basic one!
There will be no IC's. It has to be basic one with inductors, capacitor, resistor

 

my converter will be a basic one!
There will be no IC's. It has to be basic one with inductors, capacitor, resistor

How are you gonna control the mosfet?

 

Our OP doesn't show a MOSFET in their circuit design.
They show a mechanical switch, some generic rectifier, some generic inductor, and some generic polarized capacitor.

Wonder how quickly they can flip that switch on and off with their fingers? Oh, no fingers used - they aren't in the schematic....

 

How are you gonna control the mosfet?

There will be a control system. I'll have a MOSFET driver and will program it to controll the switching. This will come later on. For now, I have to figure the values of the components L and C for the buck circuit. After the circuit simulation is working, I'll work on the controlling part.

 

The frequency depends on the inductor size,the load current, and the input voltage. The duty cycle depends on the frequency, the inductor, the load current, and the input voltage. All these things are interconnected and your question gives so little information that nobody could guess what the answers are. If they did, it would require several pages to answer.

If you go to national.com and do some experimental designs with their computer generated switching designs, you can get an idea what range of frequencies and duty cycles are available and useful.

 

I built one of these for bucking voltage from a solar panel. You can find my design here. http://www.backyardsolar.blogspot.com/
schematic here: http://1.bp.blogspot.com/-Qr8J8iBYL.../x0sYyAA02BE/s1600/Buck_Converter_4-14-11.png

This design suffers from shoot through in the driver circuit which I plan to fix in the next rev. But it works well (has been working for months) charging a battery that runs some architectural lights.

The input in that schematic is driven from a PIC's PWM peripheral.

Figuring out the math for your design is a good start, and using the simulators to test your circuit is useful too (LT Spice). I've found that, with the magnetics, results vary and it comes down to having a good control loop to get out what you want. Specifically I mean testing the voltage on the output and adjusting the pwm based on that.

I would build a circuit, breadboard, perf board, whatever and just run 500mA through it to start then work your way up to 5 Amps. There will be a lot of problems that don't show up until you increase the power and 5 Amps is a lot of power. In my experience the parasitic L&C on the breadboard is not such a big problem to prevent the circuit from working. Things are slightly cleaner with a PCB but the same basic problems exist that existed with the breadboard with the big looping wires, etc.

Also, you need a decent scope to solve problems with the magnetic voltage transformations.

 

I built one of these for bucking voltage from a solar panel. You can find my design here. http://www.backyardsolar.blogspot.com/
schematic here: http://1.bp.blogspot.com/-Qr8J8iBYL.../x0sYyAA02BE/s1600/Buck_Converter_4-14-11.png

This design suffers from shoot through in the driver circuit which I plan to fix in the next rev. But it works well (has been working for months) charging a battery that runs some architectural lights.

The input in that schematic is driven from a PIC's PWM peripheral.

Figuring out the math for your design is a good start, and using the simulators to test your circuit is useful too (LT Spice). I've found that, with the magnetics, results vary and it comes down to having a good control loop to get out what you want. Specifically I mean testing the voltage on the output and adjusting the pwm based on that.

I would build a circuit, breadboard, perf board, whatever and just run 500mA through it to start then work your way up to 5 Amps. There will be a lot of problems that don't show up until you increase the power and 5 Amps is a lot of power. In my experience the parasitic L&C on the breadboard is not such a big problem to prevent the circuit from working. Things are slightly cleaner with a PCB but the same basic problems exist that existed with the breadboard with the big looping wires, etc.

Also, you need a decent scope to solve problems with the magnetic voltage transformations.

This is very helpful. Thanks!
I understood that the frequency is somehow arbitrary. To get small components, a higher frequency would be better. I don't know if it's going to affect my circuit if frequency is higher though!

I have a designed driver, and a program already made for a buck converter. However, we did them for a higher power converter (2KW 83A buck converter! it takes 48V step it down to 25V), Do you think it's going to work on my design if I modify the code? (Ton, Toff, frequency,..etc) ?

 

I'm no expert but putting together the buck topology usually does buck the voltage down, in my limited experience anyway. My guess is that you'd pay a lot more for the parts to do 83A.

 

I'm no expert but putting together the buck topology usually does buck the voltage down, in my limited experience anyway. My guess is that you'd pay a lot more for the parts to do 83A.

Yeah man, we paid around $500. We didn't even find a 250mH inductor and had to build it on our own!

But I did my calculation my current one. ( My current buck receives a variable input 15-30V and outputs 5 Volts). The thing is my components values differ when considering 30V input or 15V input, and not sure what to pick?

I used LTspice to test them and found out that if I use either 15V or 30V, it steps it down with a ratio of 1.7! So, if it was 30 it outputs 17V. If it's 15V input it outputs 8.5Volts!

Even if I make the inductance and capaticance values larger, it's still the same ratio!

 

You need a feedback circuit to control the switch PWM percent.

If you use a constant PWM percentage, your output voltage will depend upon your input voltage and your load current.

 

You need a feedback circuit to control the switch PWM percent.

If you use a constant PWM percentage, your output voltage will depend upon your input voltage and your load current.

You're right! my professor told me that today! Hey asked me to use a specific MOSFET driver though!

Please guys take a look at this datasheet! I couldn't figure out how to determine the right frequency to use for the switching!

http://www.irf.com/product-info/datasheets/data/irf9540ns.pdf

 

Hi haalamri,

Figure 11 in the spec shows test result with pulse widths down to 10uS. Therefore if the dutycycle was 10% then the maximum frequency would be 1/100uS=10KHz. The part can switch much faster however.

The td, tr, & tf specs suggest a reasonable upper limit of 1MHz, but switching efficiency would be poor. A frequency of 100KHz would allow for good efficiency.

Pick a frequency between 10KHz and 100HKz depending on what other circuit limitations there are.

Higher frequencies are not out of the question, but like I said before, it is arbitrary until you have a reason for it to be more specific.

Regards,
Ifixit

 

Hi haalamri,

Figure 11 in the spec shows test result with pulse widths down to 10uS. Therefore if the dutycycle was 10% then the maximum frequency would be 1/100uS=10KHz. The part can switch much faster however.

The td, tr, & tf specs suggest a reasonable upper limit of 1MHz, but switching efficiency would be poor. A frequency of 100KHz would allow for good efficiency.

Pick a frequency between 10KHz and 100HKz depending on what other circuit limitations there are.

Higher frequencies are not out of the question, but like I said before, it is arbitrary until you have a reason for it to be more specific.

Regards,
Ifixit

Thank You! This is very helpful!

I was trying to simulate my circuit, but it seems the output mostly depends on what Tr, Toff, Ton values I put! I changed Tperiod, inductance, capacitance, but it didn't change much in the output! However, it dramatically changes if I change the rise, off and on times!
Now, Is there a reason for that?
Plus, I see in the datasheet those values are in nanoseconds, (tr, toff,ton). My circuit wouldn't work properly till I put values in Microseconds!

 

Guys,
What would be a good driver to control this MSOFET switching?
I'm having hard time trying to search one

http://www.irf.com/product-info/data.../irf9540ns.pdf

Thanks

 

Your datasheet link did not post properly; here it is:
http://www.irf.com/product-info/datasheets/data/irf9540ns.pdf

Can you tell us the reasons/selection criteria why you chose that particular MOSFET?
Why a P-channel instead of an N-channel?
Why that high of a Vdss?