To come back to these voltage regulators for a second.....
This is powerful. So this is a 3-leg component, all you do (in my case) is punch one leg down into the 12v power plane, punch the other leg down into the ground plane and draw a trace from the remaining output to whatever circuit I want to power?? Unreal. That frees up a lot of my current board design restrictions...
Up until now I've been doing all sorts of gymnastics with my traces to feed power to separate circuits located in different places on the board. So the design is getting complex because I've been tapping into the 5V VCC line and running it (in all sorts of odd ways to ensure it doesn't cross over other lines etc.) Now, instead, if my understanding is correct, I can just punch 3 holes in the board, right next to the circuit I want to power, and then punch a leg down to 12V, another to ground, and then a short line to power my circuit?
Yes, that's right. Those regulators need decoupling capacitors - one between the input and 0V, and one between the output and 0V - for stability, so that's three components for each local regulator, but the idea is right. It's not normally done though - normally you have only one regulator for the board, and distribute the 5V around the board as needed. But you can do it.
This is too cool
The implications are wide-ranging when I think about that and my current layout....This really frees me up on the design side and enables me to precisely power different circuits with specific amounts of power and mA based on the draw of the cricuit - too cool!
"... specific amounts of power and mA ..." is an odd way to put it. You can provide different voltages to different parts of the circuit, and each section will draw a different amount of current. The 78(L/M)xx series does not normally include a 3.3V version, but STMicroelectronics make an L78L33 (
http://www.digikey.com/product-detail/en/L78L33ACZ/497-7288-ND/1038304) and Texas Instruments make a UA78M33 (
http://www.digikey.com/product-detail/en/UA78M33CKCS/296-21633-5-ND/1494026).
The 78(L/M)xx devices come in three types. The 78Lxx is in a TO-92 transistor package and is rated for 100 mA output current maximum. The 78Mxx comes in a TO-126 package, or similar, which is flat and larger, and is rated for 500 mA output current. And the 78xx series (no letter in the middle) comes in a TO-220 package (a very common package for power transistors and MOSFETs) and is rated for 1A or 1.5A output current depending on the manufacturer.
These devices are very old, and still widely used, and are made by many manufacturers. They have a "dropout voltage" of about 2V. That means that the input voltage must be at least 2V higher than the output voltage, otherwise the regulator will "drop out" of regulation, and the output voltage will be lower than it should be. The regulator needs at least 2V across it to operate properly. This is no problem in your case where there is a 7V differential, but I thought I would mention it anyway.
Because the 78(L/M)xx devices are an old design, their output voltages aren't highly accurate. For example, a 7805's output voltage could be anywhere between 4.75V and 5.25V. Again this isn't normally a problem.
The 78(L/M)xx devices also need a decoupling capacitor on the input and another on the output, for stability and good performance. Download a manufacturer's data sheet and have a look through it for more information, and Google how to use three terminal regulators to see if there are any tutorials - there probably are lots.
For future reference... if you need better performance than you get from a 78(L/M)xx regulator, there are many LDO (low dropout) regulators that need less than 1V difference between input and output - some have dropout voltages of 0.5V and less. And being more modern designs, they also have more accurately regulated output voltages.
Couple more questions though, now that I suspect I've learned the power of the voltage regulator. The Bluetooth module that I was going to power from the 5V VCC line (which shouldn't because I think there's only 28mA overhead above and beyond what the module uses) can now be powered by a voltage regulator
It can operate with an input variance of 3.3V to 5V. My guess is that the smart thing to do is match the regulator to the low side of the power requirement and that this will generate less heat and be 'easier' on the module then forcing more power into it (I did notice that they said the higher the power, the less stable connections get).
That depends on the design of the module. If it uses a regulator inside it to drop the 3.3~5.0V down to 3V, for example, and uses that rail to power everything, including the RF transmitter (which will be the main current drain), that would be right, because the lower the input voltage, the less voltage will be dropped across that regulator, and the less heat it will dissipate. And if they recommend using the lower voltage, then that's a good enough reason to use it.
