I keep seeing different values used for the caps on the input and output of LM317/337 regulators.
I know that most people don't use the tantalums recommended in datasheets but...

What values and types of caps should be used on the input and output?
I'm building a power supply to give me at least +/- 18V and these little capacitors are the parts I'm not sure about.

 

What "little capacitors"? Can you show a circuit schematic?

When you see 0.1μF caps, these are non-polar ceramic capacitors.

1μF to 10μF commonly shown are polarized tantalum electrolytic capacitors.

What is equally important is where and how you solder these decoupling capacitors. Put the 0.1μF ceramic caps as close as possible with the shortest leads to the pins of the regulator. Put the electrolytics as close as you can place them to the pins of the regulator as well.

SMD capacitors will provide the shortest leads, i.e. no leads at all.

 

You will never memorize the values for every brand of regulator. That's what the datasheet is for. With millions of datasheets available on the internet, it is easier than ever to look it up and do it right.

 

What "little capacitors"? Can you show a circuit schematic?

When you see 0.1μF caps, these are non-polar ceramic capacitors.

1μF to 10μF commonly shown are polarized tantalum electrolytic capacitors.

What is equally important is where and how you solder these decoupling capacitors. Put the 0.1μF ceramic caps as close as possible with the shortest leads to the pins of the regulator. Put the electrolytics as close as you can place them to the pins of the regulator as well.

SMD capacitors will provide the shortest leads, i.e. no leads at all.

So you're saying the .1μF cap on the input can be the plain old little round (usually orange-ish) cap labled 104 like you find everywhere?

The caps connected directly the ADJUST pin (oops, I said output in my fist post) are generally shown as 10μF polarized caps. Should these just be the regular everyday cylindrical aluminium electrolytic caps?
Could I use something bigger like a 22μF?

I've also noticed that I don't see Radial Box Metallized Polyester capacitors when people build their own power supplies.
Are they just not well suited for this?

 

Plain old ceramic caps are pretty good for stuff like this. X7R or NPO types are stable and work well at high frequencies.

Don't go nuts on the adjustment cap or you'll need an extra diode so it can discharge without going backwards through the regulator when you turn the power off.

Poly caps work just fine, but they are more expensive and bulkier.

Tantalums are more reliable in the long run, but aluminum caps are dirt cheap.

 

Yep. You are correct. The plain old little round cap marked 104 will work just fine for 0.1μF. Any of these styles of packages are good, for example:

SMD work even better.

I believe it was bountyhunter who said some time ago that if the datasheet calls for 10μF, then put 10μF and not 22μF. He knows what he's talking about. He worked for National Semi.

 

I pretty much learned everything after vacuum tubes from National Semiconductor datasheets and app notes. If I had ever taken a course in electronics, I would probably say the datasheets and app notes provide a better education, or at least more practical education than most of what happens in a school.

 

Why would the data sheet say not to use tantalum caps ???

 

Yep. You are correct. The plain old little round cap marked 104 will work just fine for 0.1μF. Any of these styles of packages are good, for example:

SMD work even better.

I believe it was bountyhunter who said some time ago that if the datasheet calls for 10μF, then put 10μF and not 22μF. He knows what he's talking about. He worked for National Semi.

The ON Semi datasheet calls for
C in = .1uF
C Adj = 10uF
As for a cap on the output I think a 22uF electrolytic should be good considering the datasheet says...

External Capacitors​

A 0.1​

​

​

​

mF disc or 1.0 mF tantalum input bypass capacitor​

(C​

​

​

​

in) is recommended to reduce the sensitivity to input line​

impedance.​

The adjustment terminal may be bypassed to ground to​

improve ripple rejection. This capacitor (C​

​

​

​

Adj) prevents​

ripple from being amplified as the output voltage is​

increased. A 10​

​

​

​

mF capacitor should improve ripple​

rejection about 15 dB at 120 Hz in a 10 V application.​

Although the LM317 is stable with no output capacitance,​

like any feedback circuit, certain values of external​

capacitance can cause excessive ringing. An output​

capacitance (C​

​

​

​

O) in the form of a 1.0 mF tantalum or 25 mF​

aluminum electrolytic capacitor on the output swamps this​

effect and insures stability.

The only tantalum I happen to have IS a 1uF say maybe I'll try that too.

One place I've noticed people don't follow the LM317 datasheets to the letter is R1 (out to adj) which is shown as 240ohm but in reality lots of people use 120ohm.

 

Why would the data sheet say not to use tantalum caps ???

Probably because they don't work well in that position.
Which datasheet are you referencing?

 

One place I've noticed people don't follow the LM317 datasheets to the letter is R1 (out to adj) which is shown as 240ohm but in reality lots of people use 120ohm.

That's because the 117 version of the chip is guaranteed to less than 5 ma minimum load.

Watch the National Semi numbering system. 117, 217, 317, are the same chip, slightly different guarantees on their performance.

 

Probably because they don't work well in that position.

Not trying to be difficult but do you or anyone know why ???

 

Not trying to be difficult but do you or anyone know why ???

Not without naming the part.

This is an odd situation that depends entirely on exactly which chip you are talking about. I have never seen that statement, "Do not use tantalum here." so I really need to know where you saw it.

 

I'm resurrecting this old post because I recently got some 0.1uF X7R caps, some big 4700uF caps and some stripboard so I'm finally thinking of building a power supply for real and not just on a breadboard.

The last thing I'm wondering is if there's any difference in using the little X7R caps connected to the input of an LM317/337
http://www.newark.com/vishay-sprague/1c20z5u104m050b/capacitor-ceramic-0-1uf-50v-x7r/dp/17F2165
OR these box type polyester ones
http://www.newark.com/cornell-dubil...-film-0-1uf-100v/dp/81K6293?ost=160104J100C-F
I know they're not really that important in the first place but I guess I'm just curious.

 

Same, same, as far as the chip is concerned. I would prefer the ceramic cap because it's smaller, and an SMT version would be even smaller than that! Small minimizes inductance in the leads and circuit board traces, so small is good.

 

Same, same, as far as the chip is concerned. I would prefer the ceramic cap because it's smaller, and an SMT version would be even smaller than that! Small minimizes inductance in the leads and circuit board traces, so small is good.

Great, thanks

 

So you're saying the .1μF cap on the input can be the plain old little round (usually orange-ish) cap labled 104 like you find everywhere?

The caps connected directly the ADJUST pin (oops, I said output in my fist post) are generally shown as 10μF polarized caps. Should these just be the regular everyday cylindrical aluminium electrolytic caps?
Could I use something bigger like a 22μF?

I've also noticed that I don't see Radial Box Metallized Polyester capacitors when people build their own power supplies.
Are they just not well suited for this?

As others have stated; you need low value - and more specifically low ESR caps as close as you can get to the chip itself - that's for regulator stability.

You probably also need bigger electrolyitics if the regulator is supplied by a full-wave rectifies mains transformer - and an even bigger still electrolytic if its a half-wave rectified supply.

If you're running from a battery or well smoothed SMPSU, you can probably get away with not much more than the stability capacitors.