Power Supply Voltage Tolerances and Measurements

Electronics Computer Programming Q&A
User avatar
haklesup
Posts: 2934
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by haklesup » Fri Oct 09, 2009 9:12 am

On a bright note today's processors use considerably less power to do more work than their counterparts of a few years ago. This results in less heat being generated by the CPU. Intel has dozens of white papers on that subject listing the power demands of their CPUs.
How true, My original comment about rapid swings in supply current are about 6-7 years old. I got it from a discussion I had with some AMD engineers during the clock speed wars with Intel. They were talking about how PSUs now had much higher current slew rate requirements (AKA response time) then then previously because of the higher clock speeds and strategies being employed to lower system power requirements.

Since the speed wars ended, there has been a quieter war going on to lower supply current due to efforts to capture more of the notebook market and now portable devices (and Green pressures more recently). Since lower supply current isn't "sexy" from a marketing POV, we don't hear much about it but the executives who make choices about what to put on motherboards are keenly aware. Smaller feature size and things like dual and quad core processers have allowed this current reductions. To see a really hot CPU you might need to go back and look at a single core 3Ghz system a few years old.

Relodron, that was a good example. I may look at similar performance specs on my own P4 celeron system this weekend, I don't have as many internal HDDs though. You might be able to push the power a little higher by running a defrag along with those three clients. And you're absolutely right about high performance GPUs being power hogs (at least for rich gamers).

User avatar
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron » Fri Oct 09, 2009 10:16 am

Now that's only a total of 324 watts, so i would think a power supply marked
with "450 watts" should be able to do it, right?
I would think so. For what you are looking to get done I would say those loads should work out fine. See what you have kicking around and let me know.

Ron

User avatar
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron » Fri Oct 09, 2009 10:24 am

CeaSaR wrote:Ron,

Where did you get your Kill-A-Watt? My local "The Shack" has them on the shelf (ok, rack).

CeaSaR
A few years ago (maybe 3) I was ordering parts for a computer build for the wife from New Egg and they were selling them for $19.95 so I ordered one. They were $19.95 because the newer revision of the same unit was out. I forget what was improved. However, I have used that little gizmo many times and it works great. Considering the price it was a good investment. I just wish they made one with an analog out of say 0 to 10 volts based on display. :)

Ron

User avatar
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron » Fri Oct 09, 2009 10:43 am

haklesup wrote:
On a bright note today's processors use considerably less power to do more work than their counterparts of a few years ago. This results in less heat being generated by the CPU. Intel has dozens of white papers on that subject listing the power demands of their CPUs.
How true, My original comment about rapid swings in supply current are about 6-7 years old. I got it from a discussion I had with some AMD engineers during the clock speed wars with Intel. They were talking about how PSUs now had much higher current slew rate requirements (AKA response time) then then previously because of the higher clock speeds and strategies being employed to lower system power requirements.

Since the speed wars ended, there has been a quieter war going on to lower supply current due to efforts to capture more of the notebook market and now portable devices (and Green pressures more recently). Since lower supply current isn't "sexy" from a marketing POV, we don't hear much about it but the executives who make choices about what to put on motherboards are keenly aware. Smaller feature size and things like dual and quad core processers have allowed this current reductions. To see a really hot CPU you might need to go back and look at a single core 3Ghz system a few years old.

Relodron, that was a good example. I may look at similar performance specs on my own P4 celeron system this weekend, I don't have as many internal HDDs though. You might be able to push the power a little higher by running a defrag along with those three clients. And you're absolutely right about high performance GPUs being power hogs (at least for rich gamers).
Oh man you nailed it! Among the last of the Pentium 4 processors from Intel was the P4 Prescott core. You could fry eggs on the thing. Interesting enough was there was also a 3 GHz Northwood core which came slightly later in the P4 family. The Northwood core was a cooler running single core CPU that did the same thing and more using less power. I think the Intel P4 Prescott core CPU was the last of the real power using CPUs.

