voltage regulator help - ot

[docwisdom]

Sorry for the OT message, but you guys are great at answering my questions, and pretty quick at it.

Right now I am using LM317 voltage regulators to power my mosfets because each color of my RGB LED requires a separate voltage. The LM317 is only 1.5 amps & I would love to go up to the LM350 which is 3 amps. My main problem is that the heat generated by either one is enough that it requires a heatsink. I am using an aluminum enclosure and would love to bolt to it as a heatsink, but the tab shares output voltage which would cause problems with the grounded case.

Could anyone recommend other adjustable voltage regulators that might solve these problems? I see that the LM2650 doesnt require a heatsink, but it is SOIC and right now I am prototyping with all through hole stuff.

Anyway, I have only been working in the IC world for a few months now, so if anyone could shine some light on other magic IC's It would be great. Diging through the Digi-Key and Mouser catalogs is pretty tedious.

much thanks

brian

[Travin77]

Why not change to a plastic case? You can buy them at radioshack for about $3 and can get some pretty large ones.

[mister_e]

Why not changing for a plastic case??? Probably because the plastic case will never sink (dissipate ... i don't know the translation... bah anyway) any heat. So useless in this case.

There's still some insulator kit for most package to avoid that in case you really want to screw the regulator on the metal casing. Most local supplier should have them in stock for few cents.

As i remind, LM317P shoul be a FP(FullPack) casing. No need for insulator. Is it still available??? i don't know, go... search

[eoasap]

maybe try a small fan to promote forced-convection heat transfer ?

[docwisdom]

I will give the insulator angle a shot.
Steve, thanks for the info on the fullpack. Have you ever heard of a larger amperage unit available in fullpack. I couldnt find a LM350 version specifically.

Any DIP varieties that someone would recommend? I am running DIP mosfets and I am liking them so far.

I would love to run a fan, but I am trying to make things weather resistant.

cheers

-brian

[mister_e]

There's still another way before jumping to another regulator. The solution is to use a transistor to boost the current. In this case, you can use a TO-92 as regulator and almost any TO-220 in Full-Pack (or isotab someting like that). About the efficiency? fair to bad depending the transistor you'll use. Some have a huge drop between Collector and Emitter when the current rise across.

OR you can even build your own regulator with a simple transistor, zener diode and a resistor. Problem is almost the same as previous

Build one with a PIC or an OP-AMP to monitor the output and apply according correction is a better solution. That's what i did on some occasion.

BUT the ultimate, look at LT1085 but the TAB is often the output so... it solve a problem but not the other. www.linear.com

About a DIP solution? Yish, i can't recommend any if it exist anyways. I feel they'll mostely burn your PCB before.

In many case, you can place a heatsink on your Regulator and secure it on your PCB. This work really well in many case. This avoid the insulator stuff.

As you want to make it 'water-proof' Less holes, and screw have to be consider... except if you really want to add some scellant.

[Acetronics2]

Hi, Doc

Your question is a bit surprising ...

As Leds are to be driven at constant current ( your Leds are supposed to need a lot ...), the evident solution could be to drive a switching regulator ... current regulating, and just voltage limiting .

And may be no more voltage to adjust, then ...

Alain

[mister_e]

mmmpfff can't agree 100% with you Alain. Desolé

Switching regulator are great... but linear too. No more improvement over the previous ... oh well maybe a bit less of heat... but much expensive... at the end, you never ever save a cent.

Just rethink of your reply.. If you regulate the current... it's still a 3 LED driving plah plah. What will be the result when all 3LEDs have to be Full on to produce a WHITE color? Nothing else than a dimmed white

[Acetronics2]

Hi, Steve

I think we could know better what Doc' wants to do ...

a "simplissime" current regulator can also be achieved with a Red Led as volt ref, two resistors ( one for LED and one for emitter current ) and a transistor.

That makes 3 cases ... if TO 220 ... makes min. 3 watts to be dissipated.

