Solid Capacitor Motherboards Introduced

jckrbbt writes with news that Gigabyte has introduced solid capacitor motherboards in its Intel 945 chipset products. From the article: "[S]olid capacitors have a higher tolerance for higher temperatures and they also perform better with higher frequencies and higher current than electrolytic capacitors. The superior heat resistance and better electric conductivity will allow PC enthusiasts to tweak the highest levels of performance from their system without fear of excessive capacitor wear or exploding capacitors."

Comments

Average

[sporkme]

Solid capacitors also last longer with an average lifetime of 23 years compared to only three for electrolytic capacitors, according to Gigabyte.

I have seen popped caps on motherboards, but 3 years seems a little short as an "average."

Additionally, solid capacitors have a higher tolerance for higher temperatures and they also perform better with higher frequencies and higher current than electrolytic capacitors.

Yay for overclockers and NASA.

Re:Average

[Anonymous Coward]

I have seen them last less than a year. All you have to do is check where the capacitor is from. If it is from China (which is likely), then it has a high probability of failing very quickly. This is due to their stealing the formula from a Japanese company who became aware of the attempted theft and fed the women a recipe from the early 60's (and well known to hold up for only a year).

Re:Average

[thedarknite]

I'm curious as to how they calculate these numbers, I've never had a capacitor fail and I play with a fair number of boxes that are beyond their 3 year "average". Then again I've never used a Gigabyte board, so they may well have had a shoddy supplier at one point.

Re:Average

[Anonymous Coward]

The standard life test for an electrolytic capacitor is 1000hr at rated temperature. For most consumer equipment, this is 85C oe 105C depending on which electrolyte is used (and price paid). The life of electronic components doubles for every 10C reduction in temperature. At end of life, an electrolytic caacitor is allowed a 50% loss of capacitance and doubled ESR. For most cases these caps work fine (provided the designer included margin). In a PC application, it is reasonable to expect operating temperatures of about 45C to 55C. This would mean lives between 8000hr and 32,000hr.

I find it interesting that the solid electrolyte caps have finally found a home. These have been around fo about 20 years in one form or another. Maybe now the price will start falling and the small wet electrolytics will go the way of the vacuum tube.

Re:Average

[multipartmixed]

I suspect your version of "fail" and "3 years" is different from theirs.

I'm guessing by "fail", they mean that N percent of them are Y percent out of spec, and by "3 years" they mean "3 years run-time", not "3 calendar years".

That said, I seem to recall electrolytic caps on digikey typically being rated for around 2,000 of use.... and their definitions of "fail" are exactly as I've said above.

Caps can (and often do) work in their intended application well after they have ceased to behave as the spec sheet says they should. Sometimes, they are not that critical; other times, the design engineers know how to derate parts to get a reaonable lifetime out of whatever it is they are building.

Re:Average

[Phleg]

and fed the women a recipe from the early 60's

That doesn't seem so bad. My mom still has some of her cookbooks from the 60s...

Re:Average

[ArsenneLupin]

and fed the women a recipe from the early 60's

That doesn't seem so bad. My mom still has some of her cookbooks from the 60s...

And how often do her dishes explode?

Re:Average

[ArsenneLupin]

The author was even kind enough to link to reputable sources that confirm the story. The story is pretty well-known, and has even been featured on Slashdot. Here are them links:

• The Great Capacitor Scare of 2003
• Leaking Capacitors Muck Up Motherboards
• Illicit Leaky Capacitors Killing Motherboards

==> only error of GP: the formula muck-up was not intentional, but the thieves themselves made the error while copying it.

Re:Average

[markh1967]

For the full story see www.badcaps.net

Re:Average

[EveLibertine]

and fed the women a recipe from the early 60's

That doesn't seem so bad. My mom still has some of her cookbooks from the 60s... And how often do her dishes explode? The average lifetime is around 3 years.

Re:Average

[Fordiman]

As an ex electrical engineer, I can say there are two things you remember about capacitors when designing a circuit:

1) If signal consistency is important to you, never use an RC oscillator
2) If you need it to last at least as long as you work there, always use a cap with a voltage of ceil((inputSignal.average+inputSignal.stddev*3)*1. 25)

These ifs are usually rhetorical; you always want signal consistency, and you always want it to last as long as you can be held responsible for it. Period.

