converting 12V to 24V

[Simon]

Vehicle does not mean car, it could be a bulldozer, an off road building site truck, a combine harvester, a tractor, nothing that we deal with at work is a car unless it's some stupid little thing.

Manufacturers make what customers require. few customers require this power at 12V for the same good reasons fan manufacturers don't like making very powerful 12v fans

Who's chosen any components yet ? whole idea of the thread.

We live and work in a competitive market, you just don't turn round and say that it can't be done or quote so much money it's stupid, unless you want to be out of business.

I'd work to find the best solution.

[DanielS]

You cannot make a switching converter (boost or push-pull) regulate in a CAR as the output power is proportional to the battery current that's always less that what is was when you attached the converter.

A properly designed switching regulator covers both input voltage variations and output load swings. Line voltage for universal power adapters can regulate across the 90-250VAC input range while maintaining output regulation usually better than 5%. All they need to do is sense output voltage and adjust duty cycle accordingly.

If your step-up converter does not regulate, it is just a dumb inverter; a crude switcher operating in open-loop.

[Precipice]

We live and work in a competitive market, you just don't turn round and say that it can't be done or quote so much money it's stupid, unless you want to be out of business.

Attempting to deliver demented projects is also a good way to go out of business.
Learning to steer customers towards sane solutions is an important skill, especially if you can do it with tact, and then bid for the new solution.
Often you'll have to tell the customer, then wait a while, after which the customer will, all on their own, come up with a new requirement which sounds suspiciously familiar. It's a bit of a surprise the first few times, but just go with it.

[David Hess]

I agree with the high frequency inverter suggestion.  This gets rid of the requirement for a massive inductor that a switching regulator would require.

Flux balancing may be an issue however so I would use some variation of a saturable core inverter.

[Simon]

We live and work in a competitive market, you just don't turn round and say that it can't be done or quote so much money it's stupid, unless you want to be out of business.

Attempting to deliver demented projects is also a good way to go out of business.
Learning to steer customers towards sane solutions is an important skill, especially if you can do it with tact, and then bid for the new solution.
Often you'll have to tell the customer, then wait a while, after which the customer will, all on their own, come up with a new requirement which sounds suspiciously familiar. It's a bit of a surprise the first few times, but just go with it.

Have you ever heard of people who won't be told ? clearly the solution is that given the size of the vehicle and it's subsystems it should be 24V in the first place, They are obviously not that clever. I have already been recommending to find a fan that is 12V and have already told my colleague that boosting nearly 1KW is not simple or cheap. However I'm still interested. The market is flooded with similar converters so it's not exactly impossible........

[DanielS]

I agree with the high frequency inverter suggestion.  This gets rid of the requirement for a massive inductor that a switching regulator would require.

That depends on the frequency. At 200kHz, a 40A output inductor would be fairly small. This also depends on maximum permissible output current ripple: if you do not care about output current and voltage ripple, you may even be able to forgo the output filter altogether aside from a small one for the PWM feedback loop if your load can safely withstand the transformer's raw secondary output voltage.

Computer power supply manufacturers are cramming 12V rails with over 120A output capacity in ATX-sized power supplies (150 x 86 x 180 mm) and this includes the surge/EMI filter, AC-DC and active PFC front-end, galvanic isolation, the 5VSB supply, three other secondary output voltages and an output inductor large enough to handle over 120A with 1% regulation and less than 50mV ripple.

Ripping out all the unnecessary stuff for a DC-DC converter would likely free up enough space to implement a 1kW unit as a pair of 500W modules to make primary-side currents more manageable.

[diyaudio]

A properly designed switching regulator covers both input voltage variations and output load swings. Line voltage for universal power adapters can regulate across the 90-250VAC input range while maintaining output regulation usually better than 5%. All they need to do is sense output voltage and adjust duty cycle accordingly.

NOT when the source is a battery, you cannot make more energy OR maintain the same energy than what you started out with, that's my point, regulation is a battery thing the current source, explain how you keep 24 volts stable, if the current is depleting at a rate proportional to a heavy load. that's my point. not in relation to 90-250VAC input source but to a battery.

Another misconception is that a these converters I spoke of isn't closed loop and are dumb (far from it) , how do you define close loop? their is a few compensation networks with a pole/zero at certain points that improve step response turn on,  no outer loop exists from the output, that dynamically programs the controller to push the PWM to say another 5% duty cycle width to regulate. wont happen.



 

[DanielS]

NOT when the source is a battery, you cannot make more energy OR maintain the same energy than what you started out with, that's my point, regulation is a battery thing the current source, explain how you keep 24 volts stable, if the current is depleting at a rate proportional to a heavy load. that's my point. not in relation to 90-250VAC input source but to a battery.

The source being a battery makes no difference: tons of regulated switching voltage regulators operate off batteries in cellphones, tablets, laptops, calculators, remote controls, cameras, UPS, cars, LED flashlights, etc.

