SMPS boost converter, want more power

26 posts / 0 new
Author
Message

Greetings,

I have incorporated an AVR microcontroller to act as a feedback system (Voltage Sample -> PWM) in a SMPS boost converter to get a regulated boosted voltage. Trying it out on a very low power scale, it worked great.

Once I decided to go up in power, I quickly hit a ceiling (~100mA output).

My circuit is just the simple model for a boost converter you can find on wikipedia or anywhere. The inductor is rated for 2.2A continuous, the power MOSFET rated for 20A continuous, with a switching frequency way past my PWM frequency and drain to source voltage higher than what I need, diode rated for 5 amps continuous. The voltage sampling divider used large resistor values that shouldn't significantly effect the load. Everything SOUNDED fine to get at least a third amp out of the system.

The boost ratio I wanted was large, but still seeming plausible: 9V in, 50V out. So 50V * 0.3A = 15W. 15W/9V = 1.66A.

My suspicion was that my feedback algorithm was going screwy, so I just did an open-loop, pot adjustable PWM signal, but I still had no luck.

So here I am, asking for some advice. Is there some inductor property I misinterpreted for switching? Is there an optimal PWM frequency I should be hitting? Does the transistor need to be rated much higher than the desired output to handle inductor spikes?

Any help would be appreciated, and please spare me the "you're going to die" speech, because I haven't yet!

What inductor value, what switching frequency, and what switching duty cycle?

Be aware that the inductor is used for energy storage. Since you're asking for more amps at the same volts, you need more energy stored in each cycle. Thus you need more inductance.

As above, the inductor value, switching frequency and duty cycle and voltage all factor into the equation.

And the math for it is even on wikipedia - and yes, the equations can be trusted, they're the same I used when taking a course in SMPS's. Sit down, take your time, and understand them. It'll make everything much more clear - it'll be much easier if you know the theory.
Also, which kind of SMPS are you using? I'm just assuming boost, since you want to go up in voltage, but there are many possibilities. As always: Schematics!

Can the 9 volt supply put out 2 amps average, ~20 amps peak? Inductance in the feed lines can limit the current at high switching speed. Is there a capacitor close to the FET? With a big capacitor to supply the surge you should be able to fry the circuit with a high enough duty cycle. (if you don't kill yourself first :))

For a boost converter operating with an 8MHz AVR and no prescaler, the PWM frequency would be 31.25KHz

From the following website:

Vin_min = 9.0V Vin_max = 9.0V Vin = 9.0V
Vout = 50.0V Iout = 0.3A f = 31.25kHz
L = 350.4uH Î”IL for Vin_min = 0.68A

IpK=2.03A, On and off times shown in graphs.

Inductor selection page is also provided with a gapped Siemens E25/13/7 core suggested.

Please take a look at that site, as we cant be much more help with out knowing the switching frequency, Inductor value, power supply capacity, I's not a 9V battery is it?
Ron.

Thanks guys, I spent so much time making sure the parts could handle the current, and not enough time thinking about what the parts actually do.

switching frequency: I've tried a few, I believe it's about 25kHz right now with a duty ratio of 75%

Inductance: 130uH

I thought at first the inductance was a bit too small, but I've seen some other designs out there boosting voltages and getting a high power output using a smaller value than this....

Give us a complete schematic with all details
and component values, and perhaps a photo.

Not enough inductance, the duty cycle should be around 50% at your desired current. Leave plenty of overhead for production of magic smoke. And you do know that the energy is stored in the air gap?

Quote:
I thought at first the inductance was a bit too small, but I've seen some other designs out there boosting voltages and getting a high power output using a smaller value than this....

Did you see the frequency too? Higher frequency smaller parts... :lol:

Regards,

Bruno Muswieck

Well in making a schematic for you guys, I went ahead and simulated it. According to the simulation, I should be able to get what I want out of 120uH, and 25kHz frequency...unless the simulation is being too "ideal" on some of these parts.

