High current power supply

[cyberfish]

I'm currently looking for a high current low voltage supply - ~7V, 30A. 120V input.

Doesn't need to be adjustable, and voltage regulation is not terribly important (6V-8V is fine). I'm going to use it to test my quadrocopter project (replacing a LiPo battery), hence the high current requirement.

Much to my surprise, I can't seem to find an off the shelf one.

So I tried to build one myself. However, I don't have any experience with this kind of things, so would love some advice.

First of all, is it better to use transformer then rectify (unregulated), or rectify then buck converter? What are the trade-offs? Transformer would provide isolation from the mains which is nice, but high current transformers seem to be VERY expensive (eg. http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=237-1089-ND $20 for 8.8A, and I need 4 in parallel!). On the other hand, for an SMPS I would only need a beefy inductor, diode, and transistor, all of which I imagine wouldn't be nearly as expensive. How do people generally do this kind of things?

Thanks

[Psi]

One option could be to buy a 700W computer powersupply for ~ NZ$130  

Get a good brand one that is rated over like 50A at 12v, (ya want a good brand so the amp rating is not fake or peak).
Some power supplies come with dual rail 12v, i don't know if you can connect both rails togeter or not, this might be something you'd need to check. Or just make sure to get a psu with single 40-50A 12v rail.


Take it apart and locate the feedback opto for the 12V rail and change the voltage divider to something that gives 8V instead of 12V
(if you only modded one 12v rail to 8v be careful not to connect it to the other 12v rail if its a dual rail unit)

If you google you can probably find lots of plans to mod computer power supplies to output different voltages, so it shouldn't be that hard.

NOTE: If you're thinking of connecting between the psu's +5 and +12 rails to get your +7 you *can't* do that unfortunately. (not unless you have a separate load on the 5V rail drawing more current than your 7V current, otherwise your 7V load would try to pull the 5V rail above 5V and will damage the psu)

[Psi]

Another option maybe a variac +  mains to 6VAC transformer. It would have to be a massive transformer though, to handle 30A

[oPossum]

7.5 volt (6.8 to 8.2 via internal adjustment), 27 amp switching power supply

Spec sheet

Available from Newark

[Psi]

that looks like a better option

[Reuben]

Get a car battery charger with the 6v option for motorcycle batteries.

[alexwhittemore]

Actually, if you're using it to test a design based around generally standard RC stuff like off-the-shelf ESCs, voltage regulation actually does matter, as you might get out of the range of the ESCs thinking they're attached to 2 cells.

We had this same problem. For a while, we were actually playing with motors that could spike 100A. The solution I came up with was half and half of the suggestions above. Since we wanted to be able to play with 1, 2, and 3-cell configurations, I whipped up a beefy-ass linear regulator: think op-amp with 6 50-or-so amp rated mosfets in a voltage follower configuration. The mosfets were mounted to old pentium 2 heatsinks, 3 per.

That monstrosity was supplied by a Corsair 1200-w PSU with a single, 100A 12v rail. I opened up the PSU and the only modification I made was to squeak a screw terminal block onto it. You could still fit it in a computer for normal use, even. This has served us pretty well. You can probably get away with a few fewer mosfets and a smaller ATX supply, but in general that's the way to go.

[Psi]

This reminds me of another option.

If the duty cycle of your testing is less than 100% (ie, maybe 20sec of 30A then a few minutes of 0A)
Then you could use a lower current powersupply and some super capacitors.

I recently got some 2600 Farad 2.5v boostcaps from The Electronic Goldmine
They really are awesome and give a whole new meaning to the word capacitor.

If you had say, a 10A powersupply, charging three of these capacitors in series (866Farad @ 7.5V) then you could draw 30A out of the capacitor bank for 45 seconds and the voltage would only have dropped to 6V (still within your voltage range).

They stock lots of different capacitor configurations but here is the 7.5V series one
http://www.goldmine-elec-products.com/prodinfo.asp?number=G18109

Also, here is a capacitor discharge calculator which might be useful
http://www.circuits.dk/calculator_capacitor_discharge.htm

[cyberfish]

Computer power supply sounds like a good idea! I'll either mod it or build a buck converter and use the 12V line.

Super caps are a neat idea. Unfortunately I do need the 30A for extended periods of time.

I built all the circuitry (including integrated ESCs) from scratch, so I know it doesn't care much about voltage regulation.

[alexwhittemore]

Are you using brushless motors? Building 3-phase brushless ESCs from scratch is pretty badass.

[cyberfish]

Yeah it's a little harder than I imagined. I got it to work for the most part, just some trouble with starting.

[Zad]

It depends how extended extended time is. How about using a 6V VW or motorbike battery with a trickle charger?

[cyberfish]

The whole point of this power supply is to allow me to do extended testing without a battery. It will eventually run on a lithium polymer battery (which would only run for about 10-15 minutes).

I think I found the cheapest solution -
$30 computer power supply with 48A on 12V, big fat NPN transistor with a zener and resistor as a linear regulator, and an old GPU heat sink and fan to take the 100W (30A, 12-8.5V) heat off the transistor (I know it's big enough because the GPU had a TDP >100W).

Should be able to do this within about $50, and not have to play with mains.

[Psi]

Be careful with the current ratings written on computer powersupplys the cheap brands often lie about their current ratings.