Boost converter vs. SMPS

Most car audio amps use a bipolar supply, so a torroid with multiple secondary windings is used.

For an amp with a unipolar supply, a boost converter may be a practical solution.

Note that a unipolar supply implies bridged or cap coupled operation. The potential saving in power supply costs may be more than offset by the increased amp costs.

Self oscillating torroid supplies can be very simple. I have yet to see a simple boost/buck converter design. A torroid supply will also tend to have less ripple, assuming full wave recitification is used.
Dc-dc Smps


Try the Elliott Sound Products (ESP) website:

There are alot of great schematice for doing +/- DC-DC converters to run almost any size amp you want off a +12VDC source (like your car). The ESP circuit uses an SG3525AN voltage-mode PWM chip, which is VERY easy to use. (There are also several other, better chips, but the '3525 is an excellent starting block).

I have made MANY +12V DC-DC units using this controller, all with very good results. This chip can drive N-Channel MOSFETS directly, as it has a drive capability of +/- 400mA. Just add a few components for true PWM operation, and you will have a very high quality REGULATED power supply for your amps. (Better than the unregulated converters running in most of the mass-produced amps like Alpine, Kenwood, etc.)

For excellent noise isolation, try using an optocoupler (like a 4N36) to sample both the (+) and (-) outputs equally. This is alot easier to implement that one might think.

Here is a schematic for the MOSFET-based SG3525 converter, running wide-open with no feedback. If you don't need regulation, then this might do the trick. If, however, you want some regulation, try looking at schemtaice of some other DC-DCs on Elliot's Pages.

Best of luck with your project,

Steve :D :cool:


  • esp sg3525-based smps.gif
    esp sg3525-based smps.gif
    11.2 KB · Views: 450
Unregulated push-pull topology allows for easily obtaining very high power outputs with a small transformer and producing little EMI. They also provide galvanic isolation to the output and require little filtering.

On the other hand, buck and boost topologies allow for regulation but require a big inductor, that must not saturate even for the highest power output, and are prone to EMI due to diode reverse recovery. Furthermore, buck converters draw a discontinuous input current, so peak current is higher than average thus increasing switch stress and requiring careful filtering, and boost converters work with discontinuous output currents, so more filtering is required.

For a boost converter working in continuous conduction mode with Vin=12V, Vout=36V and Iout=10A:

- Output current will be actually 30A for 33% of the time and nothing for the resting 66%, a lot of capacitors with high ripple current rating are required to filter that.

- Input current will be 30A continuous.

- An inductor rated at 30A is required.

Discontinuous mode gets rid of diode reverse recovery but things get even worse:

- Output current will ramp from 60A to 0 for 33% of the time and will be 0 for the rest of the time.

- Input current will ramp from 0 to 60A for 66% of the time and from 60A to 0 for the resting 33%.

- An inductor rated at 60A is required.

On the other hand, 36V 10A unregulated are easily produced with a push-pull transformer, without involving discontinuous currents or much higher peak than average current values.
Another thing you can do is mate a high/low driver like the IR2103 to the SMPS controller chip and run a full-bridge supply. The core would be a little easier to wind and the full bridge has the potential to output many kW of power.
The drawback is that the circuit is more complex.

Many people now are making smart power supplies using microcontrollers. The 18-pin PIC18F1220 for example has an enhanced PWM module that outputs all of the signals necessary for full bridge or push/pull modes [or any mode, for that matter] I know a guy that builds power supplies with one of the 8-pin PICs as well.

WIth a microcontroller you could do interesting things like log the high/low voltages for a period of time, change the bus voltages when the volume gets turned up, etc..
I find it interesting but that may just be the EE control freak in me :cool:
The selection criteria between full bridge, half bridge or push-pull is something that has been discused here in detalil several times.

The point is that for such a low voltage as 12V, push-pull is actually advantageous over full-bridge because it allows for the same power output using less components, less board space, less heatsink space, and at a lower cost.

Remember that MOSFETs rated at 30V are actually harder to find and more expensive than the ones rated at 60V, and their Rds-on is only about 30% lower, so you end using more devices than in push-pull. Also note that high side driving uses board space and requires expensive ICs or discrete charge-pumps. The layout for full bridge is by far more complex, requiring double sided PCB, while push-pull is much simpler and easy to implement in a single sided PCB. On the other hand, the cost of an additional primary winding of only 4 to 6 turns on a toroid core is almost negligible.