Electronics > Projects, Designs, and Technical Stuff

Topology and controller choice for DC-DC switcher

(1/2) > >>

I'm in the process of building a Lab power supply, liberally taking ideas from Daves design. One thing I'd like is a switching pre-regulator, to reduce power dissipation and to have some independence of the upstream supply.

So I'm looking to build a variable output DC-DC switcher, taking between 10-20 V input and providing from 2-30 volts output at 3A. The input would come from a random laptop supply, these are readily available at reasonable cost and put out around 16V.

After looking at tons of data sheets of controllers out there I see that there I have two choices:
- A single switch design with a transformer, a good example is the LTC3873. The drawback is that I need a transformer and these are not easily available from distributors.
- A buck-boost design with single inductor, for example using a LM5118. The drawback is that you need more components, like MOSFETs diodes, etc.

I don't really know on the advantages/drawbacks of either and could use some hints here.

One problem I have is that all manufacturers have nice parametric tables, but all presume that you need a fixed output voltage. So there is a lot of manual clicking to evaluate the individual controllers. I feasible I'd like to use a 'standard' controller, like the 7805 of the switching world. But there seems to be no 'well known' switching controllers or I missed the obvious choices.

I'd like a design which is easy enough to build and as low cost as possible. Any hints on where to start, good examples, sample designs etc ?

Ok since everyone else is silent, allow me.
Switchers are less easily applied to this kind of requirement. For starters the 2-30 volt output range implies a wide duty cycle swing in the switch, whatever the topology. While that in itself may not be a problem it does mean that many components, especially the magnetic ones will operate at widely varying points. Ensuring stability under all conditions of varyin output voltage, current and mutual phase would be a challenge. For example a switcher in current mode and continuous conduction will lose stability when D > 0.5 unless specific stabilization strategies are applied. The typical one is ramp compensation and that is built in to many controllers. That is only one easy example, there would be other issues needing attention.
I would try a transformer topology, preferably one that is insensitive to the switch operating modes. Full H-bridge comes to mind but it may be overkill for your requirements.

I like SEPIC Converters for this type of application.
  (10-20 V input and providing from 2-30 volts output at 3A) - 16 Nominal

16 to 4~32 (then on to regulation) is easily accomplished in a SEPIC design, but three amps is on the heavier side. You could try a 2-phase SEPIC controller, you'll need 2 MOSFETs, 2 Diodes, 4 inductors (or 2 coupled inductors).
Otherwise, a Buck/Boost design would work, but again you'll have to cascade these stages and will have to play with two feedback loops....

I understand that I'll be moving up to 100W through the switch in one extreme case (30V, 3A out) or <1 W in the other case (3V, couple of mA). On the other hand I don't really require it to go down that much under light load. If I have a too high output voltage under light loads I'll burn some more mW in the linear regulator afterwards. That's not a big deal.

@Kremmen: Here an example, straight from the linear website, with a transformer:
This looks reasonably clear to me. The biggest question is how to identify a suitable transformer and switching controller.

@PeteH: Here an example, straight from the linear website, SEPIC

Again, looks reasonably clear to me. I would calculate the necessary inductors from the highest amount of power I have to move (30V, 3Amps).

I don't know which of the two solutions are better for me, though.

General stuff: The LT3080/3083 datasheet has an example how to control a switcher up-front:
I plan to adapt the above schema to my situation, any reason this will fail ?

First link after a search "SEPIC vs. flyback"


Not sure if you care about any of the issues discussed, but they present some interesting points. In particular, the topologies are very similar, but offer different coupling mechanisms. Since the flyback is good for isolated topologies (where the SEPIC is not), we lose the big advantage over SEPIC converters when isolation is not of concern. The extra coupling capacitance provides efficiency gains....


[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod