Seriously, they should feature you in one of those 'people are awesome' videos. Your explanation is impressive, but what I find truly exceptional is how effortlessly you manage to make nerd sound cool. Awesome stuff.
My fucking ears when the video started. But when i turned it down and listened, this video is rock solid. The MSI x399 Pro Carbon AC motherboard says it's 5 + 5 doubling.
you have a couple good points. 1. Yes if incorporated correctly they are effective in reducing the load per phase, which will result in a much cooler operation. However i was talking about 4 doubled phases isn't the same as having 8 true phases, the performance is higher with the true implementation. 2. That was the OTP guidelines not being correctly set, something you can't update and fix if you used an analog PWM with a resistor controlling OTP. Since then have you seen any blow?
Cute vid there. Some things I noted: 1. Using only a single doubler per phase is not as bad as you make it to be. Potential side-effects are non-existent in real life scenarios if the implementation is proper. 2. OTP for the transistors. Most companies implement it on their PWMs but on "digital" implementations they don't always react fast enough to save a transistor from smoking (see the x79 ud3 vid). Renesas should be noted here as one company that has 2-stage OTP on their newer transistors.
great work. would love to see an updated version of this, preferably for AMD. i find the labelling of 'soc vrm' slightly confusing given it's not a system on a chip but i guess they've packaged all the other things needing power into one or something. i'm interested in all the new infineon/renesas smart powerstages and pwm controllers too, just saying. oh wait, just noticed part 2 in the autoplay queue too...
Thanks for this video man! Very informative video when choosing the right motherboard for overclocking ! you're like motherboard guru or something, btw are you electronics engineer of some sort?
Great video, really helped. But i have question, i have Asrock 960GM - VGS3 FX, the shitty one right ? And i have Phenom II 965 BE, so i tried to overclock it on 3,9ghz, and raising voltage to 1,475 volts in bios. Run the Prime95 for 2 hours and everything was stable and max temp was 49c, cause i have Hyper 212 Evo cooler, but the problem is the motherboard, it doesn't have any VRM coolers and it's pretty cheap, is it safe to keep that settings on that motherboard ? Btw. voltage on 3.9ghz is 0.5V bigger than on stock speed 3.4ghz, so it's not something drastic, but still cheap asrock mobo ?! please reply tnx !
my MSI x99a SLI plus seems to have a vrm temp sensor that hwinfo64 reads and I've been having some concern about it. my 5820k is at 1.3v core and 1.3 ring with 1.9 input and 1.35v DRAM. I see temps under load between 100-106c . I'm sure it's good for up to 120c but it just bothers me that the heatsink seems to insulate instead of transfer the heat away, even with fans blowing directly on it. opinions?
A truly superb explaination of the VRM & PWM. I also followed your guide on OCN ( Im also a member ) on Bulldozer OC'ing and that was equally as detailed & educational. I enjoy your talks/work very educational and informative. Can you tell me if my Sabertooth 990FX 8+2 is any good or is it a false selling point ?
my z77 mpower says it has CPU Power: 16 Phases PCH Power: 1 Phase Memory Power: 2 Phase from a UPI uP1618A, a 6+2 phase controller. how did they manage to get 16+2+1 out of 6+2? do they do 4 with double doublers to get 16, use 1 for the PCH and leave then other out of the 6 bit, then the memory uses the +2 bit?
Hi ,I really liked your video and the way you explained everything.But i didn't find any information about ASRock H87 Fatal1ty anywhere.Can you tell me how many Phases it has?
I have a Asus P8z68 V pro motherboard , with an Nvidia GTX 660 FtW card. i just reinstalled the card because i thought it was bad so i had it for a back up. However i notice i was getting a lot of TDR error codes causing it to freeze during gaming BF3. Now i notice the feature on the board about DIGI VRM and made some adjustments . and now i have NO CRASHES. with my other card a ASUS GTX 760 OC i was getting Crashes ,BSOD errors . Havent reinstalled the card yet with the new settings for the DIGI VRM i found ,so im thinking the card is still good (only 4 years old). So can updates for windows cause the settings to change, because i never had to tweak this in the past?
Thanks... I wanna learn more about VRM after watching your video. 17:50 was hilarious tho... the way you counted with your fingers. It was like "oh shit... how can I show 24? Fuck it.. here is 10."
In fact - for me, the most important is that the MBO is stable with no VDrops. On the AMD MBO's there is no need for Digi PWM. (maybe Piledriver will change that, but i doubt it) However, i have no issues with that ISL chip, its OK, fine and old school, and the VRM is running cooler with analog PWM - best thing for budget PC builds with limited cooling power. High end class - Digi is great, but not all that counts. Good VRM components are needed too (lets not forget the MSI AMD 8xx series).
So I watched this video and had a question for you. I have the Gigabyte 990FX Gaming mobo and I thought it was a true 8+2 phase VRM, but apparently it is two 4+1's using doublers. I'm having a hard time with overclocking past 4.7Ghz on my 8350 despite VRM's being cool and CPU/Socket temps being under 60C. It basically just freezes while stress testing. My question is, is the fact that the VRM is using doublers contributing to the lack of vcore stability?
