Variable input and Constant output(Sepic)

[samson.bunty1986]

Hi all,

I need to design a circuit, with input voltage of 18V-36V(Nominal 24V), Let the input from 24V battery. And i require Constant output voltage 24V,3A. Kindly support me to design this circuit. Can i go with Sepic? If yes kindly help me to design?

Thanks in advance.

 

[WimRFP]

Hello,

As you may know there is lots of info on the SEPIC topology on the web. I think best is to try to figure out how it functions (what are the voltages and currents in each component for your output requirements). When done, you have some idea of the components you require. the inductors will be the hardest ones. If you have specific problems, you can post them here.

I assume you have a simulator to simulate the proposed circuit. Besides normal operation, also check transient behavior (power up and down, short circuit, etc).

For the output diode, use a schottky to avoid a large transient current during on-switching of the switch and if isolation is not requried, SEPIC can do this job.

 

[samson.bunty1986]

Hi,

Thanks for your reply.

For simulation, i had LT Spice. In that i cant able to simulate properly. Because,we need to generate pulses to drive the Gate of Mosfet by sensing the output voltage.It means by taking feedback from output voltage. Where these and all i cant able to do it in LT spice. Kindly tell me is ther any software to simulate perfectly.

Thanks in Advance.

 

[WimRFP]

Hello,

You shojuld be able to use LT spice for the simulation, but you have to design your own control block (with driver for the mosfet). So if you want to turn-off the mosfet on current, you may add a small resistor and feed the voltage to a circuit.

You can use (for example) the voltage or current controlled switch as hysteresis/comparator block to design a pulse width modulator ((together with a triangle wave). As far as I know, LT spice has also the "B" device to make arbitrary functions (using non-linear function). So I can hardly imagine that it can't be done.

 

[samson.bunty1986]

Hi WimRFP,

Thanks for your helpful reply.

Can you suggest me how to generate pulses to drive mosfet? Which IC can i use for simulation?? My duty cycle range is 0.46-0.58. I have calculated every component used in the circuit. I just want to simulate will my circuit works if i switching the mosfet. Kindly support me on this.

 

[WimRFP]

Hello,
From your previous mail I conclude that you want to simulate the complete converter (power stage, driver, PWM and feedback). If you want an IC solution, you may try:
https://www.linear.com/product/LT3758 as very likely this chip is available in LT spice (as there is a demo circuit on the LT spice website). It enables current mode control and from my experience, it gives good transient behavior.

 

[samson.bunty1986]

Hi WimRFP,

Thanks you soo much for your useful reply.

I have simulate my circuit with that IC. But am getting ripples of 1V. How to reduce that ripples??

 

[WimRFP]

Hello,

Please post circuit diagram and details about ripple (for example image).

When the ripple has same frequency as switching frequency, the output capacitor may be too small.

When the ripple has a much lower frequency (then the switching frequency), your control loop is not stable, so you need to experiment with the feedback parameters. You may check the converter scheme itself by applying a fixed duty cycle PWM signal to the mosfet and observe all voltage and current with varying load.

 

[samson.bunty1986]

WimRFP,

Here is my circuit.If you have LT spice ,run this circuit.Just go through it and tell me how to reduce the ripples. Also vary the input voltage.

 

[samson.bunty1986]

Sorry ,i forget to attach. Here the circuit and ripples image

 

[WimRFP]

From the graph, you are running at around 200 kHz with 50% duty cycle. Your output capacitance is about 78uF. I assume the graph shown is the voltage across I1 component with value 2.5

Assuming 3A load, you can calculate de ripple because of the capacitance. From U = I*t/C and using t = 2.5us, the ripple would be:
3*2.5u/78u = 0.1Vpp.

when I look to your gaph, you have steep edges, so probably your capacitor model has ESR in it (series resistance), otherwise the output voltage couldn't jump. It looks like ESR is round 1V/6A = 0.17 Ohms, that is too high for this application.

Can you check whether yourcapacitors are ideal capacitors or subscircuits containing device parasitics? I don't use LT spice over here.

 

[samson.bunty1986]

The I1 Component is Load with current 2.5A. My output capacitor is from Nichicon. What should be the ESR value of the capacitor????

---------- Post added at 19:04 ---------- Previous post was at 18:49 ----------

Hey,

If i reduce the ESR of the capacitor, the ripples also reducing. But how can i get Required capacitor??? For 35V,50V,63V capacitor voltages the ESR minimum is 0.15Ohm. Can you please tell me is there any capacitor for my output side. Voltage range of capacitor should be 35V-63V.Since my output voltage is 24V.

