- I am designing a low power controller for a DSP board and I needed a

power circuit that would generate the 3.3 and 2.5 volt supplies the

MCU needs. This also needed to be as efficient as possible, but cost

and size are both issues. The circuit I ended up selecting is a

switcher, but not an inductive switcher. TI makes the TPS60500 family

of switched capacitor stepdown regulators that automatically configure

the topology to optimize efficiency. 3.3 volts is two thirds of 5

volts and 2.5 is half of 5 volts. Turns out you can configure a

switched capacitor converter to generate an output at 1/2 Vin, 2/3 Vin

or 1/3 Vin using the same components. So this circuit generates 3.3,

2.5 or 1.67 volts very efficiently.

There are some losses in the switches, so it helps if the input

voltage is just a bit higher than 5 volts or the outputs are just a

bit lower than 1/2, 2/3, or 1/3 Vin. But according to TIs data and my

tests, they work quite well at up to 100 mA or so, depending on the

exact input and output voltages.

You can get capacitors in 0805 packages up to 22 uF at 6.3 volts. So

this circuit can be very small. I have not been able to find an

inductive switcher this small or cheap.

So if efficiency is important, check out the TPS60500.

You can also power the Philips line of ARM controllers this way, but

the 1.8 volt output loses a bit of efficiency. Still it is twice as

efficient as an LDO. :) - --- In OKI-ARM-mcus@yahoogroups.com, redsp@y... wrote:
> I am designing a low power controller for a DSP board and I needed a

cost

> power circuit that would generate the 3.3 and 2.5 volt supplies the

> MCU needs. This also needed to be as efficient as possible, but

> and size are both issues. The circuit I ended up selecting is a

family

> switcher, but not an inductive switcher. TI makes the TPS60500

> of switched capacitor stepdown regulators that automatically

configure

> the topology to optimize efficiency. 3.3 volts is two thirds of 5

Vin

> volts and 2.5 is half of 5 volts. Turns out you can configure a

> switched capacitor converter to generate an output at 1/2 Vin, 2/3

> or 1/3 Vin using the same components. So this circuit generates

3.3,

> 2.5 or 1.67 volts very efficiently.

my

>

> There are some losses in the switches, so it helps if the input

> voltage is just a bit higher than 5 volts or the outputs are just a

> bit lower than 1/2, 2/3, or 1/3 Vin. But according to TIs data and

> tests, they work quite well at up to 100 mA or so, depending on the

Hi Rick, thanks for starting this group.

> exact input and output voltages.

>

If you were interfacing an SDRAM to the OKI part, what would you do

for the 3.3 volt supply to the SDRAM? I know they can draw quite a

bit of current during burst operations...

Rick - --- In OKI-ARM-mcus@yahoogroups.com, "skykotech" <rick@s...> wrote:
> Hi Rick, thanks for starting this group.

Yes, but the current scales with operating frequency. With a max freq

>

> If you were interfacing an SDRAM to the OKI part, what would you do

> for the 3.3 volt supply to the SDRAM? I know they can draw quite a

> bit of current during burst operations...

>

> Rick

of 60 MHz, they should be in the range of 60 mA or so which is not out

of reach of the TPS60500. The current draw is spikey with large

transients at the times of activity and much lower currents when not

being accessed. The frequency of operation is much higher than the

switching freq of the 60500. So using a larger cap on the Vdd rail

will help to supply current to the SDRAM when being operated.

The worse that will happen when the 100 to 150 mA boundary is crossed

is that the TPS60500 will act as an LDO and not give you the better

efficiency. But you need to consider the power being

dissipated in the little MSOP10 package.

Actually, we *will* have a single 16 bit wide SDRAM on the OKI part as

an option on our boards. It will not be needed for normal operations,

but I wanted to allow for people to run a fancier OS like uCLinux.

Our board will have to limit the SDRAM operation to 70C.