Hey sin0822 this is an awesome video. Listen i got a question to ask you, hope you can help me. I have this Biostar TF8200 mobo and one of the cpu mosfet was burn out. I got this this way. So i replace the mosfet, the burn was real bad. Now the mobo is working but the mosfets the one i replaced and one more are hitting like crazy. My chocks are hitting as well, the middle two and the capacitors as well. What puzzles me is that Vcore is stable, all MOBO temp in windows are under 45c. And the mosfets are around 80c with heatsink on. And i also hearing some squealing noise and am not sure is it from fets, chocks or some capacitors. What can cause the fets to boil like this? Can u pls shade any light on this one, huge thanks in advance!
I have a question, the 12 volt supply is switched on and off, and the longer it is on the higher the output voltage. Since we are going down from 12 to 1.2 it is not on long. The inductors filter these current pulses to something constant right? What are the caps doing here? is it the ripple current/voltage they suppress? or is there some varying voltage from the mosfets to be filtered?
Loved watching your video, I have a GTX460GS graphic card and like to make a mod in PWM NCP5395T but it´s hard to find the resistor to weld the trimpot could make a video on the subject.
hi, do you think it would make sense to take the VRM-heatsink from my Asrock z77 extreme 3, srew them loose and then put new thermal compound on it? Maybe new pads or paste? Because its 2 years working now and maybe the contact between mosfet and heatsink is not so good anymore? Just an idea. Because i have anything here, pads, paste, even copper for building my own heatsinks. i think under my heatsinks there are even this "DPAK-Mosfets" which you say are awful.
Hi i have problrm with my crosshair v. Every some time board freez when i make reset no boot just cpu red led fans full speed no read bios.. When i press heatsink on board (mosfet (or left pc off for night from socket) after this board power up. Works ok some time and again make this poo. This is vrm problem ?
ASRock was always part of pegatron. ISl6330 is a pretty standard analog PWM, LLC is controlled through a resistor set between 2 pins, pretty standard stuff. But yea it isn't the ISL6366 and ISl6336G that are use don intel boards. I just don't think AMD gets as much ove inthe VRm area like Intel side. One issue now that ASRock lost the CHL3828(great PWM BTW), is that they have to redo a lot of their new boards, like x79 extreme11 VRM had to be changed. Check out the LLC on the new IR PWms.
Some Asus EVGA and even Asrock boards also have 8 "real" phases. MSI boards will probably appear too in the near future since they started using digital modulators on some of their boards. Now, the number of phases and if they are doubled or not does not matter in practice unless you push the regulator to its output limit. For this to happen the coils (chokes) must be able to handle the amperage the transistors give too. No point in bundling a 50A transistor with a 20A coil.
Well ASRock belong to Pegatron for years now, since then they improved a lot. I'm more into the AMD marketing for the boards and I have to tell You, the UD7 rev1.2 uses the same ISL6330 and ISL6617 like on rev1.0 and rev1.1.Nothing changed, they only added the LLC components for greater good. Anyway, the board is OK, but surprise surprise, the ECS is rock solid board. Don't get me wrong, I own the Fatal1ty, but this ECS is sure kicking some ass.
Well i am not saying you are wrong, I am just saying that board is a year old. In my VRm video i was referring to ASRock Z77 boards, from Z68 to X79 ASRock used Digital CHiL PWM, however since ASUS pulled out of Pegatron(ASRock parent company), ASRock is no longer using the CHiL Digital PWM, and has switched to an analog one from Intersil, the same guys who make the PWm on the 990FXA-UD7(which is a year old). GIGABYTE no longer uses Intersil on any UD3 or above(I am not sure about AMD side).
You told when one phase is on all other phase are in off mode.... I know... one phase can carry 70A off curret.....in this scenario each phase have to carry 70A but not at a time or rather in a periodic manner ...... So what if when cpu need current more than 70 A ....does the pwm chip increase or decrease the frequency in that case.... Please tell me....i want to know... please reply
Also one thing I forgot to mention, even tho ASRock isn't using a digital PWm on any Z77 board, they are stating they use digital PWM, and that is bothersome. You work for marketing AMD board?
Question... I have an old Dell Motherboard with 3 chokes and 3 mosfets per choke next to the CPU. How much power can they reliably deliver? How can I figure that out? Why does this board only have 3 chokes and more mosfets than your example?
Also, the physical 12v power connector for the processor must be able to deliver the required power. Using one EPS connector that power ranges from 300 to 400 watts (depending on wire thickness and connection sturdiness). No point in using a regulator able to handle 1000 watts (or something) on a board with a single EPS connector. Gigabyte's z68x-ud5 and Asrock's fatality z77 professional are major offenders here since they use like 15-20 phases... but a single EPS connector.
On the phases, it used to be that with economics in mind you could get 15-20 amps per phase and at 1 volt about 15-20 watts per phase. On a low power cpu (up to 65 watts) you need 3-4 phases and medium cpu (95w) about 5-6. There was 125w-150w cpu for a while so 8-10 phases. Power trend is level to down so maybe that was the peak. Can modern MOBO do better that 20amps per phase reliably?
i will make a build asus p8z77-v is this a good mobo? im not in overclocking but games,ill just add a good gpu on it,i know this board has a doubling mosfets,is there any affect in doubling power phases in performance? thanks
Does the pwm control the boot up sequence for the processor as well, or is that another chip? I'm looking all over trying to find diagrams for the mobo I'm trying to repair/understand and not having any luck, and not able to even find any diagrams for similar mobos that would tell me what the power on/post sequence is as far as what happens electrically. My board is just dead, no power, no fans, nothing when you press the switch so trying to figure out if there's a short somewhere or what.
+Madderhatter The PCH controls the boot up sequence of the CPU and it controls power on for the PWM the last I heard. You will have trouble finding information on these things about specific boards because manufacturers keep a very tight lid on them (they are afraid other board makers will copy them). Intel also doesn't give out much, but you should be able to find the datasheet for the chipset on your motherboard, it should have startup info.
I am actually glad you did that, because it is false marketing. GB used ISL6366 they didn't say it was digital, they embraced the fact that it was analog. But now ASRock is using ISL6367. If you google ISL6366 or 67 you will see no PID, no ADC to convert the voltage and current sensing signals.. Intersil is now calling it Green hybrid-digital, which is still BS, but better than fully digital, but still not digital. MSi says hybrid digital and ihave no fault with them, asrock needs to do this.
I've known about your VRM list for a long time, and recommended Z87/97 boards based on the details contained in your list. However, a lot of people dismiss my suggestions as Gigabyte fanboyism; it's not hard to see why this happens, since GB has maintained surprisingly high quality in the VRMs of their midrange to high end boards for a long time. Except for their recent slight slip up in their 9-series boards, supposedly due to Haswell's lesser VRM requirements. Not enough people care to choose boards based on how their VRMs really work, and it's been made worse by Haswell's FIVR. When people hear that the CPU has an integrated VR, they immediately think that the motherboard's VR no longer matters, and shitty doubled analog uPIs on MSI's boards are on par with GB's IR3563A PWMs. People still regularly dismiss the Z97MX Gaming 5 in favor of the Z97M Gaming. One question for you Sin, what role does the output capacitance play in boards that have IR digital PWMs for example (and let's just say not doubled at all, for a fairer playing field)? You mentioned that ASRock's implementation of analog ISL6367 necessitated lots and lots of output caps; in a digital, true phase board like the Z97MX Gaming 5 or the SOC Force, does the total capacitance matter much for overclocking? Does it matter as much as MOSFETs' current rating or the PWM controller? And lastly, why would the Z77 Extreme4 need as high a rating as 8200uF? My first (albeit regrettable) purchase was a Z77 Extreme3, and I'm assuming that it also has a ISL6367, and what's got me scratching my head is how ASRock managed to omit a phase and bring it down to 7 from 8 doubled?
Not really seen any since then, even including those really bad low end asus boards that would easily die otherwise. In any case, what I wanted to note was Renesas' decision to focus more on the protection of their transistors instead of the output. There are always companies that do bad implementations and users that don't know what they're doing. Protection on the transistor itself with no attachment to other components can save you even from a hacked bios with disabled otp or some bug on it.
Explanation is great... But can you connect the pwm ic and mosfet with inductor and capacitor with feed back to pwm ic making the above circuit for power supply to CPU. The explanation on drawing board is ok for literally. But no explanation for VRM section with live practical is showed ever. Can you make a video with all the components in VRM Section and showing the output voltage for cpu.
Why a pwm chip is required .....after all at last cpu is getting constant voltage.......so this switching arrangement... please tell me... please I want to know..... Is this only sake of strain the ac components from the supply voltage
First off the CPu doesn't operate off constant voltage, it goes up and down to save power and adjust to different conditions, second the controller is required when the CPu takes huge current steps, which can be around 100A.
+sin0822 can you tell us what is the max temp for VRM, cause one of mine pc freeze when playing game for some time, all usb devices switchs off and only current screen that was on monitor stays in freeze, when i go back in bios i can see that VRM temp was 61*C, well this is cause i have two gpu with custom fans not reverence, that makes case hotter.
+iPlay4K -Adi™ | 4K UHD Gaming! So the VRMs usually go to about 120C until any type of over temperature protection will trigger. When OTP triggers, two things happen. First thing is that it will try to throttle the CPU speed, and on most platforms if the VRM gets hot it will throttle the CPU to 800MHz. The second thing is that it will trigger a shutdown, an immediate shutdown. The type of freeze you are seeing isn't really an overheating of the VRM, I would look for another cause, it seems like overall instability (not due to VRM).
Also, 990FXA-UD7 doesn't use a digital PWM like gigabyte is using now, nor does it use the new power stages or chokes. GIGABYTe 990FXA-UD7 is basically 8 phases of a GIGABYTE Z68 or X58 G1.Sniper 2, 4 phases doubled. I don't really follow AMD, but what is funny is that their 990FXA-UD3 has no vdroop. Are you use you have LLC on your 990FXA-UD7? What level did you set it at?
