PDA

View Full Version : Power supply woe's again...



sccoupe
01-30-2012, 08:30 PM
I had some problems with noise causing wield issues with my microprocessor projects a while back and the community got me setup with the correct capacitors and inductors to get my LDO regulator running smooth power. That was with automotive 12 volt+ power coming down to 5 vdc at about 50mah draw with no issues. Now I have a project pulling 350mah and found that an LDO regulator gets too hot. I even tried an old 7805 with a Radio Shack heat sink and its still too hot for me. So I ordered a switching regulator (TL2575-05IKV) as I read that they are more efficient. I get it all hooked up and it works but also gets too hot. What is one to do to easily get automotive voltage down to 5vdc at 350mah draw without creating so much heat that a sink is needed and without a ton of parts?

Leon
01-30-2012, 08:39 PM
Post the schematic!

sam_sam_sam
01-30-2012, 11:40 PM
Take a look at this board http://www.futurlec.com/Mini_Power.shtml for about $7.00 USD

I have used these board and they work very well for the money

average joe
01-31-2012, 06:23 AM
A series-pass transistor should work well.
http://www.ecircuitcenter.com/circuits/opreg1/opreg1.htm
http://tangentsoft.net/elec/opamp-linreg.html
My other thought would be to use a few diodes to drop the input voltage. This way the regulator wouldn't have to drop all that voltage. I would say 5 diodes in series. 5 diodes x .7v drop per diode = 3.5v drop. This would mean the voltage regulator only has to drop 3.5v. Remember when the car is running and the alternator is charging the battery, voltage should be about 14v. I'm not sure that this would work, but it's a thought. Schematic would be nice. If you are only pulling 350ma, a 1A regulator should be fine on air. Have you tested the actual draw against what you think it should be. I know I've made some mistakes that became apparent that way. Make sure you use a digital ammeter so you can catch spikes analog meters can't track.

Mark_T
01-31-2012, 03:05 PM
If you need to dissipate several watts a fan is usually needed... Alternatively if the heatsink can be a whole aluminium box it might be big enough.

Leon
01-31-2012, 03:08 PM
A properly designed switcher shouldn't run hot.

RDL2004
01-31-2012, 04:15 PM
A properly designed switcher shouldn't run hot.

This is true.

If what you say is correct, your 7805 is dissipating a bit over 3 watts, this is too much without a heatsink. I doubt you need a fan, but those little heatsinks Radio shack sells are not going to be enough.

sccoupe
02-04-2012, 02:27 PM
Ok, sorry to take so long to get back to this thread. I ordered a couple of those boards from futurlec to see how they done it, but they havent arrived yet. Attached is the schematic that I am trying to use and getting way to much heat even with a 3"x2" aluminum heat sink. This how its wired minus the ripple filter, and capacitors as im not worried about noise at this point. I also omitted the diode. The inductor that im using is a 330uH 1.19A 0.543ohms power inductor.

Lawson
02-04-2012, 02:50 PM
I'd use one of these drop in regulators (http://search.digikey.com/us/en/products/V7805-500/102-1709-ND/1828602). (the rest V-Infinity non-isolated line is nice too) They do get warm in free air, but a metal tape flag or gluing the case to a ground plane will keep it cold even packed tight to other components. They're pretty quiet electrically and will pass reverse currents back to the source if needed. (i.e. say with regenerative braking of a motor)

Lawson

sccoupe
02-04-2012, 03:03 PM
I'd use one of these drop in regulators (http://search.digikey.com/us/en/products/V7805-500/102-1709-ND/1828602). (the rest V-Infinity line is nice too) They do get warm in free air, but a metal tape flag or gluing the case to a ground plane will keep it cold even packed tight to other components. They're pretty quiet electrically and will pass reverse currents back to the source if needed. (i.e. say with regenerative braking of a motor)

Lawson

This could be a good alternative. Price is a little high, but by the time i pay for the other stuff needed for my current unworking solution it may pay off. At least with this, I just change my board design to accept a through hole regulator from a surface mount and my problem is solved. Thanks Lawson, I'll order a few and see how it works.

PJ Allen
02-04-2012, 04:48 PM
I also omitted the diode.
You reckon that's an optional or vestigial component?

Do you have all of those grounded leads going to a central point (as you ought) or do you have them connected helter-skelter with bus wire or something?

>>> the device's PDF emphasises the utility of the catch diode:

"As with other external components, the catch diode should be placed close to the output to minimize unwanted
noise. Schottky diodes have fast switching speeds and low forward voltage drops and, thus, offer the best
performance, especially for switching regulators with low output voltages (VOUT < 5 V). If a high-efficiency,
fast-recovery, or ultra-fast-recovery diode is used in place of a Schottky, it should have a soft recovery (versus
abrupt turn-off characteristics) to avoid the chance of causing instability and EMI. Standard 50-/60-Hz diodes,
such as the 1N4001 or 1N5400 series, are not suitable"

There's a bit about "Grounding" in there, too.

http://www.ti.com/lit/ds/symlink/tl2575-12.pdf

sccoupe
02-04-2012, 07:59 PM
You reckon that's an optional or vestigial component?

Do you have all of those grounded leads going to a central point (as you ought) or do you have them connected helter-skelter with bus wire or something?

>>> the device's PDF emphasises the utility of the catch diode:

"As with other external components, the catch diode should be placed close to the output to minimize unwanted
noise. Schottky diodes have fast switching speeds and low forward voltage drops and, thus, offer the best
performance, especially for switching regulators with low output voltages (VOUT < 5 V). If a high-efficiency,
fast-recovery, or ultra-fast-recovery diode is used in place of a Schottky, it should have a soft recovery (versus
abrupt turn-off characteristics) to avoid the chance of causing instability and EMI. Standard 50-/60-Hz diodes,
such as the 1N4001 or 1N5400 series, are not suitable"

There's a bit about "Grounding" in there, too.

http://www.ti.com/lit/ds/symlink/tl2575-12.pdf


Im not sure about the diode. Thats why I ask here for those more knowledgable on the subject. It seem as though that is to smooth out the power and not having so much to do with heat or efficiency. Grounding is all soldered and to a common spot.

Leon
02-04-2012, 08:14 PM
The diode should be included, it reduces noise.

Phil Pilgrim (PhiPi)
02-04-2012, 08:30 PM
The ripple filter is the only thing marked "optional." Why would you even assume that you could omit the diode and the caps? They're essential to the circuit's performance, else they'd be listed as "optional" too. See this article on buck regulators:


http://en.wikipedia.org/wiki/Buck_converter

-Phil