Pros and Cons of Digital PWM?

[HKPolice]

I'm surprised to see that Asus hasn't made the switch to Digital PWMs yet, even with their enthusiast mobos. And yet DFI and Abit who are more enthusiast oriented made the switch long ago for their high end products.

Why? I'm pretty sure DFI wouldn't have switched unless there is a net benefit to overclocking and yet I can't seem to find any comparison threads.

From what I remember, the pros are:

-More efficient power conversion = less heat pumped into the case, less stress on the PSU and your electricity bill
-Faster response times? So when CPU load changes rapidly, spikes in power demand are met = more stable overclocks

The cons I've read are:

-Massive Vdroop? I've done some research in the P35 Lanparty thread and it seems like there is a bios bug of some sorts where the vdroop option had it's effects reversed (turning it on = off and vice versa) so is this really an issue?
-The controller controller chip gets REALLY hot but I'm pretty sure it's designed to operate upto 105C... is this an issue?

[Brother Esau]

Vdroop on DFI P35 is non existent and the minute amount there is is so insignificant its stupid.

[strange|ife]

isnt ip35 analog pwm? iam not all up on the diffrences between the two. i would assume digital is the way to go.

glad your likein that DFI brother. dfi are some of the best boards still imo. I used to have a nforce 2 DFI and asus a7dlx which still runs 5 years later.

the dfi was better. I hope they get an x38 out at launch...

[pak]

-Massive Vdroop? I've done some research in the P35 Lanparty thread and it seems like there is a bios bug of some sorts where the vdroop option had it's effects reversed (turning it on = off and vice versa) so is this really an issue?

If digital PWM causes massive vdroop, I would guess that is why everyone hasn't moved over. The big plus to Asus's top of the line boards is there is no vdrop.

But like already mention, DFI's top of the line board has no vdrop.

Where are you getting your sources for your pros and cons?

--pak

[cabbspapp]

My Asus Commando with digital PWM has more vdroop the higher the voltage is set.

[Lestat]

i think no one uses it because it sounds good on paper but looses its appeal during the implimentation.

DFI started it and keeps using it, maybe beause they have a contract with the mfg of the pieces?

frankly with the massive heat from digital pwm i wouldnt use it either.

droop is something that can be controlled with some fo the cheapest parts on a motherboard but mobo makers just continue, after years of doing it, to keep right on doing it.

and asus is king of vdroop. i dont think there is another company out there who is more well known for their horrible vdroop than asus.

and the fix is just as cheap as the cause. 1 single resistor placed in the right spot is all thats needed, but they just dont do it.

[pak]

i think no one uses it because it sounds good on paper but looses its appeal during the implimentation.

Good on paper and bad on implimentation sounds like something that just needs a few bugs worked out.

--pak

[DEVIL K-ce]

My Asus Commando with digital PWM has more vdroop the higher the voltage is set.

Commando not have dig. pwm.

I tested only Quad GT with dig. pwm and it's ok with high vcore ( 1.85V+ ) no huge vdroop.

[Lestat]

Good on paper and bad on implimentation sounds like something that just needs a few bugs worked out.

--pak

well dont forget pak thats just 100% speculation.

given the way most mobo makers insist that they try to keep prices low, which means they use CHEAP parts i would believe no one uses it because digital pwm is expensive.

[dinos22]

one con is if the mobo manufacture hasn't provided enough vcore you can't mod it for higher vcore

that's what i read here somewhere anyways

[Brother Esau]

Well multiply a 1/3 of cent by 30 million and that their line of thinking!

Abit IP35 Pro was 4 phase Analog PWM and it had horrible Vdroop and ran much hotter to this digital PWM Abit 4phase Analog PWM = 42c idle 70c & Full Prime Load .........D.F.I 37c Idle 51C Full Prime load

Both with Q6600 G0 Abit @ 3.2ghz .......D.F.I @ 3500MHZ

[Lestat]

Well multiply a 1/3 of cent by 30 million and that their line of thinking!

Abit IP35 Pro was 4 phase Analog PWM and it had horrible Vdroop and ran much hotter to this digital PWM Abit 4phase Analog PWM = 42c idle 70c & Full Prime Load .........D.F.I 37c Idle 51C Full Prime load

Both with Q6600 G0 Abit @ 3.2ghz .......D.F.I @ 3500MHZ

no one ever said its bad aside form the heat and they do get hot.

analog pwm gets hot too but most mobo's dont have a heat removal system on them so they just sit there and cook in their own heat.

most people also dont use a fan on their PWM are which also needs to be done. heatsink or not.
a simple low rpm 60,70, or 80mm fan just blowing air across it helps out alot.

