Very high capacitance before or after regulators ?

[Eva]

Any LC filter can potentially form a resonant system with very high output impedance at the resonant frequency. The capacitance, inductance, capacitor ESR and inductor DC resistance values have to be carefully chosen to form a damped 2nd order system.

 

[jean-paul]

The risk with 78xx/classic regs using very large capacitors ( likely with low ESR ) after the regs is that motorboating/oscillation very probably will occur. I have seen commercial machines with these design errors.

A 10 to 100 uF electrolytic cap can be used directly after the reg. The 68,000 uF could be used before the reg but I wonder if there will be a difference opposed to standard values *( calculated correctly that is ).

Modern low drop regs don't have these "problems".

 

[dantwomey]

On a similar note. I'm going to go all out and put a BG after my most important regulator. Do I use a polarized (i.e FK) or a non-polar (i.e. N)? All the factory schematics use polarized but a few of our forum members are recommending non-polar for the ultimate in post regulator decoupling.

Regards,
Dan 😀

 

[jean-paul]

You can use non polar BG ( N series ) electrolytic caps instead of polar electrolytic caps in whatever the application may be. Just do not do it the other way around 😉

More specifically, you can use N series BG after regs with good result. An expensive way though but when you need only one it won't hurt your wallet too much.

 

[AndrewT]

Hi,
be sure to read the datasheet.
There are some "no go" areas for capacitance and ESR on the output of regulators and it varies with load.

Unfortunately it is worse for low drop out (LDO) regulators.

 

[jean-paul]

Consulting the datasheet is rule nr. 1 so I won't even mention it anymore. From a PM from Dantwomey I understood he will use small value BG caps after the regs which won't cause problems. I could have been more specific.

Modern low drop regs are designed to withstand low ESR caps....

 

[CarlosT]

The National datasheets are very interesting...man...so many hidden pearls in these things 😀

This datasheet discusses how low ESR caps create problems

http://www.national.com/an/AN/AN-1148.pdf

Our friend from Turkey brought this up in an ancient CarlosFM PS thread in the chip amp subforum...Why are tantalum caps specified after regulators? The basic LM338 datasheet seems to support using small tantalum caps on the output

http://www.national.com/ds/LM/LM138.pdf

 

[jackinnj]

The National datasheets are very interesting...man...so many hidden pearls in these things 😀

This datasheet discusses how low ESR caps create problems

http://www.national.com/an/AN/AN-1148.pdf

Our friend from Turkey brought this up in an ancient CarlosFM PS thread in the chip amp subforum...Why are tantalum caps specified after regulators? The basic LM338 datasheet seems to support using small tantalum caps on the output

http://www.national.com/ds/LM/LM138.pdf

There's a somewhat parallel discussion going on about the LM317 -- -- from Bob Pease's "Trouble Shooting Analog Circuits" Appendix C, a low ESR capacitor like a polypropylene will have a higher Q than a higher tantalum cap, so there will be a more noticeable peaking around the resonant frequency of L and C (L being the "in fact" inductive impedance of the series reulator). In essence, you can tune the regulator to generate noise in the audio band if you aren't careful.

 

[Bernhard]

There's a somewhat parallel discussion going on about the LM317 -- --

Where ?

I just removed precious parts from a huge pcb where every small section had a pair of local LM317/37 regs with lots of Tantals...

 

[CarlosT]

Thanks, Jack...I'm a total noob as you know. However, the LM338 datasheet seemed to be pointing to using a lower value tantalum thereby keeping the noise out of the relevant range? I dunno...you know...I'm just an echo of knowledge 😀

Bernhard...precious PCB? You mean that tantalum is a semi-precious metal...right? 😀

 

[jean-paul]

There are people crossing red lines for it.....

http://www.un.int/drcongo/war/coltan.htm

 

[CarlosT]

Makes damn good warheads, too 😀

 

[TechGuy]

If your goal is to reduce ripple, you can either use a pi filter or a coupled inductor with the regulator. An inductor/capacitor filter is much better solution to reducing ripple if that is your objective. A main purpose of a coupling capacitor for handling transient loads. A very large capacitor electrolytic capacitor would not have any significant impact on transients since the ESR is too large and only a small part of the charge would ever be used unless there is a short or an overload condition.

The only advantage I see with a really large capacitor would be to provide power for a short period after a power loss or shutdown. It would be best to connect the large capacitor on the regulators input side so that regulator can be used to mantain a steady voltage while the capacitor discharges. Although you may need to include a blocking inductor or diode to prevent the capacitor from rapidly discharging into the shutdown power source.

The link below is an article on how to use coupled inductors to reduce or eliminate ripple.

http://www.hamill.co.uk/pdfs/azrtatad.pdf


I hope this is useful

 

[jackinnj]

and it varies with load.

Unfortunately it is worse for low drop out (LDO) regulators.

In an LDO the capacitor is in the feedback loop,

 

[Jack Bruce]

caps

I think the best thin to do with capacitors and regulators would be.
68000uF > 4800uF > .1uF > regulator > 1000uF > .1uF +
| | | | | |
-------------------------------------------------------------- ground
| | | | | |
68000uF > 4800uF > .1uF > regulator > 1000uF > .1uF -

 

[QSerraTico_Tico]

Any LC filter can potentially form a resonant system with very high output impedance at the resonant frequency. The capacitance, inductance, capacitor ESR and inductor DC resistance values have to be carefully chosen to form a damped 2nd order system.

Right, see here fig 4a.
http://www.maxim-ic.com/appnotes.cfm/appnote_number/883/