PSU voltage instability

S

Slater

Guest
Hello, I have an ATX power supply. 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?
 
"Slater"
Hello, I have an ATX power supply. 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?
** 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
 
Slater wrote:
Hello, I have an ATX power supply. 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?
Often only one of the voltages, for example the the 5volt, is regulated,
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.
 
On Sun, 05 Dec 2010 02:57:32 GMT, nontelo@dico.no (Slater) wrote:

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 jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
Jeff Liebermann <jeffl@cruzio.com> wrote:

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!
 
"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
 
On Sun, 05 Dec 2010 04:28:25 GMT, nontelo@dico.no (Slater) wrote:

Jeff Liebermann <jeffl@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.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
On 12/4/2010 10:57 PM 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"?

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:

http://213.251.145.96/
http://88.80.13.160/
 
On Sun, 5 Dec 2010 17:57:44 +1100, "Phil Allison" <phil_a@tpg.com.au>
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.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
"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
 
"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
 
On Sun, 5 Dec 2010 19:28:39 +1100, "Phil Allison" <phil_a@tpg.com.au>
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.

--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
"Jeff Liebermann = a lying AUTISTIC ASSHOLE "


** Drop dead you pathetic fuckwit.




...... Phil
 
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 ...


--
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:

http://213.251.145.96/
http://88.80.13.160/
 
"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
 
On Mon, 06 Dec 2010 08:12:23 +1100, Phil Allison wrote:

"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
 
On Mon, 6 Dec 2010 06:51:43 +1100, "Phil Allison" <phil_a@tpg.com.au>
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.



--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
 
"Jeff Liebermann = a lying AUTISTIC ASSHOLE "



** Drop dead you pathetic FUCKWIT !!!




...... Phil
 
On 12/5/2010 6:22 PM Phil Allison spake thus:

"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 ...


--
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:

http://213.251.145.96/
http://88.80.13.160/
 
"David Nebenzahl


** This vile, autistic cunt need one in the head.



...... Phil
 

Welcome to EDABoard.com

Sponsor

Back
Top