Sign in with
Sign up | Sign in

Gigabyte Doubles Up Copper in Mobos

By - Source: Tom's Hardware | B 21 comments

Gigabyte announced this week a new technology which it calls Ultra Durable 3, promises that motherboards featuring the new technology will achieve better overclocking, reduced heat and increased energy efficiency. The premise of the new technology is actually rather simple; increasing the amount of copper used in a motherboard will lower the impedance, reduce heat and improve signal quality.

According to Gigabyte, Ultra Durable 3 motherboards will be the first consumer desktop motherboards to feature 2-ounces of copper for each power and ground layer, doubling that of traditional motherboards. A bit like how a larger copper heat sink often works better than a smaller one, having more copper throughout the motherboard allows for more efficient spreading of heat. As such, heat created by critical components on the motherboard will spread out across the entire motherboard more effectively, which Gigabyte claims can deliver 50°C cooler working conditions than traditional motherboards. Somehow that claim seems to be a bit misleading, but it should be interesting to see in tests how much cooler the motherboard components actually do get.

Doubling the amount of copper should also help to reduce the amount of impedance, which Gigabyte says is lowered by a factor of two. Impedance is like resistance in that is opposes the flow of electrical current, so reducing the impedance will also reduce wasted power. Less wasted power means less generated heat, cooler components and a lower power bill. Gigabyte claims the extra copper cuts the total amount of PCB electrical waste in half, which probably still does not amount to actually a whole lot, but every bit helps.

One last aspect in which the added copper apparently comes in handy is with improving signal quality, which results in a more stable system and greater overclocking margins. Of course, all these claims still need to be substantiated, but the theory behind adding extra copper looks sound enough. For some users, the need for extra copper in motherboards may not be really all that important or necessary, especially if it comes at a premium cost. For serious overclockers and enthusiasts however, the little extra copper found in those motherboards will likely be welcomed, even if the effect turns out to be minor.

Discuss
Display all 21 comments.
This thread is closed for comments
Top Comments
  • 15 Hide
    jliu83 , September 25, 2008 6:17 PM
    As an electrical engineer who makes PCB's professionally, I think these are all marketing schemes. Impedance characteristics of high speed signals are usually controlled, not reduced. Lowering the impedance of certain critical signals will get you a motherboard that does not work (signals get reflected if impedance is not matched). Increase the ground and copper planes will have a small effect on heat dissipation. This is because the standard 1oz copper is already a very good conductor of heat, with very low resistance. Increasing the copper to 2oz cannot improve performance by a significant amount (diminishing returns as copper is increased). The way you control the impedance is by changing the signal's trace width, and height with respect to the ground plane. It is NOT controlled by the thickness of the ground plane itself. As for low frequency signals, impedance doesn't really matter. This is because as a signal's frequency reduced, the DC characteristics matter more than the AC characteristics.
Other Comments
  • 0 Hide
    drysocks , September 25, 2008 5:44 PM
    "new technology"
    Very funny. I've improved my impedance matching, VSWR, Power handling, for years now with 2oz copper in certain circuits.

    "Gigabyte claims can deliver 50°C cooler working conditions". Do they mean 5°C?
  • 0 Hide
    mr roboto , September 25, 2008 5:48 PM
    Yeah up to 50c across the whole board. Like up to 5c on the mosfets, up to 5c on the northbridge, up to 5c on the Southbridge, etc... Up to 50c cooler combined across the whole motherboard. Very misleading and seems to be below what Gigabyte is capable of in terms of marketing. Maybe they think people will be stupid enough to believe it.

