LM338 (x3) power supply cuts out on high loads

keebler65 · 2006-09-12T03:19:10+00:00

I recently built an adjustable power supply with three LM338sessentially in parallel. The design was copied directly from National'stech notes PDF; all values are exact. I'm powering it with a 26v 15Atransformer through a full wave bridge rectifier. I'm actually usingthis supply to charge a large capacitor bank (700,000uF charged up to12-20 volts). I realize this creates a large load for a short amount oftime, so to help pad the load I put a 1 ohm power resistor in serieswith the capacitor bank. This worked for probably 20 charges or so,then something changed.I noticed the DC voltage at the bridge rectifier rose up to 40v (butonly when the regulator circuit was connected, so I know nothing iswrong with the transformer). With just a small light bulb (less than1/2 amp) it works fine throughout the whole voltage range. However,when I add about 5A of light bulbs, the output drops to zero instantlyuntil I remove the large load. This never happened before.I'm assuming it shuts off because it's now having to regulate from 40Vto 15V which is a rather large spread. I guess my main question is whatis causing the voltage at the bridge to rise from 26V to 40V when Iconnect the regulator circuit? (it never did this initially). Is itmore than likely that the LM338s are toast? I find it hard to believesince they still work great at very low currents.Note: I've read the app notes pretty thoroughly, and thereforeunderstand the basics of the LM338 (filter caps close to theregulators, protection diodes, reference wire connected directly tocase for low resistance, etc).I'd appreciate any suggestions / ideas that you may have about thissituation. If the LM338 is not the way to go for charging a large capbank, what is? I'm trying to keep this design relatively simple andcheap.THANKS!Kevin

Sep 12, 2006 23 Replies

Rate this thread:

keebler65 19 years ago

The 700,000uF worth of capacitance is connected to the output of the regulator not because I'm try to filter the output, but because I am trying to charge the bank of capacitors for a completely different purpose.

I am now finding that. With this discussion especially, I know I need to lower the voltage into the regulator. Is there an easy way to do this without having to buy a new transformer? They are somewhat difficult to locate for a reasonable price in my area, at least at surplus prices which is about the only option at this point.

Would half-wave rectification be an option? I know I would probably need even larger filter caps, but if it lowers the voltage to a reasonable amount (but not too low...) I would save myself the trouble of trying to find a different transformer.

Thanks Kevin

Vote

Michael A. Terrell 19 years ago

Please ignore the troll. Better yet, just kill file him.

Service to my country? Been there, Done that, and I\'ve got my DD214 to prove it. Member of DAV #85. Michael A. Terrell Central Florida

Vote

colin 19 years ago

National's

what

load,

very

make a

limit.

the lm317 ciruit you mentioned will probably do quit well, it needs one pnp transistor wich drives the 2n3055s. carefull to ensure its got adequate current limiting.

or roughly something like this ... used to apear in many op amp application notes ... however linear supplies are probably considered old hat now.

30v___________________ | | | | ____|/ | | | | |e | | | | |__|/____|/___|/ | |e |e |e | | | | | Re Re Re |__Z/|____|_____|____|________ op | Z\\| | R3 R1 | /|______| |_____/-| | \\+|->pot | \\| R2 | 0v______________|

4 x 2n3055 on suitable heatsink. (first transistor could be something smaller) Re=3ohm and zener = 3v for aprox 3 amp current limit, change to suit, max current =3x5 amp due to SOA Re needs to be suitable power. op amp could be lm358, pot needs to be a potentiometer conected to a reference voltage chose r1/r2 for the right op range. R3=1k ish.

also a second op amp could monitor the current ie average voltage accros all

3 Re and take the place of the zener.

If you not capable of debuging its best to find a design with more detail, im sure there must be some out there, or simulate it first with ltspice etc.

ooh a quick search showed up this ..

formatting link

Colin =^.^=

Vote

ehsjr 19 years ago

If charging speed is not an issue, just use a big R - say

50 ohms (2 25 ohm resistors at 25 watts each, in case you short the output somehow) - between the bridge and the regulator. That will be kinder on the bridge & the xformer, and you won't need 3 parallel 338's. It'll take a few minutes to charge. (You could use a much smaller xformer and save the 15 amp beast for a better purpose.)

Vote

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required