Replace electrolytic for ceramic caps

[JanJansen]

Hi, can i replace all electrolytic for ceramic caps ?, not for recapping, for new designs i mean.
Those electrolytic need replacement and going to leak after many years, is it better to use 10uF multilayer ceramic where possible ?,
they both cost 4 cent.



They keep good forever ?, or is this ebay quality not good ?, can there be quality problems with such parts ?, or are they always good ? ( i,m not willing to pay 50 cent for 1 10uF condensator )
And is it better to have ceramic for decoupling, or what is the big difference ?

For audio going thru : i will use low ESR Panasonic 22uF ( 6 cent ), no ceramics ( i heard quality is bad for audio with ceramics ).

So what do you think ?

[madires]

Which type to use depends on the purpose and the requirements. Just for bypassing multilayer ceramics should be fine. But watch out for the ESR. I would buy components from trusted sources, not ebay, because it's nice when the component matches the datasheet. For coupling audio signals a standard electrolytic would be sufficient, or use a film cap. 22µF seems quite high. Special requirements for passing very low frequencies or do you got an unsusual input impedance?

[CraigHB]

Electrolytic and ceramic caps have different characteristics and one may or may not work well in place of the other.  There are various types of ceramic caps.  The ones shown the photo appear to be the MLCC type.  MLCC caps are made in a surface mount package most commonly though they can be found in leaded packages as shown in the photo.  Also there are some big differences between electrolytics themselves such as the standard aluminum can type versus the organic aluminum polymer type.   Let me assume you're talking about the standard aluminum can electrolytic caps that have been around as long as the mountains versus the common MLCC type.

Personally I like to use MLCC caps in place of electrolytics where possible since they are just much tougher components and use less PCB real estate.  MLCCs have higher charge capacity by volume than electrolytics, in other words they have more capacity for their size.  You can save a lot of PCB space using MLCCs.  They pretty much last forever and can even handle voltage and ripple current over tolerance for limited periods where electrolytics can not.  Of course the lack of polarization is a big advantage since it allows them to be used for both AC and DC.

MLCCs typically have about a tenth the ESR and ESL of electrolytic caps (equivalent series resistance and equivalent series inductance).  Depending on the application these factors may or may not be a big consideration.  For power filtering applications it usually is, the advantage being held by the MLCC types.  For signal filtering it usually is not unless frequencies are high were ESL becomes a factor.

Having lower ESR, MLCCs can usually handle higher ripple currents than electrolytics.  It's typically a pretty good amount so that gives them a bigger advantage in power filtering applications where ripple currents can be pretty high.  Not only that, but they have much better longevity.  You don't see many MTBF ratings for MLCCs because they pretty much last forever.  Electrolytics usually have a lifetime rating because it can be rather limited.  When electrolytics fail it can be somewhat catastrophic. In some cases they can actually explode causing damage elsewhere.

It sounds like there would be no reason to use electrolytics over MLCCs, but electrolytics do have their advantages.  MLCCs get cost prohibitive when capacitance and voltage ratings get high.  They have a limit of about 100uF in capacitance value and you don't see them much over 100V tolerance.  Typically the higher the voltage rating the lower the limit on available capacitance.  It's a pretty significant difference in cost so it's not a small consideration.

MLCCs have a characteristic called DC bias which other types of caps don't suffer from.  When subjected to higher DC voltages, capacitance falls off.  It can be quite significant.  For exampe a 16V rated MLCC cap may only maintain ten percent of its rated capacitance at maximal DC voltage.  You can usually find the DC bias curve for an MLCC cap from the manufacturer web site, but it can be a rather elusive specification. For AC applications the DC Bias characteristic is usually not a problem, but it almost always is when used for power filtering.

There are other considerations I didn't mention, but those are the biggest ones.  Using arrays of several MLCC caps in parallel is a way to get around some of their limitations and it's a commonly employed technique.  Also the COG dielectric does not suffer from DC bias, but it's very limited in maximal capacitance.  Otherwise there are situations where an electrolytic is going to be the most cost effective and performance effective solution.  If one were to solve the DC Bias problem with MLCCs they would be the superior solution in most cases, but that's beyond the state of the art at this point.