The GD70 is a good MBO, but the 890FXA-GD65 is superior in OC. And again, MSI "locked up" some stuff in BIOS for the GD65 -> You can modify the GD70 BIOS and load it to GD65 -> then You can do stuff that You cant imagine with GD70. So I sold the GD70 after 2 RMA's. There are no god/bad motherboards, because every motherboard has been placed fine for some market value. I like ASRock boards - they run stable and long (entry/mid segment) - not like ASUS who likes to die a lot in entry level.
I had the 890FXA-GD70, nothing good there. VRM hotter than hell. I don't want to see any entry leven ASUS boards, they are a disaster. The same counts for Gigabyte, but they have the "bad and cheep" chip-set cooler problem - but they have improved over the years. Still, there are always beater solution in every class. ECS/BIOSTAR/ASRock for entry level are great. MSI/ASRock/FoxConn for mid-range are very good. ASUS/GB/ASRock/EVGA for high-end are superb. Intel boards are great overall too.
"Good VRM components are needed too (lets not forget the MSI AMD 8xx series)." Components had nothing to do with it. It was the design being sub-par. Not to mention that the components of boards like 890FXA GD70 were so good that they could handle a bit more than supposed to. It's just that "a bit more" is not the same as "way more". You can't power up a car with a set of 1.5v AA batteries. Low end Asus boards use crappy components yet they survive bad treatment solely because of their design.
So you're saying that you know what were the parts used on that 890FXA-GD70 and that they were somehow bad? Could you be more precise? You can start by explaining what were the components used and what should be used instead. Note that I'm not talking about amounts here, just the components themselves. Anyway, considering how things were and how things are right now there are no bad companies (well, maybe excluding Asrock :p) and good companies just bad and good motherboards.
Ok. I used this setup: MSI 990FXA-GD80 (8+2) Gigabyte 990FXA-UD7 (8+2) ECS 990FXM-A (6+1) ASRock 990FX Fatal1ty (12+2) Best OC was the ECS and ASRock. Most vDrops -> the Gigabyte WITH LLC. How can You explain that? So ASRock is to stupid to fix something that works well? Or Gigabyte is to stupid to create first series of MBO's without LLC, and then fix it with LLC - and the results are the same? Explain it to us please. Haters gonna hate.
All ASRock Intel boards are greater then AMD's - because 80% of all ASRock market is based on Intel systems. So is it and so will it be for next 2+ years and more. CHL3828 is sure more powerful than ISl6330 - but on the other side - i like analog PWM more than digital ones. Why? No analog PWM has failed me anytime - but digital have. I hope to see and hear what ASRock has to say about that - I sent them email about their Z77 boards, so maybe they can explain their Z77 "digi PWM".
I have watched many videos about mother board repair and learned something from it but I didn't learned anything from this video. This is not for a layman like me. You shouldn't have mentioned it in the beginning. It could have saved 30 minutes of my life learning basics.
+sin0822 Didn't expected such silly and cheap comment from a "technical" person. I do watch entire videos to learn 1 or 2 new points from it and I watched yours coz you mentioned "layman terms" and continued that something I can learn. You might be a "technical" person but not a good teacher/instructor. There are much better people who know better than you and good instructors and better persons as well. Don't post silly comments and don't waste my time.
+Sanjay Prasad Holy crap that was about as basic layman as you get. I followed it and I'm just starting out on mobo components, how more basic can you get. There are no technical people, just idiots and people that use their brain.
Gigabyte is like Samsung used to be, and asus is like sony used to be with propietary and expensive shit, guess who is number 1 on electronic devices today?
the way this video is explained is a little masterpiece
if all your video are like this one i'll want to see them all for sure.
Seriously, they should feature you in one of those 'people are awesome' videos. Your explanation is impressive, but what I find truly exceptional is how effortlessly you manage to make nerd sound cool. Awesome stuff.
My fucking ears when the video started.
But when i turned it down and listened, this video is rock solid. The MSI x399 Pro Carbon AC motherboard says it's 5 + 5 doubling.
Verrry informative! Thank you good sir.
awesome video.
Great video
you have a couple good points.
1. Yes if incorporated correctly they are effective in reducing the load per phase, which will result in a much cooler operation. However i was talking about 4 doubled phases isn't the same as having 8 true phases, the performance is higher with the true implementation.
2. That was the OTP guidelines not being correctly set, something you can't update and fix if you used an analog PWM with a resistor controlling OTP. Since then have you seen any blow?
Cute vid there. Some things I noted:
1. Using only a single doubler per phase is not as bad as you make it to be. Potential side-effects are non-existent in real life scenarios if the implementation is proper.
2. OTP for the transistors. Most companies implement it on their PWMs but on "digital" implementations they don't always react fast enough to save a transistor from smoking (see the x79 ud3 vid). Renesas should be noted here as one company that has 2-stage OTP on their newer transistors.
You talk like an electronics engineer. Bravo!
When i say pulled out, i am referring to ASUS selling its 30% share.
Hi Steve. Does ASUS M5A97 EVO r2.0 has a TRUE 6+2 Phase or ASUS has used doubler? What about Gigabyte GA-970A UD3, which says 8+2 Phase. True?