As to graphics cards it has not changed a bit with newer graphics cards drawing more and more power, well maybe a slight cut back but nothing like what has happened with CPUs. Some time ago in a computer related forum I am active in the issue was discussed to some extent. Someone asked how to supply power to some of the newer GPUs when paired and I showed them how it should be done:

Image

Image

All you need is an additional PSU plus a boost from a good 12 Volt source. :smile:

Ron

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Fri Oct 09, 2009 11:20 am

Hey Ron, i could have used that idea instead of buying a new case for a new power supply ! <chuckle>

You know you can also use one of those relatively new Hall Effect current sensors right?
You could connect the output directly to an analog meter and measure power input all day long if you like,
or go with a digital panel meter, or whatever you choose. You can also go with uC and display or
just a blinking LED to indicate power. The main thing is that the Hall Effect current sensors
also isolate the measured system from the measuring device (uC or meter) and they are accurate
enough for this kind of application. Because of the isolation, you could rig up a uC to measure
the HE current sensor output and feed it back into the serial port or whatever and use some
simple software to display the power right on screen. Would be neat i think.
The sensors are priced just under five dollars US and some can be used AC/DC.

BTW the power reading is very close to 120*Current reading for 120vac lines for example,
so 1.1 amps would mean 132 watts. You can approximate in your head or have the uC or
the software do the conversion for you.


As to testing the power supply, i think i'll just go with some slightly lower rated resistors and
test for less than one minute that way, then connect to an older computer i have around
to test overnight, possibly with some resistors also to load a little better.
This wont be extraordinarily expensive and will still allow decent testing i think.
You know what bites is, before i fixed this power supply i forgot to measure the ripple
voltage under some load (3.3v and 5v were loadable). That would have given me a good
reference point for how well this repair improved the supply overall. On the other hand,
the caps were so bad i bet the ripple was way out there. Now that it is fixed though
i guess i could still measure it under some load and see if it is within tolerance. If it is,
or is close, i would bet this power supply is now finally good to go.
I cant wait to try it in a computer now, but dont really want to use my good motherboard
as a test bed just in case something goes wrong :smile:
LEDs vs Bulbs, LEDs are winning.

User avatar
haklesup
Posts: 2934
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by haklesup » Fri Oct 09, 2009 1:11 pm

i think i'll just go with some slightly lower rated resistors and
test for less than one minute that way,
Resistors are rated for power dissipation in air which isn't necessarily moving. By making some sort of heat sink, you should be able to go a bit beyond the rated power without damage to the resistor. immersion in a bottle of DI water should make a decent heat sink for 12VDC even with uninsulated leads. So maybe you get a little electrolysis if it's tap water but that additional resistance should be quite a bit higher than your load resistor.

Come to think of it, two metal paddles in salt water would make a decent load which is variable by adding more salt. Its not necessarily very predictable though without some annoying calculations and careful measurements. Safe enough at 12V, just don't try it with anything more powerful or higher V or AC.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Fri Oct 09, 2009 3:55 pm

Hi Hackle,


I'm not sure i would want to use the electrolysis alone as the load, but i sure do like the idea about
immersion in distilled water. I really really like that idea, so much that i just have to try it.
I'll try it with the non insulated leads first, then if need be i'll paint the connections before
next immersion. Anyone ever try this yet?

A while back i thought about running coolant through the center of one of those big power
resistors, the ceramic kind with hollow axial center. It's like a pipe anyway :smile: so
why not run some coolant through it. Yeah it would take a pump too though and some
careful connections, and a radiator to get rid of the heat.

I'll try that power resistor immersion experiment tomorrow hopefully, if i can get some
distilled water by then. Would be very cool if i could get away with using a 10 watt
resistor for a 100 watt application :smile:
I wonder how well vegetable oil would work.
LEDs vs Bulbs, LEDs are winning.

