PDA

View Full Version : 4.5 to 25volt 15Amp power supply circuit



GeorgeD
04-08-2010, 08:23 PM
Was thinking of building this...then I found the price of a mains transformer to run cost a bomb....
http://www.circuit-projects.com/dimg/dual-power-supply-with-maximum-25V-at-15A-3A.gif


http://www.google.co.uk/imgres?imgurl=http://www.circuit-projects.com/dimg/dual-power-supply-with-maximum-25V-at-15A-3A.gif&imgrefurl=http://www.circuit-projects.com/power-supply/Page-2.html&usg=__2oADCWXbseAnu8kNPa0Erx5uhUI=&h=727&w=879&sz=22&hl=en&start=16&um=1&itbs=1&tbnid=lNmeF7sQvUcNmM:&tbnh=121&tbnw=146&prev=/images%3Fq%3Dlm350%26um%3D1%26hl%3Den%26client%3Df irefox-a%26sa%3DX%26rls%3Dorg.mozilla:en-GB:official%26channel%3Ds%26tbs%3Disch:1

tumutbound
05-08-2010, 11:56 AM
Did you look into toroidal transformers? They're not that expensive.

GeorgeD
05-08-2010, 05:33 PM
Yeah,but only on Farnell components site...pretty penny they're.

Jonathan
05-08-2010, 11:29 PM
You might want to try Rapid Electronics I've got a toroidal transformer from them. They'll probably have the other components too.

ecat
06-08-2010, 01:49 AM
Don't forget to price in the heatsinks !

Power wasted = (Voltage in - Voltage out) * current. Calculate this for worst case.
Required thermal resistance of heatsink = (Max heatsink temp - Ambient temperature) / power wasted. Say, ambient = 20degC and you're happy with a heatsink temp of 40degC then heatsink C/W = (40 - 20) / power wasted.

N.B.
Check the regulator data sheet for the minimum dropout voltage, eg LM338 requires Vin to be at least Vout + 3v
Check the regulator data sheet for safe operating temperatures.
Check the regulator data sheet for info on the inbuilt current limiter.
Double check the heatsink data sheets to see if the quoted C/W is for natural convection or when used with forced air (ie a fan)

My personal thoughts on this design: attempting 21v range on a high power linear regulator is stupid.

irving2008
06-08-2010, 06:20 PM
I agree... this circuit is a good example of doing it because it can be done, rather than good or efficient practice...

Incidentally I'd always go back to calculating junction to ambeint thermal resistance and dissipation - just because your desired heatsink temp is 40degC doesnt mean the junction isnt running out of spec...

LM338 TO3 package TRjc = 1degC/W so input 20vAC * 1.4 = 28v, output 4.5v @ 15A. Pdiss = (28 - 4.5)*5 (per device) = 117.5W. Tamb=25degC, TJmax = 125degC - not possible! Assume 0.5deg/C per watt heatsink, total TRja = 1.5degC/W, max pDIss - (125-25)/1.5 = 67W, so max output @ 4.5V is 66/(28-4.5)A per device = 2.8A, 8.4A total

Best price I could find for such a heatsink was 21.50 (and you need 3 of them). Alternatively a 1.5degC/W and an 80mm fan would give the same result... at about the same price...

Oh and just in case you're thinking "I could make a heatsink"... thats equivalent to around a 300mm length of 100mm x 50mm U channel with forced air down its length...

ecat
06-08-2010, 07:24 PM
21.50! The ones I found in a quick search were 110 each. lol. Of course you could DIY... http://sound.westhost.com/articles/diy-heatsink.htm . Beauty isn't it?

The lm338 is quite a bullet proof design, at Vin - Vout = 20v for example, the internal current limit will kick in at around 2.5A. This protects the chip but makes a nonsense of this designs claims. As always on the internet, if something takes your fancy do your best to verify everything about it, including the credibility of the author. When it comes to electronics, the link at the top of this post is good place to start.