Re:Average

[OrangeTide]

I've worked for companies where we had to RMA an entire run of our product because the caps we received from the supplier were below spec and were exploding in the field (DSL modem). I've also have replaced people's motherboards because they were running pretty unstable (even in memtest86) and had some lumpy looking caps on the board. The PSU seemed large enough so I can only point the finger at out of spec caps.

Most motherboards are Taiwanese not Chinese. Although I'm sure the government of China would like us to believe the two places are equivalent. I still can't believe GW Bush would rather play around in Iraq than help foster democracy in Taiwan which actively asked us for help in leaving the ROC and the influence of the PRC. *sigh* I guess Taiwan didn't have enough oil interests.

finally

[Swimport]

Capacitors having the shortest lifespan of most electrical components means if this catches on there will be less electronic waste, and more reliable machines. Although I bet these cost twice as much....

Re:finally

[iminplaya]

...more reliable machines.

Don't count on it. Planned obsolescence isn't going away anytime soon. Make the machine "too reliable" and the industry will be crying about lousy sales.

Re:FUD

[snowgirl]

Yeah, you know, because that's *the* biggest complaint you see on enthusiast/overclocker message boards. Exploding capacitors.

It could be worse... they could be a company selling a network card to reduce network lag... lol.

3 YEARS?

[vistic]

"Solid capacitors also last longer with an average lifetime of 23 years compared to only three for electrolytic capacitors, according to Gigabyte."

Motherboards may get obsolete fast, but I still would expect a longer life than just three years.

If this is true, I'm amazed so many old computers work so well. Maybe this is a bit off. In either case, it seems with such a huge difference in life span, unless there's a huge change in cost, the extra reliability offered by solid capacitors should make them standard in every motherboard. I'm not an electrical engineer though (or an economist).

Re:3 YEARS?

[Anonymous Coward]

so many old computers work so well.

Capacitors are generally used for filters and timing circuits. The motherboard I'm running right now has a capacitor at the edge of the AGP slot that I accidentally crushed (hey, I thought the new video card was just tough to push in, sue me). I only noticed it because the computer refused to boot until I cleaned the guts of the cap off the motherboard, and it runs just fine since then. Perhaps it won't deal with certain line noises anymore, or some USB port hardware runs 15 times as fast, but I'd say that I'll never notice either with this getup. My mother had a TV that was perhaps 20 years old. One day a cap (audibly) blew, and the only difference was that the scan controls no longer kept the picture entirely within the screen, sort of like a permanent 125% magnification, with the extra running off all four edges. She watched that tv another year or two before finally buying a new one.

Re:3 YEARS?

[alienw]

The little caps near the expansion slots are for power decoupling. There are usually lots of them and the loss of one will not affect anything as long as it does not short out. The bigger switching converter caps near the CPU are a critical component of the switching power supply, and losing one would definitely kill the mobo and possibly the CPU. They are also the ones most likely to explode or leak, since they do a lot more work and are exposed to much higher temperatures.

Re:3 YEARS?

[Max Romantschuk]

I'm amazed so many old computers work so well.

Actually, if a computer is old enough that helps too. I use a floppy drive from a 286 due to the fact that it's mechanically bulletproof. I went through 2 floppy drives before gutting an old external floppy enclosure for the drive. Newer hardware isn't built to last.

Also, thee years for a capacitor is probably three years of continuous usage.

Re:3 YEARS?

[noidentity]

Power supply filter capacitors are generally added in excess so that there is plenty of margin. Losing one will reduce this margin for that area of the board, kind of like overclocking where you increase the chance of error. When you crushed the capacitor, you probably also shorted its terminals together, and when you cleaned it up, you probably un-shorted them again.

Dell

[Cherita Chen]

The three year average doesn't surprise me, Dell has had a multitude of problems with bad/bulging caps

EH?

[SinGunner]

"superior heat resistance?"

Doesn't resistance CAUSE heat?

I'm sure the ...

[SloWave]

Gold Plated Speaker wire crowd will love this.

Re:I'm sure the ...

[nightfire-unique]

Gold Plated Speaker wire crowd will love this.

Ahem. :)

Speaking as an "audio dick," I feel I should come to the defense of both "solid capacitors" and gold plated speaker wire.