    Gigabyte has gone downhill in recent times. Not the same company that once made some outstanding AMD 939 boards a few years back. Now they're associated with high failure rates and poor QC, oh and now add BS marketing to that.
  • 1 Hide
    exiled scotsman , September 25, 2008 6:09 PM
    They'll probably ad a nice $30-$50 price premium to boards featuring this "break-through". I suspect when the benchmarks are made, that there won't be much of a difference. Wow, I can gain 100Mhz extra on my already OC'd C2D...it improved my framerates by 1%...
  • 15 Hide
    jliu83 , September 25, 2008 6:17 PM
    As an electrical engineer who makes PCB's professionally, I think these are all marketing schemes. Impedance characteristics of high speed signals are usually controlled, not reduced. Lowering the impedance of certain critical signals will get you a motherboard that does not work (signals get reflected if impedance is not matched). Increase the ground and copper planes will have a small effect on heat dissipation. This is because the standard 1oz copper is already a very good conductor of heat, with very low resistance. Increasing the copper to 2oz cannot improve performance by a significant amount (diminishing returns as copper is increased). The way you control the impedance is by changing the signal's trace width, and height with respect to the ground plane. It is NOT controlled by the thickness of the ground plane itself. As for low frequency signals, impedance doesn't really matter. This is because as a signal's frequency reduced, the DC characteristics matter more than the AC characteristics.
  • -4 Hide
    ram1009 , September 25, 2008 6:39 PM
    I was going to make the same criticizms but you all beat me to it. Good work.
  • 5 Hide
    ossie , September 25, 2008 8:47 PM
    Actually the real part of impedance (aka resistance) is dissipating power. By doubling the thickness of ground/power planes, the resistance of the connection is halved, and also the lost power. Thermal resistance might also be reduced somewhat. All this has effects only at low frequencies/DC. At higher frequencies, there will be practically no change due to the skin effect, and also no change in transmission line impedance.
  • 0 Hide
    505090 , September 25, 2008 9:43 PM
    ossieActually the real part of impedance (aka resistance) is dissipating power. By doubling the thickness of ground/power planes, the resistance of the connection is halved, and also the lost power. Thermal resistance might also be reduced somewhat. All this has effects only at low frequencies/DC. At higher frequencies, there will be practically no change due to the skin effect, and also no change in transmission line impedance.


    Impedance and resistance are two separate principles. Impedance which can be created by an inductor resists a change in current, either stabilizing a DC circuit or causing resistance to an AC circuit; because of the constant shifting of current flow inherent to AC. Resistance which can be created with a resistor resists the flow of current to either DC or AC. Being we are discussing a motherboard that functions entirely upon DC we should be careful not to confuse the two.
  • -3 Hide
    ZootyGray , September 25, 2008 9:48 PM
    I appreciate facts, truth, accurate information. I appreciate people with direct experience, sharing what they know.

    I am disgusted and fed up with useless, negative, uninformed, opinionated, self-important morons with misconceptions about the relative locations of their brains vs other body parts designed for elimination.

    It is increasingly difficult to acquire information from this site without being bombarded by passing manure spreaders that apparently have undergone some out-of-control process of spontaneous combustion.

    Simultaneously, it is increasingly difficult to simply come here and feel good while being entertained by info on recent technological changes.

    Oh so obviously, there is a chance that this might be something good. Oh so obviously, this is a marketing feature. Oh so obviously, they warranty what they sell and are not going to produce a liability for themselves. (speaking of which) >>> And oh so obviously, some of you have created for yourselves a very sad world of unhealthy negativity. It is not your fault; but it IS yours to recreate a world of joy where you can be happy. If you have not yet had enough pain, then continue as you are.

    The degree to which you find this offensive is a reflection of your immersion in the problem. You might be surprised to hear that you really do not have to live that way. It is your choice and is the result of past choices - but your future choices are all yours. Change is difficult until you have seen better - that, in itself, is the beginning of new and improved; and as it continues, there is no place for disease. This negativity is more dangerous than you know - if you knew that, you would not be doing what you are not - because, you are really not that! Your land of the free, is not supposed to be a deathtrap. And your home of the brave is useless if you are unknowingly just hurting yourself and others. It begins and ends with you. Peace. Have a nice day; unless you have other plans.

    Contrary to some other opinions, I have seen some fine products from Gigabyte; particularly their AMD boards featuring the 790GX chipset. Your mileage may vary; but I hear good feedback from people with AMD systems with Giga mobos.

    Please note - this is not a gauntlet thrown as an opportunity for yet another lame useless flamewar. I am simply a person who prefers AMD and would like to share info with others who have similar interests. You are entitled to your preferences. I have mine for my own personal reasons. I would prefer this talk to calling you all morons and suckers for ever being seduced by spintel in the first place - so you see, I have to keep myself on the right path also :)  and it is not always so easy. Sometimes it is very easy.

    Thanks for nice article. We shall see how this develops for Gigabyte and whether or not others follow.
  • 1 Hide
    johnbilicki , September 25, 2008 10:59 PM
    exiled scotsmanThey'll probably ad a nice $30-$50 price premium to boards featuring this "break-through".


    Which should still keep it a hundred dollars cheaper then the over priced Asus boards that you'll get used replacements for when you have to RMA them.

    jliu83's comment's are pretty much on the mark, what do you think your PSU's are? Same thing as a power brick or the black blocks for things like your answering machine: AC to DC power converter.

