SEPIC converter design question

Hi,

I used this calculator to calculate my needed inductor.
**broken link removed**

the thing is, it takes Vin max and Vin min,
and an absolute Vout.

I need to have a variable Vout as well.

The application is to charge a big output capacitor (3000 uF+) to 250-700 volts - variable. sometimes to 250, sometimes to 700, and sometimes in between,
the Vin rang is to accept 110vAC or 220vAC which gives when rectified about 150v min and 350v max.

my parameters would be:
Vin min 150v
Vin max 350v
Vout min 250v
Vout max 600-( even 700v)

I out (wanted) about 1A. average.
say 1.5A at 250v out and 0.8A at 700v out (if it is possible to vary, I would love to have a fixed input current, so the device always take about 500 Watts from mains, let's say)

My problem is
for Vout 250v it gives 2794uH
for Vout 700v it gives 1296uH

what inductor should i use ?
should i choose the biggest uH inductor ? the smaller ? or an average between ?

Also,
these values are huge, and coilcraft has maximum 1000uH coupled inductors to fit my needs,
i tried and put 3A in the max output current, then the value goes to ~400uH which is definitely easier to find.
the question is, what is the consequence if this current will not be met ?

I remind that we charge a big capacitor. at start i would need the maximum I out to charge the cap when empty, say 1A, but then it is charged and the current falls to nearly 0.
so can i take a smaller uH inductor (which means higher I out) but actually limit the current of the primary using a larger resistor, and thus preventing it to draw to much power from the mains ?

switching frequency will be max 30KHZ because i will use a 1200v IGBT that works at max 20KHz

any tips and hints would be highly appreciated !

sepic not much good for that high power level (250W).

Sepic rings too much when you switch on at that high power level.

use transformer isolated smps

- - - Updated - - -

sepic not much good for that high power level (250W).

Sepic rings too much when you switch on at that high power level.

use transformer isolated smps

grizedale said:

sepic not much good for that high power level (250W).

Sepic rings too much when you switch on at that high power level.

use transformer isolated smps

Click to expand...

Oh,
bad news.

what topology would you suggest for this ?
Flyback converter ?

For this application it might be fine to operate it in DCM, meaning you don't need such high inductance. So long as you can tolerate the higher ripple currents on the input and output it should be fine. And you don't absolutely need a coupled inductor. In fact I'd be surprised if you could find one that would work at such high voltages.

mtwieg said:

For this application it might be fine to operate it in DCM, meaning you don't need such high inductance. So long as you can tolerate the higher ripple currents on the input and output it should be fine. And you don't absolutely need a coupled inductor. In fact I'd be surprised if you could find one that would work at such high voltages.

Click to expand...

OK, the prototyping will surely be with regular inductors, not coupled.

How do i set it to work in DCM mode and not CCM ?
I am using the UC3842 chip.

what does the high ripple currents actually mean ?
what consequences will it have on the input and output ?

How do i start here ?
seems like i have too many variables, that can change during operation.
Vin can change, Vout can change, I out always changes because i am charging a big capacitor....
I am totally confused. the calculators sometimes give me 6000uH and sometimes 400uH. all with possible scenarios.

do i have a too-wide input range ?
will it help make things simpler and more efficient if i make 2 versions (one for 150Vin and one for 320Vin) ?

Thank you so much for guiding !
looking forward to your answer
I have ordered a bunch of 1200v IGBTs meanwhile

Well....
mtwieg - what do you think ?

moshik3 said:

How do i set it to work in DCM mode and not CCM ?
I am using the UC3842 chip.

what does the high ripple currents actually mean ?

Click to expand...

DCM just means that the inductor current falls to zero during a switching cycle. DCM/CCM aren't really "set up." They just occur depending on the various operating conditions and component values in the converter:
1. Input voltage
2. Output voltage
3. Duty cycle
4. Operating frequency
5. Inductance
6. Load current

what consequences will it have on the input and output ?

Click to expand...

