[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [N8VEM-S100:5179] Tantalum capacitor replacement?



On Tuesday, September 9, 2014 2:15:22 PM UTC-5, Don Caprio wrote:
I've wondered why most of the S-100 designs call for Tantalums for
the filter caps on the power source (LM7805). Is there a good reason to use tant's?

You'd have to ask the designer to know for sure.  Non-LDO regs don't care per the mfgrs.   I spot checked some data sheets specifically the ON 78 series and there isn't much discussion other than lower ESR generally being better (no surprise there), and the "best" cap belongs on the input or rephrased the input cap is more important to performance than the output cap.  Technically our designs with +/- 15 volts, if any, need reverse bias protection diodes as the 78 series can be hurt if Vout is more than 6 volts or so higher than Vin without that protection diode.  Not an issue with 5V regs that being a bit less than 6V under all conditions assuming you don't slam reverse voltage to the reg input LOL.

LDOs are a difficult beast not as agonizing as switching designs but not as easy as simple linears.  TI has a nice app note AN-1148 about LDO ESRs.  I just finished reading it and its interesting that some LDO regs but not all will break out in oscillation if the output has an ESR BELOW or above a certain ESR and it reminded me of the rule of thumb that about an ohm of ESR is about right for a LDO.  LDOs do make great oscillators, like a super high power tunnel diode.  And as the AN mentions, TI does make special secret sauce chips that have much better tolerance than most LDOs to unusual ESRs, etc etc.  Well, it is a TI app note, they're not going to compliment ON or Vishay for obvious reasons.

It might be a lifetime spec as explosion stories aside, electrolytics aren't spec'd to live as long as anything else including tantalums.  Its just that when electrolytics fail they generally do not detonate and blow a hole in the board, so its much less impressive.  Personally what I do is stick a very high quality ceramic on there and call it good plus or minus whats laying in the junk box (for analog work I'm a little more careful than digital).  For a LDO I'd check the app note for ESR first, because some ceramics are too LOW for stability, which is interesting.  Of course a little 1 ohm series resistor will fix that pretty well.

For a good time check some Kemet MLCC ceramic cap data sheets.  Their 0.1 uF MLCC caps are spec'd to about a tenth of an ohm ESR which isn't bad.  Ideally I use those.  They cost more than the stereotypical "penny decoupling cap" like maybe a whopping 50 cents, but you only need two per board at the regulator.  In the real world as opposed to ideally, I'll admit to just slapping on some random cap because its usually good enough.  I need to order another batch of those caps to have laying around.

I looked at the data sheets for some of those niobium caps, 50 miliohm ESR like what is that made with solid gold lead wires?  Either thats impressive or its a typo.  Actually too low ESR to keep some LDO's happy.