I'm looking for one right now, the specs on the module say to use one rated for 3.3v and 150mA. This brings up some q's;
Is my understanding above generally correct?
Yes, I think so.
Should they be heat sinked?
You mean the regulator?
You need to calculate the power dissipation in the regulator. This is equal to the voltage dropped by the regulator (which would be 12V - 3.3V, I guess) multiplied by the current flowing through the regulator, which I guess is 150 mA (0.15A), right? In that case the regulator will be forced to dissipate (12 - 3.3) × 0.15 = 1.3W. This will make any small component pretty hot and some heatsinking will be needed, even if it's just an area of copper on the PCB.
Read Steve's excellent resource:
https://www.electronicspoint.com/resources/do-i-need-a-heatsink-how-big.29/
It doesn't look like you can heat sink the SMD versions?
Some SMT parts have an "exposed pad" on the underside that's meant to be soldered to an area of copper. SMT parts also transfer heat to the copper through their leads. But generally that's right.
Here is a link Digi-Key's linear voltage regulator selection table, filtered to include only THT devices with 3.3V output, maximum allowable input voltage of at least 15V, and a TO-220 package (heatsinkable).
http://www.digikey.com/product-search/en?FV=fff40027,fff80182,40011f,40103c,40203a,402209,40316e,4031c6,c00016,1140050,1fc0fec,1fc11be,1fc11d9,1fc11f3,1fc11f5,1fc11f7,1fc11fe,1fc123e,1fc1622,1fc162b,1fc16d8,1fc16e1,1fc178a,2040001,2040006,2040007,204001b,204003c,204003d,2040043,204004c,2040053,2040068,204007f,204008f,2040093,20400aa,2040104,204041f,10a40002,1360001b,1360008e,136000d1,136000de,13600122,1360018b,136001be,136001d5,136001e3,13600211,13600244,13600261,1360029b,136002a7,136002b6,1360032f&ColumnSort=1000011&stock=1&quantity=1&pageSize=250
Any of those will be suitable. But beware, they are all LDOs and LDOs almost always have strict requirements for the output decoupling capacitor. These are described in detail in the manufacturer's data sheet.
Now here's something interesting that I'm finding. Some through-hole components are dirt-cheap (resistors). Others, like a 150mA power regulator that is rated -40c to 125c are nearly impossible to get in a through-hole design. On the other hand, SMD versions are plentiful and are cheap. And are much smaller. I was originally going to go through-hole with the original prototype with the idea that perhaps I would do a SMD version of the board in the future for fun and as a re-design because I will have learned more and may have a better way to design the board. So I was always intending on doing incremental re-designs in my spare time in keeping with my philosophy of
define,
design and
refine
Today I cleared off my desk as it was getting too hectic (overload!) and I went for a drive and thought about my decreasing real estate on the board. I then got to thinking, once more, about SMD. I haven't pulled the trigger on the DigiKey order yet...and I do have a SMD rework station here on my desk and I've done a lot of playing around removing components from spare circuit boards - with great success using both the hot air gun and the iron & wick.
I just played with my CAD program (ExpressPCB) and I found that they have a pre-made SMD component for the 3.3v regulator. In fact, it's got lots of SMD components....
The one thing that bothers me about SMD components is that I took a bunch of resistors (?) off of a computer board with no issue, but they are the size of fleas! I guess I could try to put them back on but I suspect I'll need a larger magnifying glass....
SMT resistors aren't generally reused. In fact, SMT components, full stop, aren't generally reused, unless they're really expensive. You can get a kit of standard resistor values in SMT - Digi-Key even have them, I think.
Yes you're right. In the last six months or so, I've seen Digi-Key greatly reduce the THT components in their catalogue. The writing has been on the wall for over a decade. Over two decades, actually. THT components will remain available for a while yet, but for a long time now, most newly introduced devices have been manufactured in SMT only. MOSFETs are a prime example, but also modern ICs - digital and analogue. Microchip have been struggling valiantly to continue to offer some devices in THT and I suspect that's a big reason for their popularity outside professional circles. But if I was a gambler, I'd be buying shares in companies that make prototyping boards for SMT components!