DCM is associated with higher ripple current, which is generally associated with lower efficiency, but not always. In an application like yours where you're operating at relatively high voltages and low currents, and you're operating under a wide range of line/load conditions, avoiding DCM in all cases is pretty much impossible without having enormous inductors or operating at very high frequency (both of which could easily kill efficiency).

The other main effect of high ripple currents is that it can cause additional heating in capacitors in the converter. In sepic converters, you especially have to make sure that the blocking capacitor is capable of handling the ripple.

A third effect of DCM is that it changes the transfer function of the converter in a couple ways. On one hand, the transfer function becomes more nonlinear (making things a bit more difficult) but on the other hand it eliminates the RHP zero in the transfer function (which is a good thing). But you should probably worry about control later.

How do i start here ?
seems like i have too many variables, that can change during operation.
Vin can change, Vout can change, I out always changes because i am charging a big capacitor....
I am totally confused. the calculators sometimes give me 6000uH and sometimes 400uH. all with possible scenarios.

Click to expand...

A lot of those simple calculators are only useful for DC/DC converters with relatively fixed line/load conditions. And they generally assume you want to operate in CCM all the time. But that's just not true for your case (unless you really like the idea of having huge 6mH inductors). So your approach should be a bit different.

First I want to ask a bit more about the application. You say this is for charging a capacitor in a controlled manner, does that mean the capacitor voltage will start at zero? How fast do you want the charging to be? What kind of capacitance are we talking? What rep rate?

Thanks I got it.
so if i understand right, i CAN use a lower inductor, say 470uH (just throwing a number), and then it will be in DCM, which is not too bad in my case...
what effect do have higher ripple currents ? i still do not understand that.

also, I have an understanding that the capacitor size between the two inductors is setting the converter current.
and this capacitor has an average voltage of like Vin (say 320v) - but what should be the MAX voltage rating for it ?
and it has to be a low ESR type to avoid heating, so i thought of using a CBB Film capacitor rated at 400v (is it enough?)
any idea what size of capacitor i need ? are wqe talking about 4.7uF or 47uF ? (a ~47uF Film thing at 400v is huge and expensive)

my application:
1.
I will be charging a 3000uF capacitor (for the bigger model) to 600 or 700 volts,
the capacitor starts at 0 volts because the appliance wasn't connected to power and the capacitor is empty
this is 700 Wattsecond energy, which i would want to charge in 2 seconds say, which is 350 Watts to the capacitors. (does it count as 0.5A at 700v ?)
2.
The discharge phase: in about just a few miliseconds the capacitor will be discharged, until it reaches about 80 volts. which means the next charge cycle starts from ~80v in the capacitor.
Here i need to integrate a system that will totally disable the SEPIC converter during the short discharge, to prevent power supply during the discharge, otherwise the discharge device won't turn off.
3.
A new charge cycle begins. the minimum interval between charge is done to the next charge is 1 second.
of course this is for some time, about 1-2 minutes, then it is allowed to rest a bit.
or sometimes the interval between two charge cycles will be 5 seconds or 10 seconds. it really depends on the user.
but it is understandable to me that a shorter intervals between charging and charge to a higher voltage will lead to higher heating which will require cooling.
i'm not looking for 99% efficiency, but neither can i put a water-cooling on it


The 700 wattseconds capacitor charged to 700 volts in 2 seconds is the "worst" case scenario.
we can slow it down a bit if needed, we can use a smaller capacitor and charge to higher voltage, etc.
i will also have a version that charges 1000uF capacitor "only"...

i really CAN accept a "110vac" and "220vac" versions for the circuit, if it makes things simpler, so the voltage input range would be reduced significantly (300-350v for 220vac and say 140-180v for 110vac)

what do you think of the IKW40N120H3 IGBT ?

