1
\$\begingroup\$

I am currently designing a buck converter that takes in 100 V and outputs 50 V.

I want to drive a 25 Ohms load resistance so the current is 2 A. It is in CCM mode and has 1% ripple on output and input.

How does one calculate the switching frequency? Since it is needed to calculate the inductor and the capacitor values.

\$\endgroup\$
  • 1
    \$\begingroup\$ To avoid irritation of seasoned and professional electrical engineers here, is it possible to re-formulate your question, instead of saying "I am currently designing ... " to "I am currently learning about basics of buck converter design"? Then you will likely get some better advises. \$\endgroup\$ – Ale..chenski Oct 14 '16 at 21:43
3
\$\begingroup\$

There's no formula that you can use to get some "correct" switching frequency, only some guidelines.

My approach is to use the lowest switching frequency that meets all of the other requirements. The reason for this is to minimize switching losses, EMI, driver requirements, etc. and to maximize ease of PCB layout.

One primary requirement is usually size. The higher the frequency the smaller the magnetics, so size can be a big constraint.

Most designs these days are well above the audio range in CCM mode, but 25kHz might be a good lower bound. I don't go below 100kHz for a simple buck without good reason.

So start by picking an inductor that meets your output current requirement and size requirement.

Then figure out the switching frequency you need for maybe <40% ripple current in the inductor, or whatever you need to meet your 1% output voltage ripple given your amount and type of output capacitance.

| improve this answer | |
\$\endgroup\$
2
\$\begingroup\$

One does not "calculate" the frequency - one chooses the frequency. You have three variables: frequency, inductance, and capacitance. If you fix one of them (the frequency) you can calculate the others.

The choice of frequency can be influenced by such factors as EMI emissions, circuit switching noise (especially relevant in audio circuits), etc.

In general higher frequencies mean smaller inductors and capacitors but increased EMI and thus more careful board layout.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ So basically I need to choose a frequency and calculate the values then look at the datasheet to figure out if an inductor or a capacitor can handle, for example, the current and so on? \$\endgroup\$ – apathak Oct 14 '16 at 20:02
  • 1
    \$\begingroup\$ @apathak Yep. Or pick an inductor then work out what frequency and capacitance would meet your output criteria. Or pick a capacitor and work out which inductor and frequency (though you'd never really work that way - only choose the frequency or choose the inductor). \$\endgroup\$ – Majenko Oct 14 '16 at 20:04
  • \$\begingroup\$ Yes. Basically, you choose the frequency by footprint size, then select an appropriate converter IC and inductor. \$\endgroup\$ – Janka Oct 14 '16 at 20:05
  • \$\begingroup\$ Personally I do all my designs in TI's WebBench - it does it all for you. ti.com/lsds/ti/analog/webench/overview.page \$\endgroup\$ – Majenko Oct 14 '16 at 20:06
  • \$\begingroup\$ Incidentally, WebBench gives me all designs in the 25-33KHz range for your criteria. \$\endgroup\$ – Majenko Oct 14 '16 at 20:09
1
\$\begingroup\$

You are making a DC output of 100 watts .Lets say that you want 90% efficiency which is not unreasonable .The heat losses of your proposed buck converter will dictate the size putting a minimum constraint when cooling is not forced .In other words if there are no fans or cold plates or oil baths or ethylene glycol the higher frequency will make L and C smaller But wont make the PSU smaller.When I see little coils and big heatsinks I know that things can be better.If you want to use hard switched off the shelf current mode chip then you would be best to ballpark at say 50KHz to keep switching losses low .If you use a switching loss reduction scheme like I always do then the frequency does not matter much because other losses dominate .Remember that the buck converter has only one coil and an input cap and an output cap .Sure increasing F can get ripple down .Often other considerations dictate capacitance like holdup time transient response stability etc.So its back to a single coil where the incentive to raise F is low .You stated that increasing F reduces the size of L which is true but it is not in practice linear so to say halve L size you would have to more than double F .The size of L also depends on current ripple factor so allowing more than your 40% gets less microhenries needed for a given F .At low currents a bigger ripple factor makes sense in view of readily available ceramic caps .If you are careful you should get 95% efficiency which should mean SMD devices and no aluminium heatsinks.

| improve this answer | |
\$\endgroup\$
0
\$\begingroup\$

This depends on whether your footprint is very limited, then you have to go for a high switching frequency to keep the inductor small. If you have plenty of space, you could stay in the 100kHz range and keep out of the way of having unwanted MHz ringing all over your circuits. Plus, lower frequency converters tend to better efficiency.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ The off the shelf offering wastes 12W .When you factor in the size of the heatsink it is not small at all . \$\endgroup\$ – Autistic Oct 14 '16 at 21:11
0
\$\begingroup\$

If a designer of a 100W switcher asks what frequency to use, then the only practical solution is to select an industrial-grade offering, like this one, if your management wants to have the project done.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ The off the shelf offering than you propose wastes 12 Watts .If cooling is natural then the heatsink will be much larger than the small DCDC convertor . \$\endgroup\$ – Autistic Oct 14 '16 at 21:16
  • 3
    \$\begingroup\$ Efficiency of 85% - 90% is pretty standard for this class of devices, especially if the range and variability of load is not known. If you can do better (starting form asking about selection of frequency), then good luck. \$\endgroup\$ – Ale..chenski Oct 14 '16 at 21:21
  • \$\begingroup\$ I always do better by not using hard switched current mode control.If you do one yourself using orthodox hardswitching current mode you will probably do worse than the bought one . \$\endgroup\$ – Autistic Oct 15 '16 at 7:36
  • \$\begingroup\$ Did you try multi-phase buck switchers with coupled inductors? \$\endgroup\$ – Ale..chenski Oct 15 '16 at 17:31
  • \$\begingroup\$ For these tests I used the normal Buck convertor topology .P channel main switch and Schottky diode ,Input caps and output caps were ceremic in parallel with electrolytic.I have locked two buck converters together to multiphase in prototype form .I think that multiphasing would be good for low noise at high power.I have prototyped a coupled inductor version that is not multiphase.Is there any advantage in the coupled inductor when you dont need to boost? \$\endgroup\$ – Autistic Oct 15 '16 at 21:41
0
\$\begingroup\$

There are many different methods of Buck regulators that are available with online solutions.

Your criteria ought to select , what is more important to you;

  • 1) something that works reliably
  • 2) easy to breadboard, or PCB layput
  • 3) cost
  • 4) footprint
  • 5) EMI
  • 6) number of unique parts
  • 7) efficiency , isolated or non-isolated
  • 8) parts readily available in your area
  • 9) designs that tell you how to choose every part so you can learn
  • 10) designs that teach you more about critical component selection and layout
  • 11) designs with soft-start and improved ripple rejection
  • 12) designs that have greater peak current capacity for non-linear loads such as battery charge or motors
  • 13) designs with adjustable current limit and voltage output
  • 14) designs with internal fault detection + protection such as OTP, OCP, OVP
  • 15) benchmark comparisons with off-the-shelf open frame switchers at $1/watt

    sample design of a 100W 50 V Buck regulator 92% efficient enter image description here

This is the topology of a forward converter, ( transformer + rectifier with optical feedback isolation) perhaps the most common reliable low cost type with similar topology used in many ATX PSU's .

| improve this answer | |
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.