Buck Converters

[yngndrw]

Hi,

Well one thing I've never really understood is Buck Converters, especially multi-phase Buck Converters.

I found this which really helped:
http://www.ocforums.com/showthread.php?t=264512

On that thread, there was this diagram:
**broken link removed**

That has tought me a bit. The thread also linked to this:
**broken link removed**

But that page doesn't really help because it starts off jumping into "In our switched inductor example we used the rule of thumb of designing L to get a peak-to-peak ripple current of 10% of the full load ......." which is jumping the gun for me.

Oh and I also found this:
http://en.wikipedia.org/wiki/Buck_converter

Which seemed to help a bit.

Anyway, I was wondering if some of you can help me learn about inductors and Buck Converters.

Also how are they driven ? Can you show me a simple analogue driver for one ? Not some fancy IC which does it all for you, as that isn't really useful for learning about them.

Also, on the diagram above, how could that be driven ? I simulated it in Livewire but didn't get very far.

Finally, how do you do simple calculations for the values of L and C ? The webpage above has some formulas but they are way over my head.

Thank you,
-Andrew.

 

[Overclocked]

L and C depend on your output voltage, frequency and currents. And unfourtantly, thats All you will find is Controller IC's, but all the chip does is generate PWM signals, compare output voltages (feedback); which also limits the PWM signal going to the inductor; and compare currents; which limits current to your output (if the transistor is internal). Other than that, the rest is just bells and whistles

**broken link removed**
555 Timer SMPS.

A good book will help, http://search.barnesandnoble.com/booksearch/isbnInquiry.asp?z=y&EAN=9780750673297&itm=18

That book mostly focuses on buck converters and multiphase converter, but it does give you a nice start to switching power supplies.

 

[yngndrw]

Thank you for your reply.

So looking at that 555 SMPS - To turn that into the dual MOSFET version, you'd replace the diode with a second MOSFET and drive it with the output of the 555, but inverted ?

I'll look around for that book.

 

[Oznog]

You should check on the stability issues on a buck controller.

Most people want a fixed voltage output, although there are apps where a fixed current output is much more desirable such as an LED driver.

Buck controllers are a bit difficult to regulate until you understand the mathematics and use some advanced features in the controller. For example if you just used a fixed 50% duty cycle that doesn't mean the output voltage will be 50% of the input. We could use a feeback loop, when voltage is high we reduce the duty cycle and if it's low we increase it. However, you can see it will take more than one cycle for the result of the "new" duty to become apparent, also it's have to charge the cap so the capacity of the cap affects the response time too. The end result is way the rippling output voltage is sampled, the time constant of the feedback circuit, and the gain of the feedback circuit are all important. If we respond too slowly though then sudden changes in Vin or Iout will result in significant deviations from the intended output voltage that it will be slow to correct.

 

[yngndrw]

Hmm, so am I right in assuming that a Buck Converter is just the smoothing after a standard PWM circuit ?

I simulated the 555 circuit in Livewire and on Livewire's built in oscilloscope, the output ripple was massive. If you was to add more L and C filters each with smaller values after the initial set, do you think that would improve the ripple ?

What's the best way to calculate L and C ?

Thanks.

 

[Oznog]

Hmm, so am I right in assuming that a Buck Converter is just the smoothing after a standard PWM circuit ?

I simulated the 555 circuit in Livewire and on Livewire's built in oscilloscope, the output ripple was massive. If you was to add more L and C filters each with smaller values after the initial set, do you think that would improve the ripple ?

What's the best way to calculate L and C ?

Thanks.

No. The "buck converter" at a minimum generally consists of an inductor transistor, a diode, an output cap, and a PWM controller of varying complexity.

The buck converter MUST have a load to get a useful measurement.

 

[yngndrw]

Ah okay then, thank you very much for your replies.

 

[Overclocked]

Hmm, so am I right in assuming that a Buck Converter is just the smoothing after a standard PWM circuit ?

I simulated the 555 circuit in Livewire and on Livewire's built in oscilloscope, the output ripple was massive. If you was to add more L and C filters each with smaller values after the initial set, do you think that would improve the ripple ?

What's the best way to calculate L and C ?

Thanks.

The only way to decrease Ripple would to be to increase the Output Capacitor. Remember why caps have to be large in Linear PSU's? Its because you need to filter the output, the more capacitance you have, the lower the ripple.

However, If there is a suddent change in output voltage, the capacitors will be slow to change, so theres sort of a high and low of what capacitor you need. Generally, Cout (min)is found by:

Iout(max)*(1-DC (min))
Fsw*(Vripple-pk to pk)

Fsw is the switching freq, and V ripple is the desired ripple on the output. Usually its in mV. DC min (as my book shows it) is Input Voltage/ Output Voltage (max)So for the sake of example let;

Input Voltage: 24V
Vout Max = 12V
Iout Max=1 A
Fsw=100 Khz
Vripple = 10mV pk to pk

1*(1-24/12)
100k*(10m)

which gives:

Cout (min) = 1000uF

Its better to round up, than down.

 

[yngndrw]

Thank you very much for that explanation.

Could you explain the calculation for the inductor in the same way please ?

 

[yngndrw]

Well I've got the book you suggested, but looking through it there's some things that I don't quite understand:

1) What do you do for the calculations (For Lout and Cout) when you have a variable voltage supply ? The formulas mention "The estimated on time of power switches at the highest input voltage (30% of 1/f is a good guess" for the Lout formula and "The smallest estimated duty cycly at high input line and light load (An estimate of 0.3 is good)" for the Cout formula, but what do you do for variable output voltages ? Same for the Vout in these formulas.

2) You mention DC as "DC min (as my book shows it) is Input Voltage/ Output Voltage (max)" but the book mentions it as "The smallest estimated duty cycly at high input line and light load (An estimate of 0.3 is good)", which would hint that it is the same as Input voltage / Output Voltage (min). Which is correct ?

3) Can you run me through a simple buck converter calculation for Lout and C out where let's say the output will be an LED so: The input voltage is 12V, the output voltage will be 0V to 5V through PWM of the transistor / MOSFET, and the max current drawn will be 500mA (@ 5V).

4) To drive a high power LED through PWM, is the L and C actually needed ? Would it make a difference if it was left out (and the diode for that matter).

Thanks,
-Andrew.