Good video
So please tell me is the chip work in such a way...
When cpu need more power it increases the pulse width ....
Lower when not requird
Please reply
great work. would love to see an updated version of this, preferably for AMD. i find the labelling of 'soc vrm' slightly confusing given it's not a system on a chip but i guess they've packaged all the other things needing power into one or something. i'm interested in all the new infineon/renesas smart powerstages and pwm controllers too, just saying. oh wait, just noticed part 2 in the autoplay queue too...
sick, thanks a lot
Thanks for this video man! Very informative video when choosing the right motherboard for overclocking ! you're like motherboard guru or something, btw are you electronics engineer of some sort?
Hey dude i was thinking of buying gtx 960 armor 2x and it has a 3+1 pwm...Willnit affect overclocking?
10:58 thats quick af boi
thx for the video thought
Great video, really helped. But i have question, i have Asrock 960GM - VGS3 FX, the shitty one right ? And i have Phenom II 965 BE, so i tried to overclock it on 3,9ghz, and raising voltage to 1,475 volts in bios. Run the Prime95 for 2 hours and everything was stable and max temp was 49c, cause i have Hyper 212 Evo cooler, but the problem is the motherboard, it doesn't have any VRM coolers and it's pretty cheap, is it safe to keep that settings on that motherboard ?
Btw. voltage on 3.9ghz is 0.5V bigger than on stock speed 3.4ghz, so it's not something drastic, but still cheap asrock mobo ?! please reply tnx !
my MSI x99a SLI plus seems to have a vrm temp sensor that hwinfo64 reads and I've been having some concern about it. my 5820k is at 1.3v core and 1.3 ring with 1.9 input and 1.35v DRAM. I see temps under load between 100-106c . I'm sure it's good for up to 120c but it just bothers me that the heatsink seems to insulate instead of transfer the heat away, even with fans blowing directly on it. opinions?
A truly superb explaination of the VRM & PWM. I also followed your guide on OCN ( Im also a member ) on Bulldozer OC'ing and that was equally as detailed & educational. I enjoy your talks/work very educational and informative. Can you tell me if my Sabertooth 990FX 8+2 is any good or is it a false selling point ?
my z77 mpower says it has
CPU Power: 16 Phases
PCH Power: 1 Phase
Memory Power: 2 Phase
from a UPI uP1618A, a 6+2 phase controller. how did they manage to get 16+2+1 out of 6+2? do they do 4 with double doublers to get 16, use 1 for the PCH and leave then other out of the 6 bit, then the memory uses the +2 bit?
Asus z97-a uses ASP1252 (IR3584) controller which i think have 4+1 phases.. Is it good ?
Hi ,I really liked your video and the way you explained everything.But i didn't find any information about ASRock H87 Fatal1ty anywhere.Can you tell me how many Phases it has?
nyc work
I have a Asus P8z68 V pro motherboard , with an Nvidia GTX 660 FtW card. i just reinstalled the card because i thought it was bad so i had it for a back up. However i notice i was getting a lot of TDR error codes causing it to freeze during gaming BF3. Now i notice the feature on the board about DIGI VRM and made some adjustments . and now i have NO CRASHES. with my other card a ASUS GTX 760 OC i was getting Crashes ,BSOD errors . Havent reinstalled the card yet with the new settings for the DIGI VRM i found ,so im thinking the card is still good (only 4 years old). So can updates for windows cause the settings to change, because i never had to tweak this in the past?
No, i'm working in a hardware power-shop -> repair department.
But i'm familiar with all ASRock AMD MBO's because I use them a lot there.
Thanks... I wanna learn more about VRM after watching your video.
17:50 was hilarious tho... the way you counted with your fingers. It was like "oh shit... how can I show 24? Fuck it.. here is 10."
In fact - for me, the most important is that the MBO is stable with no VDrops.
On the AMD MBO's there is no need for Digi PWM. (maybe Piledriver will change that, but i doubt it)
However, i have no issues with that ISL chip, its OK, fine and old school, and the VRM is running cooler with analog PWM - best thing for budget PC builds with limited cooling power.
High end class - Digi is great, but not all that counts. Good VRM components are needed too (lets not forget the MSI AMD 8xx series).
Question. Does M4A89GTD PRO AM3 Asus has True 8+2 or not?
So I watched this video and had a question for you. I have the Gigabyte 990FX Gaming mobo and I thought it was a true 8+2 phase VRM, but apparently it is two 4+1's using doublers. I'm having a hard time with overclocking past 4.7Ghz on my 8350 despite VRM's being cool and CPU/Socket temps being under 60C. It basically just freezes while stress testing. My question is, is the fact that the VRM is using doublers contributing to the lack of vcore stability?
Hey sin0822 this is an awesome video. Listen i got a question to ask you, hope you can help me. I have this Biostar TF8200 mobo and one of the cpu mosfet was burn out. I got this this way. So i replace the mosfet, the burn was real bad. Now the mobo is working but the mosfets the one i replaced and one more are hitting like crazy. My chocks are hitting as well, the middle two and the capacitors as well. What puzzles me is that Vcore is stable, all MOBO temp in windows are under 45c. And the mosfets are around 80c with heatsink on. And i also hearing some squealing noise and am not sure is it from fets, chocks or some capacitors. What can cause the fets to boil like this? Can u pls shade any light on this one, huge thanks in advance!
