I require assistance with buck converter calculations.

Thread Starter

Kaykaykabobo

Joined Nov 11, 2021
3
I feel inadequate because I've been out of school for a while and I'm struggling a bit. I need assistance in tackling a buck converter calculation.
How do I calculate the minimum inductor (L) size for a buck converter so that it is always in continuous conduction mode and peak to peak inductor current variation does not exceed 30% of the average value for all operating conditions. Output current varies at different loads (ie I output = 10A to 25A) Voltage ripple <1%. Attached are my attempts at the question. Honestly I do not feel confident in those attempts.
 

Attachments

Papabravo

Joined Feb 24, 2006
21,158
10-25 amps out of a buck converter is a power level that requires more than an ordinary amount of care. The primary feature of such a design is not so much the inductance, but how can you realize that power level without saturating the core. Since you might have some trouble acquiring such an inductor off the shelf you need to find a core material and a construction method that will give the required performance. Once you find a core material and choose a wire size that will carry the current you will pretty much be limited in your choice of inductance. You need to ensure that the requirements don't make the realization impossible.

Let's start with some realistic estimates for input power, output power, and efficiency. Then at least we can take a look at the feasability of those requirements.
 

Thread Starter

Kaykaykabobo

Joined Nov 11, 2021
3
10-25 amps out of a buck converter is a power level that requires more than an ordinary amount of care. The primary feature of such a design is not so much the inductance, but how can you realize that power level without saturating the core. Since you might have some trouble acquiring such an inductor off the shelf you need to find a core material and a construction method that will give the required performance. Once you find a core material and choose a wire size that will carry the current you will pretty much be limited in your choice of inductance. You need to ensure that the requirements don't make the realization impossible.

Let's start with some realistic estimates for input power, output power, and efficiency. Then at least we can take a look at the feasability of those requirements.
The issue isn't practicality. It's purely mathematical. I'm not actually physically designing a buck converter. Just maths.
 

Irving

Joined Jan 30, 2016
3,843
Your calculations would be better if you got your time period correct 10kHz = 100uS not 10mS

For those sort of currents in real-world siuation you'd probably go for a multiphase approach to give more flexibility on inductor choice as @Papabravo pointed out...

One quick way to verify the numbers is to run a simulation....

1636668758098.png
1636669312816.png
 
Last edited:

MisterBill2

Joined Jan 23, 2018
18,167
There are several semiconductor manufacturer websites that have the complete application process available, and as the intention is to sell their products, the designs should work very well. I suggest using one of those sites.
And some serious advice, which is that switching regulator supplies are one of the most complex analog creations around, and every single part is critical. And even after you have a circuit that works perfectly in simulation, creating the physical design will be a challenge because every connection and conductor has an effect. That is why I always buy the switcher power supplies, I don't have the time to do a good design.
 

Irving

Joined Jan 30, 2016
3,843
And some serious advice, which is that switching regulator supplies are one of the most complex analog creations around, and every single part is critical. And even after you have a circuit that works perfectly in simulation, creating the physical design will be a challenge because every connection and conductor has an effect. That is why I always buy the switcher power supplies, I don't have the time to do a good design.
In principle I agree; high current SMPS, like high-current motor drives, are very susceptible to layout and parasitic issues and can involve a significant amount of trial and error in development.

However, lower current, say <2A, moderate voltage (2 - 50v), smps are readily developed around off-the-shelf chips using the manufacturer supplied layouts given in application notes with few problems.
 

MisterBill2

Joined Jan 23, 2018
18,167
In post #1 the TS mentions 10 amps to 25 amps as the current range, and so this is not a simple low-power system. And regardless of the voltage, it will take a fairly massive inductor to not have the core saturate at 25 amps load current. If this is a one-off design I suggest a purchased supply, if it is for production I suggest getting some help.
 

Irving

Joined Jan 30, 2016
3,843
Well as the TS says this is a thought experiment not a practical build. In real life this would be a multiphase design, to avoid the inductor saturation issue, probably 8-phase at 3A each...
 

Irving

Joined Jan 30, 2016
3,843
28uH a bit small, gives too much ripple, ΔI = 9A. 77uH gives ΔI = 3.2A, closer to TS spec. Needs to be saturation rated at > Iout + ΔI/2 i.e. > 26.5A with some margin.
 

Ian0

Joined Aug 7, 2020
9,667
28uH a bit small, gives too much ripple, ΔI = 9A. 77uH gives ΔI = 3.2A, closer to TS spec. Needs to be saturation rated at > Iout + ΔI/2 i.e. > 26.5A with some margin.
He said 30% ripple back in post #1.
I'd have preferred 10% (but I would also have gone for polyphase)
 

Ian0

Joined Aug 7, 2020
9,667
30% for all conditions - ΔI = 3A @ 10A and its only marginally higher at 25A.
His spec is not entirely unambiguous! It could be interpreted as 30% ripple @ 25A.
Polyphase makes some things easier like inductor design but adds complexity.
You can say that again - I am currently designing a 30A buck circuit. Every time I find the ideal toroidal core no-one stocks it. When I find the ideal E-core it is a nonstandard size and no-one stocks the bobbin, and when I try an amorphous C-core Miles Platts will give me one sample bobbin but won't sell me less than a thousand.
 
Top