Anciliary caps for voltage regs

Gentlemen,

As a mere hobbyist, every time I need to use a 5V or 12V fixed voltage regulator such as the L78 series in TO-220 package I have to look up the data sheet to remind myself which size and type of cap goes across the input and ditto the output. What I'd like to know is, do you hot-shot designers also need to look these up or can you deduce the values and types needed from first principles? Here's an example (on page 5 of the datasheet linked to below) in which they show a 33:1 ratio between input and output but don't specify the dilectric of the caps in question for some reason (which seems like a serious omission to me but WTF do I know?)

formatting link

Reply to
Cursitor Doom
Loading thread data ...

78XX regulators are very forgiving as long as you use at least a microfarad or so right next to the input. The output is an NPN emitter, so is pretty low-Z at all frequencies of interest. You can make one oscillate fiercely if you use 10 nF on the input and feed it with long floppy wires. 79XX chips are much more like LDOs because the output is the collector of an NPN. With those, use a regular aluminum or tantalum cap on the output.

You size the output cap to get the transient load regulation you need.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

Thanks for the tip, Phil. I'll keep it in mind for next time I need to make a regulator oscillate fiercely. ;)

But seriously, you usually see these devices with a tantalum on one side and an electolytic on the other. I tried looking up to see why they often specify tants, but found some article which stated their higher ESR than other types of the same size was advantageous in this application, although that seems heavily counter-intuitive to me, I have to say!

Reply to
Cursitor Doom

You already got a good answer from Phil Hobbs. Here's my 2 cents: My own rule of thumb is to place a 78xx regulator close to the main filter - within a few inches of copper track at most - in which case there's no need for any other input capacitor. When that's not possible, I place an electrolytic cap of 10uF or higher close to the regulator.

Most manufacturers say that the output cap is not needed for stability but improves transient response. A 0.1uF ceramic cap works fine unless more is needed due to other considerations.

When I have to place the regulator on a heatsink away from the PCB, I solder one ceramic cap each from input and output to ground directly onto the leads.

Reply to
Pimpom

Isn't 0.33/.1=3.3?

The answer to your question is in figures 34 and 35. The larger input capacitor attenuates the ripple applied to the input thereby improving output transient response. It's usually a 10:1 ratio. They just don't want to scare people away with a 1uF.

Reply to
bloggs.fredbloggs.fred

There is no such thing as "first principles"; everyone designs their

3-terminal regulator differently, so some require very little bypassig, and others require a lot.
Reply to
Robert Baer

I don't think CD was talking about conservation of momentum or classical equipartition of energy. I think most of us probably start from rules of thumb to choose a capacitor type, search our in-house parts catalogues and/or Digikey, and then check the datasheets before choosing the exact part. Capacitor datasheets are horrible.

The parameters that matter are the capacitance and ESR. There's also ESL, but linear regulators are slow, so that rarely matters for loop stability. (It can matter a lot if there's some ugly switching supply upstream.) Time/temperature ratings also matter for wet electros.

Within each family, ESR generally decreases as the capacitance goes up. For a given capacitance and voltage rating, wet aluminum electros have the highest ESR, then tantalums. Multilayer ceramics have the lowest. Film caps and alpos are between tants and MLCCs. Alpos are an interesting case--you can choose quite a wide range of ESRs for the same value and voltage rating, which is frequently quite useful for stabilizing LDOs.

Most of the time I use ceramic output caps and put a zero-ohm jumper between the regulator output and the cap, as I posted in the "off-label uses" thread.

Regarding the horribleness of capacitor datasheets, high-valued MLCCs are the worst. You absolutely have to look at the _characteristics_ sheet (Samsung) or the online tool (Murata, Kemet, Taiyo Yuden, several others).

I have a blog post about this from a couple of years ago, most of which is still relevant:

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

It pretty much depends upon the frequency response of the load and the amou nt of ripple on the incoming dc. If you have a full-wave rectified input th e cap will have to supply the power between the half sine waves, which can be up to 8 msec (4 msec min) for a 60 Hz power source. Size the cap to prov ide this current, which is where the load instantaneous current demands com e into play.

Reply to
Flyguy

Oh, and of course MLCCs with high-k dielectrics can also be piezoelectric, which is loads of fun.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
 Click to see the full signature
Reply to
Phil Hobbs

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.