DC:DC converter - Any rough power vs volume calcs?

[max_torque]

So, without diving into the proper calculations, are there any quick "ready reckoners" that estimate the physical volume of a DC:DC converter based on it's peak power?

I'm looking at possibly needing a 300W (output) and was just wondering how much space that would take up on a pcb/ in an enclosure?  (obviously, the size of inductor / capacitors depends upon the fundamental frequency)

(nominally 12Vdc input, 0-30V out, 20A peak, 300W max out)

[johansen]

if you want to throw money at it, i found a dc-dc converter manufacturer in some magazine advertizing up to 1253 watts per cubic inch.
A friend of mine found some 0.9 volt, 32 amp power supplies that are about 2 cubic centimeters in volume, they are 24$ each if you buy 100 of them.
these: http://www.power-one.com/sites/power-one.com/files/documents/power/datasheet/qme48t35120_finalds_rev-02c_051811.pdf are 420 watts and you might be able to buy them for 50$ each.

Not many companies make dc-dc converters with an output spec of 0-30 volts like that. it certainly is possible, it just means the energy stored in the inductors (or leaky transformers) will need to be significantly more and the cost will reflect that.

[rob77]

(nominally 12Vdc input, 0-30V out, 20A peak, 300W max out)

those specs.. hmmmm... that will be a challenge

[mikeselectricstuff]

So, without diving into the proper calculations, are there any quick "ready reckoners" that estimate the physical volume of a DC:DC converter based on it's peak power?

Yes - look at the size of a few commercial modules of similar spec

[jmole]

One reason I'm really excited about this: https://www.kickstarter.com/projects/215201435/dart-the-worlds-smallest-laptop-adapter

Yeah yeah, it's AC/DC, but it's really the same principle when it comes down to it.

I'm sure VHF power is old-hat (to some extent) to those in the field, but I'm really excited that we're seeing it come to market. From an energy-conversion density standpoint, it will blow the prior solutions out of the water.

[NiHaoMike]

Is isolation needed? If not, the design can be quite a bit smaller.

[rob77]

guys.... seriously ...

you can't assume that a 420W buck converter with fixed output is a good reference for calculating the size/volume needed for a converter with the specs required by the OP.

12V input and 0-30V output  @ 20Amps is quite hard to achieve - it will be not as simple design as a buck converter.

[peter.mitchell]

guys.... seriously ...

you can't assume that a 420W buck converter with fixed output is a good reference for calculating the size/volume needed for a converter with the specs required by the OP.

12V input and 0-30V output  @ 20Amps is quite hard to achieve - it will be not as simple design as a buck converter.

LT3780, modern fets and good quality passives, you have it in around 2 cubic inches.

[rob77]

guys.... seriously ...

you can't assume that a 420W buck converter with fixed output is a good reference for calculating the size/volume needed for a converter with the specs required by the OP.

12V input and 0-30V output  @ 20Amps is quite hard to achieve - it will be not as simple design as a buck converter.

LT3780, modern fets and good quality passives, you have it in around 2 cubic inches.

no way.... that LT3780 is a buck-boost with seamless transition - that fits the 12V in and 0-30V out , but doesn't fit the 20A out.
you can use whatever mosfets you want.. it won't work for 12V in and 30V out @ 20A - you have to build a polyphase boost converter for high current boost conversion. there is no way to do it with a single inductor while keeping the ripple on the output low enough to be usable.

[max_torque]

Thanks for the replies!  I understand the power handling figures are quite hard, but note that it is 20A peak, so that as the output voltage rises i shall derate the current required to stay under the power figure!

No isolation is required, as it will just be driving a brushed dc motor, so the load is pretty "static" and the system will not require a lot of transient control authority.  An easier option would be to "rewind" the motor, but unfortunately this part i cannot change, so i need a voltage boost instead.  Operation time when running in this voltage boosted region will be shortish (<20sec) and intermitent / non repetative.

I guess a critical early decision is what maximum drive frequency to use, as this is critical for the inductor sizing and output ripple (output ripple can be quite high, as the dc motor has significant inductance and mechanical inertia).

Realistically, what sort of drive frequencies should i be aiming at for this sort of project without pushing the boundaries too hard (in term of dI/dT / gate drive requirements etc)??

[mikeselectricstuff]

Could you avoid the 0-30v range requirement using PWM of a fixed voltage, or at least a 12-30v range that avoids the need to go above and below the input?

[NiHaoMike]

What about a push pull boost converter followed by a buck converter? For running a brush DC motor, you won't need any inductors for the buck converter and only a small cap (to catch leakage inductance transients) for the push pull boost converter.

The push pull boost would run on the order of 40kHz while the buck can run as low as a few kHz. Note that despite its name, a push pull boost actually operates as a nonisolated forward converter.

[johansen]

guys.... seriously ...

you can't assume that a 420W buck converter with fixed output is a good reference for calculating the size/volume needed for a converter with the specs required by the OP.

12V input and 0-30V output  @ 20Amps is quite hard to achieve - it will be not as simple design as a buck converter.

what i posted is not a buck converter, it probably runs at 500Khz though.

the biggest problem with 0-30 volt output for a dc-dc converter is the output ripple.
a synchronous rectified forward converter is probably the way to go.

[rob77]

guys.... seriously ...

you can't assume that a 420W buck converter with fixed output is a good reference for calculating the size/volume needed for a converter with the specs required by the OP.

12V input and 0-30V output  @ 20Amps is quite hard to achieve - it will be not as simple design as a buck converter.

what i posted is not a buck converter, it probably runs at 500Khz though.

the biggest problem with 0-30 volt output for a dc-dc converter is the output ripple.
a synchronous rectified forward converter is probably the way to go.

input 36-75V , output 12V => buck converter

and yes the 1:3 step up @ 20Amps output is the problem you can't do that with single inductor and low ripple regardless of the frequency.

[max_torque]

Thanks for the ideas!  It looks like this is probably a non-starter in terms of size/complexity etc  In fact, i think i will just take the opportunity to move to a brushless motor and use  FOC/Field Weakening to get the speed range.  Afterall, i have all that inductance sitting in the motor, so it makes sense to use it for something productive ;-)

[johansen]

input 36-75V , output 12V => buck converter
and yes the 1:3 step up @ 20Amps output is the problem you can't do that with single inductor and low ripple regardless of the frequency.

I know some of us have thick skulls, but explain how its spec'd for 1500 volts isolation between the input and the output.

anyhow, as far as the 1:3 step up, Sepic and Cuk converters have serious issues running near zero volts output due to control issues, though it would be possible to use either of those topologies if no isolation is required, however, the issue is synchronous rectification. i don't know of any off the shelf control chips that will let you do that.

I would rather stick with a forward converter, as it offers operation across that zero bound better and can be controlled as a buck converter, or a boost converter if you want to use a flyback topology.

alternatively, you can use a boost buck topology, rig up a 3 phase boost converter and stick a buck converter on the other end of it. such a system would require about as much silicon as a flyback, but what the hell, it gets the job done

[rob77]

I know some of us have thick skulls, but explain how its spec'd for 1500 volts isolation between the input and the output.

sorry i missed the fact it's isolated  my apologies sir