[Brother Esau]

Hey, Lestat whats up?

Don't get me wrong the Abit IP35 Pro is a Stellar Board and I was regretful when a USB port burned out with the all of the 4 they provide....lol...and I had no choice but to order another board just for that one port ...LOL...I know thats gay but I really needed all four to function for what I have!
Outside of that I never had a moments issue with the board at all

[Heidfirst]

isnt ip35 analog pwm?

it is but the IX38 series is going to the next generation of digital PWM.
The earlier type used on IN9 32X, AB9 QuadGT runs very hot (although it is specified to).

[NickS]

For now I'm sticking with Analog. Doesn't get _as_ hot and a good mobo won't droop. Digital has no advantage at this point in time IMO. I acutally try to stay AWAY from Digital. Every implementation I've seen of it, doesn't work real well.

[Adamantine]

I've read that analog is generally better unless you've got more phases with digital. IIRC, you'd need at least 8 phase digital for current motherboards to be good. Less can obviously get by, but it's not preferred.

[Pillo-kun]

Like someone already mentioned the v-drop gets larger if i bump up more volt.
Usually it is in 0.02 range but it can be as high as 0.06 over 1.6v. I think it was that high, remember seeing 1.54/1.55 in uguru at 1.6. i can get into win at 3.9 but a "no go" for anyhing more than superpi.
hope abit x38 will be better.

[OSKAR_WU]

1 . VDroop not Vdrop , Vdoop is a necessary power design spec in order to prevent voltage overshoot too much when the processor switching from heavy load to light load ... It's not a Bug or Poor Design caused Vdroop ...

2 . The volterra chip we used is so called Digital VRM technology , including a controller chip communicate digitally with some salve chip which has MOS integrated in it ... Because of this architecture , the whole VRM can operate at much higher frequency , the higher frequency the VRM can respond , the less the capacitor the VRM need for supply the surge current ...

3 . Digital VRM can supply more current when there is restricted area we can put the power circuit on the board ... Most of the ATI/nVidia high end display card use Digital VRM solution because there is not too much space in the display card , but the GPU needs lots of power ....

4 . The Slave chip with MOS integrated in it can handle 35~38A continuously per chip and it has 155 degree celsius temperature self protection , it will enter into protect mode when the current or temp limit is over .... Each chip is individual , that means if there is single chip over temp or current limit for a short time , it will protect itself and other chip will continue to work ...

5 . The cost of using digital VRM is much much higher compare to traditional VRM ...

6 . In light load , digital VRM generally has better power efficiency than traditional VRM ... In heavy load , digital VRM and traditional VRM power efficiency is almost the same ...

7 . Digital VRM won't generate much heat than traditional VRM , what makes you feel that is most of the circuit was put in a very small area ... If we put the Digital VRM in the same area with the same heat sink size as traditional VRM , it's temp will be close to traditional VRM ...

8 . In my personal opinion , with proper design in heat dissipation , the only cons is the price too high ... The other problem is the supply is not too good compare to traditional VRM , maybe this is why 1st tier player can not use it , they can not use a design which has supply problem ...

[STaRGaZeR]

So, Oskar, what happened with the iCFX3200 digital VRM? 0.08v~0.1v vdroop with low volts with dual cores @1,35v, let alone quads and high volts. Man, with that droop, it´s impossible to get any CPU stable, so I sold the board quickly. Since then, I really hate them.

[OSKAR_WU]

So, Oskar, what happened with the iCFX3200 digital VRM? 0.08v~0.1v vdroop with low volts with dual cores @1,35v, let alone quads and high volts. Man, with that droop, it´s impossible to get any CPU stable, so I sold the board quickly. Since then, I really hate them.

I don't know how you measure 0.08V ~ 0.1 Vdroop , but even so , it's not way tooooo big to cause system unstable ...

Check this to have a good understanding about Vdroop ... http://www.xcpus.com/forums/intel/14...epository.html

[virtualrain]

4 . The Slave chip with MOS integrated in it can handle 35~38A continuously per chip and it has 155 degree celsius temperature self protection , it will enter into protect mode when the current or temp limit is over .... Each chip is individual , that means if there is single chip over temp or current limit for a short time , it will protect itself and other chip will continue to work ...