Thanks so much,
can't wait for your next reply you are very helpful !

meanwhile i ordered these:

3mH 6A common mode:


1mH 2.4Adc single:


1mH Isat:1A single:
470uH Isat:1.4A single:

moshik3 said:

Thanks I got it.
so if i understand right, i CAN use a lower inductor, say 470uH (just throwing a number), and then it will be in DCM, which is not too bad in my case...
what effect do have higher ripple currents ? i still do not understand that.

Click to expand...

It increases the RMS ripple current in the converter. That's... just what ripple current is. I'm not sure how to simplify it any more. It influence the losses in all the capacitors, and also increases the peak currents the inductors must withstand.

also, I have an understanding that the capacitor size between the two inductors is setting the converter current.
and this capacitor has an average voltage of like Vin (say 320v) - but what should be the MAX voltage rating for it ?
and it has to be a low ESR type to avoid heating, so i thought of using a CBB Film capacitor rated at 400v (is it enough?)
any idea what size of capacitor i need ? are wqe talking about 4.7uF or 47uF ? (a ~47uF Film thing at 400v is huge and expensive)

Click to expand...

The capacitance should be enough so that the resonant frequency it forms with the inductors is much lower than the switching frequency (by about a factor of ten or more). It must also sized in order to take the required ripple current in the converter. Generally when you satisfy the ripple current condition, you'll also satisfy the minimum capacitance requirement (because higher ripple current generally requires higher capacitance). At 20KHz, and L=470uH, 4.7uF might be enough. For the voltage rating, it should be greater than the maximum Vin. However under some conditions that's not enough. Like if the converter suddenly stops operating when at a high load, the circuit may ring with a large amplitude, potentially greater than Vin. This situation should be simulated to see its severity.

1.
I will be charging a 3000uF capacitor (for the bigger model) to 600 or 700 volts,
the capacitor starts at 0 volts because the appliance wasn't connected to power and the capacitor is empty
this is 700 Wattsecond energy, which i would want to charge in 2 seconds say, which is 350 Watts to the capacitors. (does it count as 0.5A at 700v ?)

Click to expand...

Depends on what charging profile you want. If you charge it with constant current, then your power will range from about 0W to 700W within a charging cycle. But constant current isn't absolutely necessary...

2.
The discharge phase: in about just a few miliseconds the capacitor will be discharged, until it reaches about 80 volts. which means the next charge cycle starts from ~80v in the capacitor.
Here i need to integrate a system that will totally disable the SEPIC converter during the short discharge, to prevent power supply during the discharge, otherwise the discharge device won't turn off.

Click to expand...

Shutting off the converter will take some care. It should be allowed to reach a steady state before discharging the capacitor.

i really CAN accept a "110vac" and "220vac" versions for the circuit, if it makes things simpler, so the voltage input range would be reduced significantly (300-350v for 220vac and say 140-180v for 110vac)

Click to expand...

You could just have a switch on the input which changes your rectifier between a full bridge (for 220Vin) or voltage doubler (for 110Vin). This is done in lots of instruments, and it works fine so long as the switch is correctly set...

mtwieg said:

Shutting off the converter will take some care. It should be allowed to reach a steady state before discharging the capacitor.

Click to expand...

this is quite problematic.
because the discharge is unpredictable.

what i can do, is preventing the user to discharge the capacitor while charging after a previous discharge.
but say it is charged, and the voltage drops in the capacitor, it will start to re-charge, and at this stage i can't prevent a discharge.
the circuit must be ready to discharge in any given time after the initial re-charge

- - - Updated - - -

why would it need care before turn off ?
if i just raise the voltage on the current sense pin of the UC3842, it would stop the converter in a very natural way, just like it turns off the IGBT in every cycle, no ?

a voltage doubler is not ideal for me. really not.
electrolytic caps will heat severely,
and Film capacitors will, again, need to be huge.

i could just use voltage multipliers for my application, instead of all this converter mess, but to achieve decent power from a doubler i just need big Film caps (40-60uF),
and also some way to limit their charge current from the mains, because circuit breakers don't like seeing a 60uF film capacitor suddenly directly connected to the mains....