boost converter design

[evandude]

here is my current design schematic.

**broken link removed**

For all practical purposes, all the stuff between V1 on the left, and the inductor and MOSFET on the right, comprise my MOSFET driver circuit and can be thought of as just a PWM pulse source.

First off, I'd like to say that i'm reasonably knowledgeable in electronics, but very new to power supply design, so bear with me.

The first problem I thought I'd ask about is the large currents I'm seeing. I am using Micro-Cap for simulation. at a PWM frequency of 2kHz at 10% duty, i am seeing about 27 amps through the inductor (average value) with a 5 amp load on the circuit. However, that is with an input voltage of 12v, and an output voltage of about 11v.
22 extra amps to boost less than 1/2 volt above the output voltage with no switching... (just a diode drop below 12v) that's pretty outrageous.


at a duty cycle of 20%, I am only boosting enough to overcome the diode drop (12v in, 12v out) and yet the inductor current averages over 30 amps.

I don't entirely trust Micro-Cap at simulating this type of circuit... So i want to know if this seems realistic, ie - if there is something really wrong with my design.

Mind you, I am only simulating for the first 5 mS since it takes ages to simulate much further. However over that period the output voltage stabilizes quite well as does the inductor current... so unless it's going to settle down to a current that is around 1/5 or 1/6 of its present value over some much longer period of time, then I need to figure out the problem.

Is it my design?
Is it because I'm only simulating for a short duration?
Would switchercad give me some drastic improvement in simulations?

thanks in advance.

 

[Schneibster]

Sounds like its your simulation program- I've heard a lot of them have problems with switchers. I'd do an analysis using Dr. Middlebrook's Canonical Model and see what it looks like.

 

[k7elp60]

All the switchers I have seen have a voltage reference and a feedback loop, I don't see either in your schematic.

 

[evandude]

well i just did the sim in switchercad. certainly does a much better job of simulating the circuit... however the current is still 30 amps or more (and growing) for even small boost.

the reason there's no feedback shown is that I had been going under the assumption that my leaving out the feedback control system would be okay just for this part of simulation, since i can manually tweak the duty cycle and wait for the output to settle at some value... but it's obvious it's not going to happen. so I guess i'll have to put in the control circuitry and see if the current levels will settle down. Without it it looks like the current is just running away to infinity.

 

[Schneibster]

My bad- yes, you will need your control circuit for the simulation to work.

I have been doing some research, and I have found some things that you will want to consider. A boost converter has pulsating output current (noisy! :? ) and you never want to allow 100% duty cycle in a boost converter, because it will destroy the switch. =:0 You will need to design your controller to suit.

You might want to consider the implications of the fact that most automotive alternator/regulators put out 13.7V or more- up to 18V in some cases :shock: - and your boost converter design, or perhaps your load, may have trouble with that. Make sure you simulate it, not to mention the noise that comes out of the regulator- you'll be surprised that auto manufacturers call them "regulators" when you see it.

Because the inductor and diode are in series with the load, with no blocking, you will have noise rejection problems as well as no overvoltage rejection- and the cap will not help with either, because it will already be charged to 12V.

You might want to take a look at this material.

If you want a common ground, the boostbuck converter is probably more usable than a straight or isolated Cuk converter. I am actually looking at converting my design to a boostbuck; I have identified an International Rectifier P-channel power MOSFET part that is available from my favored supplier. But I'm not sure I want to increase the parts count; nor am I sanguine about the efficiency with two switches instead of one. For your application, it's probably the best bet, because if you create a ground loop, it won't matter- it's common-connected anyway. Not to hijack your thread, I'll put the details in my thread. :wink:

 

[evandude]

the thing about my design is this: The load already has a linear regulator in it, so it's good for up to like 20+ volts. The DC-DC is just to make sure the voltage always keeps the output above about 12 or 12.5 volts. when the voltage is above that value just by virtue of the alternator, then my converter will not need to operate at all (duty cycle=0%)

Therefore, it will only have to kick in when the voltage droops (ie-starting the car) I realize it might be more elegant and/or efficient to just regulate directly to 12v with a buckboost or other design... but that seems like it'd be more challenging, and I feel like I'd be reinventing the wheel. As long as my efficiency isn't absolutely awful, input current isnt' that big a problem... after all, a few amps is a drop in a bucket in a car electrical system, compared to the car audio systems some people run... so even losing power in the DC-DC, and the linear regulator, and the other DC-DC that generates the rest of the computer voltages, isn't that big an issue to me... So overall my goal is enhancing my current equipment. when I have time I would love to build a full power supply, that inputs 12v and outputs all the normal computer power supply voltages, so I can replace 3 different devices with one. but for now I just want to get this working.