I have a question, the 12 volt supply is switched on and off, and the longer it is on the higher the output voltage. Since we are going down from 12 to 1.2 it is not on long. The inductors filter these current pulses to something constant right? What are the caps doing here? is it the ripple current/voltage they suppress? or is there some varying voltage from the mosfets to be filtered?
Loved watching your video, I have a GTX460GS graphic card and like to make a mod in PWM NCP5395T but it´s hard to find the resistor to weld the trimpot could make a video on the subject.
Alvaro Gomes any electronic copying different types
hi, do you think it would make sense to take the VRM-heatsink from my Asrock z77 extreme 3, srew them loose and then put new thermal compound on it? Maybe new pads or paste? Because its 2 years working now and maybe the contact between mosfet and heatsink is not so good anymore? Just an idea. Because i have anything here, pads, paste, even copper for building my own heatsinks. i think under my heatsinks there are even this "DPAK-Mosfets" which you say are awful.
SIN08222 is the hero we need right now....but not the one we deserve.
Hi i have problrm with my crosshair v. Every some time board freez when i make reset no boot just cpu red led fans full speed no read bios.. When i press heatsink on board (mosfet (or left pc off for night from socket) after this board power up. Works ok some time and again make this poo. This is vrm problem ?
ASRock was always part of pegatron. ISl6330 is a pretty standard analog PWM, LLC is controlled through a resistor set between 2 pins, pretty standard stuff. But yea it isn't the ISL6366 and ISl6336G that are use don intel boards. I just don't think AMD gets as much ove inthe VRm area like Intel side. One issue now that ASRock lost the CHL3828(great PWM BTW), is that they have to redo a lot of their new boards, like x79 extreme11 VRM had to be changed. Check out the LLC on the new IR PWms.
Some Asus EVGA and even Asrock boards also have 8 "real" phases. MSI boards will probably appear too in the near future since they started using digital modulators on some of their boards.
Now, the number of phases and if they are doubled or not does not matter in practice unless you push the regulator to its output limit. For this to happen the coils (chokes) must be able to handle the amperage the transistors give too. No point in bundling a 50A transistor with a 20A coil.
Well ASRock belong to Pegatron for years now, since then they improved a lot. I'm more into the AMD marketing for the boards and I have to tell You, the UD7 rev1.2 uses the same ISL6330 and ISL6617 like on rev1.0 and rev1.1.Nothing changed, they only added the LLC components for greater good. Anyway, the board is OK, but surprise surprise, the ECS is rock solid board.
Don't get me wrong, I own the Fatal1ty, but this ECS is sure kicking some ass.
Well i am not saying you are wrong, I am just saying that board is a year old. In my VRm video i was referring to ASRock Z77 boards, from Z68 to X79 ASRock used Digital CHiL PWM, however since ASUS pulled out of Pegatron(ASRock parent company), ASRock is no longer using the CHiL Digital PWM, and has switched to an analog one from Intersil, the same guys who make the PWm on the 990FXA-UD7(which is a year old). GIGABYTE no longer uses Intersil on any UD3 or above(I am not sure about AMD side).
You told when one phase is on all other phase are in off mode....
I know... one phase can carry 70A off curret.....in this scenario each phase have to carry 70A but not
at a time or rather in a periodic manner ......
So what if when cpu need current more than 70 A ....does the pwm chip increase or decrease the frequency in that case....
Please tell me....i want to know... please reply
Also one thing I forgot to mention, even tho ASRock isn't using a digital PWm on any Z77 board, they are stating they use digital PWM, and that is bothersome. You work for marketing AMD board?
Question...
I have an old Dell Motherboard with 3 chokes and 3 mosfets per choke next to the CPU. How much power can they reliably deliver? How can I figure that out? Why does this board only have 3 chokes and more mosfets than your example?
Also, the physical 12v power connector for the processor must be able to deliver the required power. Using one EPS connector that power ranges from 300 to 400 watts (depending on wire thickness and connection sturdiness). No point in using a regulator able to handle 1000 watts (or something) on a board with a single EPS connector. Gigabyte's z68x-ud5 and Asrock's fatality z77 professional are major offenders here since they use like 15-20 phases... but a single EPS connector.
On the phases, it used to be that with economics in mind you could get 15-20 amps per phase and at 1 volt about 15-20 watts per phase. On a low power cpu (up to 65 watts) you need 3-4 phases and medium cpu (95w) about 5-6. There was 125w-150w cpu for a while so 8-10 phases. Power trend is level to down so maybe that was the peak. Can modern MOBO do better that 20amps per phase reliably?
