Why are specific types of capacitors called for in terms of buffering an LDO?

[jnz]

I have an LDO I'd like to use. It calls for 10uF on the input and 22uF on the output. I have the output covered, but on the input side I can't fit the electrolytic it calls for. I have a height limitation of 3.8mm, there are no electrolytic in that 10uF and 40V+ range.

IIRC electrolytic suck at high frequencies, but I don't remember if a ceramic, a tantalum, MLCC, could effectively replace an electrolytic or not for input buffering (assume you met the voltage, ESR, and capacitance requirements)?  Does my datasheet say to us a 10uF aluminum electrolytic because it usually makes the most sense to do so? How much freedom do I have for low frequency input buffering?

[spec]

+ jnz

It would not be possible to answer your OP without knowing what LDO voltage regulator chip you intend to use. LDOs are normally sensitive to the input and output capacitor characteristics. Some LDO voltage regulators even go unstable with too lower ESR!

[jnz]

+ jnz

It would not be possible to answer your OP without knowing what LDO voltage regulator chip you intend to use. LDOs are normally sensitive to the input and output capacitor characteristics. Some LDO voltage regulators even go unstable with too lower ESR!

Ignoring ESR for a moment. I can make that work. Lets say all the options have the same 5mOhm ESR.

Is there a reason an MLCC or Tant or Ceramic all of equal values buffer the input to an LDO like an electrolytic would?

edit: I'm interested in the question right there because I'm finding it extremely difficult to find a 40V+ rated 10uF that fits in my height requirements. If I can looked at other styles it would help. Unless someone has a 3mm tall 10uF electrolytic that's cheap!

[T3sl4co1l]

Does it have to be a picky LDO?  Are you really getting 39.5V out from a 40.0V input?

Use a HDO like the old fashioned LM317 or 7805 or what have you (it'll have to be an -HV version most likely, but those exist, too) and you're fine.

Tim

[jnz]

Does it have to be a picky LDO?  Are you really getting 39.5V out from a 40.0V input?

Use a HDO like the old fashioned LM317 or 7805 or what have you (it'll have to be an -HV version most likely, but those exist, too) and you're fine.

Tim

Not really. It's 40V LDO, 12V automotive, 30-35V claming TVS, so I can get away with a 35V electrolytic. Nichicon makes a 3.95mm which is a hair too tall (literally, about a human hair). I MIGHT be able to lower the clamping on the TVS and use a lower rated cap, but this just risks blowig the TVS.

That said, I'm not planning on this thing being plugged in when jumping, or even driving. It's a test tool for another device. So I may be planning on voltage spikes it'll never see.

I could look into a HDO, but it's nice if this thing keeps working down to 3V like it's supposed to now. Goes with another question I posted here about having enough time to write to EEPROM if someone unplugged it when they shouldn't.

So for the space issues I have, it would really help out if I could replace the electrolytic with a ceramic just to buffer the LDO on either/both sides. But can't seem to get that answer.

[tooki]

So for the space issues I have, it would really help out if I could replace the electrolytic with a ceramic just to buffer the LDO on either/both sides. But can't seem to get that answer.

You're not getting the answer because you chose to ignore the critical question in the very first reply: which LDO?

As was explained to you in that reply, different LDOs have different requirements, so it's not possible to provide a general answer.

[mariush]

You could have a cutout in the circuit board and have the body of the electrolytic inside the cutout (and lay it flat like a quartz / crystal oscillator thing)
You could also have two capacitors in series to get your higher voltage by using capacitors rated for less voltage (like 25v)

[MiDi]

Possibly manufacturer can give you advice if other caps would fit.
Or thrown in an mlcc or tantalum and give it a smd series resistance to fit the esr of elko, but with risk that it may not work as expected.
Another option is to drill a hole with diameter of cap and mount through pcb.

[MiDi]

You could also have two capacitors in series to get your higher voltage by using capacitors rated for less voltage (like 25v)

You have to be careful with caps in series, different leakages will rise the voltage on minor leaky one!

[T3sl4co1l]

Not really. It's 40V LDO, 12V automotive, 30-35V claming TVS, so I can get away with a 35V electrolytic. Nichicon makes a 3.95mm which is a hair too tall (literally, about a human hair). I MIGHT be able to lower the clamping on the TVS and use a lower rated cap, but this just risks blowig the TVS.