Firstly, gold plated speaker wire isn't gold plated to improve the capacitance or resistance properties of the wire - it's done to prevent corrosion. If you've ever heard the crackling sound an old car stereo tends to make, it's often because of corroded copper wires. It's particularly noticeable when you live near saltwater areas or in marine applications in general.

Secondly, there is no outstanding debate in the industry on whether or not polypropylene, film, or even tantalum capacitors (what they're referring to as solid, though they're probably talking about tantalums) are of superior quality to electrolytics for audio applications. Electrolytics have changing thermal characteristics, worse tolerances, and tend to introduce a small amount of phase shift into whatever AC signal you're passing through them. Yes, these properties are measurable with the right equipment and are not generally questioned.

And yes I am an electrical engineer! :D

Re:I'm sure the ...

[PeterBrett]

Secondly, there is no outstanding debate in the industry on whether or not polypropylene, film, or even tantalum capacitors (what they're referring to as solid, though they're probably talking about tantalums) are of superior quality to electrolytics for audio applications. Electrolytics have changing thermal characteristics, worse tolerances, and tend to introduce a small amount of phase shift into whatever AC signal you're passing through them. Yes, these properties are measurable with the right equipment and are not generally questioned.

Agreed. Tantalum capacitors have much better performance than electrolytics in most circumstances. However, there is outstanding debate about whether the use of tantalum capacitors is ethical, as tantalum is just about the rarest element that's actually used in the electronics industry and most of the deposits are in developing countries. Accusations have been levelled that electronics manufacturers are going to inordinate lengths to secure tantalum deposits, and the people who live there are the losers (especially since the by-products of processing tantalum ore are decidedly unpleasant).

I try to avoid using tantalum capacitors in my own designs as far as possible, trying to keep to NASA's guidelines for component derating when using electrolytics. Where I need precision capacitances I design the circuit so that a ceramic NP0 or similar EIA Class 1 capacitor can be used instead. I haven't had any capacitors fail yet.

Re:I'm sure the ...

[vojtech]

The capacitors in question are not tantalum, but solid polymer capacitors. A tantalum capacitor design would be possible, but would be very expensive and also rather bulky.

Compared to tantalum capacitors, these capacitors reach much higher capacities at the same physical volume, and the same or better ESL/ESR.

See for example here:

• United Chemi-Con PSA series
• Famous Sanyo OS-CON

These aluminium electrolytic capacitors, with a solid conductive polymer electrolytic find their way even on most current mainboards, most often in the CPU DC-DC convertor circuits. They're usually easily recognizable from classic electrolytics by their small size and metal casing without a plastic sleeve.

A benefit from an all-solid-polymer capacitor mainboard is dubious, since classic alimuium electrolytic capacitors work just fine in many roles they're needed for, particularly in low-ripple-current situations.

Re:I'm sure the ...

[DrDitto]

Yet another NASA engineer who potentially compromises the design of a spacecraft in the name of environmentalism. You wanna know why the Space Shuttle is having all these recent foam problems that crashed the Colombia? Because they switched the formula to remove freon in order to be more environmental even though the EPA gave NASA a special waiver. The original design worked, but it used freon in the foam. Don't believe me? Here is a journal entry from NASA's website in 1997!!!: http://quest.arc.nasa.gov/people/journals/space/ka tnik/sts87-12-23.html

During the STS-87 mission, there was a change made on the external tank. Because of NASA's goal to use environmentally friendly products, a new method of "foaming" the external tank had been used for this mission and the STS-86 mission. It is suspected that large amounts of foam separated from the external tank and impacted the orbiter. This caused significant damage to the protective tiles of the orbiter. Foam cause damage to a ceramic tile?! That seems unlikely, however when that foam is combined with a flight velocity between speeds of MACH two to MACH four, it becomes a projectile with incredible damage potential. The big question? At what phase of the flight did it happen and what changes need to be made to correct this for future missions? I will explain the entire process.

Yet somehow this never got mentioned by the mass media back when Colombia disintegrated over Texas.

Re:I'm sure the ...

[Ancient_Hacker]

>Firstly, gold plated speaker wire isn't gold plated to improve the capacitance or resistance properties of the wire - it's done to prevent corrosion.

Wrongo, Bub. Wire strands are usually tin-plated to reduce corrosion and improve solderability.