How can you tell if you mobo has a doubler?
i will make a build asus p8z77-v is this a good mobo? im not in overclocking but games,ill just add a good gpu on it,i know this board has a doubling mosfets,is there any affect in doubling power phases in performance? thanks
this is exactly why ichose the z77xud5h and to this day even with a 4770k given to me i wont give up this board
Does the pwm control the boot up sequence for the processor as well, or is that another chip? I'm looking all over trying to find diagrams for the mobo I'm trying to repair/understand and not having any luck, and not able to even find any diagrams for similar mobos that would tell me what the power on/post sequence is as far as what happens electrically. My board is just dead, no power, no fans, nothing when you press the switch so trying to figure out if there's a short somewhere or what.
+Madderhatter The PCH controls the boot up sequence of the CPU and it controls power on for the PWM the last I heard. You will have trouble finding information on these things about specific boards because manufacturers keep a very tight lid on them (they are afraid other board makers will copy them). Intel also doesn't give out much, but you should be able to find the datasheet for the chipset on your motherboard, it should have startup info.
Could you find out how many true phases the GA-970A-UD3P has?
yes.. i'm quite curious too since i use it ... what does the "split power plane" on it really mean?
What PWM does Biostar use in their TZ77XE4's? Thx
I am actually glad you did that, because it is false marketing. GB used ISL6366 they didn't say it was digital, they embraced the fact that it was analog. But now ASRock is using ISL6367. If you google ISL6366 or 67 you will see no PID, no ADC to convert the voltage and current sensing signals.. Intersil is now calling it Green hybrid-digital, which is still BS, but better than fully digital, but still not digital. MSi says hybrid digital and ihave no fault with them, asrock needs to do this.
I've known about your VRM list for a long time, and recommended Z87/97 boards based on the details contained in your list. However, a lot of people dismiss my suggestions as Gigabyte fanboyism; it's not hard to see why this happens, since GB has maintained surprisingly high quality in the VRMs of their midrange to high end boards for a long time. Except for their recent slight slip up in their 9-series boards, supposedly due to Haswell's lesser VRM requirements.
Not enough people care to choose boards based on how their VRMs really work, and it's been made worse by Haswell's FIVR. When people hear that the CPU has an integrated VR, they immediately think that the motherboard's VR no longer matters, and shitty doubled analog uPIs on MSI's boards are on par with GB's IR3563A PWMs. People still regularly dismiss the Z97MX Gaming 5 in favor of the Z97M Gaming.
One question for you Sin, what role does the output capacitance play in boards that have IR digital PWMs for example (and let's just say not doubled at all, for a fairer playing field)? You mentioned that ASRock's implementation of analog ISL6367 necessitated lots and lots of output caps; in a digital, true phase board like the Z97MX Gaming 5 or the SOC Force, does the total capacitance matter much for overclocking? Does it matter as much as MOSFETs' current rating or the PWM controller? And lastly, why would the Z77 Extreme4 need as high a rating as 8200uF? My first (albeit regrettable) purchase was a Z77 Extreme3, and I'm assuming that it also has a ISL6367, and what's got me scratching my head is how ASRock managed to omit a phase and bring it down to 7 from 8 doubled?
What revision is your gigabyte UD7.
Not really seen any since then, even including those really bad low end asus boards that would easily die otherwise.
In any case, what I wanted to note was Renesas' decision to focus more on the protection of their transistors instead of the output. There are always companies that do bad implementations and users that don't know what they're doing. Protection on the transistor itself with no attachment to other components can save you even from a hacked bios with disabled otp or some bug on it.
jeez that audio sounds like mine dimonds
@sin0822, what is your educational background?
Explanation is great... But can you connect the pwm ic and mosfet with inductor and capacitor with feed back to pwm ic making the above circuit for power supply to CPU. The explanation on drawing board is ok for literally. But no explanation for VRM section with live practical is showed ever. Can you make a video with all the components in VRM Section and showing the output voltage for cpu.
Why a pwm chip is required .....after all at last cpu is getting constant voltage.......so this switching arrangement... please tell me... please
I want to know.....
Is this only sake of strain the ac components from the supply voltage
First off the CPu doesn't operate off constant voltage, it goes up and down to save power and adjust to different conditions, second the controller is required when the CPu takes huge current steps, which can be around 100A.
1.2 with LLC and AM3+ BIOS support "out of the box".
fckin awesome shit
is that a typo "CHL3828"?
'couse i found on your vrm list CHL8328 for MvG..
great vid btw
+sin0822 can you tell us what is the max temp for VRM, cause one of mine pc freeze when playing game for some time, all usb devices switchs off and only current screen that was on monitor stays in freeze, when i go back in bios i can see that VRM temp was 61*C, well this is cause i have two gpu with custom fans not reverence, that makes case hotter.
+iPlay4K -Adi™ | 4K UHD Gaming! So the VRMs usually go to about 120C until any type of over temperature protection will trigger. When OTP triggers, two things happen. First thing is that it will try to throttle the CPU speed, and on most platforms if the VRM gets hot it will throttle the CPU to 800MHz. The second thing is that it will trigger a shutdown, an immediate shutdown. The type of freeze you are seeing isn't really an overheating of the VRM, I would look for another cause, it seems like overall instability (not due to VRM).
sin0822
thank u very much, what about south-bridge?
+iPlay4K -Adi™ | 4K UHD Gaming! Modern motherboard don't have the north/southbridge anymore. Everything into the CPU and the PCH.