    Most people will stick to affordable motherboards. Not going to pursue a multiple-GPU solution? A Via chipset won't give you much of a performance hit then a high end system, just fewer features.

    Now if they included a bar of gold with the motherboard for an extra $30 or $40 I'd be sold!
  • 0 Hide
    Anonymous , September 25, 2008 11:09 PM
    Copper is very precious, it's price is rivaled only by tin and lint.

    I do not think using 2oz copper has anything to do with impedance matching, it says that 2oz copper is used for the ground and power layer, nothing about the signal layer.

    Pretty much anything PCB related is cheap, 1oz vs 2oz, 10 layer vs 12 layer, is just pennies and cents. No doubt though gigabyte is going to make a spectacle out of this to make it seem that the premium that consumers pay is justified. Can't blame then though, Money talks and BS walks.


  • 0 Hide
    dagger , September 25, 2008 11:36 PM
    johnbilicki


    That's a good point. Although you have to keep in mind that Asus is so overpriced to begin with that it makes everything else look good in comparison. Gigabyte gives you some extra copper for a small premium, compared Asus, who gives you nothing but an Asus sticker for a large premium.
  • 0 Hide
    jliu83 , September 26, 2008 1:26 AM
    ossieActually the real part of impedance (aka resistance) is dissipating power. By doubling the thickness of ground/power planes, the resistance of the connection is halved, and also the lost power. Thermal resistance might also be reduced somewhat. All this has effects only at low frequencies/DC. At higher frequencies, there will be practically no change due to the skin effect, and also no change in transmission line impedance.


    To ossie: You are correct in that, within the PCB itself, the currents creating the most heat is from the DC currents. HOWEVER, you need to consider the fact that MOST of the heat comes from the components soldered to the PCB. On a CMOS process, power only dissipates during switching. It does NOT switch when a digital signal is held high or low (except for leakage currents etc.). Well it doesn't take a genius to figure out that an IC switching at 1 GHz is going to dissipate more power than an IC switching at 1 MHz. A good example of this is VGA vs DVI designs. The DVI IC's are going suck a lot of power, and create most of your heat. The heat dissipated by the resistance within the copper layer is smaller in comparison, by several order of magnitude.

    The equation for power is P=R*I^2, or P=IV. The resistance of a copper trace 10 mils wide, on a one oz copper, 4 inches long is something like 1.4*10^(-12) Ohms. Well, I can tell you that between two IC pins, the voltage differential is going to be in the mili-volts, if not smaller. The power dissipation is VERY VERY small, even if you have hundreds, or thousands of traces. The main bulk of the heat will come from your fast switching IC's, and your linear/switching regulators.

    The higher frequencies needs to be pushed through with MORE CURRENT. In the power equation, the current parameter is squared! You create more heat at higher frequencies. As a rule of thumb, and PCB design will create more heat at higher switching frequency.

    By doubling the thickness of ground/power planes, the resistance of the connection is halved, and also the lost power.


    The bulk of the resistance comes from the signal layer. This is because the thin traces has a smaller cross section compared to a ground plane. The result is that the ground plane resistance is very small. By doubling the ground plane, you are reducing an already small resistance by half, but your signal trace's thickness remains the same (and hence the resistance remains the same). The resistance therefore reduced by a small amount, but NOT HALVED.
  • -1 Hide
    khaydin , September 26, 2008 5:51 AM
    I wonder what stupid name Asus will call this when it rips off Gigabyte's new idea/feature.
  • 1 Hide
    ossie , September 26, 2008 7:26 AM
    @505090 Actually impedance and resistance are closely coupled concepts.
    Impedance is a complex quantity while resistance is the real part of it:
    Z=R+jX
    An ideal inductor/capacitor would only be represented by the imaginary part (reactance). In reality there are no ideal components, one of RLC is the primary quantity, but here are parasite components in every real one.

    @jliu83 I treated strictly the aspects of PCB power trace losses. Of course the influence on total power consumption is minimal if not negligible.
    Don't confuse signal with power. Because both ground and power planes have the cross section doubled, this actually means half the resistance _on the power bus_. I've already mentioned that doubling the thickness of the ground plane will have no practical influence on signal traces.
    Another useful side effect would be the halving of ground bounce when IC outputs switch (noise induced in ground).
  • 0 Hide
    neiroatopelcc , September 26, 2008 11:00 AM
    I'm amazed by the amount of electrical knowledge some of you possess! But I still have a question!
    Even when the positive effect of doubling the cobber is only significant with fast switching ic's I have to wonder if it at all affects overclocking. For all I know, the chipset itself is usually the limiting factor, and since they're not doubling the internal paths in the chips it would remain so? the only difference is a marginaly lower amount of wasted energy that would contribute to heat output of the chip - but that wouldn't been easier to negate with better cooling in the first place?