Is that all you're worried about?  A 25V electrolytic will handle that fine.  It won't even get noticeably hot or bulged during such a brief event.

Don't do the same with polymers -- they'll fail shorted.  (Apparently OS-CONs can self-heal, but I wouldn't count on it.)

Ditto tantalum, of course, with more fireworks.

I could look into a HDO, but it's nice if this thing keeps working down to 3V like it's supposed to now. Goes with another question I posted here about having enough time to write to EEPROM if someone unplugged it when they shouldn't.

Draining every last volt from your reservoir isn't going to save you an order of magnitude on capacitance required for a given hold-up time.
https://www.seventransistorlabs.com/Calc/PSHoldUp.html
For the reg, use the CC case.  For a switcher, use the CP case.

If you need milliseconds and milliamperes, you'll likely need hundreds of uF.  If you can't fit it, and can't figure out some other mitigation... that's just how it is.

Or maybe you can squeeze a thin pouch cell into that enclosure, and have minutes to hours of hold-up time. [emoji848]

So for the space issues I have, it would really help out if I could replace the electrolytic with a ceramic just to buffer the LDO on either/both sides. But can't seem to get that answer.

Oh, no problem, you can put resistors in series with 'em, don't have to be stuck with the monstrously low ESR of a naked ceramic cap after all...

Tim

[GeoffreyF]

Try different capacitors, watch the ripple with an oscilloscope.  Experiment. Don't be afraid to fry something.

[spec]

Ignoring ESR for a moment. I can make that work. Lets say all the options have the same 5mOhm ESR.

Is there a reason an MLCC or Tant or Ceramic all of equal values buffer the input to an LDO like an electrolytic would?

edit: I'm interested in the question right there because I'm finding it extremely difficult to find a 40V+ rated 10uF that fits in my height requirements. If I can looked at other styles it would help. Unless someone has a 3mm tall 10uF electrolytic that's cheap!

OK, as you have the capacitance value and the ESR values established for your design, here are some points about the two capacitor types that you mentioned as possible replacements for aluminum electrolytic capacitors.

Tantalum (solid)
Tantalum capacitors would make good replacements as far as capacitance and ESR are concerned. But there is one problem. Tantalum capacitors should not be used where the current is not limited to a low value, or tantalum capacitors may fail and sick their guts all over the place- that is the general view anyway. This means that tantalum capacitors would not be suitable for your application.

Ceramic
Ceramic capacitor plus points
p1) high capacitance/volume ratio
p2) low ESR
p3) long life (practically infinite, unlike aluminum electrolytics)

Ceramic capacitor minus points.
m1) capacitance very dependent on applied voltage
m2) capacitance dependent on temperature
m3) microphonic

The degree of the above minus points is highly dependent on the dialectic material of ceramic capacitors.
To short a long explanation, the only dialectics worth considering for your application is X5R or preferably X7R.

There is another catch. In general, the smaller the ceramic capacitor, the worse are minus points, m1 and m2.

You should work out all these parameters from the capacitor data sheet but, as a general rule of thumb, whatever minimum capacitance you want, chose a ceramic capacitor of twice that value, with a maximum of 15% tolerance and X7R dialectic.  Furthermore, chose a ceramic capacitor to be as big as possible (not good for your application) and as higher voltage rating as possible.

There is a lot of information about ceramic capacitor characteristics from the manufacturers, but to get a quick feel for ceramic capacitors have a look at page 15 of the data sheet for the LT1763 series, LDO voltage regulators:

https://www.analog.com/media/en/technical-documentation/data-sheets/1763fh.pdf

Don't let all the above put you off ceramic capacitors. They are amazing components, but you need to do a worst-case budget to ensure that a particular ceramic capacitor meets all the requirements of your application, and under all operating conditions. But, if your application needs low distortion, be very wary of ceramic capacitors!

[jnz]

Not really. It's 40V LDO, 12V automotive, 30-35V claming TVS, so I can get away with a 35V electrolytic. Nichicon makes a 3.95mm which is a hair too tall (literally, about a human hair). I MIGHT be able to lower the clamping on the TVS and use a lower rated cap, but this just risks blowig the TVS.