Nobody, but nobody gold-plates wire, even for the most demanding aerospace and military applications. I've taken apart Titan missle guidance computers, Mil-spec avionics, even old satellite guts and havent found an inch of gold-plated wire.

Now connector edges and IC and transistor leads that are going into sockets are often gold-plated. And I've even seen a very few Sprague electrolytic capacitors, MIL-spec, maybe one in a million, with gold-plated leads. But not wire.

> and tend to introduce a small amount of phase shift into whatever AC signal you're passing through them.

Wrongo again Bub. As you should know, the phase shift is proportional to the capacitance, and NOTHING ELSE. So a 100uF electrolytic capacitor will have EXACTLY the same phase shift as a polypropylene capacitor of the same value. And doubly false, as electrolytics are hardly ever used anywhere in their phase-shifting frequency range-- they're almost always used at frequencies where they're nearly dead shorts to AC and near zero phase angle. Exceptioin-- old cruddy speakers used non-polarized electrolytics in their phase-shifting range, but that's about it.

>And yes I am an electrical engineer! :D

Yep, I've met way too many EE's that don't recall the basic physical facts.

3 years???

[wiredlogic]

Solid capacitors also last longer with an average lifetime of 23 years compared to only three for electrolytic capacitors, according to Gigabyte

This is complete BS. A three year service life may be the norm for bootleg Chinese knockoffs of Japanese parts but quality Aluminum electrolytics can last far longer.

Dell GX270's

[Gates82]

Dell should start using these for their GX-270 line. I (1 out of 5 at the site I worked) have replaced a good 30 270 Motherbo--- (sorry per dell, system boards) that have video problems. All stem from bulgin and leaky capacitors. Most of these systems where between 1-2 years old (none over three).

--
so who is hotter? Ali or Ali's sister.

Re:Dell GX270's

[parasonic]

Dell should start using these for their GX-270 line. I (1 out of 5 at the site I worked) have replaced a good 30 270 Motherbo--- (sorry per dell, system boards) that have video problems. All stem from bulgin and leaky capacitors. Most of these systems where between 1-2 years old (none over three).

On the GX270's, there is now a lifetime replacement warranty on the motherboards. Capacitors blow on these things, whether that is six months down the road, or four years. We got a bunch of 270's around 2003 and still have several dozen of them here at the company where I work. A few months ago, we called Dell, and they sent two boxes complete with new motherboards and return labels for the old ones.
 
As a CmpE (currently working in I.T.), I will tell you that electrolytics are absolutely fine. I have electronics from the 60's and 70's with electrolytics that hold up. If the manufacturing process is botched, something may go wrong. But you can end up with a mess also if you manfacture tants, micas, polypropylene, even ceramic disc capacitors incorrectly. "Solid" capacitors are more of a sure-fire thing, but they can fail, too.

About time!

[Orphaze]

As someone who has painstakingly replaced all the capacitors on two separate motherboards, I can definitely see why this is a good idea. The most recent was my Epox 8kra2+ board (with an Athlon XP 2600+, not over clocked.) I noticed the caps beginning to bulge slightly on top and develop some crusty electrolyte "dandruff" on the heads after 2 years of use, but I decided to hold off on major surgery until I began to notice any problems.

About a year later the system began to lock up mysteriously, and after ruling everything else out (this was my main system after all) I grabbed my soldering iron and began an hour or so of some rather nerve wrecking soldering. Every single 1000F and 1500F cap on the board needed replacement, so an old PIII board became the donor.

I measured the bad caps after removing them and most of them were off by about 300-700F, way outside of tolerance. After I finished I booted the system up, ran memtest for a few hours successfully, and never had a lockup since.

Doesn't really do any good for a computer though

[Sycraft-fu]

I see VERY few computers failed due to a cap problem before they are retired on account of being too old to be useful anymore. The most common component I see fail is the HD, which is no surprise given that it's mechanical. This could be useful for devices that are good for 25 years, but comptuers tend to get thrown out after 5.

23 years?

[ScaryMonkey]

Solid capacitors also last longer with an average lifetime of 23 years compared to only three for electrolytic capacitors, according to Gigabyte

I guess a longer lifespan is good, but do I really need a motherboard to last for 23 years? I just might get around to upgrading the processor in that time frame...

Re:23 years?

[gbobeck]

I guess a longer lifespan is good, but do I really need a motherboard to last for 23 years?

