using a synchronous rectifier in a boost circuit

I wonder what is this for.

Make a transformer buck from 12 to 1.5V, then connect it in series to the input 12V?

VLV

See if this one fits your needs, comes with LTSpice simulator model and the whole nine yards:

Hi,

I am using an off the shelf ATX power factor corrected 750Watt powersupply as the 12VDC input, and then boosting this to ~13.5VDC for charging lead acid batteries at 600watts+. This is part of my "ATX power tower" project, a cheap home energy system in a PC case :)

Its going to be open source, and designed in eagle cad. Here are some of the specs:

grid input (charges batteries and/or powers load)

internal auto transfer relays to auto-select power sources

solar/wind/hydro compatible MPPT input (charges batteries)

modular inverter interface (turns on when grid is down)

optional battery desulphator outputs

fully microcontroller controlled to set charging voltages and monitor currents etc

I am putting this circuitry on an ATX rev 2.2 compatible PCB so that it can be installed into a surplus ATX PC case.

cheers, Jamie

Thanks Joerg, that IC looks great! I have a sim of it running with

cheers, Jamie

Two things are wrong here. Firstly, the 'powersupply' that is your input has low ripple and excellent regulation, both of which are not required for charging lead-acid batteries. Second, your 'boost circuit' just adds a second layer, altering the 12VDC power setpoint would have the same effect. There's already a heatsinked rectifier or two for the

I'd either get a 'brute force' 13.5V power supply, or a lead-acid charger instead of doing this kind of modification. I hope, too, this is outdoors? Six hundred watts of heat will likely make a sulphuric mess if your charger stays connected too long. Commercial chargers do sensing and foldback limiting to avoid that.

That is more than 42 A.

Are you sure that the ATX power supply will deliver that amount of current _at_ 12 V ?

Paul makes a good point.

It's usually the 5V rail thats good fo 30 to 50A the 12V is usually only good for 12A to maybe 20A. At least the ones I've seen.

Oh and I wouldnt count on the supply to be able to delivery it for extended periods of time.

Or do it at all with no load on the +5V

Play with the Rdson of the sync rectifier FET versus PWM frequency versus gate drive capacitance, might be able to push the dissipation down some more.

Thing is, you'll have to obtain models for more modern FETs. The ones in the LTSpice selection table are a bit long in the tooth. Or just whip up a prototype.

Another thing that helps is to zoom in and look at when exactly the body diode comes on and when the FET channel starts conducting. The FET channel conduction window must be inside the diode conduction window, fully. Now two things happen that don't go in your favor:

a. The body diode behaves like a Si-diode, meaning a corresponding voltage drop. This hits when the channel hasn't come on yet or has already been turned off.

b. The body diode is kind of sluggish.

Not much you can do about b but you can do something about a: Connect a big Schottky in parallel if the BOM budget can stomach it.

It'll cost efficiency but I agree. The 12V rail diodes in most PC supplies are sort of skimpy. Probably because 12V only has to deliver peaks and once the hard disks or CD have spun up consumption goes back down.

Of course Jamie could install bigger diodes on the 12V side but then he might as well swing the regulation loop over to 12V, disable everything else and set the new voltage divider resistor so that 13.5V comes out.

Why not just tweak the ATX switcher directly to get to "13.5VDC"?

Which, BTW, may not be adequate to charge a cool/cold lead acid battery. ...Jim Thompson

There's certainly a voltage/temperature relationship, but I'll be damned if I can find my handy battery book!

Hacking the supply sounds good, unless the intent is to have something that is wholly external to the thing, to allow you to use any old ATX supply of sufficient drive.

It seems that if you're going to go to the trouble to misuse an ATX power supply by building a 600W switcher to run off of an already-regulated output, you could without too much more time and expense just build an off-line switcher.

Only if he could parallel more windings, I doubt the cooper could take two to three times the rms current they were originally designed for without excessive losses and causing thermal runaway on the transformer.

At this point it would probably be less trouble to buy an off the shelf solution or build your own llc Half-bridge converter if your feeling ambitious. ;-)

------At 600 watts I'm sure you are talking about bulk charging.

--------------------------------------- Posted through

Hi,

The ATX supply I am working with is rated at 700watts and "80 Plus" efficiency, model OCZ700MXSP. It has dual rail 12V outputs (25Amps max on each rail, 552Watt max load rated on 12V) and the 5V rail is rated at

cheers, Jamie

Make sure your switchers are well current-limited if you do that, and don't motorboat against each other.

Probably not if you swing the regulator feedback from 5V over to the 12V rail.

...Jim Thompson

PIC processor will do what you need, as will the Arduino. I have no experience with the PIC but I'm learning the Arduino.

A universal battery charger using an Atmega8. That's pin compatible with the Arduino ATmega328.

The code is copy and pastable into the Arduino compiler interface.

An Arduino PWM driver

A model railway controller, showing one pin driving a transistor:

These fine people have many fine links:

The Home page:

The Arduino USB prototyping board is around 30 dollars US, with a microcontroller chip included. When you need more chips, they run anywhere from 3 to 6 dollars each.

Youtube.com has a ton of Arduino related videos.

I would first try as Joerg suggests switching feedback to the 12V.If it does work you saved a lot of time and money. See his post below.

If you want universal usages its pretty much a given PSU's once they start getting over 500W has some type of PFC. Not just because of THD regulations.Residential lines are only good for 15 to 20A.

The typical SMPS with no PFC has a PF OF .55 TO MAYBE .65. At .65 the input cap is going to be small--> large 2f line ripple---> low DC input means high conduction losses.