Help Needed to Solve SEPIC Converter

First, welcome aboard. But a word to the wise, observe the forum rules and don't double post.

Now I'm just a dumb ole ME, but have you put an ammeter in to see how much current is really flowing? If less than 2 - 2.5A, there's obviously a problem in the circuit, which I can't help you with. But if it's 2.5A or more, 2 possibilities

1. Are you sure it's a 24V bulb and not 12V? If it's 12V, the resistance would be less and overload your circuit. This is an ordinary incandescent bulb, right?

2. Check the resistance of the bulb (cold) with an ohmmeter. Tungsten filaments have lower resistance at low temperature than they do at normal operating conditions. So they draw more than rated current at start-up. Your circuit may not be able to supply this extra 'juice'.

Hi ,

Thanks for your prompt reply.

Excuse me for double posting.

I have connected 22ohm,10W and 22ohm,5W resistors in series at load. now am getting 270-300mA with voltage dropping to 13-14V.

In case of bulb, It is 24V,ncandescent bulb. It is not AC bulb. It purely glow in 24VDC. So i have connected it at load side. Thant time my output voltage drops to 8V.

One more thing i dint understand. If my output current less than 2-2.5A why there will be problem??Kindly explain me regarding this.

Here is my circuit. Kindly help me out from here.

Thanks in advance

The diagram is too small to be read properly. Please re-post at a larger size.

I Cant able to attach the file..So only itz coming very small..how to attach the files??

It looks like the problem is in your circuit, which I can't help you with. But, there are others here who can.

You obviously figured out how to attach the file. To make it bigger so it can be read, after you insert it in the editor, just drag the lower right corner out. Then preview/edit the comment and submit it.

Quote "I have connected 22ohm,10W and 22ohm,5W resistors in series at load. now am getting 270-300mA with voltage dropping to 13-14V"

Normal amperage for a 40 watt bulb is 40w/24v= 1.67 amps. Bulb resistance = 24v/1.67a = 14 ohms. You have a series ckt that has 14 + 22 ohms.

therefore amps = 24 v/ 36 ohms = 0.67 amps. That is a little off because the bulb is not at normal temperature. Your ckt is not able to put out enough amps to light the bulb. VD at 22 ohms = 0.67 x 22 = 15 volts. Voltage at bulb = 24 - 15 = 9 volts. These values will be a little off because of bulb temp. If you did not have the 22 ohm resister in series, the required amps for 40 w bulb = 24 v/ 14 ohms = 1.67 a. You need to take a look at the ckt again. Do you need the 22 ohm resistor?

I can't read your circuit, but obviously Wareagle can and what he says makes sense

TIP.

If you cannot read your diagram when posted, neither can we......

None of your diagrams can be read, only guessed at....it may be better for people to send you a private CR4 email with their "proper" email address and you can send a proper copy....

Just a thought.

My personal take is that you haven't taken into full consideration the low cold element resistance of a light bulb.....but, why do you use old fashioned, soon to be mostly done away with incandescent lamps????

Why not try using something modern based on LEDs for example? Problem solved!!!! I expect your circuit to work with the far lower current draw of the LED with no problem.

OK, here's the problem with using a circuit simulator. Most people do not model the core material for the inductor. The simulator works on the assumption that the inductor really has the inductance value under all modes of operation. It is very likely you are saturating the core and the effective inductance drops (and hence the ability to store and deliver energy). Another possibility is the core material or construction is not appropriate for the switching frequency of the converter. If your inductor is a laminated iron core, the eddy current losses will severely degrade the performance.

We have no way of telling from your schematic. Get the inductor current waveforms (peak, not average or RMS) from LTSpice and then check the specifications on the inductor you are using. If you are exceeding the rated current you will need to select another inductor. Usually if ferrite is used for core material in the inductor, it is gapped to prevent saturation. Mag-Inc. offers some nice Cool Mu cores with distributed gaps. Old school iron powder cores can work too but they are generally falling out of favor.

The app notes that come with these switching converter controllers often times will recommend a particular inductor part. Pay attention to that and see what their app engineers found that will work well. Unfortunately, for the uninitiated, it looks very easy to implement a switching converter solution, but with switching converters, the devil is in the details.

I get paid a LOT of money to make these things work well and if it were easy they wouldn't have to pay me nearly as much as they do.

Good luck with your quest and let us know how you're doing.

Cheers !!

Dear Brave,

Kindly go through this inductor waveform.

I have used 47uH, Isat= 4.9A , I rms,2A.

The part number of the inductor is RDR1215-470K-LF

Suggest me why my voltage id getting drop if i connect even 44 Ohm ,20W resistor also.

Is it because of inductor??

Thanks in Advance

Big bucks aside, post #12 is a good starting point.

