Overclock.net › Forums › Components › Power Supplies › 620HX Voltage Question
New Posts  All Forums:Forum Nav:

620HX Voltage Question - Page 3

post #21 of 31
Quote:
Originally Posted by shinji2k View Post
It sounds like you've seen more individual computers than me. I've yet to see one that had noticeable discrepancies between software voltages, right now Everest and Speenfan are reading exactly the same for me (not that that is any kind of proof). When you say wide range of voltages, what do you mean? Are we talking, hundreths, tenths, more?
It can happen with unsupported chipsets. The software developer just releases an updated to interpret the sensor correctly.
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #22 of 31
I think we can all agree that software monitoring isn't exactly reliable and I just realized we are debating on which software (Windows vs Windows vs BIOS) reads the incorrect value better. I'm tired . We're splitting hairs here.

I still say that Windows software (when programmed correctly and on a supported chipset/sensor, these programs are free usually ) can theoretically read the same value as BIOS. But this can't ever confirmed since you can't be in BIOS and read Windows software at the same exact moment.
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
post #23 of 31
Quote:
Originally Posted by shinji2k View Post
It sounds like you've seen more individual computers than me. I've yet to see one that had noticeable discrepancies between software voltages, right now Everest and Speedfan are reading exactly the same for me (not that that is any kind of proof). When you say wide range of voltages, what do you mean? Are we talking, hundreths, tenths, more?
The 12V Rail for instance. I've seen ranges from over 12 down to 0 volts - and many in between.

Quote:
Originally Posted by shinji2k View Post
I think we can all agree that software monitoring isn't exactly reliable and I just realized we are debating on which software (Windows vs Windows vs BIOS) reads the incorrect value better. I'm tired . We're splitting hairs here.
More specifically, I'm trying to point out to you that 3rd party software can be programmed incorrectly and subsequently be less accurate then BIOS readings. From my experience, this does actually happen in real life.

Quote:
Originally Posted by shinji2k View Post
I still say that Windows software (when programmed correctly and on a supported chipset/sensor, these programs are free usually ) can theoretically read the same value as BIOS. But this can't ever confirmed since you can't be in BIOS and read Windows software at the same exact moment.
And theoretically, Crysis properly utilizes all 4 cores on a C2Q CPU.

Of course the values from the BIOS and 3rd party software can read the same thing (or very close to it). The software must be coded properly to do this. At least you've stopped thinking in absolutes here. But if the software isn't programmed correctly (which DOES happen), there is a good chance it won't be reading the sensor right - and subsequently show a value different of that from the BIOS. I've seen this happen, in wide ranges, with my own two eyes.

If you don't want to believe that I've seen certain voltages, readings etc., that's fine. Because I don't actually have proof. But even though I don't, not accepting it as a possibility is illogical. I think you may understand that now.
Edited by Choggs396 - 5/25/08 at 2:37am
    