[ajb]

MLCCs have a characteristic called DC bias which other types of caps don't suffer from.  When subjected to higher DC voltages, capacitance falls off.  It can be quite significant.  For exampe a 16V rated MLCC cap may only maintain ten percent of it's rated capacitance at maximal DC voltage.  You can usually find the DC bias curve for an MLCC cap from the manufacturer web site, but it can be a rather elusive specification. For AC applications the DC Bias characteristic is usually not a problem, but it almost always is when used for power filtering.

It's worth noting that this effect varies quite a bit with dielectric type, and even between different capacitors of the "same" dielectric.  X7Rs tend to be much better than Y5Vs, but some X7Rs are better than others.  The effect can be compensated by choosing a capacitor with a higher voltage rating, though of course that may mean a larger package.  C0G/NP0 are the most stable, but have the lowest capacitance per volume.

[CraigHB]

There was a time where I had to use care in capacitor selection for chips like regulators and converter controllers, but I'm finding most newer chips support the MLCC type now.  That's good because MLCCs are much better to use when they can fit the bill.

It seems most of the projects I do are either size sensitive or weight sensitive or both.  MLCCs can make a big difference there.  I'm doing a BLDC motor controller design right now where I really wanted to get away from electrolytics mainly because of their tendency to fail and also to reduce footprint.  Having much lower ESR you can get away with less capacitance, but only to a point.  I need a rating of 100V.  To do that I have to use a crazy looking 20 count arrray of 2.2uF electrolytics in a 1206 (3216 metric) package.  The package is large since the size of the cap mitigates the DC bias, but even so it's still smaller and lighter than the electrolytics I would use in their place.  The main advantage in my application is the more robust nature of MLCCs. 

What I've found is you really need to pay attention to those DC bias curves.  They can vary quite a bit from one MLCC to the next.  There is an amount of trade-off they do in producing them, in other words you can trade off ESR for DC Bias to a small extent.  Though in cases at the higher end such as 50V compared to 100V, rating does not always reduce DC Bias.  I've seen curves for parts with different voltage ratings where the shape is not much different other than the curve extending further out.  It looks more like the curve was truncated at a lower voltage rather than bias reduced at higher voltages. 

I've found DC Bias to be one of the more convoluted things to evaluate in MLCC capacitor selection.  You really have pay attention to that curve and it's not something commonly published by manufacturers.  Sometimes I have to disqualify a make because I simply can not find a DC bias curve for their product line.

[JanJansen]

Hi Graig, thanks for your big posts.

I have a question about the DC bias :
how do you notice about this ?, i normally use 12 volt to flatten with the cap, and 6 volt for half voltage ( opamp center point ).
Is there no problem for me ?, how does DC bias works ?, is it some short circuit ?

Oh, and by the way, if i use a 10uF ceramic, do i still need to add a 100nF ceramic ?, do i need that frequency or something ?,
or can i replace the 10uF electrolythic cap + 100nF ceramic with the 10uF ceramic ?

And i can not find bigger then 10uF for trough hole, those 100uF i find SMD only.

Anyways thanks for all the reply`s, i will use those MLCC in the future,
one Panasonic 1500uF ( 10000 hour ) first, then the rest 10uF MLCC`s.

[JanJansen]

22µF seems quite high. Special requirements for passing very low frequencies or do you got an unsusual input impedance?

I like good bass, with 22uF there is still sub bass < 50Hz, and they are the cheapest panasonic`s.
Normally i took 10uF, now i upgrade my qualty with low ESR panasonics, 22uF where just cheaper.
I like to be sure the cap will not influence the audio frequency, so 22uF is even better then 10uF.

Because i use single supply i gotto have capacitors in the audio signal, i look at the scope to keep a square wave really flat,
sometimes i need a very big cap, sometimes not, it might have to do with the impedance, only i am no schooled electronic person so i dont know about that,
just testing some sounds to solder for making music with.

[suicidaleggroll]

I have a question about the DC bias :
how do you notice about this ?, i normally use 12 volt to flatten with the cap, and 6 volt for half voltage ( opamp center point ).
Is there no problem for me ?, how does DC bias works ?, is it some short circuit ?