---------- Post added at 19:29 ---------- Previous post was at 19:04 ----------

WimRFP,

If you dont mind please tell me How you have calculated ESR?? What is that 1Volt and 6Amp??

 

[WimRFP]

As you may know, sepic has low input ripple current (due to the series inductor), but high output ripple, so you need a large capacitor that can handle the rms ripple current. You should think of a ripple current of about 4Arms. The actual size of the capacitor you should determine based on ESR. To reduce the ripple towards 0.1V, you need an ESR of 10 times less (so in the range of 15 mOhms). For the very fast current transients, you need some multi layer ceramic capacitors.

To further reduce the ripple, you may add another LC section, based on a low Q (EMC) inductor that can handle the output current and an electrolytic capacitor or multi layer ceramic.

---------- Post added at 15:28 ---------- Previous post was at 15:14 ----------

when runing a sepic at 3A output around 50% duty cycle and large inductors, the peak-peak ripple current through the capacitors is about 2*output current. so that is 6App, You have around 1Vpp ripple, so the ESR is around 1/6 = 0.166. You have a relative small inductor, so the output peak-peak ripplecurrent is somewhat more, hence the actual ESR is less.

You can "measure" the ripple current in simulation by adding a current meter in series with all the output capacitance, or by displaying the current through the output capacitors.

 

[samson.bunty1986]

Hi WimRFP,

I dint find the capacitor of value 78uF or 100uF with low ESR(0.015 ohm) of voltage rating 35V. I have checked in KEMET capacitors, but i found the capacitor with valus 100uF,ESR 0.05 ohm,25V. Can i use this capacitor??Why am asking this is my output voltage at starting for few micro-seconds it reching 25.9V. Will this voltage damage the capacitor??? Kindly help me to get exact capacitor of values Capacitor78uF or 100uF, ESR 0.015ohm, and voltage rating 35V.

---------- Post added at 12:11 ---------- Previous post was at 12:10 ----------

One more thing. How to add LC section??what value i have to use??

---------- Post added at 12:20 ---------- Previous post was at 12:11 ----------

Hi WimRFP,

The part number of that capacitor is T510E107K025A from KEMET. Kindly go through that datasheet and suggest me can i use this one.

 

[samson.bunty1986]

Dear WimRFP,

The cost of each capacitor is too high, if we want less ESR(0.015 ohm),4Arms. Kindly suggest me about the LC filter. I think this is only one we can reduce the ripple to 0.1V with low cost.

Thanks in Advance

 

[mtwieg]

Look at polymer electrolytics, which have ESR much lower than standard electrolytics. They are a bit more expensive, but still very reasonable. Put three of these in parallel and you get an effective 141uF with 10mohm ESR and 5.7Arms, for a total price of ~6$.

http://search.digikey.com/us/en/products/PLV1V560MCL1TD/493-3866-1-ND/2294175

If you increase your inductor a bit such that your ripple current is less than 3.8Arms, then you can get away with just two in parallel.

This is almost certainly a better option than an output LC filter.

 

[samson.bunty1986]

Hi mtwieg,

Thanks for your reply.

How can i get 141uF??itz coming nearly 168uF(Since capacitors are in parallel).As per my calculation ,my ripple current is 1.086A(del I). So, can i use two capacitors in parallel??

 

[mtwieg]

Whoops, I thought they were 47uF not 56uF. Yes you'll get ~168uF, but overall capacitance isn't really the point anyways.

Is 1.086A your inductor ripple current? The ripple current on the output isn't equal to the inductor ripple current, since the output current is discontinuous.

 

[WimRFP]

Dear WimRFP,

The cost of each capacitor is too high, if we want less ESR(0.015 ohm),4Arms. Kindly suggest me about the LC filter. I think this is only one we can reduce the ripple to 0.1V with low cost.

Thanks in Advance

You need an output capacitor that can handle the ripple current, whether or not you add a post LC filter to reduce the ripple further. The SEPIC topology has an output ripple that has to go somewhere. You may see this in your simulation also.

I think you need bigger capacitance (or more smaller capacitors in parallel) as larger size capacitors have lower ESR. Using an output capacitor with larger capacitance is no problem for the SEPIC topology.

As Mtwieg says, you may increase the output inductor, this reduces the ripple current somewhat.

So first step is to select the capacitor that can handle the current at your design frequency. If after this the ripple is still too high for your requirements, you may add an LC filter.

 

[samson.bunty1986]

Hi mtwieg,

The 1.086A is my inductor ripple Current(del IL).