help on inductor design for buck type dc-to-dc converter

[markking]

hi. im designing a buck type converter to stepdown a 24v battery supply to 12v 15amps to supply my radio transceiver for my field tractor. Ive tried several designs including the transformer type design but later on realized that the size of my transformer's too bulky to fit into my tractor. somebody suggested the buck type design which i think is perfect for my application. now the problem, im stuck w the inductor design. i need at least 120uH that could run through 15amps of current.
can anyone help me? my resources is only limited to air core or laminated e-i core design. i hav a no.12 awg double insulated magnet wire.

 

[Glyph]

Don't know if this helps but here's my two cents:

For a radio transceiver application the interference from a switching regulator may be problematic if you use an air core inductor. Heck, even the drum-core inductor of my cell-phone car charger ruins the reception on my FM radio (particulary in the 99-108 Mhz band). I would seriously reccomend trying to obtain some toroidal core inductors if possible. Terribly tedious to wind but massively better than air-core when it comes to interference. In addition its fairly easy to stack several toroidal cores to get very high inductance levels.

If you're use your E-I cores do you know if they are rated for high frequency operation? if not they might heat up considerably. If they are rated and you still want high inductance can i reccomend stacking those too? placing the cores side by side sandwich style. You can get away with fewer total turns and have more inductance.

At high frequency, really thick gauge magnet wire isn't much help anyway due to the 'skin effect' (do a google search for more info). You're better off running multiple strands of wire in parallel for higher current. A friend of mine built a high current inductor using EI cores and copper foil, getting around the skin effect quite well.

Sorry i can't give any numbers, I don't know the permeability of the core you're using.

As a final suggestion: if you're willing to part with a little extra money, you might save yourself a load of time by buying an ATX powersupply that has a 24V input (yes they exist). They already have a 12V line that's buck regulated and with some VERY MINOR modifications (mainly just adding some resistors to ATX connector and a switch) you can turn it off and on at will without a computer attached, then you can use the 12V regulated output for your radio tranceiver. In addition, unless you order from a seedy looking company, the power supply should be FCC approved and cause little interference, any interference it does cause can be mitigated by simply locating some distance away from your tranceiver.

Don't know if this helps at all. good luck with your project!

 

[markking]

Actually, il be using a e-i laminated core with either 1.25in or 1.5in center leg with. il be switching it using a FET at 20khz. i dont know about the permeability of the core.

abt the atx computer power supply. i dont have very much idea about its design but does a conventional atx 300w power supply with 12v output means it can give 25A of current? (300w/12v=25A). if so, i think i can modify a 220vac input supply to my need. t.y.

 

[Glyph]

No,

a 300watt ATX only means 300 watts of total power, the individual currents on each line are less. I think that a conventional 300watt atx actually gives 10amps on the 12 volt line. Bigger supplies give higher currents.

 

[Oznog]

120uH for 15 amps at only 20KHz does not sound right. The ripple will be heavy. I did a 1 amp Luxeon driver with a drum choke and found around 80KHz to be a better freq. Make your system with an adjustable freq so you can find the highest freq that keeps the transistor and coil reasonably cool.

You cannot expect to do a whole lot of filtering with a capacitor on the output, you need the inductor to do most of the work.

Frankly, for a fixed ratio, I would advise you to use a transformer. Size is not the issue- either a transformer or buck inductor can be large or small depending on the freq used. There are toroidal cores with 2:1 windings you could use, or maybe you could wind your own. As long as you have a core which works at high freq, you can do the job with small, low inductance devices.

The design issues are- DC resistance (not trivial with 15 amps!). There are many different types of cores with diff freq capabilities, if you exceed them, the core will heat up. If the core heats up due to wire resistance or too high a freq, the inductance and saturation current suffer. If you exceed the saturation current, the inductor partly stops being an inductor and ripple goes wild- basically, the inductor stops resisting the current change and acts as a wire, so the output will go to 24v for that part of the waveform.

