I know that diodes drop a specific amount of voltage across themselves. What I'm trying to do is run an LM317 regulator from a 12V battery. I plan to output 5.6V from the regulator. Batteries tend to never actually run at their rated voltage, it could be more (upwards of 13.8V) or less (down to maybe 10.5). I'm thinking of putting a couple of 1N4001 diodes in the path to step the voltage down a little bit before it gets to the LM317 reg. This would cause less heat dissipation at the regulator. I could maybe chain several diodes up and output around 8.1 to 11.4V to the reg. Am I wrong in thinking that diodes, despite their voltage drop, do not really dissipate much energy and thus could be used to keep general system efficiency and heat loss in check?
sorry, but that power goes somewhere. If it isn't dissipated in the regulator it's dissipated by the diode.
-j
How much current do you plan to draw from the LM317?
The thermal resistance of a LM317 in a TO220 package without a heatsink is probably about 70degC/W. The absolute maximum junction temperature is 150degC. With an
ambient temperature of 25degC you could safely dissipate over 1W in the regulator.
The power dissipated in the regulator is (Vin - Vout) * Iout.
The diodes are not going to dissipate much power since the forward voltage is
not high (0.3-0.6V). The diode power dissipation is Vf * If. In this circuit If is the
same as Iout of the LM317.
If you need to dissipate more than 1W in the LM317 I would look at heatsink.
If you don't want the heatsink you would better off with two LM317's in series
than multiple diodes.
The best solution is to use a DC-DC converter. Something like the TI DCR021205
could work. The high efficiency will maximize your battery life.
(* jcl *)
Yeah, I know about DC-DC convertors. I could do that, it's just a little more money. Maybe it's worth it though.
I'm planning on drawing maybe 0.5A. This is still way more than 1W heat dissipation (at 13V the drop to 5.6 is 7.4V times a half amp is 3.7W heat dissipation)
It makes sense what you said about heat dissipation in diodes. I'm still in the learning phase and I just couldn't get my head around it. But what you said makes sense.
So the first diode (in my original idea) would see 13V at one end and 12.4V at the other for a total drop of 0.6V. 0.6V * 0.5A is 0.3W. So the best diodes could do is to spread the heat dissipation out among more components. And thus all I could hope to do is increase the cost but not the efficiency. This makes sense now that I really think about it.
I don't mind using a heatsink. Now I just have to determine whether it's worth it to use a linear reg with a heatsink or spring for a DC-DC conversion circuit to get higher efficiency.
The LM317 + heatsink will be a lot cheaper than the DC-DC converter.
You should be able to easily dissipate 4W in not too large a space.
Make sure you use a thin layer of thermal grease between the heatsink
and tab. You will not get the specified thermal performance with no grease or
with a thick layer of grease.
I would expect the LM317 + heatsink to be in the $2-$3 dollar range. The
DC-DC is probably in the $10 range.
(* jcl *)
In small quantities the DC-DC convertor is actually around $15 from digikey but I could search around for a better price. I've already got LM317's, heat sinks, and thermal grease laying around so that's the more expidient option. However, it also wastes about 57% of used energy as heat. The DC-DC convertor is about 90% efficient instead of 43% efficient. Like you said, the LM317 and heatsink is cheap. I probably really don't have more than $2-3 into the linear parts to do one unit. Choices, choices...
What is the AH rating of your 12vdc battery? Once one defines an application requiring battery power then it really does make sense to try and optimize the efficiency of the overall regulation system. Not only do you gain run time but heat build-up in small enclosures as well.
Here is pre assembled step-down switching regulator available on E-bay:
Lefty
Yeah, I do think that in the long run a DC-DC convertor will be the best option. I plan to make more than one of these units so I'll likely go with some sort of buck switching reg for the rest of them. I'll probably just use the LM317 for the initial prototype though, just to get something working. China is a long ways away and if I wait I'll be sitting here for two-three weeks.
IIRC there was an IC manufacturer that was packaging their DC/DC controller ICs + inductor + caps etc on a little PCB that was a little larger than a TO-220.
The pinout was the same as an LDO. I thought it might bee Micrel, TI or
Power Integrations.
If you can do PCB layout making a little DC/DC converter by carefully following the
reference design is not too difficult. I have done a few layouts of MHz converters.
Checkout the PCBs at Loading...
If you can get to 1-2MHz the inductors and caps are small (<10uH, <10uF).
You may be able to retrofit a TO220 device.
(* jcl *)
It was just an example, however the stull I have ordered from Sure electronics has so far always arrived in 9-11 days and their prices have always made the wait worth waiting 
Lefty
The TO-220 form factor regulator is from TI. And that was the $15 part I was referring to that Digikey had.
I did order up one of those convertors from Sure electronics. We'll see how I like it.
Thanks everyone.
Couldn't you use a high wattage zener diode?? I havent ever used a zener diode to do this, but if you get one with a high power rating (5W) and a Zener Voltage of 5.6V it should work.
This setup should work:
Again this is going to be super cheap (quick look on digikey for a diode of this spec shows hey cost 18 cents!), however you are going to be very inefficient like you noted earlier.
The zener doesn't dissipate any power the resistor does. At apx 7V and 0.5A it would
have to dissipate 3.5W. You would need to use a 7W resistor or a 3.5W resistor
on a heatsink. The high power resistor will probably cost you more than the LM317 + heatsink.
(* jcl *)
Yeah, I've had to buy power resistors before (you'd think that industrial workers would be able to understand the concept of a 30% duty cycle on a welder but NOOOOOOO) and I don't really want to do that for this application. For right now I'll hook up an LM317 to get the prototype up and running and in the future I will almost certainly use some form of buck converter to get the voltage down to the proper level.
For high power it is cheaper to dissipate the power in a FET or BJT than to use
a power resistor. The LM317 is essentially a variable resistance.
(* jcl *)
"For high power it is cheaper to dissipate the power in a FET or BJT than to use
a power resistor"
Boy isn't that the truth. I just bought on E-bay, ten 60 volt 50 amp logic level N channel MOSFETs (STP50N06L) for 55 cents each including shipping and got them in 3 days from FL to CA!
I think I'm ready to sink some impressive DC loads around here, if I can find some heatsinks
Lefty
I think I'm ready to sink some impressive DC loads around here, if I can find some heatsinks
If you are going to build some electronic loads make sure you use an independent
control loop for each FET. Do not just parallel the gates.
The transconductance mismatch, especially at high drain to source voltages, will cause
uneven power dissipation between devices.
I had to fix the design of a system that had 64 600Watt FET loads. To save money
on the current shunt power resistors they paralleled multiple FETs on a single
control loop. The system ran for a long time at high currents and low voltage.
At high voltage and low current the FET loads started to die.
The vendor saved at least $3K in resistors. The repair was quite a bit more
(* jcl *)