• Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

SEPIC converter is only 67% efficient...why so bad?

Flyback

Well-Known Member
Thread starter #1
Hello,

My 5W sepic converter (LED driver) is only 67% efficient

Vin=6V,
Vout = 31V
Iout = 0.13A
F(sw) = 51KHz
SEPIC inductor = MSD1583-223 (22uH)

Schematic:
http://i45.tinypic.com/2qx9cfr.jpg

All the capacitors are ceramic.

I suspect that due to the ceramic capacitors, there is ringing between the SEPIC inductor’s leakage inductance and the input ceramic capacitors.

How can I snub out this ringing?…..is it by putting a series damping resistor in the path of the SEPIC “primary”?

SEPIC inductor = MSD1583-223
http://www.coilcraft.com/pdf_viewer/showpdf.cfm?f=pdf_store:msd1583.pdf

(The SEPIC inductor feels hottest, of the power components..the FET and Schottky diode feel barely warm)

Peak FET current is 3.2A and the SEPIC is fairly deeply discontinuous
 
Last edited:

misterT

Well-Known Member
Most Helpful Member
#2
How did you measure the efficiency?
 

misterT

Well-Known Member
Most Helpful Member
#4
I measured voltage and current input, and same for output, then (output power/input power) * 100%
Did you measure with differen loads?

The series resistance of the inductors and capacitors affect the efficiency. So does the switching speed of the diode and FET. Also the physical layout of your components can make a difference.. do you have a picture of your circuit?
 

ChrisP58

Well-Known Member
#6
Generally speaking, the SEPIC is about the least efficient power conversion topology. It's main advantage is that it that the output voltage can be either higher, or lower, than the input. If that is important, then it is a good choice.

But if the output voltage is always higher than the input, then a boost converter will be more efficient.
 

Flyback

Well-Known Member
Thread starter #7
Thanks,
Its the SEPIC inductor that's hot, in spite of its very low series resistance.
Output can sometimes be 5V, hence the sepic , as you say.
-also, we needed output short circuit protection, so sepic ok for that.

The fet and diode and sense resistors are barely warm, the sepic inductor is hot.
 
#9
3.2 amp peak current for only 0.13 average amps?

a crest factor of 3.2/0.13 = 24.6

For discontinuous operation, the aim is to have a crest factor between 2 and 3. Otherwise the switching components are being used inefficently.

You have to increase the switching frequency, the inductance or both.
 

Flyback

Well-Known Member
Thread starter #10
What inductor are you using?
....Datasheet isted at the bottom of the first post (MSD1583-223)

3.2 amp peak current for only 0.13 average amps?
...yes, this is what you unfortunately end up with when you have an smps that must serve a range of output voltages and input voltages.

The input voltage can be anywhere between 5.25V and 7V.
The output voltage can be anywhere between 5V and 40V.
The output power can be anything from 0.5W to 5W.
 
#11
Ok here goes. You have a 22uH inductor which sounds a bit low for a 50kHz switching frequency. please see my article on the design of SEPIC converter:

http://www.simonbramble.co.uk/dc_dc_converter_design/buck_boost_converter/sepic_buck_boost_converter_design.htm

Maybe your inductor is saturating which is causing the over heating.

Are you measuring the input current by looking at the reading on the bench power supply? This is not a good idea for most switched mode power supplies since the input current to a switched mode is a current ramp (or worse) so the current reading you get from your bench power supply will probably be wrong.

To measure the efficiency of a dc/dc converter:
Put a big ceramic cap close to the input to the circuit (the top of the inductor in a SEPIC converter, the drain of the top FET in a buck converter). This will create a local low impedance path for the gulps of current taken by the power supply.

Fit an inductor equal to the SMPS inductor in series with the input rail to the circuit. This will smooth out the current spikes taken by the SMPS.

Insert a current sense resistor in series with the input rail. Probe across this with a scope (a differential measurement will ensure you dont short the power supply to ground with the earth probe)

Increase either the ceramic cap or the inductor or both until the voltage across the sense resistor is flat.

Once the voltage across the current sense resistor is flat you can then measure this voltage with a voltmeter. This will give you the input current to 2 decimal places.

Measure the input voltage, output voltage and output current then work out input power and output power and hence calculate efficiency, accurate to 2 decimal places. This is the only way to determine the efficiency of a switched mode power supply.

Other likely areas of loss in a sepic are FET ON resistance (since the input inductor current and output inductor current flow through the FET at the same time), QG of the FET limiting the turn on time of the FET (resulting in increased switching losses) or inductor DCR (but there is little you can do about this).

If you can probe the current in the inductor, this would be good. If it is a straight ramp, this is good. If its gradient starts to increase, this is a sign it is saturating..

