Z370 / Z390 VRM Discussion Thread - Page 413 - Overclock.net - An Overclocking Community
Forum Jump: 

Z370 / Z390 VRM Discussion Thread

Reply
 
Thread Tools
post #4121 of 4250 (permalink) Old 07-01-2019, 12:06 AM
New to Overclock.net
 
elmor's Avatar
 
Join Date: Mar 2014
Posts: 876
Rep: 354 (Unique: 158)
Quote: Originally Posted by AlphaC View Post
Wouldn't it be better to test with MCE-like clocks? I don't think anyone sane buying a $400 Gene is going to be running 4.7GHz @ 1.2V. Maybe 5.0-5.1GHz with AVX offset.


Also is that a typo? Should be SDS1104X-E, no?



Good to see some results though
Quote: Originally Posted by Falkentyne View Post
Sweet Mother of God.
That Siglent looks so much better than the Rigol and that scope TIN (on the gamersnexus video with Steve) had with the GPU loadline thing!
Thank you Elmor!

With you and buildzoid's video, this is scope heaven !
*edit*
Just read your post again.
Those spikes and dips are just sitting idle in the BIOS?

Oh dear... 80mv dip at LLC8 at idle... :/
The dip with prime95 FMA3 running would probably by guessing be close to 150mv (!!)
That's scary....
It's possible to test with tougher settings, but the point was to visualize how the transients look in practice and what impact different load-line levels has. With higher currents (frequency + voltage), the signal will look the same but have larger undershoot/overshoot. The load is periodical at 60 Hz due to CPU interrupts when running tasks like updating the GUI. It's difficult to get a good capture when running something like Prime95, the signal is too noisy and irregular. It's also not an instantaneous load as it takes time to prepare data sets, initialize threads etc.

In this test case, the load is only for ~150 µs. At Level 1 the load-line is set to 2.08 mohm, (1.200-1.060)/0.00208 = 67A load step.

I did a short re-test at 5.2 GHz with 1.500V set point, in this case the load step is (1.500-1.290)/0.00208 = 101A. I had to change the vertical scale to 50 mV/div for these measurements so the amplitude is scaled down by 2.5 times.







edit: Yes, SDS1104X-E is correct.

Last edited by elmor; 07-01-2019 at 12:12 AM.
elmor is offline  
Sponsored Links
Advertisement
 
post #4122 of 4250 (permalink) Old 07-01-2019, 12:38 AM
Facepalm
 
Falkentyne's Avatar
 
Join Date: Dec 2007
Location: Riverside
Posts: 7,822
Rep: 455 (Unique: 316)
Quote: Originally Posted by elmor View Post
It's possible to test with tougher settings, but the point was to visualize how the transients look in practice and what impact different load-line levels has. With higher currents (frequency + voltage), the signal will look the same but have larger undershoot/overshoot. The load is periodical at 60 Hz due to CPU interrupts when running tasks like updating the GUI. It's difficult to get a good capture when running something like Prime95, the signal is too noisy and irregular. It's also not an instantaneous load as it takes time to prepare data sets, initialize threads etc.

In this test case, the load is only for ~150 µs. At Level 1 the load-line is set to 2.08 mohm, (1.200-1.060)/0.00208 = 67A load step.

I did a short re-test at 5.2 GHz with 1.500V set point, in this case the load step is (1.500-1.290)/0.00208 = 101A. I had to change the vertical scale to 50 mV/div for these measurements so the amplitude is scaled down by 2.5 times.







edit: Yes, SDS1104X-E is correct.
Thank you very much for your hard work, Elmor.

[email protected] ghz, RX Vega 64, 32GB DDR4, Gigabyte Aorus Master, Seasonic Platinum 1000W, Corsair 760T
Alt: MSI GT73VR Throttlebook with 7820HK @ 4.7 ghz, GTX 1070 MXM TDP mod to 230W, 32 GB RAM

Quote: Originally Posted by sakete View Post
Well, I want you to know I have an academic degree in speculation.
Falkentyne is online now  
post #4123 of 4250 (permalink) Old 07-01-2019, 03:46 AM
New to Overclock.net
 
Luck100's Avatar
 
Join Date: Oct 2013
Posts: 458
Rep: 29 (Unique: 20)
Quote: Originally Posted by elmor View Post
It's possible to test with tougher settings, but the point was to visualize how the transients look in practice and what impact different load-line levels has. With higher currents (frequency + voltage), the signal will look the same but have larger undershoot/overshoot. The load is periodical at 60 Hz due to CPU interrupts when running tasks like updating the GUI. It's difficult to get a good capture when running something like Prime95, the signal is too noisy and irregular. It's also not an instantaneous load as it takes time to prepare data sets, initialize threads etc.

In this test case, the load is only for ~150 µs. At Level 1 the load-line is set to 2.08 mohm, (1.200-1.060)/0.00208 = 67A load step.

I did a short re-test at 5.2 GHz with 1.500V set point, in this case the load step is (1.500-1.290)/0.00208 = 101A. I had to change the vertical scale to 50 mV/div for these measurements so the amplitude is scaled down by 2.5 times.
Great stuff elmor! What's the horizontal axis scale? 1 Microsecond per grid box?

Battlestation
(15 items)
TV Buddy
(15 items)
CPU
Intel i9-9900k
Motherboard
Asus ROG Maximus XI Hero (Wifi)
GPU
Gigabyte Aorus 1080ti
RAM
Corsair Vengeance LPX 2 x 8 GB
Hard Drive
Samsung 970 EVO NVMe 1 TB
Power Supply
Seasonic PRIME Ultra Gold 1000 W
Cooling
Thermalright Le Grand Macho RT
Case
Cooler Master MasterCase H500M
Operating System
Windows 10 Pro
Monitor
Acer Predator XB270HU
Keyboard
CM Quickfire TK
Mouse
Logitech G500s
Mousepad
Logitech G440
Audio
Logitech G533 Gaming Headset
Audio
StudioSpares Powered Studio Monitors
CPU
i5-4690
Motherboard
Asus Z97-A
GPU
EVGA GTX 670 FTW
RAM
Crucial Ballistix 4GB x 2
Hard Drive
Western Digital Caviar Green 3 TB
Hard Drive
Mushkin Chronos Deluxe 240 GB SSD
Optical Drive
Pioneer BDC-207DBK Blu-Ray Reader, DVD/CD Writer
Power Supply
SeaSonic G550
Cooling
Noctua NH-L12
Cooling
Scythe Gentle Typhoon 37.0 CFM 120mm x 2
Case
Silverstone GD09B
Operating System
Windows 8.1 Pro
Keyboard
Logitech K400
Other
FLIRC USB Dongle for Media Centre
Other
Harmony 650 Remote
▲ hide details ▲
Luck100 is online now  
Sponsored Links
Advertisement
 
post #4124 of 4250 (permalink) Old 07-01-2019, 05:46 AM
New to Overclock.net
 
elmor's Avatar
 
Join Date: Mar 2014
Posts: 876
Rep: 354 (Unique: 158)
Quote: Originally Posted by Luck100 View Post
Great stuff elmor! What's the horizontal axis scale? 1 Microsecond per grid box?
Each division (grid box) is 20us horizontally and 50mV vertically (20mV in the first set).
elmor is offline  
post #4125 of 4250 (permalink) Old 07-01-2019, 11:41 PM
New to Overclock.net
 
Imprezzion's Avatar
 
Join Date: Jul 2010
Location: Gorredijk, Holland
Posts: 2,774
Rep: 111 (Unique: 96)
Well, the MSI Z390 Ace VRM seems plenty for 24/7 clocks at least.

I'm running a 9900K not delidded (yet) at 5.025Ghz core, 4.7Ghz cache, 1.315v with stock SA, IO and 1.36v VDIMM. LLC Level 5, didn't touch the response times of the DigitALL Power settings yet.

In Prime95 with AVX and FMA3 enabled on small FFT the CPU sits in the high 80's with a power draw of about 235-240w and VRM (MOS sensor in Hwmonitor) gets up to about 90c after 30 minutes and stays there. This is with totally zero airflow in the VRM sinks as the rad and fans covers them partially.

Gaming runs them around 60-65c MOS temp and high 60's for the CPU as well. (~150w load)

So far, no thermal issues.

