5
\$\begingroup\$

How does one go about choosing a capacitor for a LDO regulator which specifies a min. ESR? Capacitor datasheets that I've seen only contain the Max. ESR parameter. One could just choose a cap with a sufficiently high ESR and hope it never goes below the stability margin but that feels like cheating - besides, a unnecessarily high ESR would add ripple.

I know that nowadays LDOs exist that are stable with ceramic caps as they add their own 'zero' for stability but it would like to know how to work LDO's that don't do this.

\$\endgroup\$
6
\$\begingroup\$

The best way to deal with a LDO that requires a minimum ESR output cap is to use a different LDO.

These LDOs were designed when anything more than a few 100 nF was too big for ceramic or any other technology that has very little ESR. Tantalum was the usual choice for a few µF to a few 10s of µF. These had some inherent ESR, so LDOs were designed to be stable with them.

Somewhere around 2000 multi-layer ceramic caps became small, cheap, and available enough to replace tantalums in these applications. Newer LDOs were designed with these in mind, and are stable even if the output cap has 0 ESR. This latter arrangement is better, because the cap does it's job better the lower the ESR is. You want the LDO plus cap to look like as a low a impedance as possible to the rest of the circuit, and having some resistance in series with the cap defeats that to some extent.

If you really really need to use one of the ancient LDOs that require a minimum ESR, you can use a ceramic cap and add a resistor in series with it. You can try to find a tantalum cap that specifies the minimum ESR required by the LDO, but those are getting harder to find in the range ceramic can nowadays cover well.

\$\endgroup\$
4
\$\begingroup\$
  1. As Olin says, change LDO or add series R, but

  2. Why are you using tantalum caps?
    Tantalum caps are a disaster waiting to happen.
    They are an excellent component if you can guarantee a crucial requirement can be met. If you cannot then they are a maintenance hazard waiting to destroy your equipment.

    • Use only where you can GUARANTEE that applied voltages will NEVER exceed rated voltage even for the briefest of moments.

Tantalum capacitors have an oxide film grown on the tantalum metal to serve as an electrolytic barrier. This layer is very (very very) thin and can be punctured by voltages only slightly above rated voltage. If it is punctured it does not self heal. If there is enough energy to cause a minor thermal event then the capacitor will oblige and will self destruct with some or all of flame, smoke, smell, noise and explosion. I've seen one do all of these on the same occasion.

A very short voltage spike - even in the microsecond range - can start this process.

Instead, use solid Aluminum or ceramics.

Lots of material on solid Al caps.

Halt and catch fire !!! - a 16V tantalum run on 20V !!!

The photo below is "cheating" in that they used reverse polarity BUT you can get effects every bit as "good" just by using slight overvoltage

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Suppose, I'd like to use a 47uF cap for bulk capacitance for a few ICs, would you still recommend I go with ceramic/solid Al.? Even if the power is regulated? \$\endgroup\$ – Saad Feb 29 '12 at 14:16
  • 1
    \$\begingroup\$ Are you feeling lucky ? :-) If it doesn't matter if the tantalum melts down then use one. They usually end up short circuit so if the equipment is fused not much harm ensues, usually, if the tantalum capacitor is mounted safely, your insurance is paid up, you don't scare anyone to death if it explodes, your ... . ie why do it? Unless you want super long unattended lifetime even a std Al electrolytic may do. / Mounting tantalum away from LDO may provide enough resistance and inductance. Can you GUARANTEE a spike or surge free environment? \$\endgroup\$ – Russell McMahon Feb 29 '12 at 14:22
  • \$\begingroup\$ Thanks, Russell. I'll go with solid alum. from now on and use a more modern LDO. Thanks! \$\endgroup\$ – Saad Feb 29 '12 at 14:27
  • 1
    \$\begingroup\$ @Saad, 47 uF only needs five 0805 packages to do with ceramics (depending on WV). Reasons to go with electrolytic anyway would be extreme space constraints or a high-vibration operating environment. \$\endgroup\$ – The Photon Feb 29 '12 at 16:22
  • \$\begingroup\$ Actually the OP never said anything about tantalum. I mentioned tantalum, but only to explain how things used to be and why LDOs exist that expect some ESR of the output cap. Nice picture though. \$\endgroup\$ – Olin Lathrop Feb 29 '12 at 23:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.