PSU question

[Larry Roberts]

I'm upgrading from an Nvidia 7600GT to an ATI HD3850. These
newer cards require more power than a 7600GT, and I'm wondering if my
PSU is gona handle it. I'm using a Rosewill RP550 550W (Single Rail).
The Rosewill RP models seem to have decent reviews for powering Nvidia
8800, 9600, & ATI HD3800 series cards. Althogh I'd look to something
better fo do multi-card, or multi-gpu setups.
Mine works great with my system now, but I discovered
something that has me worried. I ran Everest Home Edition today, and
on the Sensor page, the +12v is reading 7.68. However, when I boot out
into the BIOS, it reads 11.99. Does this mean my PSU is defective? My
system is not highend:

ATI Xpress 200 chipset mainboad
Athlon X2 5600+
4GB DDR2 (2 x 2GB)
7600GT PCI-E
Soundcard PCI
TV Card PCI
1 x SATA HDD
1 x DVD-RW
1 x FDD
2 x USB game controllers

 

[RobV]

I'm upgrading from an Nvidia 7600GT to an ATI HD3850. These
newer cards require more power than a 7600GT, and I'm wondering if my
PSU is gona handle it. I'm using a Rosewill RP550 550W (Single Rail).
The Rosewill RP models seem to have decent reviews for powering Nvidia
8800, 9600, & ATI HD3800 series cards. Althogh I'd look to something
better fo do multi-card, or multi-gpu setups.
Mine works great with my system now, but I discovered
something that has me worried. I ran Everest Home Edition today, and
on the Sensor page, the +12v is reading 7.68. However, when I boot out
into the BIOS, it reads 11.99. Does this mean my PSU is defective? My
system is not highend:

ATI Xpress 200 chipset mainboad
Athlon X2 5600+
4GB DDR2 (2 x 2GB)
7600GT PCI-E
Soundcard PCI
TV Card PCI
1 x SATA HDD
1 x DVD-RW
1 x FDD
2 x USB game controllers

If +12V were really 7.68 volts, the system wouldn't be running for you
to see it.

It's an erroneous reading. The best way to measure the voltage coming
from the PSU is with the system powered up and measuring the voltages at
the ATX connector with a digital volt meter. Failing that, the BIOS is
going to be more accurate than a software program running in Windows.

 

[John Doe]

Mine works great with my system now, but I discovered
something that has me worried. I ran Everest Home Edition today, and
on the Sensor page, the +12v is reading 7.68.

Does this mean my PSU is defective?

You need to figure out why it's reading 7.68. Try a different measuring
utility.

 

[John Doe]

The best way to measure the voltage coming from the PSU is with
the system powered up and measuring the voltages at the ATX
connector with a digital volt meter.

Probing inside of the case when the computer is plugged in and
turned on is a bad idea for the vast majority of users.

 

[Paul]

If +12V were really 7.68 volts, the system wouldn't be running for you
to see it.

It's an erroneous reading. The best way to measure the voltage coming
from the PSU is with the system powered up and measuring the voltages at
the ATX connector with a digital volt meter. Failing that, the BIOS is
going to be more accurate than a software program running in Windows.

I think hard drives are pretty sensitive to getting the full 12V on the
motor rail, and wouldn't work properly if the voltage dropped to 7.68V .

The utility I use on this computer, gets the 12V readout wrong, and I've
verified with a good multimeter, and the voltage is perfect when measured
that way. Using a multimeter, and grabbing a spare disk drive connector,
to check the 12V, is one way to do it. Just be careful not to short the
meter probes together while measuring the voltage.

7600GT 36 to 38 watts.

http://www.xbitlabs.com/articles/video/display/xfx-7600gt-xxx_5.html

HD3850 60 to 63 watts.

http://www.xbitlabs.com/articles/video/display/gigabyte-radeonhd3870-3850_5.html#sect0

The RP550 single rail version, is good for 12V @ 35A, and 5 amps to run
the video card, isn't likely to matter. That is, if the supply actually
delivers according to its label ratings.

http://c1.neweggimages.com/NeweggImage/productimage/17-182-010-08.jpg

Paul

 

[Larry Roberts]

You need to figure out why it's reading 7.68. Try a different measuring
utility.

I did. I also used SpeedFan, which had the same reading. If
what the others are saying is true, then I'm not worrying about it
till I get the HD3850 Monday, which should have been getting here
Friday if not for the 4th of July.