Is that all you're worried about?  A 25V electrolytic will handle that fine.  It won't even get noticeably hot or bulged during such a brief event.

Don't do the same with polymers -- they'll fail shorted.  (Apparently OS-CONs can self-heal, but I wouldn't count on it.)

Ditto tantalum, of course, with more fireworks.

How can you be sure a 25V cap will hold up to automotive transients? My ESD should clamp at 32V or so, but the electrolytics I've seen that even specify surge/spikes usually say something like 1.15 * Rated Voltage.

Noted on polymer and tantalum. Yea, on tantalum I've decided that's at best on the 5V side of things if I use one at all.

I guess overall I'm trying to find out if MLCC / Ceramic / Tantalum / Polymer have the same low frequency characteristics as aluminum electrolytic - assuming apples to apples capacitance. For instance, if I have to buy a 25V 22uF MLCC to get 10uF actual and I can make the ESR work, that's fine, but the LDO mfg's specifically call out an electrolytic for a reason I would think.

[jnz]

Try different capacitors, watch the ripple with an oscilloscope.  Experiment. Don't be afraid to fry something.

Nice idea, but I'm up on a deadline, will only have one shot at a prototype board, and space is an issue so applying many empty pads for different package types would be an iussue. There is a chance I'll be able to fit say a 1210 inside a 2917 for example so I can try those big polymer electrolytics and a handful of MLCCs and Tantalum at the same time. Not ideal though.

[T3sl4co1l]

How can you be sure a 25V cap will hold up to automotive transients?

'Cuz I've done it.

And there's all those times people have plugged 120V equipment into 240V mains, and the caps get hot but they don't bulge or vent for quite some time.  I've even heard of e.g. 200V capacitors being reformed to operate at 350V or more (not that they'll necessarily meet any of their specs after such treatment).

Electrolytics are kind of like vacuum tubes: old, reliable (heh, well, "reliably unreliable" would perhaps be a better way to put it), able to take abuse for seconds, minutes even at a time.  Just don't make a habit of it, and the overall lifetime won't be seriously impacted.

If you need design assurance -- no, I don't think you're going to find that in a datasheet or appnote anywhere, and the safe thing is to follow the datasheet as far as possible, and get manufacturer approval where the datasheet comes up lacking.

Hmm, on that note, this at least hints at it;
https://cn.tdk-electronics.tdk.com/download/540988/5f33d2619fa73419e2a4af562122e90c/pdf-generaltechnicalinformation.pdf
but the graph isn't to scale so I wouldn't draw any conclusions from it in regards to surge or transient rating.  They do say they can provide transient ratings as needed, which may be a custom part request, so if you need a million pieces...  Never hurts to ask if they have standard parts carrying such a rating, though.

Tim

[jnz]

How can you be sure a 25V cap will hold up to automotive transients?

'Cuz I've done it.

And there's all those times people have plugged 120V equipment into 240V mains, and the caps get hot but they don't bulge or vent for quite some time.  I've even heard of e.g. 200V capacitors being reformed to operate at 350V or more (not that they'll necessarily meet any of their specs after such treatment).

Electrolytics are kind of like vacuum tubes: old, reliable (heh, well, "reliably unreliable" would perhaps be a better way to put it), able to take abuse for seconds, minutes even at a time.  Just don't make a habit of it, and the overall lifetime won't be seriously impacted.

If you need design assurance -- no, I don't think you're going to find that in a datasheet or appnote anywhere, and the safe thing is to follow the datasheet as far as possible, and get manufacturer approval where the datasheet comes up lacking.

Hmm, on that note, this at least hints at it;
https://cn.tdk-electronics.tdk.com/download/540988/5f33d2619fa73419e2a4af562122e90c/pdf-generaltechnicalinformation.pdf
but the graph isn't to scale so I wouldn't draw any conclusions from it in regards to surge or transient rating.  They do say they can provide transient ratings as needed, which may be a custom part request, so if you need a million pieces...  Never hurts to ask if they have standard parts carrying such a rating, though.
Tim

Ok, that's fine, and it's what I figured. There is a lot of experience that isn't in datasheets. I figure that part of the reason electrolytic are bad at high frequency stuff is probably why they seem to be fairly good at resisting transients.