Can someone please point me to the math behind why that inductance and switching frequency will not work?

Here are some datasheets for my parts:

Attachment(s):

Did you see the currents? Inductor's currents?
Place it too...

Regards,

Bruno Muswieck

Here ya go. Looks like it's a bit above my inductor's rating, but in LTSpice I also put "Peak Current" of the inductor at 2.2A...

Attachment(s):

A bit above? You have 3,1A as maximum, it's 40% above the current (2,2A).
You need a power inductor.

Regards,

Bruno Muswieck

If you've crunched the numbers and it still doesn't work then you may have some real world problems that are making themselves known. First up, if you're not putting enough energy into the inductor, then you're not going to get enough out. Power supply source impedance is something to check. Use an oscilloscope. Inductor current is the next thing to check. Is the inductor saturating? Is the mosfet turning fully on? Inductance and resistance in the drain circuit affects the mosfet drive. Diode losses and electrolytic capacitor ESR are other things to check. Have you ever wondered why on many switchmode supplies they put a number of electros in parallel? ESR is the main reason.

Does spice simulate the AVR output and mosfet input impedances? I think you need a transistor driver between the AVR output and the mosfet.

It does not look like the logic-level output of the AVR is sufficient for your mosfet. You probably need a driver for it to switch properly.

Do you have a scope ?

What does the signal on the mosfet output looks like ?

Markus

Alright, I know the gate charge has to do with turning the transistor on and off, but is there any way to calculate what size current spike it will produce so I can determine what size transistor to get for the driver?

Will a simple single transistor driver be sufficient? I've seen other drivers online that used a 2-transistor "push-pull" configuration.

If you switch at 25kHz -> 0.04ms time period - lets say you want your switching to happen in 0.5% of the time period, i.e. 0.2 micro seconds. (the longer your switching period, the more power you lose during the switch)
Then roughly - if you have a 100nC total gate charge (from your FET datasheet), you will have an average current spike of (using Q = I*t) 100n/0.2u = 0.5A.

Thorough discussion is at http://focus.ti.com/lit/an/slua054/slua054.pdf. Lots of oscope porn :)

Just goes to show that in many cases, it's easier just to go with a specific chip that does the proper job rather than use a micro.

Which chip should I replace the micro with? That driver chip? I don't think so...

but adding it to the mix sounds like a good idea.

Alright, got a little impatient tonight waiting on MOSFET driver chip, so I tried designing my own...Are there any objections that this will work? Just trying to make something to keep me occupied...

The AVR-PWM source does not go above 30mA...but I'm getting that big push/pull on the MOSFET gate (modeled with a cap...can't see it in picture)

Attachment(s):

There's a huge array of devices for power supplies these days. For something old school,there the UC3842.. series that drive the mosfet directly and have the usual pwm circuitry.

Just thought I'd give an update:

1) My power source was limiting me from getting some power out. It claimed to be capable of producing the power but after some tests I found out it was not. Bummer

2) The mosfet driver chip came in, but before it did I found out about bjt common-emitter driving circuits that worked fairly well for what I was doing, so I haven't even taken the chip out of the bag.

3) I switched it up a little bit, and now my SMPS is being utilized for a constant voltage SLA battery charger. Pretty cool! A lot less work on the SMPS (input 12V, output ~15V) but I plan on using a solar panel to provide the power.

4) I've been thinking about designing a power inverter. Is it realistic to try to utilize a SMPS to boost the voltage from 12V to 170Vpk and still get some power out of it? Maybe 2-3 Amps?

My plan of attack would be: Boost -> Sine Wave Modulated PWM H-Bridge Inverter -> Filter -> 1:1 isolation transformer.

Or is it just more practical/efficient to use a step-up transformer:

Sine Wave Modulated PWM H-Bridge Inverter -> Filter -> 1:14 Step Up Transformer?