Switchmode (buck,boost,sepic,cuk,etc.) power supply design/debug/optimization is NOT simple. Nonlinear inductor operation may be your problem. Poor layout, bad feedback path and/or inadequate switch operation may also be factors. You need to examine and understand ALL the switching waveforms at each component to find where your problem lies. If you supply a readable schematic, a detailed photo of your actual prototype circuit, and the oscilloscope traces of the critical switching waveforms we could probably help you.

For hi-res images, your best option is to join a free photo-sharing site and upload your images to that site. Then provide "links" to your images on your CR4 posts.

hi,, there.. the circuit you mention i think it is not a current regulated.it gives you a voltage 24 volts without load,,but not the exact current of 2.5 amperes, and also automotive bulb of 24 volts are inductive load, and need to be more current,about 5 amperes, thats why it glows low only.

you mention that its input voltage varies from 21 to 28 volts,so the circuit should tend to play the the current to adjust automatically in order to remain the power output at constant power. power input = power output

I must admit that the last time I checked an incandescent bulb out for inductance is a long time ago, but it was a pretty tiny value, what has changed? What sort of values have you measured?

Thanks in advance.....

(By the way, I am totally unconvinced!)

OK..so then,if that is the case the circuit you have build are in the problem of the power input.meaning all dc power conversion pass through diodes and have filter capacitor in it, so if you the brigde type design its better,,so check your diodes and try to replace them with new one.. better IN5402..,,FOLLOWS by the capacitor,try out..50volts 2200 microfarad.

Hi XAS35,

Where i need to add bridge diodes and capacitors?? am sorry,i cant understand what you are going to tell..In my circuit where are these bridge diodes and capacitors???

Thanks in advance

hello.. what am i trying to say is, your primary supply, the 21- 28 source voltage,,did you examine its circuitry what kind is it..try to connect the bulb there if it works as you adjust its voltage,the brightness of the bulb varies accordingly.. the diodes and capacitor i mention is there because that your primary power converter..

Review...

supply design:

Single Ended Primary Inductor Converter (SEPIC)
24V at 2.5A output
21V-28V at 4A input (is your source supply current adequate?)

target load:

40W 24V bulb (1.7A, 14.4ohm nominal)

test load:

44 ohm (2x 22ohm) resistor
measured output was only 13V at 0.30A

Inductance of incandescent bulb filaments is usually very small and should not be significant for this design UNLESS the prototype is VERY unstable (bad feedback path).

I've measured cold filament resistance to be ~6x smaller than hot filament, but other sources indicate the value can be even lower (http://en.wikipedia.org/wiki/Incandescent_light_bulb#Current_and_resistance).

The resulting surge current is brief, but it can be a big problem for some power supply designs. Since this prototype doesn't meet the design goals even with a constant resistive load, cold filament startup is not the primary problem (yet). I'd get the circuit working with a single 22 ohm (30W, 1.1A) resistor first, then work on optimizing the design over full range (0A-2.5A plus transient) load conditions.

As a general rule, I start a SMPS design using Vout*Iout*1.25=Vin*Iin. Please verify your input supply is adequate. Buck, boost, flyback converters are somewhat easier to design/debug/optimize. SEPIC and CUK converters are more complex and difficult to implement. I wish you good luck!

Dear mjb,

Thanks for your support.

My actual load is printer(apprx 40W).(But still i have not connected)

I have taken bulb as test load. Once my voltage is dropping with bulb, i started testing with resistors. Even then my voltage is dropping.

One query i had..Will the input power supply source current should have 4A??

Why am asking is, my input power supply source is 22-26V variable and 2.2A max current it can give.

Kindly suggest me what to do now..

Thanks in advance

Regarding your input power supply:

When output is 24V at 2.5A
Pout = 60W
estimate loss at 20% (will be much less WHEN design is debugged and optimized)
Pin = 60/0.8 = 75W
Vin = 22V, Iin = 75W/22V = 3.4A (Minimum)

You can test lighter loads using your existing input supply, but to test the full load you initially specified you will need a larger input supply. These are very basic considerations when designing any type of regulated power supply. I don't want to discourage you, but without the basic knowledge of simple power supply design it is unlikely you will be able to debug a SEPIC design with information I can provide on this forum. However, I'll give it one more try.

Please provide links to:
a readable schematic.
detailed photo(s) of your prototype PCB.
oscilloscope traces of critical waveforms under no-load and resistor load (see below).

ref this wiki schematic since I can't read yours http://en.wikipedia.org/wiki/File:SEPIC_Schematic.gif
Oscilloscope traces of Voltage at:
node L1-Cin
node L1-S1-C1
node C1-L2-D1
node D1-C2-RL