Reducing heat in 7805

[treitmey]

Sorry this is off topic.
I'm trying to reduce the heat in my 7805 regulator.
If I do my pwm for my LED back lighting using a N-FET
And run the + from raw 12VDC instead of the +5 from 7805
Wouldn't that reduce the current through the 7805 and the heat?

[rnielsen]

One thing I do, sometimes and it's been stated here before, is place a power resistor in series with the 12V input to the 7805. I calculate, or measure, the maximum current that will be drawn and then calculate the largest value resistance that will drop the input voltage to just above the drop-out voltage of the 7805. As your circuit draws more current, more voltage is dropped across the resistor which will help reduce the power dropped across the 7805.

It's sort-of a dirty trick to use but it works.

As far as pulsing your LED from the 12V, it will work. Just need to figure out the maximum pulse widths to keep it from over heating. Also, not knowing how your back lighting is set up, you can run your LEDs from the 12V without pulsing as long as you adjust your resistance to keep the current to safe levels, assuming that it's possible to change the resistance of your current limiting resistor for the LEDs.

Ronald

[asmboy]

i know what you want to do:

basically use the N fet as a low side quasi- buck regulator - right ?

the bad side of this solution is that you can make considerable harmonic RFI if the design is not filtered .

ALSO

w/o feedback - as the LED load varies - the voltage can swing rather widely at the low duty cycle you will be using to drop 12 down to 5V or less .

adding a properly calculated storage inductor ( for your load current vs PWM freq) , a splash diode and tantalum output filter capacitor will go a long way towards making it play well in the RF universe.

but basically it will work pretty well if the LED load does not vary - ie is JUST backlighting frinstance.

keep in mind that it could be done cleanly and simply as a HIGH side buck too with very nice hi efficency - especially compared to a linear regulator .

- and then the LEDS would all be sinked ( sunk ?) to ground potential - but at the disadvantage of needing an extra part like a 2n7000 /7002 or 2n3904 - gate pulldown driver for a P fet switch

HOW MUCH current do you wanna draw?

give me an email address and i'll share a small PDF of the very simple - tested circuit with you .

[newguy]

What package (TO-220, TO-92, etc)? I always design my PCBs (usually just 2 sided) with top & bottom copper pours over almost all of the unused space - they become my ground plane. I just make sure that the regulator has a good thermal bond with the plane and as long as the board is more than approx 1"x1" heat isn't an issue. If you're using a small (ie 100mA) regulator then I agree that heat will be a problem.

[asmallri]

[MikeW]

this is a good solution, but expensive, there are standard TO220 or smd types


http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=0416862


or

http://uk.farnell.com/jsp/search/browse.jsp?N=500006+1002564&Ntk=gensearch_001&Ntt=pt78s&Ntx=

Texas Instruments PT78ST105ST is a good part, but $13

http://focus.ti.com/docs/prod/folders/print/pt78st105.html

Mike

[treitmey]

Thanks for all the good ideas.
I am actually using a mc34063 switching regulator. I want to
keep the inductor as small as possible. But the theory is the same.
Reduce current & power. ((don't drop voltage as heat))

(1) I've got to start with 12V, that can't be changed.
(2) I think the resistor is just dropping the voltage as heat in a similar fashion to the 7805(I think).
(3)The quasi- buck regulator idea is interesting, but it adds an inductor.

I will bread board the design and test its functionality. I am using the PIC pwm to drive gate of bss123 N-FET that low side controls the back light of my LCD at 12VDC. I was just curious if I'm low side switching 12VDC. if the current somehow gets into the mc34063?

[asmallri]

[treitmey]

Thanks.