Hope this helps

Simon
 

Flyback

Well-Known Member
Thread starter #12
Thanks Simon, i think youre right in that the inductor is too low value, but with peak current just around 3.5A, its not saturating.

This will smooth out the current spikes taken by the SMPS
.......i have a filter in there, so this isnt the problem.

You have a 22uH inductor which sounds a bit low for a 50kHz switching frequency.
................yes i think i should make the inductor value bigger.....but i was wondering that due to the leakage, maybe i have a lot of resonanting current flow in the inductor, greatly increasing the rms current above whats calculated?

As you allure in your article, the sepic with coupled inductor design is like the flyback design, but not quite as there is the leakage inductance to cover...how do you account for the leakage in the sepic?....in the flyback its not a problem as you just clamp it and snub it out.
 
Last edited:

MrAl

Well-Known Member
Most Helpful Member
#13
Hello,

The core loss of an inductor basically comes from the area traversed within the BH curve...the less the better. And larger active BH area is due to a larger change in current. It's not the peak current itself, it is the difference between the lowest current and the peak current that matters and this is usually directly related to the difference between the nominal current and the peak current to make it a little simpler.

To give an example here, say we have an inductor current that is 1 amp average, and say we have a 1.1 amp peak. That means roughly the current changes from around 0.9 to 1.1 amps. Now say we use a lower value inductor, we are most likely going to get a higher peak yes but also a larger difference in the current change, so say the peak is now 1.2 amps, which means it probably swings from around 0.8 to 1.2 amps. Clearly 0.8 to 1.2 amps is TWICE as much change in current as 0.9 to 1.1 amps was, so the inductor who's current changes from 0.8 to 1.2 amps will be less efficient than the inductor who's current only changes from 0.9 to 1.1 amps (all else being equal). So it's not the peak current that matters really, it's the total change in current, because that's what causes the largest number of magnetic domain flips and that eats up some energy.

That's just one of the reasons for lower efficiency however and there are many more. I'll come back with a list later.
 

Flyback

Well-Known Member
Thread starter #15
SEPICS work best if the inductors are bi-filar wound on top of each other
Thanks, i appreciate that, but we are stuck with using the cheapest off the shelf sepic inductors available, we cannot get custom wound cores so we must stick with the MSD1583 series, this has the lowset leakage of any sepic inductor (coupled)

Its so weird that the coupled inductor sepic is exactly liek the flyback, (ie if you just take the cap out of the sepic).

...when designing a coupled inductor sepic, i presume that one uses the same method as for the flyback?

..........the non coupled inductor sepic , i have found, needs such heavy damping of the resonance of the L's and C's that its unusable above about 2W....dont you think?
.......................................................................................................................................................

OK, new findings, my sepic inductor (coupled) is 22uH, and the sepic capacitor is 2u2, so these have a resonant frequency of 23KHz, and this is too close to my switching frequency, and is giving rise to the resonating currents that are overheating the coupled sepic inductor.

Coming away from the hardware and using the simulator, using a much bigger faradic capacitor value gives far less rms current in the coupled sepic inductor coils.

So here is a secret of the industry.
"If you are doing a sepic converter with a coupled inductor, then use the biggest faradic value of sepic capacitor that you can, in order to reduce ringing and losses due to the ringing currents"

....This is a "gem" of information...a real golden nugget, please feast at your leisure, as you will not find this written in any article anywhere concerning the sepic converter with coupled inductor.

(just to mote, when doing a sepic converter with separate inductors (not coupled), you should always the lowest faradic value of capacitor that you can.)

Anyway, what is the maximum voltage seen by a sepic capacitor in a sepic converter with a coupled sepic inductor.?
.....................................................................................................................
More of note regarding the sepic

SimonBramble, your article was great, to augment it further, you could add how you select the sepic capacitor value, and how the resonant frequency of the L and C cannot be anywhere near the feedback loop frequency, ....the loop crossover frequency must be much less than this resonance, otherwise it will ring like mad, especially the non coupled version.

Also, for the non coupled version, the switching frequency cannot be anywhere near the L,C resonant frequency....the switching frequency must be eiether miles above , or miles below , the L,C resonant frequency.
 
Last edited:

ronsimpson

Well-Known Member
Most Helpful Member
#16
I like that coilcraft inductor. I have used many. It is wound correctly. Tightly coupled.

................ADDED LATER............

I have a PWM much like yours running in the other room. I am using a smaller inductor. (coilcraft) The frequency is 500khz. The current is much smaller.

Any time you use an inductor at or near it ratings it will be hot. (not efficient) If the coil was at full current it would be too hot to touch.
 
Last edited:

Latest threads

EE World Online Articles

Loading

 
Top