 

[John Doe]

I think hard drives are pretty sensitive to getting the full 12V
on the motor rail, and wouldn't work properly if the voltage
dropped to 7.68V .

The utility I use on this computer, gets the 12V readout wrong,

Time to get a better utility.

and I've verified with a good multimeter, and the voltage is
perfect when measured that way. Using a multimeter, and grabbing a
spare disk drive connector, to check the 12V, is one way to do it.

It isn't worth the risk and it isn't necessary anymore. Having to
use a multimeter to determine power supply voltage is like driving a
car without a dashboard.

 

[John Doe]

I did. I also used SpeedFan, which had the same reading.

Even if using a multimeter were a good approach, it still wouldn't
explain the bad reading. I would want to know the cause. Maybe the
mainboard/chipset sucks.

 

[Paul]

Even if using a multimeter were a good approach, it still wouldn't
explain the bad reading. I would want to know the cause. Maybe the
mainboard/chipset sucks.

You realize of course, that the ADC in the SuperI/O chip, that makes
the voltage readings, has external scaling resistors per channel, installed
on the signals being measured. If the scaling resistors are wrong,
or the application used to read out the values, doesn't have the
correct scaling information, then the readings from the utility
can be wrong. Scaling resistors are necessary for any rail being
measured, where the rail voltage is higher than the max reading
range of the chip. This is why you are less likely to get an
incorrect reading for Vcore or perhaps 3.3V.

I have one utility that gets the right answer, and one utility that
gets only the 12V reading wrong, which implies that the equation
used in the incorrect utility needs to be corrected.

This is why, comparing utility readings to multimeter readings,
can set your mind at ease, when you see a problem.

What I don't know, is whether any standard exists, or mechanism
exists, for a motherboard designer to tell users or applications,
what resistors were used. While the application note for the device,
may provide guidance on what components to use on the interface,
nobody is forced to stick with the suggested values.

Paul

 

[John Doe]

John Doe wrote:
....


You realize of course, that the ADC in the SuperI/O chip, that
makes the voltage readings, has external scaling resistors per
channel, installed on the signals being measured. If the scaling
resistors are wrong,

.... you need better hardware.

or the application used to read out the values, doesn't have the
correct scaling information,

.... you need a better utility.

then the readings from the utility can be wrong. Scaling resistors
are necessary for any rail being measured, where the rail voltage
is higher than the max reading range of the chip. This is why you
are less likely to get an incorrect reading for Vcore or perhaps
3.3V.

I have one utility that gets the right answer, and one utility
that gets only the 12V reading wrong, which implies that the
equation used in the incorrect utility needs to be corrected.

This is why, comparing utility readings to multimeter readings,
can set your mind at ease, when you see a problem.

In any case, when you tell a user you do not know to open the case
and probe around with the computer plugged in and powered on, IMO
you should at least provide some warnings. Some users have never
used a multimeter and they wouldn't know how.

While home building a computer, the only reason to have the case
open while it's powered on is to make sure all of the fans are
running, and that practice is probably becoming obsolete too.

 

[Paul]

... you need better hardware.


... you need a better utility.


In any case, when you tell a user you do not know to open the case
and probe around with the computer plugged in and powered on, IMO
you should at least provide some warnings. Some users have never
used a multimeter and they wouldn't know how.

While home building a computer, the only reason to have the case
open while it's powered on is to make sure all of the fans are
running, and that practice is probably becoming obsolete too.

Well, people do stupid things all the time. And where do you
suggest we draw the line ? For example, maybe they just
shouldn't plug appliances into wall outlets, because,
the voltages on there are lethal. How many people know
you're not supposed to hold onto the metal prongs, while
inserting the plug ? I could see some dope, half plug in
an AC line cord, then drop a screwdriver across the exposed
metal.

Or use a toaster while sitting in the bathtub.

The only time I've ever had an issue, inside a
piece of equipment, was when working on something
which was patently unsafe. They used to make power
supplies, which were called "open frame", where there
is no covering over the circuitry. Even a protective
metal grillwork would have been a useful feature, but
they don't make them that way. I had a piece of lab
equipment fall into the thing, even though I was
trying to be careful (I knew the risks). By comparison,
the ATX power supply has metal on all sides, and there
is no chance of me repeating the accident with that design.

A person can be electrocuted by a 1.5V dry cell, if
done properly. Does that mean we should close
down Radio Shack ?

Paul