Is it critical that I use tantalum capacitors with my voltage regulator?

I'm at the point where I need to select bypass capacitors for my design. And I'm planning to make a lot of these boards, so I'm trying to keep the assembly and parts cost down. I'd also prefer to use smaller package sizes.

The problem with Tantalum caps is besides their using larger package sizes than the ceramic caps, they're also more expensive, and if I have to use two capacitor types then I have to pay the $7 reeling fee for both and it seems like a waste to spend $40 on tantalum caps if I already had to purchase 10,000 ceramic caps of the same size.

Here's some examples of the prices I've calculated for different capacitors in various quantities:

Ceramic:

$40 for 50 47uf capacitors? EMK325BJ476MM-T
$21 (10K) 587-1227-2-ND .1uf 10V 0402
$27 (10K) 587-1451-2-ND .1uf 16V 0402
$16 (4K) 0603YC104KAT2A .1uf 16V 10% 0603
$20 (150) C2012Y5V1A106Z 10uf 10V 0805 or
$50 (2K)
$75 (50) C3216X5R1A107M 100uf 10V 20% 1206 or
$116 (100)

Tantalum:

$46 (250) CAP TANT 10UF 10V 20% 1206 - F931A106MAA (May be unsuitable for voltage regulator... One needs to be both on input and output and input > 10V)
$80 (500) CAP TANT 10UF 16V 20% 2312 - TAJC106M016RNJ or
$73 (250) (Not worth it!)
$55 (250) CAP TANT 10UF 16V 20% 1206 - T491A106M016AT

$31-$41 (100) Avg price for 10uf 16v in 1206 package

I've already changed the voltage regulator I was planning to use as a result of the prices I was getting for caps. I was going to use this one:
http://www.diodes.com/datasheets/AP1084.pdf

Which was the cheapest 3-5A regulator I could find, but it requires two 100uF capacitors and those are really expensive in an SMD package and probably even moreso in electrolytic. (Well, they're expensive when you're trying to get 50 boards made on the cheap, and you'd need 100 of them.)

So now I'm thinking about using this TI, or Micrel, but they both want 10uF ceramic caps:

http://www.micrel.com/_PDF/mic29150.pdf

But it will cost $38 more for 50 TI regs, and $57 more for 50 Micrel regs. So it almost makes it a wash between using the cheaper regulator with the expensive 100uF ceramic caps, or the more expensive regs that require the less expensive but still costly tantalum caps.

It might not seem worthwhile to worry about saving $40 on 50 PCBs but the cost of these boards is adding up fast, so I'd like to reduce costs where I can.

Oh, and another reason I'd like to avoid using the tantalums is I've heard they can catch fire with overvoltage and these boards will be used by people whom I don't trust not to do crazy stuff with them. Also they're polarized and I'm not sure how to specify on the PCB layout which way they go for the assembler.

Oh and just to share some stuff I've been reading in my own research into this.. I found a document from TI which indicates it's possible to put a resistor in series with a ceramic cap to replace a tantalum. But I've also found something from Micrel which indicates I should just cut the farads in half. Still other sites mention that with older voltage regs instability was a problem but modern ones have extra circuitry to help prevent this.

So I don't really know who to trust or whether these regulators I'm looking at count as modern.

scswift:
Oh and just to share some stuff I've been reading in my own research into this.. I found a document from TI which indicates it's possible to put a resistor in series with a ceramic cap to replace a tantalum.

It must've been that they were showing an ESR model (a 'perfect' cap with a resistance in series).

Just my opinion, but for power bypass application usage, using .1ufd values, I would think the lower cost ceramic caps are more then good enough over the more expensive and polarized tantalum caps.

Lefty

scswift:
I found a document from TI which indicates it's possible to put a resistor in series with a ceramic cap to replace a tantalum. But I've also found something from Micrel which indicates I should just cut the farads in half.

These are related. Tantalum has much higher ESR than a Ceramic. So TI is saying that their regulator is sensitive to the ESR of the cap. If it is too low, the regulator will be unstable.

