Hello,

i would like to design a SEPIC(or buck-boost) converter which converts an input voltage with range 3-19V to 8.4V. On its output will be connected three DC-DC converters which are depicted below :



My first try was this (components calculated according to some articles i found), but cannot convert input voltages under about 8V:

Could anyone suggest how to calculate the components of the SEPIC(or buck-boost) converter?
The complete circuit :

 

The SEPIC is fairly tolerant of component values.
Just start from
\(
\frac{V}{L}=\frac{\Delta I}{\Delta t}
\)
Choose ΔI as 10% to 20% of the output current
\(
\Delta t = \frac{1}{2f}
\)
From that you can calculate a minimum value for L. Any value higher that that will work. Similarly there is no fixed value for C - just make sure it is big enough.
SEPICs and Čuks work best if wound on iron powder toroids, with both windings on the same toroid badly coupled - one winding on each half - both winding have the same number of turns.
The step up/down ratio is
\(
\frac{V_{out}}{V_{in}} = \frac{D}{1-D}
\)
If your control IC has enough duty cycle, it should work down to the minimum voltage for the controller IC, or until the MOSFET doesn't have enough drive voltage.

Have you noticed that if you remove C37 it becomes an isolated flyback circuit?

 

SEPIC converter design:
http://www.simonbramble.co.uk/dc_dc...nverter/sepic_buck_boost_converter_design.htm

 

You would be better off using 3 Buck-Boost-Converters and
just skip the first one.
.
.
.

 

The SEPIC is fairly tolerant of component values.
Just start from
\(
\frac{V}{L}=\frac{\Delta I}{\Delta t}
\)
Choose ΔI as 10% to 20% of the output current
\(
\Delta t = \frac{1}{2f}
\)
From that you can calculate a minimum value for L. Any value higher that that will work. Similarly there is no fixed value for C - just make sure it is big enough.
SEPICs and Čuks work best if wound on iron powder toroids, with both windings on the same toroid badly coupled - one winding on each half - both winding have the same number of turns.
The step up/down ratio is
\(
\frac{V_{out}}{V_{in}} = \frac{D}{1-D}
\)
If your control IC has enough duty cycle, it should work down to the minimum voltage for the controller IC, or until the MOSFET doesn't have enough drive voltage.

Have you noticed that if you remove C37 it becomes an isolated flyback circuit?

In
\(
\frac{V}{L}=\frac{\Delta I}{\Delta t}
\)
V is the input voltage(3V-19V) or the output(8.4V)?

 

In
\(
\frac{V}{L}=\frac{\Delta I}{\Delta t}
\)
V is the input voltage(3V-19V) or the output(8.4V)?i

Yes (that’s the answer to your question)

Every (non-resonant) switched mode is governed by the equation
\(
\frac{V}{L}=\frac{\Delta I}{\Delta t}
\)
When you know Which V? Which I? Which t? And Which L? You’ll understand them all.
If L is the input side inductor, then V is the input voltage (the voltage across said inductor), then ΔI is the input current ripple, and t is the time the FET is on.
Similarly, when the FET is off, the output is supplied by the output inductor, so V is the output voltage, ΔI is the change in current (technically not the ripple current as the output current is discontinuous, but the amount it changes during the time the FET is off, because it is zero when the FET is on, and t is the time the FET is off or the time the diode is conducting)

 

You would be better off using 3 Buck-Boost-Converters and
just skip the first one.
.
.
.

Yes three would be better than one and would cover the needs of the other three converters but if possible i would prefer the solution with one because of design and construction cost.

 

Yes (that’s the answer to your question)

Every (non-resonant) switched mode is governed by the equation
\(
\frac{V}{L}=\frac{\Delta I}{\Delta t}
\)
When you know Which V? Which I? Which t? And Which L? You’ll understand them all.
If L is the input side inductor, then V is the input voltage (the voltage across said inductor), then ΔI is the input current ripple, and t is the time the FET is on.
Similarly, when the FET is off, the output is supplied by the output inductor, so V is the output voltage, ΔI is the change in current (technically not the ripple current as the output current is discontinuous, but the amount it changes during the time the FET is off, because it is zero when the FET is on, and t is the time the FET is off or the time the diode is conducting)

But in this phase of design how could i know all these values (ΔΙ,t,I, except for Vin and Vout)?

 

hi Peter,
In all your proposed circuits [ sims] you show an 'idealised' isolated Gate drive pulse, using a Voltage source.

Do you have a design concept on how you would achieve this using circuit components.?
E

 

hi Peter,
In all your proposed circuits [ sims] you show an 'idealised' isolated Gate drive pulse, using a Voltage source.

Do you have a design concept on how you would achieve this using circuit components.?
E

Yes with a pwm signal from a microcontroller.

 

Yes with a pwm signal from a microcontroller.

