Best approach for a wide-range (10:1) SMPS

[Spehro Pefhany]

Hi, all:-

Have a little project that needs to operate off of DC. Any suggestions
on the best approach for a wide range (eg. 6V to 60V operation,
transients to perhaps 100V) operation?

I only need a watt or so at 5VDC (or +/- 2.5VDC). I'd like to make it
idiot-proof, so jumpers and such like are out. Isolation optional, but
it might make things easier.

Current-mode flyback to get variable Ton?


Best regards,
Spehro Pefhany

 

[Spehro Pefhany]

The first thing that leaps to mind is something synchronously rectified
-- essentially a class D amp that you happen to hold mostly at one value
for a long time.

Hmmm.. why do I care about synchronous rectification? Diode loss isn't
much with a 5V/100mA output. Am I missing something?

But you'd have a lot of loss at 5V if you wanted it to be capable of 100.

50-60% efficiency would be just peachy. Unfortunately, it's more
important at ~12V in than 50V. A linear regulator followed by a 6:1
buck regulator would probably do.


Best regards,
Spehro Pefhany

 

[Spehro Pefhany]

Unless you say somethin gabout what those spikes are, in time duration,
it is difficult to say.
For very short spikes, a series resistor and transzorb could protect a serial switcher
input.

The 100V is what gets past the TVS.

For longer duration 'pulses' you need to design for the max voltage, and have
a fast enough response time, may work out on L.
So: how long is a spike?

Milliseconds.

Looks like a variable frequency current-mode flyback could cut it.


Best regards,
Spehro Pefhany

 

[MooseFET]

Hi, all:-

Have a little project that needs to operate off of DC. Any suggestions
on the best approach for a wide range (eg. 6V to 60V operation,
transients to perhaps 100V) operation?

I only need a watt or so at 5VDC (or +/- 2.5VDC). I'd like to make it
idiot-proof, so jumpers and such like are out. Isolation optional, but
it might make things easier.

Current-mode flyback to get variable Ton?  

I suggest a SEPIC design.

Current mode or a variable Ton based on the input voltage seems
needed. At low input voltages, the inductor current will almost
certainly go continuous. At high input voltages, you would need too
much inductance for that.

At the very top of the input voltage range the frequency will need to
decrease because the switching circuit will have a minimum on time.

Making the Ton based on a current sense and then basing the off time
on the output voltage should give a system that is stable over the
full input voltage range.

The SEPIC design can be done without having to get a transformer at
all. If you can get an inductor with two windings, off the shelf, I
suggest this sort of a thing:

L1 L2A C1 D1
----)))))-----)))))-----+---!!---+-->!----
! !
O -)))))--GND
/ L2B
!
GND

L1 is a small inductor. L2 does most of the work. If the L2A and L2B
are tightly coupled, C1 can be a modest sized capacitor. D1 needs to
be big and fast.




Another way to go may be to build a simple booster that takes the 6V
input up to about 20V. When the voltage is above 20V, the transistor
in the boost would stay off and the "20V" supply would follow the
input. This would make the bucker design a lot easier.

 

[Klaus Kragelund]

I suggest a SEPIC design.

Current mode or a variable Ton based on the input voltage seems
needed.  At low input voltages, the inductor current will almost
certainly go continuous.  At high input voltages, you would need too
much inductance for that.

At the very top of the input voltage range the frequency will need to
decrease because the switching circuit will have a minimum on time.

Making the Ton based on a current sense and then basing the off time
on the output voltage should give a system that is stable over the
full input voltage range.

The SEPIC design can be done without having to get a transformer at
all.  If you can get an inductor with two windings, off the shelf, I
suggest this sort of a thing:

      L1         L2A         C1     D1
 ----)))))-----)))))-----+---!!---+-->!----
                         !        !
                         O         -)))))--GND
                          /          L2B
                         !
                        GND

L1 is a small inductor.  L2 does most of the work.  If the L2A and L2B
are tightly coupled, C1 can be a modest sized capacitor.  D1 needs to
be big and fast.

Another way to go may be to build a simple booster that takes the 6V
input up to about 20V.  When the voltage is above 20V, the transistor
in the boost would stay off and the "20V" supply would follow the
input.  This would make the bucker design a lot easier.

You don't mention what your output voltage quality needs to be (ripple
voltage etc?).

If it does not need to be short circuit safe, I would use an ultra
simple hysteretic buck (no isolation, bang-bang mode). If you need
perfect ripple, then set the buck some hundreds of millivolts above
the desired output and add a LDO to the output.

Regards

Klaus

 

[Spehro Pefhany]

Is this for automotive (or similar) use?

No, not for use in a vehicle.


Best regards,
Spehro Pefhany