The other comment is related, but different. Switching power supplies don't operate at DC, so the frequency response of the cap matters. The ceramic's low ESR gives it much more (relative) effective capacitance at higher frequencies than a tantalum. The voltage and temperature coefficients of ceramic shouldn't be overlooked. (The closer applied voltage is to rated voltage, the less effective capacitance.)

So now I'm thinking about using this TI, or Micrel, but they both want 10uF ceramic caps:

Are you sure? That's not what the datasheet from TI says: "The recommended load/decoupling capacitance is a 10uF
tantalum or a 50uF aluminum." It specifically mentions it wants a capacitance with high ESR. The only time ceramic is mentioned is on the output, in parallel with an electrolytic.

Micrel says: "in fact, extremely low ESR capacitors may contribute to instability. Tantalum capacitors are recommended..."

Neither seem to be suggesting a ceramic-only solution.

If you don't need to worry about extreme temperature ranges, Aluminum Electrolytics are a cheap alternative to Tantalum. Or as suggested, a small 1 or 2 ohm resistor in series with the ceramic.

If you do choose a Tantalum (or Electrolytic) make sure you pick a rated value at least 2X the high transient spike the cap will see.

Tantalum has much higher ESR than a Ceramic. So TI is saying that their regulator is sensitive to the ESR of the cap. If it is too low, the regulator will be unstable.

Here's the datasheet I was referencing in regards to using a resistor in series with a ceramic cap to replace a tantalum. It's not specific to this regulator:

So now I'm thinking about using this TI, or Micrel, but they both want 10uF ceramic caps:

Are you sure? That's not what the datasheet from TI says:

I misspoke. I meant to say they both want 10uF tantalum caps. The tantalum is what I'm trying to replace with ceramic.

"The recommended load/decoupling capacitance is a 10uF
tantalum or a 50uF aluminum." It specifically mentions it wants a capacitance with high ESR. The only time ceramic is mentioned is on the output, in parallel with an electrolytic.

Micrel says: "in fact, extremely low ESR capacitors may contribute to instability. Tantalum capacitors are recommended..."

Neither seem to be suggesting a ceramic-only solution.

Thanks for finding that, I'd missed it; I've been reading so many documents these past two weeks and taing pages and pages of note trying to keep all this stuff straight.

Yes, neither are suggesting ceramic only, and that's the problem. I want to use ceramic cause it's cheaper.

I guess I should look into electrolytic as that is an option, but this board needs to be tiny and thin. That's why I'm using surface mount components. All the electrolytcs I've seen are pretty large. And I'd prefer to stick with surface mount components. I think electrolytic caps come in surface mount varieties but I'm not certain.

Perhaps I should use the larger electrolytic caps though. I don't know if there will be problems driving 3-5A with only tiny ceramic caps on the regulator. Is there some kind of wattage rating like thing for caps which I should be paying attention to?

Oh, and before I forget, here is that document from Murata which indicates MLCC capactiros of 1/2 the farads can be used as replacements (Pg. 4):
http://www.digikey.com/Web%20Export/Supplier%20Content/Murata_490/PDF/Murata_TA_Replacement_Catalog_C-24-C.pdf?redirected=1

If you don't need to worry about extreme temperature ranges, Aluminum Electrolytics are a cheap alternative to Tantalum. Or as suggested, a small 1 or 2 ohm resistor in series with the ceramic.

The issues I have with the additional resistor is:

1. Takes up more space. (minor concern)
2. I don't know what wattage rating I need to use for it.
3. Adding another value of resistor to the board adds a lot more cost. Some PCB places charge extra for each unique part number. And there's a $7 reeling fee from digikey so it doesn't make sense to buy any less than a full reel of 10K. That means I'd probably end up spending $20 just to get 50-100 2 ohm resistors.

There is a possibility the PCB maker just has some onhand and can toss a few on there cheap but I don't know this for certain and I don't know what size they have, and I don't know how to specify the parameters for it like wattage and voltage without putting them on the stencil which is probably not gonna happen with parts this small.

If you do choose a Tantalum (or Electrolytic) make sure you pick a rated value at least 2X the high transient spike the cap will see.