The effect of DC bias is to decrease the effective capacitance of the cap as the voltage across it increases.  Say you have a 10uF cap.  When it's completely discharged it's 10uF.  If you charge it up by a volt, it might now be 9uF.  Charge it up to 2 volts and it's now 8uF, and so on, all the way to the point where the effective capacitance of an MLCC that's charged up to its rated voltage might only be 10% of nominal (aka: your 10uF cap now only has 1uF of capacitance because you've charged it up by 16 volts).  This is a big part of why they're so bad in audio applications (in the signal path).  Since the capacitance changes with voltage, you end up with some nasty non-linear effects when you feed a sine wave through it.

http://www.murata.com/en-us/support/faqs/products/capacitor/mlcc/char/0005

As far as I know, the voltage rating of the cap has no impact on the effect of DC bias.  The two things that matter are dielectric type and package size.  Larger packages suffer a smaller DC bias effect, and some dielectrics (eg: C0G/NP0) have no DC bias effect.

[JanJansen]

Wow, that is a big difference, having 1uF left, or are these values just for the example ?
And can you measure this while in use ? ( i dont have capacity meter yet to bad )

Having much lower ESR you can get away with less capacitance, but only to a point.

Does it compensates enought for the loss ?

[suicidaleggroll]

Wow, that is a big difference, having 1uF left, or are these values just for the example ?

Those values are realistic for small packages.  It's less severe for larger packages, but you need to check datasheets of parts you're interested in if you want numbers you can rely on.

[JanJansen]

Ebay dont has any datasheets, they are rated for 50 volt.

[suicidaleggroll]

If you want specs you can rely on, don't buy on ebay.

[madires]

Because i use single supply i gotto have capacitors in the audio signal, i look at the scope to keep a square wave really flat,
sometimes i need a very big cap, sometimes not, it might have to do with the impedance, only i am no schooled electronic person so i dont know about that,
just testing some sounds to solder for making music with.

The -3 dB cut-off frequency for a RC low-pass filter is f = 1 / (2 pi R C). C is the coupling capacitor and R the input impedance of your amp.

[madires]

Ebay dont has any datasheets, they are rated for 50 volt.

And no idea of the dielectric, the vendor, the ESR, the capacitance derating for DC bias and so on.

[DaJMasta]

If you want specs you can rely on, don't buy on ebay.

A good general rule.  It's great for cheap stuff and experimental stuff, but if you're not getting a data sheet or a reputable manufacturer, you have no guarantee of quality control.


As for the cap value, it's almost entirely dependent on the impedance of the load.  For example, if you're AC coupling an audio input into an opamp amplifier with an impedance of at least a few K, even a 1uF cap would be plenty, but if you're driving a low impedance speaker or something, you could easily need 100-10000uF worth of capacitance to get good low frequency performance.  If you know you're application, you can pick an appropriate cap.  A lot of times electrolytics are sufficient or even required because of how high the value has to be.  If small values are fine, a ceramic will probably do a decent job, but may not give you the tip top performance of some others, and the relative quality of the ceramic for your application can differ a fair bit with the dielectric, which probably isn't stated on your ebay caps.

[JanJansen]

I still have this question :
Do you need a 100nF ceramic cap together with the 10uF cap agains hi frequency noises/ripples ?
Normally with normal elkos, you need the 100nF, its about in every schematic.
Does the 10uF MLCC replace those both ?

And for less then 10 % of the original price i will try them without datasheet,
did anybody have the same caps, and measured the quality ?, it must be attractive to buy you have to admit that.

I mean : what could go wrong with making such a simple part ?
I also going to try ebay for film capacitors, since they are better for audio,
or dont you buy these also ?, can they be bad for your audio ?, even if they are film caps ?

[madires]

If you buy via eBay you'll get whatever is most cheap and available in Shenzhen. Tomorrow's caps could be different from yesterday's. You can't even expect that they are 50V types. What do you think why they are so cheap?

[suicidaleggroll]

And for less then 10 % of the original price i will try them without datasheet,

Not 10% of the original price, 10% of the actual price for that part.  You need to ask yourself why they cost 10% of what everyone else charges for the same part?  Is it really a good deal, or maybe they're not what they say they are?  When buying from ebay it's usually the latter.

did anybody have the same caps, and measured the quality ?, it must be attractive to buy you have to admit that.

It doesn't matter what anybody else has measured, because the parts you get will not be the same.  I will not admit they're attractive, because they're not.  When you look at the real cost of buying no-name parts from ebay it usually ends up being higher than the real thing, once you factor in the hours you need to spend testing them to see what their real specs are and how long they'll last, including replacement cost when they fail.  My evenings are worth more to me than the $1 I'd save by buying no-name ceramics from ebay instead of a legitimate distributor.

