PSU voltage instability
Slater
attached and noticed that the 12V rail gives me a lower voltage when
the CD tray is moving (11.79V against the usual 11.87V).
Could this be due to bad capacitors?
Phil Allison
"Slater"
** Small variations in the 12V and some other rail voltages are quite
normal - only the 5V and 3.3V rails are actually regulated with any
precision.
.... Phil
Sjouke Burry
the others are determined by transformer ratio, and are more
load dependent.
The 12v is usually one of the "raw" voltages, and as such is allowed to
vary quite a lot.
Jeff Liebermann
>Hello, I have an ATX power supply.
Right. All ATX power supplies are identical. Maker and model please?
Some power supplies are junk.
>I tested it with a CD ROM drive
>attached and noticed that the 12V rail gives me a lower voltage when
>the CD tray is moving (11.79V against the usual 11.87V).
>Could this be due to bad capacitors?
If you're getting 10mv accuracy with your DVM, you must have a decent
way of calibrating your instrument to that resolution. 100mv accuracy
is more reasonable.
See the ATX PS specification at:
<http://www.formfactors.org/developer%5Cspecs%5Catx2_2.pdf>
Section 4.1.4 specifies:
Table 6. Voltage Tolerances
Voltage Rail Tolerance
+5VDC ą 5 %
-5VDC (if used) ą 10 %
+12VDC ą 5 %
-12VDC ą 10 %
+3.3VDC ą 5 %
+5VSB ą 5 %
A tolerance of 5% would be anything from 11.4VDC to 12.6VDC.
However, if your unspecified maker and model DVM instrument is
calibrated to maybe ą3%, your acceptable range might be 11.0 to
13.0VDC.
--
Jeff Liebermann je...@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Slater
>Right. All ATX power supplies are identical. Maker and model please?
Maker: Whatsup
Model: ATX-400W P4
My concern is not how off the voltage is, it's the instability: is it
normal? The voltage changes depending on the load. The PSU inside my
PC doesnt do this.
>Some power supplies are junk.
Perhaps mine is in that league!
Phil Allison
"Jeff Liebermann"
>
>>I tested it with a CD ROM drive
>>attached and noticed that the 12V rail gives me a lower voltage when
>>the CD tray is moving (11.79V against the usual 11.87V).
>>Could this be due to bad capacitors?
>
> If you're getting 10mv accuracy with your DVM, you must have a decent
> way of calibrating your instrument to that resolution. 100mv accuracy
> is more reasonable.
** How pedantic and idiotic !!!
The OP is only asking about the small DROP in voltage he measured under
load - which NOT dependant on calibration or accuracy of the meter.
The LINEARITY of most DMMs is of a very high order - 1 part in 1000 is
typical.
.... Phil
Jeff Liebermann
>Jeff Liebermann <je...@cruzio.com> wrote:
>
>>Right. All ATX power supplies are identical. Maker and model please?
>
>Maker: Whatsup
>Model: ATX-400W P4
Sorry. I've never heard of that brand. The part number is rather
generic but matches an iCute power supply, another brand with which
I'm not familiar.
>My concern is not how off the voltage is, it's the instability: is it
>normal? The voltage changes depending on the load. The PSU inside my
>PC doesnt do this.
The drop is fairly normal for a power supply going from no load to
about 1/2A load at 12v. A drop of 80mv is about what I would expect
across the Molex connectors. With a CDROM motor load, that's
R = E / I^2 = 0.08V / 0.5^2 = 0.32 ohms
internal resistance. That's within what I guess would be normal. Also
note that only the 5V and 3.3V lines are actually regulated. The 12V
line is not.
I'm not sure why your "other brand" of ATX supply does not show a
similar drop. It might be where you're measuring it with your
unspecified model DVM (my guess is Radio Shack or worse).
In any case, what you're testing is not very realistic. If you want
to test a power supply, use a real load, or a resistor dummy load that
simulates full rated load. If there's anything wrong with the power
supply, you'll either see it with your unspecified model test
equipment, or it will blow up the power supply. I've seen power
supplies rated at 300 watts, that will shut down or die at 200 watts
(or less).
If you think you have cheap junk or defective capacitors, they will
show a drop in output voltage somewhat before full load. That not
because the regulation fails. It's because they no longer act as a
filter and you'll start to see switching ripple on the output leads.