Not sure where you take the "making more energy" bit from. If you want to drive a 40A 24V load from a 12V battery, the converter will need to draw 80A not counting converter losses regardless of whether it is a PWM, self-excited inverter or whatever else. No magically created energy there: V*A in = V*A out + losses.

The way a DC-DC switching regulator compensates for battery (or whatever its input source might be) droop is exactly the same way an AC-DC switching regulator compensates for the 10-30V ripple on its AC input filter capacitor between input peaks: it modulates its duty cycle to compensate for changes in input voltage.

If you want to design a double-forward switching regulator to step 10-14V battery input to 24V output, you need a transformer ratio of about 1:3 assuming you lose only 1V in both input and output components including wiring resistance. When the battery is fresh-off-the-charger at 14V surface charge, the PWM regulator would operate at about 32% duty-cycle per transformer half (64% for the whole transformer) and when the battery reaches 10V, the PWM should be at about 47% per half.

[ConKbot]

Gave the thread a quick skim may have missed the suggestion if someone put it out there already.  Since this doesnt need to super well regulated, set up a push-pull converter (or full h-bridge). But, have it be a 1:1 (well, 1: 1:1 on the push-pull) so the converter spits out an isolated 12V.  Put your isolated 12v in series with the existing +12v rail and your converter only needs to handle half the wattage as if it was outputting 24V.

Could be as cheap and nasty as a transformer, oscillator of choice running at a fixed duty cycle, some hard-switched mosfets, with a good gate driver to try and switch fast, and some beefy diodes for rectifying.   For a bit more refinement, you could do a current fed push-pull whichwould gain you regulation, and soft-switching of the FETs. 

[mzzj]

Gave the thread a quick skim may have missed the suggestion if someone put it out there already.  Since this doesnt need to super well regulated, set up a push-pull converter (or full h-bridge). But, have it be a 1:1 (well, 1: 1:1 on the push-pull) so the converter spits out an isolated 12V.  Put your isolated 12v in series with the existing +12v rail and your converter only needs to handle half the wattage as if it was outputting 24V.

Could be as cheap and nasty as a transformer, oscillator of choice running at a fixed duty cycle, some hard-switched mosfets, with a good gate driver to try and switch fast, and some beefy diodes for rectifying.   For a bit more refinement, you could do a current fed push-pull whichwould gain you regulation, and soft-switching of the FETs.

Not bad idea, i have used similar scheme with other power supplies. Output protection needs a little extra attention for short-circuits but i guess beefy diode and fuse can handle that.

Your cheap-and-nasty has a possible problem of flux-walking, so i would recommend something a little bit more evolved with a pulse by pulse current limiting plus the necessary extra hardware.

You can skip the filtering caps and output coil almost entirely in a push-pull converter if you run it unregulated and close to 100% duty cycle. I bet the required power supply BOM is less than 50 euros, but then you have spends thousands on design and build man many units.

So i guess it alldepend does Simon Corporation need 5 or 50000 units.

[Simon]

Gave the thread a quick skim may have missed the suggestion if someone put it out there already.  Since this doesnt need to super well regulated, set up a push-pull converter (or full h-bridge). But, have it be a 1:1 (well, 1: 1:1 on the push-pull) so the converter spits out an isolated 12V.  Put your isolated 12v in series with the existing +12v rail and your converter only needs to handle half the wattage as if it was outputting 24V.

Could be as cheap and nasty as a transformer, oscillator of choice running at a fixed duty cycle, some hard-switched mosfets, with a good gate driver to try and switch fast, and some beefy diodes for rectifying.   For a bit more refinement, you could do a current fed push-pull whichwould gain you regulation, and soft-switching of the FETs.

Not a bad idea, it won't entirely limit spike draws as I'll still need to draw in peaks of 80A but yes that would cut out a few parts and reduce costs an overall consumption

[Simon]

I suppose if an inductor does not cause too many losses a +12 / -12V DC/DC converter will do just fine as in series with +12V will give me 24V total. Although that might screw with the control system as it will put the fans ground at -12V although they could use an opto isolator to pass PWM speed signals.

[DanielS]

Gave the thread a quick skim may have missed the suggestion if someone put it out there already.  Since this doesnt need to super well regulated, set up a push-pull converter (or full h-bridge). But, have it be a 1:1 (well, 1: 1:1 on the push-pull) so the converter spits out an isolated 12V.  Put your isolated 12v in series with the existing +12v rail and your converter only needs to handle half the wattage as if it was outputting 24V.

Get out of here, that makes too much sense 

Actually, your scheme can also regulate total output voltage: use a 1:2 transformer but instead of regulating its output as isolated, sense the whole battery + PWM output voltage so the PWM can cover the battery's droop with a ~650W design instead of 1kW. This makes things a little simpler since you do not need to isolate a floating output.