You mean the voltage, right? How do I determine how high of a transient spike they might see?
I was gonna go with maybe 50v on the input voltage on the input and 10v-16v on the output because the regulators are rated up to 25v and output 5v. I don't actually expect anyone to run them at 25v as that would be a waste, but they might go as high as 12V, or whatever a 12V battery is at before you start discharging it. I think it's like 13-14v?

Okay, the datasheet specifies 50uF electrolytic, but the closest I could find was 47uF, which I assume is okay. I am curious about the ripple current ratings though, as they are different.

Here are the caps I found:
(Panasonic size D) http://search.digikey.com/us/en/products/EEE-1CA470WR/PCE3890DKR-ND/1806446
(Panasonic size E) http://search.digikey.com/us/en/products/EEE-1CA470SP/PCE3889DKR-ND/1806445

Here's the panasonic landing pad size datasheet, if you're curious about the sizes:
http://industrial.panasonic.com/www-data/pdf/ABA0000/ABA0000PE247.pdf

They're both have landing pads around .3" wide. A little big, but manageable if I only need one per board.

The datasheet seems to indicate that I need one on the output, but on the input any kind will do, so I guess I can use a ceramic cap for that?

But the ripple current concerns me. The output caps are probably okay, but I'm concerned about the input cap which will have a lot more voltage on it. I would like both inout and output to be able to handle up to 3-5A though, so maybe the ripple current rating needs to be really high on both? I'm not sure how to calculate it.

Ripple current on the output cap entirely depends on the load.

For input cap it depends on the nature of the power source... Typically the power source would have enough of the right kind of decoupling anyway...

For the output cap requirements, I note the AP1084 datasheet says "A minimum of 10uF (0.15? ? ESR ? 20?) capacitor must be connected from this pin to ground to insure stability." (I think they meant "ensure" not "insure"!) - so you need to check the ESR value of your output cap and add series resistance if necessary (between the rail and the cap, not inline with the rail!). However the load circuit itself is likely to have significant decoupling capacitance which might be low ESR - some testing is probably required to satisfy your self all is well.

If you don't need a LDO regulator then you'll find other regulators that aren't so sensitive to low output cap ESR I believe.

When I worked at one place we were banned from using tantalum capacitors in power supply circuits. This is because when a tant cap fails it fails short circuit and that is a fire risk. If you ever want to get a circuit UL approved they will not allow the use of this sort of capacitor in this situation.
So I would say it is critical you don't use one.

Poor Tantalum. Always picked on by those who don't understand him.

Grumpy_Mike:
This is because when a tant cap fails it fails short circuit and that is a fire risk.

All caps can fail short. In fact the most common failure for a ceramic is flex crack which leads to direct short. (My college calls them LECs: light-emitting caps.) That's why ceramics come in "Open-Mode" configurations and Flexible-type terminations. It is one reason why aluminum electrolytics have safety vents (or stress points designed to blow.) So tantalum isn't the only cap that can fail short and it isn't the only one that can ignite.

It is true that a MnO2-based Tantalum can flash. However, when properly de-rated (at least 50% of the largest voltage transient) the chance of failure is almost zero. The failure occurs because of weaknesses in the dielectric which are created when the parts are subjected to reflow oven temperatures.

Polymer-based Tantalums have a benign (no ignition) failure mode as their cathode contains no oxygen. Of course their low-ESR probably makes them a bad choice for this application.

Grumpy_Mike:
If you ever want to get a circuit UL approved they will not allow the use of this sort of capacitor in this situation.

I have never heard this to be the case. Proper testing after the board has reflowed can eliminate any safety hazard.

Tantalums don't fail over time. They only fail the first time the dielectric is subjected to a voltage it cannot withstand. If after reflow the maximum voltage the cap will see is applied, and it does not fail, it is extremely unlikely to fail during its life: the same as any other component.

Note: I work for a capacitor company.

All caps can fail short.

True but a tant will always fail short.

So tantalum isn't the only cap that can fail short and it isn't the only one that can ignite.

I know but it is so much more likely with a tant than any other type.

I have never heard this to be the case.

Ever tried to get anything through UL?

Tantalums don't fail over time.