EDIT: 5.025 is hard to get 100% stable so far as it seems to need a tad more for 100% stability. It BSOD's with a PAGE_FAULT after 10-15 min of Prime AVX/FMA3 and is also prime to giving WHEA errors in games. They will both go away at 1.33v but CPU gets up to 93c and that also makes it unstable due to heat.

I'm settling in on 4.925Ghz @ 1.248v, 4.625Ghz cache, 3000Mhz RAM @ 12-15-15-35-400-1T with 24960 tREFI 1.36v and 0.960 SA 1.050 IO. Temps stay in the low 80's then.

MOSFETS peak at 80c and stabilize at 78-79c according to the boards sensors. This is still with no airflow over them at all. If i point a fan at them they drop to high 60's.
Attached Thumbnails
Click image for larger version

Name:	4.925-9900k-1.248v-avx.PNG
Views:	50
Size:	1.06 MB
ID:	277696  


Cars:
-Crashed: (not my fault ) '94 Nissan 100NX GA16DE (a.k.a. NX1600). Renault Laguna Phase I facelift '99 1.6 K4M. (Broke a strut and smacked a tree).
-Sold: '00 Nissan Primera P11-144 QG18DE, '97 Honda Prelude F20A4, 03 Ignis Sport.
-Current daily: '07 Ford Focus 1.6 Futura.
-Current funcars: '03 Peugeot 206CC 2.0 RFN fully riced and tuned (Turbo coming soon)
Looking for: Gen 4 or 5 Honda Prelude, >92 Toyota Celica or >90 MR2.
Full-On Rainbowbarf
(16 items)
P45 power!
(9 items)
CPU
i9 9900K @ 5.1Ghz 1.248v (VR VOut)
Motherboard
MSI MEG Z390 ACE
GPU
RTX 2080 Ti Phoenix GS + G12 & Kraken X52 @ 2130/8100 1.093v (EVGA FTW3 BIOS)
RAM
Corsair Vengeance RGB B-Die @ 4200 16-17-17-34-320-2T 1.54v
Hard Drive
Samsung 850 EVO 500GB x2
Hard Drive
Samsung 970 Evo 1TB
Power Supply
Seasonic Prime Ultra Gold 1000w
Cooling
EK MLC-Phoenix 280 Kit
Case
Phanteks Enthoo Evolv X
Operating System
Windows 10 Pro X64
Monitor
Samsung S27B350H @ 1080p 75Hz OC
Monitor
Iiyama GB2788HS @ 144Hz
Keyboard
Logitech G710+
Mouse
Trust GXT31
Mousepad
Corsair MM300 Large
Audio
Steelseries Siberia V2
CPU
Intel Core 2 Quad Q6600 G0 @ 3Ghz (333x9)
Motherboard
MSI P45 Neo2-FR
GPU
MSI GTX760 Gaming X
RAM
6GB completely random 1 and 2GB sticks. All 4 different ones.
Hard Drive
A-Data 120GB SSD
Power Supply
AOpen 350w
Cooling
Some huge aluminum Xeon heatsink
Case
Cartboard box.
Operating System
W10 Pro x64
CPU
Intel Xeon E3-1270v1 @ 3.8Ghz
Motherboard
ASUS P8H61-i R2.0
GPU
Sapphire R9 280X Vapor-X @ 1230Mhz core.
RAM
16GB Kingston HyperX Savage 2400CL11 @ 1333CL6
Hard Drive
Kingston 240GB SSD
Power Supply
Corsair CX600
Cooling
Corsair H110i GTX
Case
Custom open test bench made from a cut-up CM690 v1
Operating System
W10 Pro X64
Monitor
Samsung 23" cheap thing
Keyboard
Cheapest one Aldi had
Mouse
Cheapest one Aldi had
Mousepad
The desk...
▲ hide details ▲

Last edited by Imprezzion; 07-03-2019 at 02:56 PM.
Imprezzion is online now  
post #4126 of 4250 (permalink) Old 07-16-2019, 09:23 PM
New to Overclock.net
 
KedarWolf's Avatar
 
Join Date: May 2013
Location: Toronto, Canada
Posts: 4,773
Rep: 306 (Unique: 194)
Quote: Originally Posted by WannaBeOCer View Post
That was completely at stock without enabling XMP. I believe the CPU was running at 3.6Ghz on all eight cores at stock with MCE disabled. Here's mine set to 5.1Ghz but the CPU runs at 5Ghz cause of some sort of AVX offset that I can't modify.
This is with the Studio Nvidia drivers that do better with Cinebench.