[Simon]

I'm actually thinking that that is exactly what most of the commercial units do, one that I saw only had 3 terminals so they are actually handling 1/2 the rated power.

[Zero999]

I didn't think voltage regulation was important, since this is driving a 24V fan which will be designed to be run off an unregulated supply. As long as line regulation, given the same input voltage, isn't too bad I don't see the problem.

The idea of using -12V is good but you could also consider a 12V to 12V converter with the secondary connected in series with the input, giving 24VDC.

A self-oscillating inverter is probably the simplest to build and will give decent efficiency.

See the links below:
http://www.seekic.com/circuit_diagram/A-D_D-A_Converter_Circuit/MOSFET_resonance_type_DC_DC_converter_circuit.html
http://www.smps.us/inverters.html
https://www.circuitlab.com/circuit/6f656p/mosfet-royer-like-oscillator-01/
http://www.circuitsonline.net/forum/view/110947
http://www.sciencezero.org/vasil/romtcb/groyer.html

The Schottky doides on the secondary can probably be replaced with MOSFETs to give synchronous rectification, if gate drive is provided by another winding.

[Simon]

Well fortunately they have found a 12V fan but the idea was interesting. I was trying to stay away from a transformer unless they are readily available off the shelf. Seems that 4+ oz copper tracks both sides and possibly metal PCB base to carry away heat is the way to make it work on a PCB. But the PCB will then become a major component cost most likely. Apparently as much as 12oz is available but that is very special and costly.

[Simon]

This is the most expensive SMPS transformer on digikey: http://www.digikey.co.uk/product-detail/en/750341145/732-4454-6-ND/3831276
250W 200-700KHz and I still haven't worked out in what ratio they are offering it, according to the drawing there are options.

[DanielS]

Well fortunately they have found a 12V fan but the idea was interesting. I was trying to stay away from a transformer unless they are readily available off the shelf.

High-power transformers are rarely available off-the-shelf since they usually have tons of application-specific requirements like secondary rails, auxiliary supplies, bias windings, duty cycle, operating frequency, etc.

DigiKey's transformer selection is more typical of stuff you would need to build the PWM regulator manufacturers' reference designs and closely related variations.

For a 1kVA transformer, your best bet would probably be a 1-1.5kVA UPS and you would need to rip the high-voltage secondary winding out to put your new primary in.

BTW, you do not need 4oz copper if you wave-solder the board without solder mask on power traces and keep traces short or make it so all the high-current parts are on chunky copper pours instead of traces.

[Simon]

BTW, you do not need 4oz copper if you wave-solder the board without solder mask on power traces and keep traces short or make it so all the high-current parts are on chunky copper pours instead of traces.

We are talking about 40-80A and solder has 10 times the resistance of copper, yes it can be a great help to squeeze more amps out on marginal designs, I spoke to a PCB manufacturer and he was talking dual traces of 4oz (140um) with 35um plating at 15mm wide each side (30mm total) allowing for a 40C rise, there wasn't any plan to wave solder and it sounds more like an inconsistent bodge that may be difficult to do repeatedly and reliably in manufacturing.

[DanielS]

We are talking about 40-80A and solder has 10 times the resistance of copper, yes it can be a great help to squeeze more amps out on marginal designs

Few amps?

Put a 0.5mm solder coat on top of those 15mm traces and you have just eliminated the need to add a second 4oz layer.

Bodge or not, practically all computer power supplies and somewhat cost-conscious power products do this.

[Simon]

Well i don't know if these would end up wave soldered, manually adding solder loading will just cost labor. The PCB's can also be heavily plated, 4oz copper is 140um usually you have another 35um of plating and more can be added on request, the PCB has to be manufactured and plated anyway.

[mzzj]

How about http://homemadecircuitsandschematics.blogspot.fi/2013/03/simple-dc-to-dc-high-current-voltage.html

Capacitor ripple current requirements are going to make you cry but no need for magnetics.
And both output terminals are swinging up and down.

[Simon]

that sort of thing was my original plan, but diodes need replacing with switched mosfets

[DanielS]

Then again, for such a relatively simple circuit as a 600W 12V DC-DC booster, it should not be too difficult to put together a layout where critical pins in high-current nodes can be bunched together and joined with solder blobs on copper islands instead of relying on PCB traces. The input filter might be a problem though... for a double-forward converter, the ripple would be in the neighborhood of 45A.

For the charge-pump idea, you might still want to use inductors in line with those diodes to take some of those ripples off and soften turn-on edges... with about 40A RMS of ripple, each of the two capacitor banks would need about a dozen capacitors.

[mzzj]

that sort of thing was my original plan, but diodes need replacing with switched mosfets

If efficiency is not your main concern you get much easier results with the diode version.

I was playing with the synchronous rectifier version on ltspice and getting >95% efficiency looked reasonably easy.  Btw, the Synchronous rectifier version works both ways so if you swap the load and supply positions it will be voltage "halfer"