 

[markking]

can anyone give give the formula for making transformer w my given specification using an e-i laminated core. i dont know the permeability nor any other specs aside from the fact that i have a 24v lead acid supply with 60amps of current. and i need a transformer to step it down to 12v 15amps. i googled for the answer but always end up with nothing because i am required to fill in data about my core which i dont have. but i heard from others (who cant also give the details) that you can get away without those infos only the ones i mentioned and the area of the center leg of the core & data of the wire. please help. thank you.

 

[TheOne]

To calculate the windings for a switching transformer, you need to know 3 basic things.

1/ Ae (effective area of a core)
2/ Peak flux swing (Tesla) that the core can safely tolerate at the various frequencies.
3/ The operating frequency and max duty cycle (on-time)

 

[markking]

oznog,
i data i gave i got from an online inductor calculator some time ago. i kept the result for this project. unfortunately i cant seem to google where it is located now. i just inputted the input voltage=24v, output voltage=12v and current=15A, the Rdson of the FET switch=0.077, Voltage drop of the ultrafast diode bridge =0.95 in the buck type configuration and switching frequency=20khz. That's it. then it gave me abt 112 uH or so of minimum inductance of the coil. i tried using higher frequencies but the microcontroller im using is only 1us per cycle thats why i settled for 20khz.
do you know any other way to computer inductance of coil using e-i laminated core or air core?
thank you.

 

[markking]

dats d problem, i dont know any other things about my core. the retailer i bought it from cannot give me any. wat i know s only wat i see. like the dimention of the e part. is there any way around this limitation?
thank you.

 

[TheOne]

I see what you want is really a choke for a buck converter, so here you need tha Al value of the core to make a calculation on the windings as shown in the picture

The on-time is in uS

 

[TheOne]

If it is a laminated core (most probaly made from silicon-steel), I doubt if it will be usable over a few hundred Hz as the losses will increase rapidly with frequency. You need a ferrite or powder-iron type of core for the choke if you ever hope to switch at 20kHz!

 

[markking]

thanx. i googled for more info regarding coils for a buck converter and it is recommended that the inductor should be at least 10 times the computed value to make it more efficient. so basing on your computation il be needing 1mH of inductance (100uH X 10). and since i cannot possibly make it neither with torroid nor powder core (no one manufactures them here in the philippines) is it possible to make it air core? perhaps i can filter further the output of the converter using another choke (a 12v, 25A ready made choke made for automobiles). in case it'l give out rf noise perhaps i can encapsulate the coil in a stainless steel box. as you can see i have to insist on using only wats available. either an ei or air core coz i have no other choices. thank you again for the help but im getting depressed.

 

[TheOne]

Really?? I find this hard to swallow. These formulas were derived from sound Engineering principles and confirmed by measurements over the years. I can accept a slight increase in the value to compensate against the inductor "running dry" on low load currents, but 10 times ??

 

[markking]

yeah, dats wat they recommend, X10 the computed value and im not taking chances so im making my inductor big. heheh. better big so i can just remove some turns in case its not working ryt. i dont have an inductance meter so il have to see the output to see if its working. and btw, i already made it. its an air core awg#11, 2" inner diameter, 2" in length and weighs about 950 grams. the size is quite small too. ive enough room for the fuse holders and switches in my 6"X8"X9" case. the varnish is about dry now and im gonna test it the moment its ready. im hoping this will be the last time il be tingkering with its electronics. the rest would be on programming the microcontroller and getting the right frequency and duty cycle.

 

[TheOne]

Mind sharing the website where you got this information?

Inductor value is calculated to give a balance between load regulation and transient response, taking into account minimum (10% of maximum), and maximum load current. Making the inductor larger than the calculated value will improve load regulation but transient response (how fast your regulator reacts to sudden load changes) will degrade.

Sure you are not somehow getting the 10% mixed up with 10 times? :?

 

[TheOne]

im hoping this will be the last time il be tingkering with its electronics. the rest would be on programming the microcontroller and getting the right frequency and duty cycle.

I am not so sure