DC-DC Boost Converter Switching Frequency

Thread Starter

amir97

Joined Apr 9, 2021
3
hi guys,
currently i'm want to design a dc-dc boost converter but i don't know what is the best frequency value that i need to use in this converter. can someone advise how to find the best pwm switching frequency? i need a justification why i supposed to use that frequency value on my design.
 
Last edited by a moderator:

DickCappels

Joined Aug 21, 2008
10,153
Is this homework or a school assignment?

It will help a lot if you explain what your input power source is (volts), what your load is (volts and amps, is it DC?), and what it does ("blender motor")
 

Ian0

Joined Aug 7, 2020
9,667
Your choice of frequency is the best compromise between transformer size and losses which are higher at low frequencies and semiconductor switching losses which are higher at high frequencies.
Avoid <20kHz because it will whistle annoyingly, unless it will be installed where no-one can hear it.
If you want a less vague answer, fill in some more details as @DickCappels suggested!
 

Papabravo

Joined Feb 24, 2006
21,158
hi guys,
currently i'm want to design a dc-dc boost converter but i don't know what is the best frequency value that i need to use in this converter. can someone advise how to find the best pwm switching frequency? i need a justification why i supposed to use that frequency value on my design.
There is no "best" frequency. Do you really believe that the universe is that simple? Why don't you start by articulating your requirements?
 

Thread Starter

amir97

Joined Apr 9, 2021
3
Is this homework or a school assignment?

It will help a lot if you explain what your input power source is (volts), what your load is (volts and amps, is it DC?), and what it does ("blender motor")
Just a homework.. I need to design a DC-DC Boost converter.. Input 9volt.. Output depends on the pwm duty cycle(MOSFET).
 

Thread Starter

amir97

Joined Apr 9, 2021
3
There is no "best" frequency. Do you really believe that the universe is that simple? Why don't you start by articulating your requirements?
I need to design a standard DC-DC boost converter. 9volt input. Output voltage depends one the pwm duty cycle.. So, now, I don't the amount or value of frequency that I have to use in this circuit.. How about 25kHz? Its okay? If okay, Why? I need a justification..
 

Papabravo

Joined Feb 24, 2006
21,158
I need to design a standard DC-DC boost converter. 9volt input. Output voltage depends one the pwm duty cycle.. So, now, I don't the amount or value of frequency that I have to use in this circuit.. How about 25kHz? Its okay? If okay, Why? I need a justification..
You need an output voltage and a current requirement. Oen thing to keep in mind, especially with a boost converter is that the power output will always be less than the power input. sometimes, if you are not careful, it will be a great deal less. There are many things that go into the choice of the switching frequency, especially what kind of magnetics you will be using. You can't just pick a frequency out of the air and call it good. That is no way to do a design. How much experience do you have with DC-DC converters? Are you able to simulate one? That would be a really good first step.
 

Papabravo

Joined Feb 24, 2006
21,158
Just a homework.. I need to design a DC-DC Boost converter.. Input 9volt.. Output depends on the pwm duty cycle(MOSFET).
Changing the duty cycle is for maintaining regulation and NOT having an adjustable output. The adjustable output will lead to a whole host of other problems. Adjustable within a fairly narrow range might work but you need to quantify what that might be.
 

BobTPH

Joined Jun 5, 2013
8,804
You have to make sure the inductor does not saturate in the on period at the highest duty cycle. You pick one the other, then compute the other.

Bob
 

crutschow

Joined Mar 14, 2008
34,280
There's a tradeoff between switching losses in the switch and at a high frequency, and the size of the inductor at lower frequencies.
You can't know that without first deciding what the control circuit design will be.
Can you use IC's for that or is it all discrete?
 

Ian0

Joined Aug 7, 2020
9,667
I need to design a standard DC-DC boost converter. 9volt input. Output voltage depends one the pwm duty cycle.. So, now, I don't the amount or value of frequency that I have to use in this circuit.. How about 25kHz? Its okay? If okay, Why? I need a justification..
Look at the frequencies that fixed frequency switched-mode controllers such as NCP1216 run at - you'll find 65kHz, 100kHz, and 132kHz. Does that give you a clue?
https://www.onsemi.com/products/pow...ollers-regulators/offline-controllers/ncp1216
 

Papabravo

Joined Feb 24, 2006
21,158
Here is another complication. MOSFETs are the preferred type of switch because they have low losses when they are on, and the harder you drive them the lower the on resistance. There is a price to pay and that is the Miller capacitance. It slows down the turn on and the turn off. During this time the switch is in it's linear region and what is worse it dissipates a good deal of power in this region. That is why you want to minimize the time it spends in this region. Unfortunately there are more compromises than you can shake a stick at. The only suggestion I have for you is to get a good reference and study it carefully.
 

MrSalts

Joined Apr 2, 2020
2,767
Most modern SMPS are 1MHz or above. The inductors (most expensive part when running less than a few hundred kHz) can be incredibly small with the higher frequencies (and much cheaper). Great advances in high electron mobility mosfets, GanFET and SiC MOSFETs enable the high frequencies. Companies like CUI have GanFET power supplies operating at 2 to 5MHz, vey small package sizes and high amperage.

Here is a MC34713 with integrated 5A MOSFET running at 1MHz.
https://www.nxp.com/docs/en/data-sheet/MC34713.pdf
 

Ian0

Joined Aug 7, 2020
9,667
At a rough approximation, the power dissipated whilst in the switching region is half the voltage x half the current.
The effect of the Miller capacitance is proportional to the drain voltage, so at 9V it can almost be ignored, but at 400V it's a very real problem. You don't see many off-line supplies running much faster than 150kHz.
Copper losses (the resistive loss of the wire in the inductor) is easy to calculate to a first order approximation, but then skin-effect and proximity-effect have to be accounted for.
Then there's core loss, which is proportion to frequency raised to the power 5/3 and peak-to-peak flux excursion raised to the power 5/2, then there's diode reverse-recovery losses, which disappear in discontinuous mode.
And once you have optimised for efficiency, you have to get it through EMC tests, and they are very much dependent on your choice of operating frequency. Of course, you can use "frequency dither" to spread out the interference spectrum to make it easier to pass.
 