I already have a bunch of filtering for the load... (which is a computer, whose main purpose is to play music) due to the high pitched alternator noise present on the power supply. (large inductor and two large capacitors)
so theoretically any noise present on the output should get decently attenuated, especially if it's far above audio frequency (which it should be, since the converter will run above audio frequency)

but thanks... I guess i'll have to throw my feedback/control circuit in there too.

 

[FRIED]

2000Hz, 10% = 50µS. VxT/L = I.
12x50^-6/18^-6 = 33A

 

[evandude]

sounds bad. but first off, i'm running the simulation at 100kHz, and will be running the real thing at somewhere around that level.

so by my calculations, that's 660 mA. but that also assumes only one cycle, whereas the cycle is being constantly repeated, that's why the current slowly climbs up to 30 amps and beyond... the way it's looking, I wouldn't be surprised if it just kept right on growing...

so like i said... I guess my simulation is going to be worthless until i stick in feedback/control.

 

[evandude]

on reading the latest reply to Schneibster's DC-DC thread, I've decided to go with a TL5001A PWM controller and a MIC5015BN MOSFET driver.

that takes care of most of the headaches... since I'll be using tried and true chips, with circuits taken from the datasheets, rather than risking things with my own cobbled-together design. not to mention, the circuit will come out a lot smaller, and power for the control circuitry won't be an issue since the MOSFET driver has an internal charge-pump for driving the MOSFET gate.

the only downside of that is that I can't simulate either of them so I'll have to just go straight for real-world building. but since i'm using THEIR schematics and such, I won't worry so much.

 

[evandude]

okay, so new question... what size inductor should I use? I plan on running at 100kHz, with a 1000uF cap on the output (since the ripple with that cap value looked decent in simulation)

in all my simulations I've just put in arbitrary values that are either guesses, or values I've seen in other people's designs.

 

[Schneibster]

Here's where the design equations and the canonical model come into play. I don't know the design equations for a boost converter; the ones for a Cuk converter are the ones I have used. When my design is complete, I will take a look into the canonical model's equations in Cuk and Middlebrook's books (Advances in Switched-Mode Power Conversion), and in Pressman's Switching Power Supply Design, and see if I can derive (or- best of all worlds- find) design-oriented equations. You'll need to consider the current-carrying capability of the coil as well as its inductance- this is clear from the fact that you have to consider the potential-handling capability of the capacitor as well as its capacitance. It might be a few days, or more depending on how much time I get to spend on my design considering I have work to do.

 

[evandude]

schneibster, you are just helping me to no end. despite the fact that i'm a college student on a campus with a huge library, it never occurred to me to look there for books until you mentioned it. turns out they have several good books on the matter. off I go...

 

[Schneibster]

The two I mentioned are the two best. You want Advances in Switched-Mode Power Conversion, volumes 1 and 2, R. D. Middlebrook and Slobodan Cuk, TESLAco 1983, or else the two papers, Design Techniques for Preventing Input-Filter Oscillations in Switched-Mode Regulators, by R. D. Middlebrook, from the Proceedings of Powercon 5, the Fifth National Solid-State power Conversion Conference, May 4 - 6 1978, San Francisco, CA, and Modelling and Design of the Cuk Converter, R. D. Middlebrook, from the Proceedings of Powercon 6, the Sixth National Solid-State power Conversion Conference, May 2 - 4 1979, Miami FL. I'm not sure you can get the two papers anywhere else. Dr. Middlebrook is a genius electronics scientist, and Dr. Cuk is a gifted electronics scientist and a genius inventor. IMHO, the advances these people have made in the science of switched-mode power conversion, and in fact in power conversion in general, stand head and shoulders above anyone else in the field. They are generally ignored for several political reasons, and deserve any promotion they can get.

For general reference, if you decide to continue with your boost converter instead, try Switching Power Supply Design, by Abraham I. Pressman, McGraw-Hill, 1998 (Second Edition). The ISBN for SPSD is 0-07-052236-7; I don't think ASMPC has an ISBN number.

If you insist upon owning ASMPC, as I did, you can get it from TESLAco at 213/795-1699. If that number doesn't work, let me know, or google up TESLAco. You'll find Dr. Middlebrook's canonical model far more useful in modeling your converter than anything Pressman has to tell you; but Pressman has considerably more material on many other SMPS designs, and it is considerably better organized, not to mention being something a little more coherent and accessible than a large collection of scholarly technical papers, which is all that Middlebrook and Cuk have to offer.

On edit: I hope it's "helping you no end," rather than "helping you to no end." I'm sure it is.

 

[evandude]

I ended up with "switching power supply design & optimization" by Sanjaya Maniktala, 2005.

our library didn't have a huge selection of SMPS books, and right now I don't want to buy any...

 

[Schneibster]

If it's any good, particularly if it covers the canonical model, or some permutation of it, please let me know. Always looking to improve my library.