Alain

[docwisdom]

Thanks for the replies.
When I arrive home today I will post a schematic. For now, here is a link to the LEDs I am driving http://www.laminaceramics.com/products/bl4000.aspx. For now I am running 1 RGB, I would like to run up to 8 in parallel, therefore the necessity for powerful regulators and powerful mosfets.
I tried using specialized LED drivers, but they all maxed out at 1amp. Cost me $60 too, I may sell them or return them.

I am a filmmaker by trade and that is the ultimate purpose for these LED's. A low power solution for car kits and mobile rigs.

P.S. My source voltage, ~12-14v will not be constant, I know that. A switching regulator that doesnt vary ouput based on input would be nice. The LT1074 looks interesting, also the LM2676. What do you think?

[docwisdom]

Here is the full schematic, it includes the LED drivers instead of the mosfets.

[Acetronics2]

Hi, Doc


Looking back to your first post ... make me think using LM350 ( T or K ) or LM 338K's ( 5A )instead of the 317's would be a very good solution ... as it exists some insulating silicone pads ( or Mica sheets ...) and insulating spacers to avoid connecting reg.case to your aluminium casing !!!

The " best " solution surely is here !!!

See here i.e. : http://www.selectronic.fr/soussousfa...0&ssfam_ref=30

or here : http://www.selectronic.fr/soussousfa...0&ssfam_ref=20


Alain

[BigWumpus]

It's not easy for us to understand your problem.

You are mixing up linear regulators and switching regulators... we don't know what voltage and current you need...

I think you want to build a luminated heater. You want to burn away a lot of the electric power in your regulators in order to give a little piece of power to your LEDs.

Think about switching regulators !

[docwisdom]

The RGB led's need 4.5, 6.7, 7.6 volts respectivly and draw 350mA per color. for a total of 1.05 amps per LED unit.

I would like to run 7 or 8 of these in parallel, giving me the necessity for 8-9 amps of total current. So 3A per color.

so to break it down, I need [email protected], [email protected], and [email protected]. Each controlled via hpwm from the pic.


hope this clears it up

thanks
-brian

[NavMicroSystems]

...to break it down, I need [email protected], [email protected], and [email protected]. Each controlled via hpwm from the pic.

Brian,
Let's assume the average supply voltage is around 13V
that gives a total loss of about 60W!

Earlier you were talking about a "low power solution"

You should think about a different and more economical approach (switching rather than linear).

Don't focus on those expensive "fully integrated" switching regulators,
there are several "low cost" SMPS-controllers available that give good results and high efficiency if you use the right external components.
Take special care on the inductors, capacitors, mosfets and diodes
and last but not least the PCB layout.

With one of my current designs I get close to an efficiency of 90% at competitive costs.
(SEPIC Converter 120W - Input 8-18V Output 12V@10A)

[mister_e]

Welcome back Ralph! We miss you!

[docwisdom]

Ralph,
Thanks for the post.
I think you went over my head a little bit with SMPS and SEPIC. Could you throw out a couple specific examples for me, or a website reference?

thanks
-=brian

[NavMicroSystems]

For the SEPIC Converter See: HERE

[docwisdom]

Ralph,
Thanks for the post. I have read the thread and reviewed the schematics. Its quite a large unit, but that is to be expected for 60w. Anyway, could you or anyone else recommend component changes so I can operate from a 4.5 to 7.6 output at 3A?

As I have mentioned in previous posts, I am pretty savvy with analog circuits, but just now getting my feet wet with digital components.

thanks very much

p.s. What is your opinion on switching regulators like the LM2676/2678? Both you & Steve mentioned that they are expensive, but digi-key lists them around $6 each. They are available in variable output versions as well.

[NavMicroSystems]

...Its quite a large unit, but that is to be expected for 60w...

Well, it is even more, it provides a stable 12V output at Input Voltages between 8 and 18 Volts, so it steps up or down at high current (with high efficiency).