KedarWolf-AMD 3950x @ CCX 44.50/44.00/43.50/43.25GHZ - Motherboard MSI X570 MEG Unify - G.Skill Trident Z Neo CL16 3600MHZ 2x16GB 15-16-10-14-30 2T - 264 tRFC @ 3800MHZ 1.42v Gear Down Mode Disabled- Gigabyte/EVGA 2X 1080 Ti SLI FTW3 Kingpin Fix BIOS - 2x 1TB Sabrent Gen 4 M.2 System/Storage Drives- PSU Corsair AX1600i - Samsung 49" CHG90 QLED 32:9 Gaming Monitor - Logitech G502 Lightspeed Mouse- Logitech Lightspeed Mouse Pad
CPU
3950x
Motherboard
MSI X570 MEG Unify
GPU
EVGA 1080 Ti SC Black EKWB Waterblock And Backplate
GPU
Gigabyte 1080 Ti FE EKWB Waterblock And Backplate
RAM
Trident Z Neo 2x16GB 16-16-16-36 3600
Hard Drive
Sabrient Gen 4 1TB M.2
Hard Drive
Sabrient Gen 4 1TB M.2
Power Supply
Corsair AX1600i
Cooling
EKWB Phoenix 240
Cooling
EKWB Predator 360
Case
Thermaltake X9
Operating System
Windows 10 Enterprise 19628
Monitor
49" Samsung LC49HG90DMNXZA
Monitor
27" BenQ GW2760
Keyboard
Logitech G910
Mouse
Logitech G502 Lightspeed
Mousepad
Logitech Lightspeed
▲ hide details ▲
KedarWolf is offline  
post #4127 of 4250 (permalink) Old 07-20-2019, 01:29 PM
New to Overclock.net
 
chinobino's Avatar
 
Join Date: Jul 2006
Posts: 20
Rep: 1 (Unique: 1)
I have a 9900K and two motherboards, a Z370 with 4 vcore phase doublers with no current balancing and a Z390 with 5 vcore phase doublers with current balancing.

I would like to know how much sustained current draw each motherboard is capable of, assuming that the VRM heatsinks are capable of doing their job.

The problem is I have no idea how efficient either board's VRM is.

1. EVGA Z370 CLASSIFIED K

VCC Controller - Infineon IR35201 (4+2)
VCC Doubler - 4 x Infineon IR3599 (interleaving only, no current balancing)
VCC DrMOS - 8 x Infineon TDA88240 35A
VCC Inductor - Unknown

VCCGT Controller - Infineon IR35201 (4+2)
VCCGT DrMOS - 2 x Infineon TDA88240 35A
VCCGT Inductor - Unknown

VCCIO & VCCSA Controller - Infineon IR35204 (1 x VCSSA power phase & 1 x VCCIO power phase)
VCCIO & VCCSA DrMOS - 2 x Infineon TDA88240 35A
VCCIO & VCCSA Inductor - Unknown

VDDR Controller - uPI SEMI uP1537P single-phase PWM controller with integrated driver
VDDR MOSFET - 1 x ONSemi NTMSF4C024N Lowside MOSFET (Single N−Channel Power MOSFET 30 V, 76 A, 4.00mΩ)
VDDR MOSFET - 1 x ONSemi NTMFS4C029N Highside MOSFET (Single N−Channel Power MOSFET 30 V, 46 A, 5.88mΩ)
VDDR Inductor - Unknown

2. GIGABYTE Z390 UD

VCC Controller - Intersil ISL69138 (5+2)
VCC Doubler - 5 x Intersil ISL6617A (with current balancing)
VCC Driver - 10 x Intersil ISL6625A
VCC Highside MOSFET - 10 x ONSemi 4C10N 46A
VCC Lowside MOSFET - 10 x ONSemi 4C06N 69A
VCC Inductor - 10 x Vishay PowerPak ~60A