Actually, the intended buyer of this motherboard happens to be Gentoo users. The idea is to sell them a motherboard which will run after everything compiles on their system.

Note: I am a happy Gentoo user, above was only a joke.

Why is this news?

[Angelwrath]

I've had a Gigabyte board with solid-state capacitors for more than 3 months now, it's based on the 965 chipset, so I was a bit confused why this article made it sound like this was a new innovation.

Re:No thanks...

[fabs64]

Conveniently they're also used for a completely different purpose in electronics. :-P

Re:But

[nmos]

the question on everyones mind is whether these even-more-overclocked PCs will be able to run Windows Vista?

Well maybe not "run" but it should at least "walk" now.

Not vacuum tubes...

[Kadin2048]

I'll stick to my vacuum tubes. Not only is the technology well-tested over the years, you can heat up the entire house if your computer room is in the basement.

What you really want are Leyden Jars.

Instead add neon lite to PC for better performance

[viking80]

The recipe for the electrolyte in capacitors is kept as a big secret similar to the secret ingredients in the sauce at a restaurant.

Chinese industrial spies stole a fake formula from a Japanese company, and started making capacitors, and the rest is history.

A combination of a smaller solid cap with good HF performance together with a cheap and large electrolyte further away, but with better LF performance will beat the solution in the article.

I use the power supply from a 25 old HP HDD as a lab supply. It has huge electrolytes that still deliver great performance.

You will probably get more performance improvement by adding neon lights to your case.

Article in ieee.org members only
http://ieeexplore.ieee.org/iel5/6/26410/01176509.p df?arnumber=1176509

http://www.burtonsys.com/bad_BP6/story1.html

Re:Doesn't really do any good for a computer thoug

[repvik]

I've seen a lot. Of those, a lot were caused by this: http://en.wikipedia.org/wiki/Capacitor_plague

Um, right

[coder111]

When was the last time motherboard manufacturers used quality Japanese parts instead of bootleg Chinese knockoffs? And Gigabyte is guilty of doing this as much as every other motherboard maker.

--Coder

Re:Mod parent up.

[PeterBrett]

I do a lot of DIY speaker building and there definitely is a difference between Film/Foil caps and electrolytics. A speaker crossover made with Electrolytic caps sounds like crap compared to one made with even the cheapest of film/foil caps.

Firstly, WTF are film/foil capacitors? As far as I am aware, the only major types of capacitors used are:

• Aluminium electrolytic capacitors (aluminium foil, tightly would in a dielectric fluid)
• Ceramic capacitors (single- or multi-layer, using EIA Class 1 or Class 2 dielectric ceramics)
• Tantalum- and niobium-based capacitors (chip or electrolytic styles)

Are you thinking of resistors? I use thin-film SMT resistors all the time...

Which part of the crossover are we talking about? Which design do you use? Do you have inductors in there? Quite a few performance issues when using electrolytic capacitors are due to an inappropriate choice of inductors, IMHO.

Also, I've got a friend who does psycho-acoustics research, and he did an interesting series of experiments a couple of years ago that indicated that systems that performed technically very well (almost perfect filter characteristics, no harmonic errors) actually were rated worse than a system that had all sorts of junk spewing out of it, when the audiophiles participating weren't told which system they were listening to...

YAWN! Capacitor FUD

[EmbeddedJanitor]

You basically get three major types of caps on motherboards, each of which have different properties:

Ceramics. Small and fast. Typically used for decoupling (small charge storage).

Electrolytic: Larger and slower. They are slower because they are highly inductive. They don't like working at very high frequencies which can make them fail.

Tantalum: Medium/large and fast. They are less inductive than electrolytics. They can dump current far faster than electolytic which can cause undesirable current rushes.

Of course I have not RTFA because that's not the point of /., but I suspect they're swapping tantalums in to replace electrolytics. With proper usage electrolytics will not typically fail, so this is perhaps FUD. Particularly the "overclocker" bit. It sounds like FUD to try generate a new "feature" to sell their motherboards.

Re:YAWN! Capacitor FUD

[splutty]

Funny, you sort of go completely past the point of the *actual* difference between ceramic condensators and electrolytic condensators, which is that one's polarized and the other isn't. They're not interchangable.