Insufficient filtering will do much the same thing. Attach an
oscilloscope to the output leads and you should see the noise level
increase dramatically with bad caps.
>>Some power supplies are junk.
>Perhaps mine is in that league!
I've never heard of a Whatsup power supply. Neither has Google. That
should offer a clue.
Grab a flashlight and look inside. If you see a crowded pile of
electrolytic filter capacitors near the output section, it's probably
a good design. If you see fairly small and few filter capacitors,
it's probably (cheap) junk.
Good:
<http://commons.wikimedia.org/wiki/File:ATX_power_supply_interior.jpg>
Plenty of output filtering.
Junk:
<http://i01.i.aliimg.com/photo/v0/364432581/computer_power_supply_ATX_12V_2_2.jpg>
Note the lack of enough output filter caps.
David Nebenzahl
> "Jeff Liebermann"
>
>> [O.P. wrote:]
Wellll ...
Isn't the correct answer here "don't worry about it"?
I mean, the O.P. is reporting a voltage drop of 80 mV. Compared to a
reference of 12 volts (nominal), that's ... lessee ... 0.67%. Much less
than any *rated* regulated variation of any computer power supply, right?
--
How To Access Wikileaks
These sites are still up as of 12/3/10:
http://wikileaks.de
http://wikileaks.fi
http://wikileaks.nl
http://wikileaks.eu
http://wikileaks.pl
And these IP addresses can be used:
Jeff Liebermann
wrote:
>"Jeff Liebermann"
>>
>>>I tested it with a CD ROM drive
>>>attached and noticed that the 12V rail gives me a lower voltage when
>>>the CD tray is moving (11.79V against the usual 11.87V).
>>>Could this be due to bad capacitors?
>>
>> If you're getting 10mv accuracy with your DVM, you must have a decent
>> way of calibrating your instrument to that resolution. 100mv accuracy
>> is more reasonable.
>
>** How pedantic and idiotic !!!
Pedantic
-adjective
1. ostentatious in one's learning.
2. overly concerned with minute details or formalisms, esp. in
teaching.
Ummmm... ok. I'm impressed with your ability to read between the
lines, especially when there's nothing there to read.
>The OP is only asking about the small DROP in voltage he measured under
>load - which NOT dependant on calibration or accuracy of the meter.
The accuracy of the DVM's ability to measure 12.000000000 VDC is
dependent on the calibration. While he could measure the relative
voltage drop (80 mv) with considerable accuracy (as you noted), he
could not determine whether the output voltage is within the specified
5% ATX tolerance (11.4 to 12.6VDC) unless the meter were calibrated to
a tolerance much better than 5%. The question was whether the power
supply was defective, which in my opinion is more a matter of keeping
the output voltage within the specifications (with any load).
>The LINEARITY of most DMMs is of a very high order - 1 part in 1000 is
>typical.
Linearity is usually +/- 1 bit or LSB. If it's a 3 1/2 digit cheapo
DVM, with a full scale reading of 1999, it probably has 10 bit A/D
converter. At full scale, on the 20 VDC scale:
1 bit error = 20.0 V / 2000 = +/- 10 mv error.
For an 80 mv difference, there's a 40 mv uncertainty due to
non-linearity (+/-10 mv error for the no-load, and another +/-10 mv
for the loaded readings). That's worst case and rather improbable for
the rather small 80 mv measurement difference.
Phil Allison
"David Nebenzahl"
Phil Allison spake thus:
>
>> "Jeff Liebermann"
>>
>>> [O.P. wrote:]
>>>
>>>> I tested it with a CD ROM drive attached and noticed that the 12V
>>>> rail gives me a lower voltage when the CD tray is moving (11.79V
>>>> against the usual 11.87V). Could this be due to bad capacitors?
>>>
>>> If you're getting 10mv accuracy with your DVM, you must have a
>>> decent way of calibrating your instrument to that resolution. 100mv
>>> accuracy is more reasonable.
>>
>> ** How pedantic and idiotic !!!
>>
>> The OP is only asking about the small DROP in voltage he measured under
>> load - which NOT dependant on calibration or accuracy of the meter.
>>
>> The LINEARITY of most DMMs is of a very high order - 1 part in 1000
>> is typical.
>
> Wellll ...