VCCGT Controller - Intersil ISL69138 (5+2)
VCCGT Driver - 2 x Intersil ISL6596
VCCGT Highside MOSFET - 2 x ONSemi 4C10N (small package)
VCCGT Lowside MOSFET - 2 x ONSemi 4C06N (small package)
VCCGT Inductor - 2 x Vishay PowerPak ~60A

VCCIO & VCCSA Controller - 2 x Richtek RT8120D
VCCIO & VCCSA Lowside MOSFET - 2 x ONSemi 4C06N
VCCIO & VCCSA Highside MOSFET 2 x NIKOSEM PK6H6BA N-Channel Enhancement Mode MOSFET 30V 46A
VCCIO & VCCSA Inductor - Unknown

VDDR Controller - Richtek RT8120D
VDDR Lowside MOSFET - 3 x ONSemi 4C06N 30V 69A
VDDR Highside MOSFET 1 x NIKOSEM PK6H6BA N-Channel Enhancement Mode MOSFET 30V 46A
VDDR Inductor - 2 x Vishay Powerpak R50

Intel 9900K
Gigabyte Z170-HD3-DDR3
16GB Corsair DDR3 2800 MHz
MSI LIGHTNING GTX980Ti
Enermax Revolution 85+ 1250W
chinobino is offline  
post #4128 of 4250 (permalink) Old 07-20-2019, 01:53 PM
⤷ αC
 
AlphaC's Avatar
 
Join Date: Sep 2012
Posts: 11,211
Rep: 913 (Unique: 592)
The EVGA Z370 one is markedly better. TDA88240 35A is a top tier powerstage albeit in 35A rating (some of the best are the TDA21472 70A). 4C10N+4C06N are run of the mill powerpaks (discrete high + low mosfet)

Powerstages tend to be around 90-92% efficient at the 35A tier, while that 4C10N+4C06N solution will be around 88-89% at best (and that doesn't count the driver losses).

Current balance is achieved at the PWM side for the EVGA board via the IR35201 PWM.


To be fair though, the Z390 UD is a budget level board, the Z370 Classified is a top tier from EVGA (supposed to be on par with at least upper midrange from other vendors such as Z370 Hero / Taichi / Aorus Pro).

► Recommended GPU Projects: [email protected] , [email protected] (FP64) (AMD moreso) ► Other notable GPU projects: [email protected] (Nvidia), GPUGrid (Nvidia) ► Project list



Last edited by AlphaC; 07-20-2019 at 02:09 PM.
AlphaC is offline  
post #4129 of 4250 (permalink) Old 07-20-2019, 02:19 PM
New to Overclock.net
 
chinobino's Avatar
 
Join Date: Jul 2006
Posts: 20
Rep: 1 (Unique: 1)
Thank you for your fast and detailed reply and the notes on efficiency, I should have checked the data sheet for the IR35201 regarding current balancing.

Do you have any idea how much sustained current each board could pull from the PSU?

Cheers.

Intel 9900K
Gigabyte Z170-HD3-DDR3
16GB Corsair DDR3 2800 MHz
MSI LIGHTNING GTX980Ti
Enermax Revolution 85+ 1250W
chinobino is offline  
post #4130 of 4250 (permalink) Old 07-20-2019, 02:57 PM
⤷ αC
 
AlphaC's Avatar
 
Join Date: Sep 2012
Posts: 11,211
Rep: 913 (Unique: 592)
How much power you can draw is also directly related to how well the power delivery is cooled. If you already own the boards, you can test when it starts reaching ~85-95°C (this is about as far as high as I would try since over that and you can expect to have greatly deteriorated component lifespans). 105°C is basically out of the question and 120°C typically trips throttling or shutdown.


Intel spec is 193A max for the i9-9900K.

► Recommended GPU Projects: [email protected] , [email protected] (FP64) (AMD moreso) ► Other notable GPU projects: [email protected] (Nvidia), GPUGrid (Nvidia) ► Project list


AlphaC is offline  
Reply

Tags
K6 Motherboard , Msi , voltage regulator , z370 vrm , z390 , z390 vrm

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off