Which capacitors for L7805 regulator?

This data sheet (pdf) includes an example circuit using a .33 uf capacitor on the input side, and a 0.1 uf capacitor on the output side, presumably both ceramic:

However, I see lots of example circuits all over the internet that use different capacitors.

Here's one using much larger 100 uf and 100 uf electrolytic capacitors:

And here's one using 100 uf and 10 uf electrolytic capacitors:

Is there a reason to use capacitors other than what the data sheet example uses? Advantages? Risks? Rules for determining which might be optimal for a given use?

Thank you.

If the smoothing cap (for a mains transformer) is too far a way from the input pin, a cap near the IC has to be used.
100n, 330n, 10uF, whatever. Not too critical.
The output needs better bypasssing.
10uF tantalum. or 10-100uF electrolytic with a 100n ceramic.
The electrolytic cap can be omitted if there are more caps in the rest of the circuit.

Note that these regulators are last century.
They drop 2-2.5volt, so a 9volt battery and a reverse protection diode gets quickly below the point of regulation.
Are you using this to supply an Arduino?
Leo..

Note that these regulators are last century.

I was brought up on them and still think there is a place for them.

Switch mode may be much more efficient but take up more room. In the right place (low in/out V difference and low current) they are the simplest thing to use.

Weedpharma

Hi,

weedpharma:
I was brought up on them and still think there is a place for them.

Switch mode may be much more efficient but take up more room. In the right place (low in/out V difference and low current) they are the simplest thing to use.

Weedpharma

Here, here... :slight_smile:

[soapbox]
If in a hurry, got to get prototype going or got lots of boards plugged into a mother board and each board needs its own 5V supply.
Taa daaaaa LM7805 and a couple of caps.
Need odd regulated voltage, bootstrap.
Also the caps aren't tortured with high freq current to the extent that exist in SMPSs.
I admit there is a place for SMPS type low current supplies, but about 15% of my work is replacing tortured electro caps in SMPSs.
Don't point out they were made with low quality caps, I didn't assemble them in the first place.

[/soapbox]
With caps needed around LM78XX, if tight for space make the caps tantalums, 10uF, at least on the output pin.
Tom..... :slight_smile:

For many years now, I have used a 47 to 220 uF electrolytic cap on the input side and a 10 to 22 uF electrolytic or tantalum cap on the output side. Also, I ALWAYS put a 0.1 uF monolithic or ceramic cap on each side for any high frequency noise present. Watch the polarities on the electrolytics and especially the tantalum - they will disintegrate violently with reverse polarity !!!!

If you want to go down from 12 or 9 Volts to 5 Volts and you do not need more than 200-300 mA, line regulators are just perfect.

If you are running on battery power things are a little bit different.

If your power source is AC, you should not worry about power losses in the regulator.

Going from 9V to 5V at 200 mA is just a power loss of 0.8 Watts. You will not even need a heat sink in that case.

At 100 mA it will be only 0.4 Watts.

The size of the capacitors depends on multiple factors.

At the input side I recommend at least 220 µF for every 100 mA. At the output side I always use 47 or 100µF if the output current is not higher than 200 mA.

Always use a 100nf ceramics capacitor in parallel on both sides to suppress high frequency spikes.

The size of the input capacitor depends on the voltage drop between input (behind the rectifier bridge) and the output.

The smaller the voltage drop, the larger the capacitor has to be. If you look at a rectified sine wave you will understand why. The capacitor has to compensate for the time when the sine wave is reaching the crossover point. And do not forget the voltage drop of the diodes in the rectifier bridge.

A DC/DC converter does not make sense in most cases when you only need a little bit of current.
And they give you high frequency spikes, which is not good when you have a sensitive circuit.
I once had to replace a DC/DC converter in a signal generator to get clean signals.

And you can hardly make a mistake when you oversize capacitors on the input side. They just cost more and need more space.

All my recommendations are referred to an AC power supply (transformer plus rectifier bridge)
Powering devices from batteries is a different world.

ArduinoAleman

1 Like

weedpharma:
I was brought up on them and still think there is a place for them.

Switch mode may be much more efficient but take up more room. In the right place (low in/out V difference and low current) they are the simplest thing to use.

Weedpharma

Actually, this ones don't take up so much more space, if you can live with 0.5 A (version with 1A output exist, don't know if much bigger). Price is quite higher, too.

Cheers, Ale.

Oki Murata has some small units too, 7805 size
http://www.digikey.com/product-search/en/power-supplies-board-mount/dc-dc-converters/4325599?k=oki%20murata

CrossRoads:
Oki Murata has some small units too, 7805 size
http://www.digikey.com/product-search/en/power-supplies-board-mount/dc-dc-converters/4325599?k=oki%20murata

Yes, and both have the same 7805 pinout, so very easy to replace.

Cheers, Ale.

If your power input is a transformer with a rectifier bridge, you will get a rectified sine wave signal.

And you need a capacitor that is large enough to bridge the time when the 100/120 Hz rectified sine wave signal delivers only low or no current (the voltage goes from peak to zero Volts 100 or 120 times per second after a rectifying bridge).

So the sizing of the capacitors on the input side is about the same for a line regulator (78xx) and a DC/DC converter (depends on the internal voltage drop). When the output current is the same, you need enough capacity to guarantee that the input voltage of the regulator (any type) does not fall below the minimum value. The higher the output current, the larger the capacitor has to be.

Without enough power reserves on the input side the output voltage cannot be constant. And the reserves are stored in a large capacitor. (Read my recommendations in my mail above)

If your input is already a stabilized DC voltage, maybe a 100nf capacitor will be enough.
However, like I said before, when your input is AC, you will need large capacitors.

However, I see lots of example circuits all over the internet that use different capacitors.

There are two reasons for this:-

  1. Different manufacturers have different minimum requirements.
  2. As long as you exceed the minimum it is not too critical what value you use. A lot depends on the application, the load and the nature of it and what your specification is for ripple you can tolerate.

Thanks for all the info. Didn't realize it was this complicated. :slight_smile:

arduinoaleman:
If your input is already a stabilized DC voltage, maybe a 100nf capacitor will be enough.
However, like I said before, when your input is AC, you will need large capacitors.

Here is what I have in the project.

AC power in.

This power supply to provide 12V DC to the project.

Some parts need less than 12V though.

I have a servo motor that can use up to 7.4 V, and a DC motor that gets its speed controlled by varying the voltage to it from about 1.5 - 5V. At most it'll need 5V.

So I'm using this to drop the 12V to 7.0 - 7.4V for the servo.

Then I was using the L7805 regulator to drop that 7 - 7.4 V to 5V for the DC motor.

I would imagine that the 7.4 V delivered to the L7805 is stabilized pretty well.

Not much current demand; the DC motor draws about 150 mA.

I've been using a 100 uf electrolytic capacitor on the input side, and a 10 uf electrolytic on the output side. Seems to work fine. Do you see any reason to change them? I don't think the DC motor is particularly susceptible to noise, so I'm not sure how much I need to care about filtering. But I'd like to do it correctly.

A motor does not require a good regulated voltage.

As long as the maximum voltage is not succeded and the input is DC (required in your case) it will run fine.

I was thinking of a sensitive electronic device.

While a motor is not susceptible to noise it does generate one hell of a lot of it. Filtering the motor supply protects the rest of your circuit.