Hi Oskar... why am I and so many other DFI 680i LT owners having troubles with boards shutting off under load (by load I mean Prime95 Small FFT's)? I have a water cooler on my slave chips that is keeping my NB and SB at 40-deg. in the same loop so the power chips can't be much higher than that. What's the real story with the shutdowns?

As a side question, is it not possible to design a board with an adjustable Vdroop so that enthusiasts can risk dialing it out if they choose. I don't see this as any more dangerous than any other BIOS setting put into the hands of the ill-informed.

[OSKAR_WU]

When the cpu load changes , no matter how fast the VRM can respond , there is always a very short period you need to use the Choke and Capacitor to maintain the voltage stable ... If we can design unlimited capacitor on the board , we can remove the droop design when the capacitor can handle all these transient state , but that's not possible ...

[virtualrain]

Thanks, back in my car audio days, we use to put a 1 Farad cap or more on the 12V line to the amps to help deal with transient power... would more capacitance on the 12V Rail that serves the CPU ATX connector be helpful in the PSU rather than on the board?

Any thoughts on my other question regarding why the DFI 680i LT may be shutting down under heavy load with a quad-core?

[OSKAR_WU]

Hi Oskar... why am I and so many other DFI 680i LT owners having troubles with boards shutting off under load (by load I mean Prime95 Small FFT's)? I have a water cooler on my slave chips that is keeping my NB and SB at 40-deg. in the same loop so the power chips can't be much higher than that. What's the real story with the shutdowns?

As a side question, is it not possible to design a board with an adjustable Vdroop so that enthusiasts can risk dialing it out if they choose. I don't see this as any more dangerous than any other BIOS setting put into the hands of the ill-informed.

First , with proper cooling and good contact and good TIM between the slave chip and heat sink , you should not encounter shutdown ... From what you have said , in your situation , it's the TIM you used on the slave chip and heatsink can not conduct the heat quick enough , the TIM we used in slave chip and heatsink is shinetsu X23-7783D which is the best TIM we can found except liquid metal ... You can trying using AS5(or any other TIM which is not electrically conductive) if you already remove all of them ...

Shutting down is caused by a chain reaction start form 1 slave chip , if all the slave chip is close to temperature limit , when one slave chip protect in temp , all the load will be shared by all the other slave chip ... This will easily caused other slave chip goes into temp protect , once all the slave chip shutdown one by one , vcore gones ... And the motherboard will shutdown because of Vcore gone , this is a protection to prevent cpu damaged ...

For future digital VRM design , we will use larger heatsink , also we will add circuit to control if Vdroop will be used or not ...

[STaRGaZeR]

I don't know how you measure 0.08V ~ 0.1 Vdroop , but even so , it's not way tooooo big to cause system unstable ...

Check this to have a good understanding about Vdroop ... http://www.xcpus.com/forums/intel/14...epository.html

Yeah, I know how Vdroop works, and I really don´t want it. Stable voltage is best for us XS members. Asus, DFI, Gigabyte... must implement an option in the BIOS to let us rock stable voltage under load, or a max of 0.01v~0.02v, like CPU Voltage Damper does in lastest Asus boards.

I measured it with the Smartguardian utility, didn´t know how to read it with a multimeter. Anyway, SM confirms what I saw in my tests: with the same CPU, Asus and my current Intel board behaviour is the same. Both measured with a well calibrated multimeter, after vdroop it needs 1,19v to a stable 3.3Ghz, a little 24/7 OC. With the iCFX3200, 1.35v in BIOS resulted in a 1,28v idle (no problems here, typical Vdrop), but at load, it dropped to about 1,20v. At that voltage it should be stable like it was on Asus and Intel, but it wasn´t, 1,25v or 1,30v under load didn´t fix it. Even at that low OC, that huge voltage fluctuations kill the stability. I played with the iCFX for about a month, trying to understand what I was doing wrong, but honestly, I think that the problem was a faulty board or something, not me. I really liked all the BIOS options, great board, but the voltage related stability issue forced me to get rid of it.

PS: Anyway, there are any proyects of VRM v2 or something like that, to fix current VRMs minor problems? I really like clear CPU area and reduced power circuit area too. If a board with it can disable Vdroop function, I will buy it, and more people too IMHO.