Papabravo

Joined Feb 24, 2006
21,158
At a rough approximation, the power dissipated whilst in the switching region is half the voltage x half the current.
The effect of the Miller capacitance is proportional to the drain voltage, so at 9V it can almost be ignored, but at 400V it's a very real problem. You don't see many off-line supplies running much faster than 150kHz.
Copper losses (the resistive loss of the wire in the inductor) is easy to calculate to a first order approximation, but then skin-effect and proximity-effect have to be accounted for.
Then there's core loss, which is proportion to frequency raised to the power 5/3 and peak-to-peak flux excursion raised to the power 5/2, then there's diode reverse-recovery losses, which disappear in discontinuous mode.
And once you have optimised for efficiency, you have to get it through EMC tests, and they are very much dependent on your choice of operating frequency. Of course, you can use "frequency dither" to spread out the interference spectrum to make it easier to pass.
Kudos! If that doesn't keep you off the streets, and up late at night, then perhaps nothing will.
Lay it on thick, then pile it higher and deeper.
 

MrAl

Joined Jun 17, 2014
11,389
hi guys,
currently i'm want to design a dc-dc boost converter but i don't know what is the best frequency value that i need to use in this converter. can someone advise how to find the best pwm switching frequency? i need a justification why i supposed to use that frequency value on my design.
There has to be more to this question than just "what is the best frequency" because that question assumes that you already have some other constraints you are being forced to work with, such as the choice of inductor and the choice of switch and other components and even possibly the allowable EMI.

There is a trend in modern designs however because devices are getting smaller, and that is to go as high in frequency as possible in order to keep component size small which also often keeps cost down too. Higher frequencies require more careful design though so it also depends what you can actually do and what parts you can actually obtain somewhere.
Back in the early 1980's we were stuck with designs around 50kHz maybe, but as MOSFETs started to grab more of the market the frequency almost immediately went up to 100kHz. Now we see designs of 1MHz and even 2MHz which is a lot higher. Note however that a lot of computer power supplies are still around 20kHz much lower. So thinking about that, we see a range of 20kHz to 2MHz which is a 100 to 1 spread.

So the frequency is not some stand alone arbitrary choice but is dependent on other factors that would have been already established beforehand. Perhaps you can describe the question in more detail or even get more details from where you got this question from in the first place. Perhaps the answer is just that given here, but i think you need more information really.
 

MrSalts

Joined Apr 2, 2020
2,767
So the frequency is not some stand alone arbitrary choice
Really, from the title of Robert Keim's article, it appears that he completely disagrees with you....
"How-to-choose-the-frequency-of-your-switching-regulator"
https://www.allaboutcircuits.com/te...se-the-frequency-of-your-switching-regulator/

...but if you actually read it, you'll realize it is a fluff piece with no real information - likely an attempt to help ACC raise their SEO ratings as "new" but useless content. In the end, the author asks the reader for advice! Crock-a!
 
Last edited:

Papabravo

Joined Feb 24, 2006
21,158
Really, from the title of Robert Keim's article, it appears that he completely disagrees with you....
"How-to-choose-the-frequency-of-your-switching-regulator"
https://www.allaboutcircuits.com/te...se-the-frequency-of-your-switching-regulator/

...but if you actually read it, you'll realize it is a fluff piece with no real information - likely an attempt to help ACC raise their SEO ratings as "new" but useless content. In the end, the author asks the reader for advice! Crock-a!
Those that can do. Those that can't write articles with no real information.
Way to skewer a member of the Glitteratti
 

MrAl

Joined Jun 17, 2014
11,389
Really, from the title of Robert Keim's article, it appears that he completely disagrees with you....
"How-to-choose-the-frequency-of-your-switching-regulator"
https://www.allaboutcircuits.com/te...se-the-frequency-of-your-switching-regulator/

...but if you actually read it, you'll realize it is a fluff piece with no real information - likely an attempt to help ACC raise their SEO ratings as "new" but useless content. In the end, the author asks the reader for advice! Crock-a!
Well i'll have to read it i guess i did take a quick look and i did see one interesting point being brought up, using an ADC in a system with a switching regulator. If the ADC sampling frequency is close to the switching regulator switching frequency it could interfere with measurements being made by the ADC.
I have found that the switching frequency can interfere with a lot of stuff. In fact, if it is an integer multiple of the switching frequency it could be very problematic. If not like that there is a chance it could actually help in an ADC system using oversampling depending on the shape of the noise.
But he is after all comparing the frequency of a converter to other things like the ADC for one example. I dont think i can agree with the efficiency issue though as modern designs seem to have this under control. I have to say though that i have not worked with any 1MHz switching converters yet most of the ones i dealt with in the past were around 10kHz to 100kHz. Back in the old days all we had were bipolars and they were slower so there was immediately a limit on the switch frequency.

You know what is kind of funny though, if you take the 's' from "switch" and place it at the end instead of at the beginning of that word you get "witchs" ha ha. Yeah it is missing the 'e' but it is kind of funny that the switching frequency can kind of be like a witch in the ADC front end.
 

Ian0

Joined Aug 7, 2020
9,667
I've seen many thread-starters on this forum who think they can accomplish the feedback to control a switching regulator with a MCU. I wonder how many of them succeed? It's all about the phase response, and calculating that becomes much less trivial once you have a sampling system.
 
Top