I see i haw 10 inductors and 10 larger mosfets + 10 smaler they in one side are blacker mybe from overclocking.
Awsome viedeo but pls fix your sound and when u have a top view camera set it so manual focus so there is no pumping of the focus motor.
How to made this video, Great Explanation
Also, 990FXA-UD7 doesn't use a digital PWM like gigabyte is using now, nor does it use the new power stages or chokes. GIGABYTe 990FXA-UD7 is basically 8 phases of a GIGABYTE Z68 or X58 G1.Sniper 2, 4 phases doubled. I don't really follow AMD, but what is funny is that their 990FXA-UD3 has no vdroop. Are you use you have LLC on your 990FXA-UD7? What level did you set it at?
So why did they make multiple phases when all you need is a constant output?
Nvm, you explained it after the pwm
The GD70 is a good MBO, but the 890FXA-GD65 is superior in OC.
And again, MSI "locked up" some stuff in BIOS for the GD65 -> You can modify the GD70 BIOS and load it to GD65 -> then You can do stuff that You cant imagine with GD70. So I sold the GD70 after 2 RMA's.
There are no god/bad motherboards, because every motherboard has been placed fine for some market value.
I like ASRock boards - they run stable and long (entry/mid segment) - not like ASUS who likes to die a lot in entry level.
I had the 890FXA-GD70, nothing good there. VRM hotter than hell.
I don't want to see any entry leven ASUS boards, they are a disaster.
The same counts for Gigabyte, but they have the "bad and cheep" chip-set cooler problem - but they have improved over the years.
Still, there are always beater solution in every class.
ECS/BIOSTAR/ASRock for entry level are great.
MSI/ASRock/FoxConn for mid-range are very good.
ASUS/GB/ASRock/EVGA for high-end are superb.
Intel boards are great overall too.
Nice job Sin...Hard to believe ASUS Z97 Deluxe has 16 phase...! It must be so inneficiant....
20:43 Shit gets real.
"Good VRM components are needed too (lets not forget the MSI AMD 8xx series)."
Components had nothing to do with it. It was the design being sub-par. Not to mention that the components of boards like 890FXA GD70 were so good that they could handle a bit more than supposed to. It's just that "a bit more" is not the same as "way more". You can't power up a car with a set of 1.5v AA batteries.
Low end Asus boards use crappy components yet they survive bad treatment solely because of their design.
zoom in components next time. great video
main pwm driver+driver+doublers i dont understand
"what if you have 6 but you want 8? do you take 4 and double then take 2 direct? fuck no you dont!"
So you're saying that you know what were the parts used on that 890FXA-GD70 and that they were somehow bad? Could you be more precise? You can start by explaining what were the components used and what should be used instead. Note that I'm not talking about amounts here, just the components themselves.
Anyway, considering how things were and how things are right now there are no bad companies (well, maybe excluding Asrock :p) and good companies just bad and good motherboards.
What motherboard is that?
X79-UP4
Ok. I used this setup:
MSI 990FXA-GD80 (8+2)
Gigabyte 990FXA-UD7 (8+2)
ECS 990FXM-A (6+1)
ASRock 990FX Fatal1ty (12+2)
Best OC was the ECS and ASRock. Most vDrops -> the Gigabyte WITH LLC.
How can You explain that? So ASRock is to stupid to fix something that works well?
Or Gigabyte is to stupid to create first series of MBO's without LLC, and then fix it with LLC - and the results are the same?
Explain it to us please.
Haters gonna hate.
20:40 lol
All ASRock Intel boards are greater then AMD's - because 80% of all ASRock market is based on Intel systems. So is it and so will it be for next 2+ years and more.
CHL3828 is sure more powerful than ISl6330 - but on the other side - i like analog PWM more than digital ones. Why? No analog PWM has failed me anytime - but digital have. I hope to see and hear what ASRock has to say about that - I sent them email about their Z77 boards, so maybe they can explain their Z77 "digi PWM".
Cool but the summary is don't buy an Asus or asrock board so I bought a gigabyte thanks
I have watched many videos about mother board repair and learned something from it but I didn't learned anything from this video. This is not for a layman like me. You shouldn't have mentioned it in the beginning. It could have saved 30 minutes of my life learning basics.
+Sanjay Prasad if it was too technical then why did you continue watching for 30 minutes? Do you just like my voice ? lol
+sin0822 Didn't expected such silly and cheap comment from a "technical" person. I do watch entire videos to learn 1 or 2 new points from it and I watched yours coz you mentioned "layman terms" and continued that something I can learn. You might be a "technical" person but not a good teacher/instructor. There are much better people who know better than you and good instructors and better persons as well. Don't post silly comments and don't waste my time.
+Sanjay Prasad Holy crap that was about as basic layman as you get. I followed it and I'm just starting out on mobo components, how more basic can you get. There are no technical people, just idiots and people that use their brain.
Sanjay Prasad electronics is not for you mate
Gigabyte is like Samsung used to be, and asus is like sony used to be with propietary and expensive shit, guess who is number 1 on electronic devices today?
Who?
people with headphones....yikes.. lower the volume.