CPUMotherboardGraphicsRAM
Intel Q6600 G0 @ 3.6GHz (lapped) EVGA 780i SLI FTW @ 1600MHz FSB 2 x EVGA 8800GTX 768MB (SLI) @ 625/1500/2000MHz 2x2GB G.Skill PC2-8000 PQ @ 800MHz (linked to FSB) 
Hard DriveOptical DriveOSMonitor
2x160GB WD RAID 0, 750GB Samsung Spinpoint F1 LITE-ON 20X SATA DVD/CD RW Windows 7 Ultimate x64 Viewsonic VA2702w 27" LCD 1920x1080 
KeyboardPowerCaseMouse
Logitech Value 100 (beer magnet) Corsair TX 750W Cooler Master ATCS 840 Logitech SBF-90 (cheapest one on Newegg) 
Mouse Pad
Some kinda rubber thing (beer magnet) 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
Intel Q6600 G0 @ 3.6GHz (lapped) EVGA 780i SLI FTW @ 1600MHz FSB 2 x EVGA 8800GTX 768MB (SLI) @ 625/1500/2000MHz 2x2GB G.Skill PC2-8000 PQ @ 800MHz (linked to FSB) 
Hard DriveOptical DriveOSMonitor
2x160GB WD RAID 0, 750GB Samsung Spinpoint F1 LITE-ON 20X SATA DVD/CD RW Windows 7 Ultimate x64 Viewsonic VA2702w 27" LCD 1920x1080 
KeyboardPowerCaseMouse
Logitech Value 100 (beer magnet) Corsair TX 750W Cooler Master ATCS 840 Logitech SBF-90 (cheapest one on Newegg) 
Mouse Pad
Some kinda rubber thing (beer magnet) 
  hide details  
Reply
post #24 of 31
I honestly hope you are right. If BIOS readings can be trusted a little more then 3rd party, it makes troubleshooting a little easier for those without multimeters. For kicks I broke out my multimeter and took some measurements (although it is a cheapo one, who knows how well it is callibrated). Everest read 5.10V (5.00V on the MM) on the 5V and 11.78V (12.00V MM) on the 12V rail. Not horrible, but it could be better. I did the same in the BIOS, with the 5V being the same as earlier and the 12V read 11.96V in BIOS and 12.00V on the MM. There's definitely an improvement over 3rd party on the 12V rail. So it looks like you may have a point. I still don't understand why 3rd party reads the sensor differently then the BIOS sometimes, but it happens and maybe that'll be the next thing I do some research on. The process seems simple enough.

I'm man enough to admit when I'm wrong (it happens more than I would like). When I first started trying to learn as much as possible about psus, I did the normal thing, listen to what the experts say. And well, they aren't always right.
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
post #25 of 31
Let's clear this up once and for all....

The BIOS and software all get their readings from the same chip.

So... Why do readings differ?

First off, there's an algorithm used to report the voltage. One algorithm may be closer than another at different loads, but none are accurate.

Second, in the BIOS you're at a zero load state. The chip that's reporting the voltages is on the motherboard, traces and traces away from the actual power input it's trying to read voltages from. With next to no load, you have almost no resistance, so the voltages read by this chip are going to be more accurate.

I will always tell someone that if they're having problems with their PC and they're showing low voltages in the BIOS that they should check the connection of their 24-pin power connector. If a wire is pulled up a little bit in the connector, the lack of a good connection could cause low voltages and it's worth checking. But BIOS is accurate, software is inaccurate.... ask anyone that's an engineer or product manager for power supplies OR motherboards. Any kind of voltage reading you get from a "Winbond" type chip is going to be off. This is why so many power supply and motherboard manufacturers WANT to latch onto Nvidia's ESA. They're checking voltages at the source and that's the best way to get accuracy.
post #26 of 31
I'm still confused.
Quote:
The BIOS and software all get their readings from the same chip.
Quote:
One algorithm may be closer than another at different loads, but none are accurate.
Quote:
But BIOS is accurate, software is inaccurate...
In that last quote did you mean that BIOS is more accurate than software, but both are inaccurate to some degree? If so, disregard the me being confused part.

That brings up a question. Have you seen any tests/reviews where power consumption was measured and compared between navigating BIOS, and then maybe the Welcome/login page of Windows and finally idle when logged into Windows? Obviously it will be different for each example, but would it be enough to account for noticeable changes in voltage? I know when I did the little test, voltages read with a multimeter were the same when I was in BIOS as when I was in Windows with a decent amount of background programs running. But I also know that on the two rails I measured (5V+12V), I won't see much more than a change of ~0.06V throughout the load range with my Corsair (based on madmat and OW's review). So for my setup, it isn't really a problem of different loads causing different voltages, it's that both 3rd party programs I used were reading the same "Winbond" data as the BIOS but are less accurate when reading the 12V rail (5V read the same in BIOS and 3rd party). Everest and Speedfan said 11.78V and BIOS reported 11.96V when the multimeter said 12.00V for both. I can imagine a 3rd party program being programmed incorrectly, but two with the same problem? Is there some inherent flaw in the way the "Winbond" chip is read while in a Windows environment?

tl;dr:
BIOS is supposed to be more accurate because loads are lower, therefore voltage is closer to the actual value.
How much difference is there in terms of power consumption between BIOS idle and Windows idle?
Enough to affect voltages noticeably?
12V rail had a discrepency of ~0.2V (max could only be ~0.06V and realistically ~0.03V) which is impossible if both 3rd party and BIOS read the same data and the difference in accuracy in based only on load affecting voltage.