>
> Isn't the correct answer here "don't worry about it"?
** NO !!
The pig arrogant arsehole made a false point.
.... Phil
Phil Allison
"Jeff Liebermann is a FUCKING CUNT"
>> If you're getting 10mv accuracy with your DVM, you must have a decent
>>> way of calibrating your instrument to that resolution. 100mv accuracy
>>> is more reasonable.
>>
>>** How pedantic and idiotic !!!
>
> Pedantic
> -adjective
> 1. ostentatious in one's learning.
> 2. overly concerned with minute details or formalisms, esp. in
> teaching.
** More fucking, idotic pedantry.
Only thing this stinking POS is any good at.
> Ummmm... ok.
** Go fuck your mother.
>>The OP is only asking about the small DROP in voltage he measured under
>>load - which NOT dependant on calibration or accuracy of the meter.
>
( Snip more pedantic, fucking SHIT )
> The question was whether the power
> supply was defective, which in my opinion is more a matter of keeping
> the output voltage within the specifications (with any load).
** More pedantic BOLLOCKS.
>>The LINEARITY of most DMMs is of a very high order - 1 part in 1000 is
>>typical.
>
> Linearity is usually +/- 1 bit or LSB. If it's a 3 1/2 digit cheapo
> DVM, with a full scale reading of 1999, it probably has 10 bit A/D
> converter.
** ROTFLMAO !!!!!
Wot a FUCKING MORON.
Regular DMMS do not use conventional A-D converters.
They use "dual slope integration " - ie an analogue technique.
The "digital" bit is the fucking display.
IMBECILE !!!!!!!!!!!
> 1 bit error = 20.0 V / 2000 = +/- 10 mv error.
> For an 80 mv difference, there's a 40 mv uncertainty due to
> non-linearity (+/-10 mv error for the no-load, and another +/-10 mv
> for the loaded readings).
** Crapology.
The same meter is used and so the only error is the +/-1 bit display
uncertainty.
GO FUCK YOURSELF you STINKING PEDANT !!
.... Phil
Jeff Liebermann
wrote:
>>>The LINEARITY of most DMMs is of a very high order - 1 part in 1000 is
>>>typical.
>>
>> Linearity is usually +/- 1 bit or LSB. If it's a 3 1/2 digit cheapo
>> DVM, with a full scale reading of 1999, it probably has 10 bit A/D
>> converter.
> Regular DMMS do not use conventional A-D converters.
> They use "dual slope integration " - ie an analogue technique.
> The "digital" bit is the fucking display.
Thanks. I didn't know that. The DVM's I play with all have SAR
(successive approximation register) type A/D converters, which suffer
from missing code linearity problems. Under ideal circumstances, the
minimum is the display uncertainty of 1 LSB.
Microchip. Analog-to-Digital Converter Design Guide
<http://ww1.microchip.com/downloads/en/devicedoc/21841a.pdf>
> The same meter is used and so the only error is the +/-1 bit display
>uncertainty.
That's also the linearity if everything else were perfect. Reading
between the lines of your copious insults, you seem to be referring to
"analog linearity", which does not include the digital conversion
part. Assuming the integrating capacitor doesn't have a dielectric
absorption problem, the analog section is quite linear.
Unfortunately, you completely missed my point. If you make two
voltage measurements with the same meter, and there was some
non-linearity error, and the two measurements were fairly close
together, then (as you mentioned) any errors in the difference of
these two measurements would probably cancel. That's fine and I have
no problem with that.
If you use either voltage measurement, to determine if the output of
the power supply is within specification (Per ATX 2.2), then the not
so minor problem of absolute accuracy, something you seem to be
ignoring, is important. DMM's are speced at +/-1% or less, so it's
not a huge error when compared to the +/-5% power supply error.
However it does become important when measuring near the power supply
limits.
Phil Allison
"Jeff Liebermann = a lying AUTISTIC ASSHOLE "
** Drop dead you pathetic fuckwit.
..... Phil
David Nebenzahl
> "Jeff Liebermann = a lying AUTISTIC ASSHOLE "
>
> ** Drop dead you pathetic fuckwit.
OK, so my guess is, oh, another week or so before Phil posts a calm,
reasonable (and correct) message here.
Guess it takes that long for the meds to take effect ...
