Multiple power supplies from 12V dc source

[Fred]

I have to provide a power module taking a 12V source to produce the
following voltages and currents:

5V +/- 10% @ 300mA
3.3V +/- 10% @ 550mA
2.5V +/- 5% @ 100mA
1.8V +/- 10% @ 50mA
1.2V +/- 5% @ 100mA

This is for medium volume and the product is cost sensitive.

I'm tempted to use two switching regulators for the +5V and +3.3V supplies
and regulate the rest from the 3.3V supply. The tolerances are quite tight
so it's difficult to use 0.7 - 1.0V diode drops to produce the 2.5V or the
1.8V supplies.

Any help would be appreciated for an efficient cheap solution.

 

[Roger Gt]

: I have to provide a power module taking a 12V source to produce
the
: following voltages and currents:
:
: 5V +/- 10% @ 300mA
: 3.3V +/- 10% @ 550mA
: 2.5V +/- 5% @ 100mA
: 1.8V +/- 10% @ 50mA
: 1.2V +/- 5% @ 100mA
:
: This is for medium volume and the product
: is cost sensitive.
:
: I'm tempted to use two switching regulators
: for the +5V and +3.3V supplies and regulate
: the rest from the 3.3V supply.
: The tolerances are quite tight so it's
: difficult to use 0.7 - 1.0V diode drops
: to produce the 2.5V or the 1.8V supplies.
:
: Any help would be appreciated for an efficient cheap solution.
:

Your plan to use switchers for the 5 and 3.3 Volt supplies is
good, then
Select LDO regulators for the 2.5, 1.8, and 1.2 Volt outputs.

I did this for a board recently but the currents were much worse,
with the 1.2 V requiring 900 mA. So I had a lot of power lost,
but the scheme is good.

 

[Ken Smith]

I have to provide a power module taking a 12V source to produce the
following voltages and currents:

5V +/- 10% @ 300mA

I'd look at making this a parasite supply on the 3.3V regulator producing
about 6V and then a LDO making it into 5V

(12-5)*0.3 = 2.1W so a linear would need a heat sink. More electronics is
often cheaper than adding a heat sink.

3.3V +/- 10% @ 550mA

This really needs to be a switcher.

2.5V +/- 5% @ 100mA
1.8V +/- 10% @ 50mA
1.2V +/- 5% @ 100mA

I'd look for a multiple LDO.

 

[Fred]

: I have to provide a power module taking a 12V source to produce
the
: following voltages and currents:
:
: 5V +/- 10% @ 300mA
: 3.3V +/- 10% @ 550mA
: 2.5V +/- 5% @ 100mA
: 1.8V +/- 10% @ 50mA
: 1.2V +/- 5% @ 100mA
:
: This is for medium volume and the product
: is cost sensitive.
:
: I'm tempted to use two switching regulators
: for the +5V and +3.3V supplies and regulate
: the rest from the 3.3V supply.
: The tolerances are quite tight so it's
: difficult to use 0.7 - 1.0V diode drops
: to produce the 2.5V or the 1.8V supplies.
:
: Any help would be appreciated for an efficient cheap solution.
:

Your plan to use switchers for the 5 and 3.3 Volt supplies is
good, then
Select LDO regulators for the 2.5, 1.8, and 1.2 Volt outputs.

I did this for a board recently but the currents were much worse,
with the 1.2 V requiring 900 mA. So I had a lot of power lost,
but the scheme is good.

Yes that was the scheme I was thinking off. Just that it's a lot of
components.

I also need some sort of supply/brown out protection to reset the CPU. The
12V supply seems the obvious determining factor.

 

[Fred]

I'd look at making this a parasite supply on the 3.3V regulator producing
about 6V and then a LDO making it into 5V

I've tried this on an earlier design which worked apart from introducing a
lot of noise on the +3.3V rail. I assume this was because current was no
longer nominally constant, as in a straighforward step down voltage
converter, where chunks of current were now taken out by another winding.

(12-5)*0.3 = 2.1W so a linear would need a heat sink. More electronics is
often cheaper than adding a heat sink.

I am not very keen on this level of heat dissipation!

This really needs to be a switcher.

I'd look for a multiple LDO.

I haven't seen any. Where can I get one? Do they really work out cheaper
than separates?

Many thanks.

 

[Ken Smith]

I've tried this on an earlier design which worked apart from introducing a
lot of noise on the +3.3V rail. I assume this was because current was no
longer nominally constant, as in a straighforward step down voltage
converter, where chunks of current were now taken out by another winding.

It is best to make the parasite winding the lower powered one and to put a
significant amount of bypass on it. There is a funny non-linear step
responce issue in supplies like this. Bepending on the sizes of the
filter caps, The 3.3V supply can't swing up as fast as it swings down or
down as fast as up. Generally the servo loop of the regulator has to be
designed so that it can have its gain cross over anywhere in a widish span
of frequencies.

Intentionally making the coupling to the parasite winding a bit sloppy can
help loop stablity.

The variations in current on the 5V supply will effect the 3.3V supply if
you are not careful. Since the 5V supplies regulation doesn't need to be
so tight, you may want to slip a resistor somewhere in that part of the
design to force any current spikes to be taken from the capacitors and not
from the windings.

I haven't seen any. Where can I get one? Do they really work out cheaper
than separates?

I think Linear has one. They cost more but the stuffing cost could pay it
back.