Yes they do. All components fail over time, that is what the Bath Tub Curve is all about.

it is extremely unlikely to fail during its life

So it won't fail until it fails, great.

the chance of failure is almost zero.

That will be an electronic engineering first then.

Is it fair to say that, after the initial test and burn-in, a tantalum capacitor has an expected life that's more than double that of an aluminum electrolytic, assuming the device is operated within "normal" temperature and load ranges, and maybe even longer than the expected life of many other components (such as surface mount ICs with thermal flex, or PTC fuses, for example)?
If so, then I think the statements by James C4S are a fair approximation of reality.

Also, after teaching myself a little bit of power supply stuff by reading data sheets and other references, I've come to realize why most products just delegate the power regulation to the wall wart. If you need 5V regulated input, make an input for 5V regulated DC, and don't worry about it in your circuit. Let the wall wart regulate and filter and provide your 5V.

jwatte:
If you need 5V regulated input, make an input for 5V regulated DC, and don't worry about it in your circuit. Let the wall wart regulate and filter and provide your 5V.

The problem is that what you get in terms of reguation from a standard "regulated" wall wart may be less than desired. This may not be much of an issue with some circuits, but for accurate sensor readings it typically is. Same goes for "drop-in-replacement" switching regulators. What you get is typically what you pay for in terms of regulation/ripple. There is a reason why linear regulators are still in demand, they perform a desired function and they do it well.

jwatte:
Is it fair to say that, after the initial test and burn-in, a tantalum capacitor has an expected life that's more than double that of an aluminum electrolytic, assuming the device is operated within "normal" temperature and load ranges

That's the irony of Tantalum. If it will survive turn-on, it will outlive most of the other components. That's where understanding had to occur. If a Tantalum fails after turn-on, it is because it was subjected to a transient it had never seen before and couldn't heal itself at that voltage. (e.g. 5V rail spikes to 5.1V... generally not that minor but that's all it takes)

The expected life of a Tantalum versus Wet Aluminum Electrolytic is significantly more than double.

Comparing wet and solid electrolytics (and lifetime):
Wet Electrolytics wear themselves out in both usage and storage. In storage the electrolytic eats away at the oxide (dielectric). If it has sat for a long period of time unpowered it can fail catastrophically at turn-on. Like Tantalum, if it survives that initial surge then the oxide will re-form (self-heal). Again, wets need to be unpowered for this type of wear-out.

The trade off is that the electrolyte is consumed when that oxide is reformed. Eventually, the cap won't be able to self-heal anymore. You'll see loss of cap, higher ESR over time. These parametric changes are referred to as end-of-life. Modern aluminum electrolytics are, generally, only rated for a 10,000 hours of operation (or so). Nothing like the caps from 30 or 40 years ago, with oil based electrolytes.

Solid Electrolytics (Aluminum or Tantalum) don't have the same wear-out. When on the shelf, the dielectric is unaffected. When powered, the dielectric does the same kind of self-healing as wets. However, once a portion of the dielectric is self-healed, it never needs to be healed again. Whereas a wet will probably need that portion re-healed if it sits again for a long period of time.

This ability to self-heal is a strength of Solid-Aluminum and Tantalum capacitors. However, it is also the weakness of Tantalum. If the dielectrics weaknesses can't heal properly, that's when the capacitor fails. This is why initial turn-on is so critical. The Tantalums that survive will have healed their weaknesses.

BenF:
The problem is that what you get in terms of reguation from a standard "regulated" wall wart may be less than desired.

So specify / use / include a wall wart that actually does the right thing! The right thing might actually be a switching regulator that outputs 5.7V, followed by a LDO regulator that outputs 5.0
In the original post, the problem was building the regulator onto the circuit board. But, if that circuit board needs power anyway, then the regulator isn't actually all that necessary on the circuit board. Using micro-USB only for power is one way of doing it -- USB power is generally well regulated, and the power connector that fits will only generate one voltage, and thus there's no risk of plugging the "wrong" wall wart in (which would otherwise be a problem if you don't have your own regulator).

Note: I work for a capacitor company.

I'm glad to see that you are now a 'Faraday Member'.

Don