NERD ALERT Pure Electronics Question

[MEXICOWANDERER]

I want to use 78xx voltage regulators in my 36 volt lighting. To power integral with heatsink 1 watt fans. Clarification: I purchased new surplus PC obsolete heatsinks with ball bearing fans for less than the cost of a new fan.

OK. Limited space for additional electronics. Fifty watt LED chips I drive at max safe current.

7812 voltage regulators have a max input voltage limit of say 30 volts. They cannot handle 36 volts.

So I came up with a MOSFET, Bipolar 100K resistor circuit that zener's incoming voltage to 18. Far less Delta T burden on the 78xx.

With the tiny fan only load imposed on the 78xx, and MOSFET (which passes full 78xx current)...

Do you think the TO220 packages of the 78xx and MOSFET will demand heat sinks? Vertical clearance above the PCB will be limited (but enough ventilation) The MOSFETS I selected but not ordered are rated 30 amps @ 50vdc.

The heatsinks I use are black anodized and need a slow boat from China. It sure would be nice to save several dollars per light and gain a lot of headroom. The push down airflow of the heat sink fan could aid fan power supply cooling.

Yes they do make DC power supplies that "could" do this. Check incoming voltage limit, price and size. I am building an 11 lamp array.

Remember this is for a very low current fan.

[MEXICOWANDERER]

Sadly all extruded aluminum seems to be pricey. And black anodizing adds 30% to the cost and cools 35% better.

[mike-s]

GordonThree wrote:
For this design, keep in mind:

Vref * Iout = Pwaste (heat), heat in the chip and the resistor, watch the wattage rating for R ... under a buck for the chip maybe, but more $ on heatsink?

Vref for the LM317 is fixed at 1.25V nominal. And, the only power related to it goes into the resistor, which can be a couple of watts at higher current settings.

Not sure what you're calling Vref here, it's a current regulator. Nothing here is referenced to ground. The dissipation depends entirely on the current and the voltage differential across the LM317 needed to maintain that current. The closer the input voltage is to the required output voltage, the less heat. Dropout voltage is about 2.5V @ 1.5A so yeah, about 4 watts minimum dissipation. It will need a heatsink if driving more than about 0.5A with a minimal voltage difference.

[GordonThree]

mike-s wrote:
An LM317 can be configured as a current regulator up to 1.5A. Costs less than a buck.

For this design, keep in mind:

Vref * Iout = Pwaste (heat), heat in the chip and the resistor, watch the wattage rating for R ... under a buck for the chip maybe, but more $ on heatsink?

[MEXICOWANDERER]

Circuit board + LM317 + heatsink + resistor + cap

Tough to come in under a dollar. If that @#$%&! heat sink came out to be 1/8" shorter I could do it but standard TO220 sinks are fat and skinny (black anodized) heat sinks cost almost two dollars then require potting material.

Thanks for your help sir, I am going to try the eBay switcher!

I also have a small box of TO247 adj regs rated for 10-amperes. They are fun to play with. I only wish they could be configured (manufactured designed) to use Silicon Dioxide MOSFETs

Can't have everything but boy you sure solved my dilemma. I am a happy camper. I was looking at switchers but finding only those with 30 volt upper limit.

[mike-s]

An LM317 can be configured as a current regulator up to 1.5A. Costs less than a buck.

[MEXICOWANDERER]

BECAUSE (A necessary caveat)

The SHARP brand of chip has proven to be so dependable, I use a blanket power supply to cover several emitters in parallel. Zero current throttling.

The heatsinks with fans are new surplus units for Pentium CPU's. The emitters stay close to ambient temp. I can therefore drive them harder than any LED driver ever thought of. And their p/n junction stays cooler. Say 85 watts worth of driver eqvt light. And years doing this has proven the concept. No failures.

A 350 watt Meanwell 36-volt unit provides six chip potential. The Meanwell has proven to be far more reliable than any current limiter I have tried. "Better" grade 50 watt drivers are pricey when cumulatively aggregated.

Compromises incorporated. With network power distribution I lose easy individual lamp switching. And easier installation.

But I always have a nagging feeling that a current limiter exists that would serve adequately at a really good price. $x6 versus one Meanwell. I just do not have the financial ability to try a flood of different drivers to proof out the cream of the crop. The lumen output difference between adjustable ma control and set driver control is to me, a real eyebrow raiser. OEM "thriftiness" in heat sink cost may be the clue for this phenomenon. Keep in mind my record of no emitter or fan failures.

[GordonThree]

depending on your application, you can get an LED driver module with a secondary output for a cooling fan. what are you using to control the current for the 50 watt LEDs?

[MEXICOWANDERER]

Thank you sirs!

Less than one minute after reading Mr. mike-s post I went to eBay and purchased all of the vendor's remaining inventory. That circuit is light-years better for my purpose than a linear regulator. Far less heat.

[mike-s]

Matt's right, but you don't need even that much. For less than a buck, about the same price as an LM7812, you can get a complete stepdown switching regulator from China/ebay, and everything is taken care of for you. At 100 mA, no need for a heatsink. And, much much more efficient than using a linear regulator.

[Matt_Colie]

Sure, you can do all that with linear parts, but I have taken up buying Chinese modules for these things. Step-down switcher
I have about 6 doing different things, some are driving LEDs, one is powering the phone hot spot and the repeater and the another the GPS. I know there are two more than I can't place at the moment.

Matt

[GordonThree]

MEXICOWANDERER wrote:
The heat sink might complain.

Or, are you suggesting free-standing the heat sink?

Free standing, or maybe with the use of a potting compound?

[MEXICOWANDERER]

The heat sink might complain.

Or, are you suggesting free-standing the heat sink?

[GordonThree]

Parallel to the board is easy enough with a TO220 just leave the legs long and bend over 🙂

[MEXICOWANDERER]

RATZ. Wish they made a single flat (parallel to circuit board) heat sink. To hold both devices.

But most important is a thank you Gordon Three, and wnii!