I would really advise against it, there are ICs which will do all for you it in the analogue domain that cost buttons.
You would need to implement a complex magnitude and phase response that is correct into the MHz region to make sure it is stable. First, you need a grasp of the works of Hendrik Bode, and a thorough knowledge of how to design the feedback of a switched-mode controller in the continuous-time domain. The with a knowledge of z-transforms and DSP you could translate it into the digital domain, if you have a good enough microcontroller. Your sampling rate will need to be >200ksamples per second, and your algorithm will have to run every 5us.
Have a read at SLUP340
https://www.ti.com/seclit/ml/slup340/slup340.pdf
and decide if you could digitise it!

 

But in this phase of design how could i know all these values (ΔΙ,t,I, except for Vin and Vout)?

I and ΔI are your maximum values.
t is determined from your choice of operating frequency, and the step up/down ratio
\(
\frac{Vout}{Vin}=\frac{D}{D1−D}
\)

I‘d suggest that if you want to see how SEPICs behave, make one with a 555 as the controller, and use a couple of pots to vary the on and off times and frequency.

 

I and ΔI are your maximum values.
t is determined from your choice of operating frequency, and the step up/down ratio
\(
\frac{Vout}{Vin}=\frac{D}{D1−D}
\)

I‘d suggest that if you want to see how SEPICs behave, make one with a 555 as the controller, and use a couple of pots to vary the on and off times and frequency.

As you told that there is no fix value for capacitors, just that they should be big enough, do you think that the values i chose in the first try (first post) will cover me?

 

Yes three would be better than one and would cover the needs of the other three converters but if possible i would prefer the solution with one because of design and construction cost.

I don't understand your statement ...........
I'm saying 3 instead of 4 inverters.

I personally wouldn't even consider using a Micro-Controller,
there are literally hundreds of excellent designs on a single Chip,
most of which can be synchronized together,
and/or, that can be run in a "3-Phase" arrangement to keep
Input-Noise and Current demands even and smooth.
.
.
.

 

I don't understand your statement ...........
I'm saying 3 instead of 4 inverters.

I personally wouldn't even consider using a Micro-Controller,
there are literally hundreds of excellent designs on a single Chip,
most of which can be synchronized together,
and/or, that can be run in a "3-Phase" arrangement to keep
Input-Noise and Current demands even and smooth.
.
.
.

Maybe i didnt understand your statement. I said that i want to supply three DC-DC converters (one boost and two buck ) with a sepic and i want to calculate the components of this sepic. Maybe you mean just to not use at all the sepic?

 

As you told that there is no fix value for capacitors, just that they should be big enough, do you think that the values i chose in the first try (first post) will cover me?

Yes, but microfarads and voltage aren’t the only specs for a capacitor.
In this case, check the ripple current. All the output goes through the capacitor, so as a rough approximation the ripple current spec should be more than the output current.
You may need more microfarads to get sufficient ripple current.

 

Maybe i didnt understand your statement. I said that i want to supply three DC-DC converters (one boost and two buck ) with a sepic and i want to calculate the components of this sepic. Maybe you mean just to not use at all the sepic?

Each conversion is a loss of energy and a source of heat.
The fewer conversions the better! Three is better than four.

 

Each conversion is a loss of energy and a source of heat.
The fewer conversions the better! Three is better than four.

Ok, yes that's true!

 

Hello,

i would like to design a SEPIC(or buck-boost) converter which converts an input voltage with range 3-19V to 8.4V. On its output will be connected three DC-DC converters which are depicted below :
View attachment 245872View attachment 245873
View attachment 245879

My first try was this (components calculated according to some articles i found), but cannot convert input voltages under about 8V:
View attachment 245876
Could anyone suggest how to calculate the components of the SEPIC(or buck-boost) converter?
The complete circuit :
View attachment 245878

Is this actually homework or is this a hobby thing?

Yes handling the power once is the general preference although there have been exceptions in the past.
In the distant past the best you could usually hope for was maybe 80 percent efficiency per stage at full load. Today some converters use two higher efficiency stages but of course one is still better if you can get away with it.

 

The trick is to completely understand the advantages, and the drawbacks, of each design.
There are very good reasons why You won't ever see a Power-Supply
configured in the manner that You have proposed.

You need to start with the overall design of the System.
If at all possible, limit the number of different Voltage Supplies to 1 or 2.
Having just one Voltage is usually impractical due to
most Micro-Controllers being either 3.3-Volts, or 5-Volts,
which is quite often inadequate for Powering external Loads.
So, if You can, make everything either 3.3 OR 5V, but not both.

If You can use separate Batteries for Logic and Load,
You may be well ahead in the game, because this means that You can
use a simple Linear-Regulator with little-to-no Efficiency penalty,
by using a single Li-Ion-Cell to power the Logic,
and a separate, much larger, 12-Volt Supply.
This will transfer almost all of your inefficiencies into the Battery-Charging-Circuitries.
.
.
.