Anyway, could you or anyone else recommend component changes so I can operate from a 4.5 to 7.6 output at 3A?

Again there are many ways, of course you can simplify the circuit as you don't need to step up, all you need is a "simple" step down switching regulator.
(Your minimum Input voltage is always higher than the maximum required output voltage)

See the UC3843 Datasheet for design examples.

p.s. What is your opinion on switching regulators like the LM2676/2678? Both you & Steve mentioned that they are expensive, but digi-key lists them around $6 each. They are available in variable output versions as well.

Sure, you could even use an LM2676, that would save some components like the external MOSFET.

I'm using the 2676 in one of my (GSM) designs where part count and PCB space was an issue, but overall this design was more expensive than others.
However, the 2676 is a somewhat expensive but reliable component.

And as mentioned earlier:
if you want efficiency, take special care of the inductors and capacitors you use!

You'll be surprised what increase (or decrease) in efficiency you get by simply changing a capacitor or inductor.

I have spent days with winding my own inductors on different cores to get maximum efficiency over the full Input voltage and output current range.

[wanhaibin]

I also had the heat problem with such linear regualtor. I was not able to find a good heat sink to remove the heat. The problem was solved after I change to a switch type regualtor (LM2575). Although the connection is more complicated, it brings good relibility to my control board.

[Acetronics2]

You'll be surprised what increase (or decrease) in efficiency you get by simply changing a capacitor or inductor.

I have spent days with winding my own inductors on different cores to get maximum efficiency over the full Input voltage and output current range.

Hi, Ralph

you're right ... new design = "some" windings to try !!! I also appreciated your converter.

by the way ... Would you know a replacement part for the MAX 879 ( low power, low voltage Step up/ Step down ) ... ???

It was a jewel for my 4 cells to 5 v applications, but no more produced now ...

Regards

Alain

[docwisdom]

I am in a holding pattern right now until I get my capacitor and inductor specs figured out. I going to go the LM2676 route. I have tried a piece of software by National that figures these values out for you, but it seems to be giving me conflicting information, or just information I dont understand (most likely) Anyway, if anyone is a wizz at this stuff, could I get a recommendation on a purchase list of caps and inductors for these three voltages?

thanks much

[NavMicroSystems]

by the way ... Would you know a replacement part for the MAX 879 ( low power, low voltage Step up/ Step down ) ... ???

It was a jewel for my 4 cells to 5 v applications, but no more produced now ...

How about the MAX 710/711?

ok, it requires a little more space plus an external schottky, but it's efficiency is better.

[docwisdom]

From the specs I see its less than 1 amp rated.

[Acetronics2]

How about the MAX 710/711?

ok, it requires a little more space plus an external schottky, but it's efficiency is better.

Thanks Ralph,

The only failure in it is .... only SMD available !!! I'm looking for through hole components ... ( I sometimes use the copper side also... but do not like it too much ... )

I'll try to get some from Maxim's ...

Alain

[Acetronics2]

Hi,Doc

There's also a boost for LM 257x ... see attachment.

Alain

[Acetronics2]

How about the MAX 710/711?

ok, it requires a little more space plus an external schottky, but it's efficiency is better.

Hi,Ralph

I Found that ...: http://www.farnell.com/datasheets/5438.pdf

may be interesting for you ( adding just a Power P-MOSFET i.e. ... hé,hé !!!)

Regards

Alain

[docwisdom]

Any resources for better calculating an inductor/capacitor match for the 2676?

[NavMicroSystems]

may be interesting for you ( adding just a Power P-MOSFET i.e. ... hé,hé !!!)

Hi Alain,

thanks for the link, but at the first glance I can't see any benefit.
With my latest design I get really good results even at high loads.

I really don't want to spend yet another number of days finding inductors and capacitors to push a different Controller to the efficiency I have already reached with my current design.

You and I know:
it takes a little more than just adding a MOSFET...
and other simple things a datasheet tells you.

Let others pick up this topic and let's see what they come up with...