Why are mainboards still so large?

[Elledan]

TX2 mainboard

The mainboard on the page I linked to is a 6 layer PCB, ATX form factor one, yet it's a lot smaller than conventional mainboards. Its part of the Cappuccino TX2 mini-PC.

Now, why can't the size of every mainboard be decreased (not necessarily as small as the TX2 mainboard)? Would it make the production too expensive? Would performance suffer? Or would it simply be very hard to design one?

Any insights are welcome

 

[Killbat]

There has to be space for the expansion slots, and they have to fit the standard mount points in cases.
The existing form factors are fine for almost all applications.

 

[Elledan]

<< There has to be space for the expansion slots, and they have to fit the standard mount points in cases. >>

You can create some kind of 'adaptor'-plate to fit it with the mount points.

As for the expansion slots, that's actually the only thing I forgot about they obviously take up the most space.

Hmm.. that makes me wonder, why are these expansion slots so large? Can't they be decreased in size, like PCMCIA?

 

[rimshaker]

Today's technologies have no problem shrinking the components of a m/b. Ever heard of SOC (System On a Chip)? The technology is there, but you have to remember, for the products to be distributed and used by the general public, computer makers set standards and such so there is some 'orderly flow' if you will. Hence, we are all stuck with the current ATX form factor... for now. You can bet the next generation will be smaller in some ways. A m/b will always have to be large enough to hold multiple PCI/AGP slots... that's another size llimitation.

 

[Killbat]

"Hmm.. that makes me wonder, why are these expansion slots so large? Can't they be decreased in size, like PCMCIA?"

Why? It's easy and cheap to make cards the way they are. A lot of cards need airspace around them for cooling, as well.
I'm sure if someone really wanted to, they could figure out a more space-efficient expansion method, but there is no reason to. The current system is more than adequate.

 

[Elledan]

So:

- there's no need to decrease the size of mainboards and components on it, mainly because it would be costly and quite some hassle to switch to different standards.

- the current set of standards works.

But how about the non-mainstream market, i.e. specialised applications?I would imagine that for many of such applications, the current standards would not be sufficient to create the right system.

 

[blahblah99]

take a look at SBC boards... they have video, disk on chip, ethernet, serial, parallel, ide all built into a small footprint the size of two credit cards.

 

[Peter]

The case form factors dictate the mainboard sizes ... and then there's a substantial jump in cost with every extra layer you have to add when squeezing stuff tighter together, smaller versions of standard components (from resistor arrays to capacitors and chips) are much more expensive than their standard size siblings, and finally there is an even larger jump in manufacturing cost when you start putting components on the underside.

So basically when you design a standard mainboard, you start out with the form factor as a given, try to fit everything on one side and put all signal traces on two layers (to fit onto four total) for the best price - and if you manage, then try to make it as small as possible.

When you have to squeeze as much stuff as possible onto a given form factor, and cost is a small problem, the rules are different - here, you have to see how all that stuff fits, use the smallest manageable passive component sizes, and use as many layers as needed.

An example of how much PC equipment you can squeeze onto 160x200 millimeters ... (my baby, BIOS-wise ):

http://www.sbs.com/pdfs/or_dsh_ct7.pdf

but don't believe you'll get this for $100 and don't think you'll be able to keep it cool with the usual fan&heatsink toys either. Cheap cooling also needs lots of room.

regards, Peter

 

[Elledan]

I found this page on SBC mainboards: Linky.... Looks... small

Thank you, Peter. Looks very interesting =)

 

[BraeBrae]

Elledan,

What exactly is your problem with the current mobo dimension? Are you seeking a smaller board for a smaller chasis? If so, try a micro-ATX board. Also, if you are looking for something specific and cannot find it, which lead to your original question ... How about telling us the reason behind the question? Enquiring minds want to know.

 

[AreEss64]

Ah. SPC or PCOC or IPC or SBC's.. pick your acronym. They're actually geared towards industrial use, which is why you see them still being produced with 386's and 486's and such. They require special backplanes and cases (which usually cost you upwards of several hundred dollars apiece!) and are intended for fire-and-forget solutions, such as automation management, robot operation, and things of that nature.

They also make GREAT routers/firewalls if you can pick them up used. Relatively low power consumption, very small footprint. Just don't plan on upgrading 'em. Personally, I -like- a larger motherboard. Larger means there's more open space between components usually, meaning that airflow over the board is less impeded, meaning my system isn't as liable to overheat if a non-monitored fan should happen to die. Personally I'd like to see some bigger dual Athlon boards.. say, SuperMicro S2DL3 sized. My case can handle it with ease. (Of course, my case is built to accept dual power supplies from PC Power&Cooling -and- an old Quad PPro mo'board..) Linkage for imagery. Supermicro's site is down, sadly. But it's 12"x13" for comparison. (A bit bigger than an Abit KT7-RAID.)

Bigger is better!

 

[Elledan]

<< Elledan,

What exactly is your problem with the current mobo dimension? Are you seeking a smaller board for a smaller chasis? If so, try a micro-ATX board. Also, if you are looking for something specific and cannot find it, which lead to your original question ... How about telling us the reason behind the question? Enquiring minds want to know. >>


I'm obsessed with everything small and efficient

AreEss64,

Thank you for that information! I don't think that I'll see any used PCB's any time soon, though

I actually had an idea of a very tiny mainboard, with PCMCIA-like expansion slots and, well, tiny connectors for I/O (mouse/keyboard, USB).

 

[Kadarin]

<< I actually had an idea of a very tiny mainboard, with PCMCIA-like expansion slots and, well, tiny connectors for I/O (mouse/keyboard, USB). >>



There's also both a driver issue here, as well as a backwards-compatibility issue. There's a lot of legacy hardware out there, and it takes time to get it off the market. We're just now starting to see ISA die out. As for the driver issue, you'd have to get OS vendors to support whatever system you come up with, as well as convince hardware manufacturers to produce devices for it. We could/would end up in a situation where incompatible competing standards fight for control of the market, and it would suck to be on the wrong team in another Beta/VHS war.

-j

 

[zephyrprime]

With heat dissipation needs going up, I wonder if computers will stop getting smaller? Maybe the notebook - desktop performance gap will also increase?

 

[Elledan]

<< With heat dissipation needs going up, I wonder if computers will stop getting smaller? Maybe the notebook - desktop performance gap will also increase? >>


Unlikely. Currently most heat is produced because of 'leaks', i.e., when electrons migrate through non-conductive material (e.g., Si) because this layer is getting to too thin to stop every single electron. Transistors aren't perfect either, so these 'leak' too. Furthermore, current CPU's work with rather high voltages, which means a lot of electrons, resulting in a lot of electricity being used. By reducing 'leaks' and making transistors more sensitive, so that they can work with less electrons, heat will decrease enormously.

Of course CPU's based on 'spin' would be wonderful as well