der_fisherman
17-08-2010, 08:46 PM
You didn't mention just how many amps at what voltage you actually need. I built about a year ago a design from Elektor that works in a similar way to a Switched Mode Power Supply, but with most of the dangers to the user removed.....also, with the chips noted, you can have 10 amps at up to 40 volts. With other chips, you can have even higher voltages......if you want full details, I can post them here for you, just let me know......it was easy to build, easy to set up and is around 85% efficient, almost as efficient as a top SM Power Supply....and with the plus that I know EXACTLY how it works and I can easily repair it if ever needed.....all the clever stuff is in two identical chips.....
Regards
Andy

irving2008
17-08-2010, 11:00 PM
Sounds interesting, tho if it was in Elektor its a copyrighted design... but you could post up the gist of it for information.... personally, with the cost of PSUs on eBay in the 20 region its hardly worth building you own in many cases... I have a 0 - 20v bench supply @ 5A I built 20y ago, when I had access to unlimited MIL spec parts lol, but hardly ever use it, since a fixed 5v and 12v supply suffices for most things PIC-based these days. Rarely do any discrete analogue stuff now....

John S
17-08-2010, 11:33 PM
A good source for heatsinks are old blown inverters, they are usually mounted on a large cast custom heatsink, often with an inbuilt fan.

.

irving2008
17-08-2010, 11:34 PM
if you have a source of those John... most of us don't... lol

John S
18-08-2010, 12:20 AM
I do a lot of work for 3 rewind companies and they save me any dead ones, as do the inverter reps. Problem is now the cheaper versions have pressed steel heat sinks, you need the older ones for the best ones.

der_fisherman
18-08-2010, 08:47 AM
Sounds interesting, tho if it was in Elektor its a copyrighted design... but you could post up the gist of it for information.... personally, with the cost of PSUs on eBay in the 20 region its hardly worth building you own in many cases... I have a 0 - 20v bench supply @ 5A I built 20y ago, when I had access to unlimited MIL spec parts lol, but hardly ever use it, since a fixed 5v and 12v supply suffices for most things PIC-based these days. Rarely do any discrete analogue stuff now....

The complete article is attached here. The PCBs are still available via the companies that Elektor uses themselves or you can get one made.
The reason that I build my own is that they are far simpler to repair if need be because I know how it works!!! Not that I have needed to up to now.
Bought Switched Mode Power Supply (ebay for example) examples are seldom supplied with a manual, nor are the parts always easy to find. Plus true SM Power Supplies are simply dangerous for the amateur, whereas the chips used in the Elektor design do all the clever stuff and the only dangerous part is the mains input to the transformer. So its far far safer.......
The supply is actually 2 x 5 amp chips, running in parallel at a high efficiency.
There are two resistors that need to be programmed to set the voltage required, as is, the power supply supplies 17 volts, not a useful voltage!!! One of mine is set for 24 volts another for 36. Both run with no problems at all.....
best wishes
Andy

irving2008
18-08-2010, 09:15 AM
Thats a neat design, but its a bit disengenuous to say its not switched mode... it is a switched mode regulator, but I take your point that its not AC line level switching, therefore safer. That chip is good to 60v so a 0 - 50v variable supply at 5A (for a single chip) with low heat output is very possible, and using tapped inductors a 0 - 25v at 10A with a single chip is possible... interesting that the Elektor approach used two chips for 6A when they could have done it in one by winding their own coil instead of buyng off the shelf... think they missed a trick there... the chips are 9.24, the ferrite a lot cheaper...

However, it still doesn't overcome the OP's initial issue - the cost of toroidal transformers...

der_fisherman
18-08-2010, 09:33 AM
Thats a neat design, but its a bit disengenuous to say its not switched mode... it is a switched mode regulator, but I take your point that its not AC line level switching, therefore safer. That chip is good to 60v so a 0 - 50v variable supply at 5A (for a single chip) with low heat output is very possible, and using tapped inductors a 0 - 25v at 10A with a single chip is possible... interesting that the Elektor approach used two chips for 6A when they could have done it in one by winding their own coil instead of buyng off the shelf... think they missed a trick there... the chips are 9.24, the ferrite a lot cheaper...

However, it still doesn't overcome the OP's initial issue - the cost of toroidal transformers...

Actually, if you look around a bit, ebay and the like, you can get hold of a transformer for very little money. I paid €20 for a brand new one for my first power supply. The second one I got given!!! He should be a bit patient!! Not in a rush!!
But each to his own....I spent many years in the computer industry and I like REALLY stable regulated and efficient supplies. Ones that I fully understand exactly how to repair......
Also many thanks for your comments about prices and other methods etc....very informative.
Best wishes
Andy