Re:FUD

[The Darkness]

how do you make a solid capacitor? Doesn't it work by keeping a charged capacitance between two nearby electrical circuits seperated by a vacuum? You're close, but I recall from my physics classes that capacitors aren't required to have a vacuum between the plates. I just checked the wikipedia entry on capacitors and found the following piece of text:

The capacitance is proportional to the surface area of the conducting plate and inversely proportional to the distance between the plates. It is also proportional to the permittivity of the dielectric (that is, non-conducting) substance that separates the plates.

So, if I understand that text correctly, you can create a smaller capacitor with the same capacity to store charge as a vacuum-gap capacitor by placing an appropriate dielectric material between the plates.

Re:YAWN! Capacitor FUD

[deevnil]

No, the real difference is the amount of capacitance. I wonder what a 100uf ceramic disc cap would look like, a frisbee?

Re:FUD

[unitron]

So, while my first instinct is "how do you make a solid capacitor? Doesn't it work by keeping a charged capacitance between two nearby electrical circuits seperated by a vacuum?"

Anytime you have two conductors separated by an insulator you have a capacitor, or at least capacitance.

If you put a sheet of wax paper between two sheets of aluminum foil, that's a capacitor. If you replace the wax paper with a layer of air, it's still a capacitor. If you replace the air with a vacuum, it's still a capacitor. You can even replace the vacuum with a non-conductive liquid. It'll still be a capacitor.

As long as whatever separates the two conductors is an insulator, then you've got a capacitor.

Electrolytic capacitors use stuff that's a sort of paste, or "goo", which, combined with a layer of oxidation on the inner surface of one of the conductors, makes up the insulator. "Solid" capacitors use something that's, well, solid.

If you have a further interest, Googling for "capacitor disease" may prove instructional.

Re:YAWN! Capacitor FUD

[AaronLawrence]

That's true but often irrelevant. The difference in size (electroytic gets much more capacitance per size, so after about 1uF they are usually used) and RF behaviour (ie. inductance) of the two is much more important in most cases.

mod parent DOWN, true but misleading

[petermgreen]

actually you can get non-polarised electrolytics too, they are often used in loudspeaker crossover circuits, i think they are basically two ordinary electrolytics in inverse series.

but anyway in most cases (especially in digital equipment like computers) capacitors are used in a way that keeps them biased the same way all the time so it doesn't really matter if they are polarised or not.

i agree with the gp that the important characteristics of electrolytics are big and slow (high ESR) while the important ones of ceramic are small and fast (low ESR).

tantalums are fairly big and fairly fast, they also have much better lifetime characteristics than electrolytics. The downside is that they are expensive and when they do go bang (tantalums are polarised) they tend to fragment into a shower of tiny hot high velocity shrapnel.

Its unclear from TFA if the "solid capacitors" gigabyte are reffering to are tantalums or some new technology.

Apologies to fellow slashdotters

[theskipper]

My electrolytes are low this morning so I currently don't have the energy or capacity post a reactive response to your highly charged and (possibly) polarizing comments.

The summary looks tantalizing though.

Re:YAWN! Capacitor FUD

[Bassman59]

Let's be pedantic here.

You basically get three major types of caps on motherboards, each of which have different properties:

Ceramics. Small and fast. Typically used for decoupling (small charge storage).

Also used in filtering applications ...

Electrolytic: Larger and slower. They are slower because they are highly inductive. They don't like working at very high frequencies which can make them fail. I suppose you mean Aluminum Electrolytic. The failure modes are not high frequency, but mainly heat (either because of the environment or because its ESR is relatively high, which means it's self heating). They also don't like reverse biases, the results of which can be very exciting. They're not "slower" because they're "inductive," they're "slower" because their values are typically much larger than ceramic types.

Tantalum: Medium/large and fast. They are less inductive than electrolytics. They can dump current far faster than electolytic which can cause undesirable current rushes.

Tantalums are also electrolytics. The electrolyte allows for larger capacitance in a given area. You REALLY don't want them to be reverse biased, and they have a bad tendency to explode if inrush current is too high. You don't want to use a tantalum as a DC block in audio circuits.

You also forgot some other capacitor types: polystyrene, mica, etc. These guys tend to be physically larger than a ceramic for a given capacitance, but they're a lot more stable. (And more expensive.) You'll find them in filter and integrator circuits, in high-frequency stuff, anywhere where stability is required.