    ps. "Gigabyte has gone downhill in recent times. Not the same company that once made some outstanding AMD 939 boards a few years back. Now they're associated with high failure rates and poor QC, oh and now add BS marketing to that." THAT is complete rubbish! In the olden days I'd stick to abit, but in the last many years all new boards have been gigabyte. The nforce 4 sli board was great, the 965p was better, and the p34-ds4 I have now definetly is no less good than the 965 - so unless the boards before the nforce 4 would operate flawlessly in 100% humidity on a poor power supply in the rainforrest I can't think of how they could've gotten worse over the years - cause the current boards are awesome - can't possibly have been much better in the past... And let's not even get started about customer support - they actually have it! unlike asus and msi ..
  • -1 Hide
    mr roboto , September 26, 2008 2:15 PM
    neiroatopelcc: The reason I say that is from my own experience. Yours may be better. I've owned three AMD Gigabyte boards from 2000-2006 and they were the best. Rock solid and all around great. I bought a Gigabyte P45-DS4 and have been very disappointed. Having a look around different forums and seeing others having similar problems with their boards leads me to believe they are either not the same company they used to be or are not real good with the new Intel boards. Having read about the Asus and Gigabyte partner split, their quality seems to have gone down since that time.

    I loved my AMD Gigabyte boards and the EasyTune software was the best Windows overclocking utility I've ever used. All other Windows overclocking software would lock up my PC.
  • 0 Hide
    TheFace , September 26, 2008 4:20 PM
    Awww... I just got my EP45-DS3R with only 'Ultra Durable 2' technology. Damn you Gigabyte for making new technology that makes mine outdated the DAY after I get it!!!
  • 0 Hide
    Anonymous , September 26, 2008 5:30 PM
    For 1333MHz current, because of skin effect, 99.961% run within 1oz depth, and 99.999985% when calculate to 2oz. For 333MHz bus, the values are 98.02% and 99.961% respectively. Means for new mainboard with 1333MHz bus, you save 0.039% lost. Where is the meaning of "reduce loss"?

    I think the only advantage is heat spreading. Small chips without heatsink which uses PCB as heatsink, will be overheat when you overclock.
  • 1 Hide
    neiroatopelcc , September 26, 2008 7:54 PM
    @ roboto
    "Having read about the Asus and Gigabyte partner split, their quality seems to have gone down since that time"
    They actually never worked together. I was asking gigabyte about this last year when I heard the rumors. Turns out the joint venture was solely with the purpose of sharing manufacturing facilities, but the deal went to the floor. In the same reply the techie told me that a recent (independant) test of motherboard thermals showed that the asus 965 boards were significantly worse than gigabytes. They had many 'very hot' zones whereas gigabytes board maintained the same heat level everywhere on the board. So asus and gigabyte never split, cause they never agreed in the first place. Anyhow, perhaps the p45 boards aren't that good - I don't personally only know one guy who's bought one, and he did so two weeks ago. Haven't heard anything negative about it yet, but then he's not done with his watercooling yet. But the 965 and the p35 boards are both top quality. Nothing to complain about there.
  • 0 Hide
    neiroatopelcc , September 26, 2008 7:58 PM
    Supplement to the last message. Here's the Gigabyte response to the Asus matter dated 16-05-2007:

    "Answer - 474215
    Answer : HI,
    Thank you for your interest in Gigabyte products. First of all I can say that the joint company (not aquisition of Asus) Gigabyte United was supposed to be a way of using the same factories for manufacturing - the idea was that Gigabyte and Asus would still have their own designs, engineers, testing etc, so that a Gigabyte MB would still be a Gigabyte MB (but maybe manufacturde in a factory built by Asus) and vice versa.

    And secondly... I say "the idea was" because the joint venture has been called off. The new company Gigabyte United still exist, but is now a wholly owned doughter company of Gigabyte Technology focused on the motherboard and graphics card markets.

    FYI: In a resent comparison between GIGABYTE P35-DS3P and ASUS P5K measuring the tempertures of different key areas and components (running applications like 3DMark) the Asus board was between 8 and 75 (!) degrees hotter - something that will have a negative impact on stability, durability and overclocking potential.

    To sum things up: Gigabyte will continue to strive for the best quality that money can buy and will not stop competing with any of the other brands.

    Best regards,

    "
Display more comments