Phil Allison
"David Nebenzahl is Donkey Brained Troll "
>
>
>> "Jeff Liebermann = a lying AUTISTIC ASSHOLE "
>>
>> ** Drop dead you pathetic fuckwit.
>
> OK, so my guess is, oh, another week or so before Phil posts a calm,
> reasonable (and correct) message here.
** Shame is pal, EVERY SINGLE TIME I do just that - a bunch of
fucking, autistic RETARDS like YOU come along afterwards and try to piss on
it.
Same all over damn usenet -
nothing but FUCKING TROLLS !!!!!!!!!!!
.... Phil
Meat Plow
> "David Nebenzahl is Donkey Brained Troll "
>>
>>
>>> "Jeff Liebermann = a lying AUTISTIC ASSHOLE "
>>>
>>> ** Drop dead you pathetic fuckwit.
>>
>> OK, so my guess is, oh, another week or so before Phil posts a calm,
>> reasonable (and correct) message here.
>
>
> ** Shame is pal, EVERY SINGLE TIME I do just that - a bunch of
> fucking, autistic RETARDS like YOU come along afterwards and try to piss
> on it.
Agree- somewhat.
> Same all over damn usenet -
Comes with the territory Phil
> nothing but FUCKING TROLLS !!!!!!!!!!!
Including you.
--
Live Fast, Die Young and Leave a Pretty Corpse
Jeff Liebermann
wrote:
>"Jeff Liebermann = a lying AUTISTIC ASSHOLE "
Sheesh. I agreed with part of what you said. I thanked you for the
correction on the type of A/D converter. But, this is what I get. Did
you even read what I scribbled?
Have you EVER thanked or complimented anyone on what they posted? Do
you even know how to do that? You know, "thank you for taking the
time to explain the problem" or "thank you for the correction". Try
it. I think you'll find that admitting a mistake and/or offering a
compliment is a fair tension release, that just might produce a
beneficial reduction in your vitriolic and venomous comments.
>** Drop dead you pathetic fuckwit.
I appreciate your concern for my well being, but I think I can
adequately take care of myself. If not, ObamaCare will keep me going
until the money runs out. Try not to worry too much about it.
Phil Allison
"Jeff Liebermann = a lying AUTISTIC ASSHOLE "
** Drop dead you pathetic FUCKWIT !!!
..... Phil
David Nebenzahl
> "Jeff Liebermann = a lying AUTISTIC ASSHOLE "
>
> ** Drop dead you pathetic FUCKWIT !!!
So you see, Jeff, it's basically a medical problem, not an etiquette issue.
Too bad. Guy seems pretty sharp on his few good days ...
Phil Allison
"David Nebenzahl
** This vile, autistic cunt need one in the head.
..... Phil
Michael A. Terrell
David Nebenzahl wrote:
>
> On 12/5/2010 11:51 AM Phil Allison spake thus:
>
> > "Jeff Liebermann = a lying AUTISTIC ASSHOLE "
> >
> > ** Drop dead you pathetic fuckwit.
>
> OK, so my guess is, oh, another week or so before Phil posts a calm,
> reasonable (and correct) message here.
>
> Guess it takes that long for the meds to take effect ...
A baseball bat is quicker.
--
For the last time: I am not a mad scientist, I'm just a very ticked off
scientist!!!
Meat Plow
> David Nebenzahl wrote:
>>
>> On 12/5/2010 11:51 AM Phil Allison spake thus:
>>
>> > "Jeff Liebermann = a lying AUTISTIC ASSHOLE "
>> >
>> > ** Drop dead you pathetic fuckwit.
>>
>> OK, so my guess is, oh, another week or so before Phil posts a calm,
>> reasonable (and correct) message here.
>>
>> Guess it takes that long for the meds to take effect ...
>
>
> A baseball bat is quicker.
Sometimes that doesn't work and you end up like Dave.
Michael A. Terrell
"Don't try this without a firm grip, and a good follow through..."
stra...@yahoo.com
What others told you about only the 5 (or 3.3) being regulated is
true. As the load varies on that supply, the other voltages will rise
of fall based on the load current on the regulated supply. This is
normal and not broken. I first noticed this characteristic around 10
years ago when my power supply fan (runs on the 12 V supply) was
changing pitch depending on the processor activity. That machine
varied between 50 and 100 Watts.
G²