*edit* - I should stop talking, I'm just making myself look like an idiot. You said that voltages within the motherboard will differ based on increased load (and resistance) which makes sense. For some reason I read is as the PSU will have different voltages based on load (which is true). But you meant that due to extra resistance between the atx connector and the "Winbond", the sensor will read a lower voltage than what is true and that the greater the load (and thus resistance) the greater this value read by the sensor will deviate from the actual voltage. I would still like to know just how much difference there is in power consumption between BIOS and Windows idle.

My reading comprehension has really dropped.
Edited by shinji2k - 5/25/08 at 6:36pm
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
post #27 of 31
Quote:
Originally Posted by jonnyGURU View Post
Let's clear this up once and for all....

The BIOS and software all get their readings from the same chip.

So... Why do readings differ?

First off, there's an algorithm used to report the voltage. One algorithm may be closer than another at different loads, but none are accurate.

Second, in the BIOS you're at a zero load state. The chip that's reporting the voltages is on the motherboard, traces and traces away from the actual power input it's trying to read voltages from. With next to no load, you have almost no resistance, so the voltages read by this chip are going to be more accurate.

I will always tell someone that if they're having problems with their PC and they're showing low voltages in the BIOS that they should check the connection of their 24-pin power connector. If a wire is pulled up a little bit in the connector, the lack of a good connection could cause low voltages and it's worth checking. But BIOS is accurate, software is inaccurate.... ask anyone that's an engineer or product manager for power supplies OR motherboards. Any kind of voltage reading you get from a "Winbond" type chip is going to be off. This is why so many power supply and motherboard manufacturers WANT to latch onto Nvidia's ESA. They're checking voltages at the source and that's the best way to get accuracy.
This is just out of curiosity, what is the deal with the innacuracies of these sensors? I honestly don't get why it is so hard. Just parts from the lowest bidder or something?

I assume that the sensors used are thin film RTD's?

We purchase some component type RTD's and they are DEAD on, always. Every single one (hundreds) I've unpacked reads between 108-109 ohms (depending on ambient), and we have an accuracy of at least 0.25C across a range of 20C-150C by using a very simple formula. Incredibly repeatable as well.

I just can't quite wrap my head around why this is so hard for manufacturers to get a handle on. This is really not that high of technology for this type of variation in performance to be even remotely excusable IMO.
post #28 of 31
I think that was what was getting to me earlier. The process isn't that complicated and the sensor should be fairly accurate. The problem is the location of the sensor, at least when talking about voltage readings, in the motherboard. Between the psu and sensor there are connectors, plugs and traces etc. All of these add resistance and will cause a voltage drop. On top of that, the more current that is drawn through these connections and traces, the more resistance increases and the more voltage will drop. That's why Jon mentioned ESA which I imagine will have a USB connection or such to the PSU. Inside the PSU will be voltage, load and temperature sensors measuring directly at the source and reporting to software via USB much like what we see in the Gigabyte's Odin PSU.
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
i7 monsta
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.2GHz] [1.325V] ASUS P6T Deluxe V2 2x XFX 5870 [1000/1250] G.Skill 3x2GB 1600MHz 
Hard DriveOSMonitorKeyboard
X25-M 80GB, 320GB WD Scorpio Black Vista x64 2007FP-3007WFP-2007FP-1080p 32" OCN Ducky Brown 
PowerCaseMouse
Silverstone ST85F-P whiteOut G9 
  hide details  
Reply
post #29 of 31
as a software developer, I can make the program read/report any value I like. If I make a mistake in my code, dealing with float values and conversions, for example, my end numbers can appear way off. The bios is just software on an eprom chip as opposed to on a magnetic disk like a windows program - the same problem still exists in the bios as well. Usually a multimeter is the best way to really know your voltages. I wouldnt sweat it too much.
Chuckle2Tits
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 2.66 45nm D0 Foxconn BloodRage x58 EVGA GTX 285 6 GB Corsair Dom. DDR3 1600 
Hard DriveOptical DriveOSMonitor
1x300 GB WD Velociraptor Asus 16/48 DVD/CD ROM SATA Windows 7 Pro 64 bit Viewsonic VX2255wmb 22" LCD 
KeyboardPowerCaseMouse
Logitech USB wired Silverstone Decathalon 850 PolymorphiX (PMX_v1s) ms optical, wired 
Mouse Pad
ancient piece of crap :P 
  hide details  
Reply
Chuckle2Tits
(13 items)
 