[Siwastaja]

I'm surprised of the always touted notion of Ebay parts being "cheap". Usually, in my experience, they are horribly expensive, usually much more than the proper parts at the proper distributor. For example, proper capacitors typically cost something like a few cents each. You are not saving anything on the components. Of course, the real distributors often ask for postage fees exceeding those of Ebay sellers, or have minimum order limits for free shipping.

Also, it's much worse than not having proper or unknown specifications. Usually, they are outright inoperable scams. Capacitors may not be capacitors at all. Just don't buy them.

So, any price comparison is totally meaningless when you are not getting the product you are buying!

I can sell you anything you want even cheaper, just send me your money! I won't send you anything, but come on, it's cheaper!

[JanJansen]

You need to ask yourself why they cost 10% of what everyone else charges for the same part?

I dont know the answer, i bought them to replace the tantalum capacitor that goes with the DSPIC chip, they dont explode.
Why would anyone buy a tantalum ?, because they are cheaper then the ceramics original price.

Also, it's much worse than not having proper or unknown specifications. Usually, they are outright inoperable scams. Capacitors may not be capacitors at all. Just don't buy them.

So, any price comparison is totally meaningless when you are not getting the product you are buying!

I can sell you anything you want even cheaper, just send me your money! I won't send you anything, but come on, it's cheaper!

Would you buy a made in Taiwan LCD diplay for 15 euro ?, while on ebay i can get a 1602 for €1,50.
Can someone explain to me what is wrong with those LCD displays ?, i think they are good.
Because they sell so much they also can sell cheaper.
I,m sure you buy them also.

Or how about rotary encoders, i buy 10 with pushbutton for under 4 euro on ebay, normally 1 cost 3 euros.
How can they be bad quality ?, its a rotary encoder.

[Siwastaja]

Rotary encoder is maybe a prime example of a part where quality matters. Shitty rotary encoders are a real plague, and that does include some higher cost, "quality" parts. For example, we have a totally unusable Agilent oscilloscope because of bad rotary encoders. I'm sure they used expensive, brand-name encoders, and I'm sure they were fine in the start. Cleaning might help, at least for some time. So it's really difficult to manufacture a reliable, high quality, user-friendly rotary encoder.

Tantalums are used in some special applications, such as linear regulator output caps when stable ESR (higher than those of ceramic caps) is needed, or in timing or filter circuits when high capacitances are necessary. For beginners, I never recommend tantalum caps, because you need to be able to calculate the transient currents of the circuit, a thing easily overlooked. Large value MLCCs in big SMD packages are equally difficult, but for different reasons (hidden damage from excessive soldering heat).

I use Ebay alphanumeric LCD displays in hobby projects. Some have failed after some time due to mechanical contact problems or whatever.

The main product I source from Ebay is the simple pin header and some simple screw terminals. I have found that the quality and price works out for me in this product segment.

Any types of capacitors or semiconductors are the worst; the chances you get inoperable products in these product segments is in the range of 50-100%, so it doesn't make sense to buy these types of products, unless you want to spend a lot of time and money by buying the things multiple times before learning. Also, unlike with connectors, in semiconductor and capacitor segments, the Ebay prices often are not even cheaper; sometimes, they are more expensive.

[madires]

Or how about rotary encoders, i buy 10 with pushbutton for under 4 euro on ebay, normally 1 cost 3 euros.
How can they be bad quality ?, its a rotary encoder.

We only tried to spare you some trouble or headaches, but you're right, you have to make your mistakes yourself.

[JanJansen]

What rotary encoders would you recommend then ?

[madires]

I've got good experience with ALPS EC11 series.

[Zero999]

MLCCs have higher charge capacity by volume than electrolytics, in other words they have more capacity for their size.

*snip*

MLCCs have a characteristic called DC bias which other types of caps don't suffer from.  When subjected to higher DC voltages, capacitance falls off.  It can be quite significant.  For exampe a 16V rated MLCC cap may only maintain ten percent of its rated capacitance at maximal DC voltage.  You can usually find the DC bias curve for an MLCC cap from the manufacturer web site, but it can be a rather elusive specification.

When taking into account the reduction in capacitance MLCCs have at near the rated voltage, are you sure they really do have a greater energy density than aluminium electrolytic capacitors?

If a 10µF 25V MLCC only has a capacitance of 1µF at 25V, then its energy capacity will be the same as a 1µF aluminium electrolytic capacitor. Whether the MLCC has a higher energy density than the aluminium electrolytic or not depends on its size. Have you looked into this?