  
CPUMotherboardGraphicsRAM
Core i7 920 2.66 45nm D0 Foxconn BloodRage x58 EVGA GTX 285 6 GB Corsair Dom. DDR3 1600 
Hard DriveOptical DriveOSMonitor
1x300 GB WD Velociraptor Asus 16/48 DVD/CD ROM SATA Windows 7 Pro 64 bit Viewsonic VX2255wmb 22" LCD 
KeyboardPowerCaseMouse
Logitech USB wired Silverstone Decathalon 850 PolymorphiX (PMX_v1s) ms optical, wired 
Mouse Pad
ancient piece of crap :P 
  hide details  
Reply
post #30 of 31
Quote:
Originally Posted by shinji2k View Post
I think that was what was getting to me earlier. The process isn't that complicated and the sensor should be fairly accurate. The problem is the location of the sensor, at least when talking about voltage readings, in the motherboard. Between the psu and sensor there are connectors, plugs and traces etc. All of these add resistance and will cause a voltage drop. On top of that, the more current that is drawn through these connections and traces, the more resistance increases and the more voltage will drop.
Ding ding ding!!! You got it.

Now going for the bonus round....

Quote:
Originally Posted by shinji2k View Post
That's why Jon mentioned ESA which I imagine will have a USB connection or such to the PSU. Inside the PSU will be voltage, load and temperature sensors measuring directly at the source and reporting to software via USB much like what we see in the Gigabyte's Odin PSU.
Correct. Getting voltages as close to the source as possible removes as much of the resistance variable as possible. At least then, if you're getting really low voltages at the connector, you know the problem is between the unit and the connector and not the unit itself.

Quote:
Originally Posted by CattleRustler View Post
as a software developer, I can make the program read/report any value I like. If I make a mistake in my code, dealing with float values and conversions, for example, my end numbers can appear way off. The bios is just software on an eprom chip as opposed to on a magnetic disk like a windows program - the same problem still exists in the bios as well. Usually a multimeter is the best way to really know your voltages. I wouldnt sweat it too much.
Also correct! You guys are on fire today. Remember, the chip doesn't just "know" that what it's reading is a temperature and knows to spit it out as a number in Celsius. It has to be PROGRAMMED. The BIOS program reads and interprets this data, as does Everest, Speed Fan, etc. The programming can be pretty damn good, but still is likely to be accurate at one voltage and not another depending on load, resistance, etc.

A good example is the PSU's with the "Total Watt Viewer" we had a while back. Acbel, Cooler Master, Thermaltake... They were inaccurate not because of bad programming, but because of environmental variables. They measured AC input and then used an algorithm to calculate DC output. Let's say they assumed the PSU was 80% efficient. Well, we've all learned that PSU's aren't the same efficiency across the board, so the Watt Viewer would only be accurate when the PSU was actually 80% efficient. If the PSU was more or less efficient depending on the load, the Watt Viewer would be exponentially less accurate.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Power Supplies
Overclock.net › Forums › Components › Power Supplies › 620HX Voltage Question