User avatar
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron » Fri Oct 09, 2009 6:06 pm

Would be very cool if i could get away with using a 10 watt
resistor for a 100 watt application :smile:
I wonder how well vegetable oil would work.
I think using a 10 watt resistor for a 100 watt load would work just fine... for a very, very short period of time. :smile:

How about some relatively pure mineral oil as a bath for cooling?

Actually MrAl I had given thought to using a hall effect sensor. The new ones are pretty cool little devices and I would venture that is how this slick little Kill A Watt works.

Ron

User avatar
haklesup
Posts: 2934
Joined: Thu Aug 01, 2002 1:01 am
Location: San Jose CA
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by haklesup » Fri Oct 09, 2009 7:53 pm

I think a 10x improvement is a bit optimistic but I don't have a feel for how much to expect. Adding a metal heatsink around the resistor and combining with the water would be even better.

The advantages of oil are resistance to high voltage arcing, the ability to rise to a higher temperature without boiling and it doesn't evaporate as easily. The right oil would probably be easier to keep free of ions since water tends to contribute to corrosion and absorb gasses. Aside from that, Water is cheaper and easier to find and easier to clean up.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Sun Oct 11, 2009 8:39 am

Hi again,


Ron:
How about a full minute? More about this later in a new topic...
Mineral oil does sound like a good idea.
Yes those HE sensors are very nice. I am going to try to get to test one
this week just to see what it can do and of course what it cant do (with
temperature rise and all).

Hackle:
Yes, optimistic but guess what? I was even able to go above that level for
some time period.
We also tried a small 12v 80ma fan, and that was able to reduce the temperature
on a 10watt resistor by as much as 85 degrees F (47 deg C).
The results using distilled water was so good that i am starting a new topic
on this subject.

The power supply tested good even with a 12 amp load on the 12v supply, and
now it stays 'on' without any load at all when before it would shut right off.


Ok im off to start the new thread about the power resistors and water cooling...
It turned out to be very interesting.
LEDs vs Bulbs, LEDs are winning.

User avatar
reloadron
Posts: 519
Joined: Sat Jun 28, 2008 8:57 am
Location: Cleveland, Ohio
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by reloadron » Sun Oct 11, 2009 1:10 pm

I have to give you quite a bit of credit MrAl. Generally those PSUs end up in a landfill and I only know a few who have actually gone through the depths of repairing one.

I have been messing with the PIC thing in some detail. That began with the battery voltage thread. Eventually I should have some questions to post in another thread. I can see using a PIC to measure current with a sensor attached.

The water cooling of power resistors should make for an interesting topic.

Ron

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Mon Oct 12, 2009 6:45 am

Hi Ron,


Well thanks, but you know i found that repairing them when only the caps are bad isnt
too hard at all. Four screws to get the top off, four more to get the board out,
then unsolder the caps (solder sucker helps a lot here) and remove them noting polarity,
and then install and solder the new ones. I've also been informed that some of the boards
are marked wrong as to polarity so we have to note the polarity when pulling them out
unless we can trace back to the output (like +12v where the positive lead would
connect to).
The only other problem really is getting the right size caps. I was lucky this time that
the caps fit in as they were a little bigger than the old ones. The diameter was only
2mm larger but that was enough to make mounting impossible the way the old caps
were installed so it took a little doing to get them to sit right. I should have taken
before and after pics.
I wasnt going to fix it at first and save it for parts alone, but then i kept looking at it
and figured i should at least pull the output caps and check them to see if they are
looking high in ESR, and sure enough, two out of the three i pulled were bad. I was
surprised that they were THAT bad, with even the capacitance being so much lower
than the case marking that i figured that had to be the problem. The extra ripple
probably triggered the over voltage circuit, so i figured it would be just a matter of
changing a few caps. I then also noticed that two other caps were bad too
(bulging top) so i included those in the order for new parts and when they got here
replaced all five caps and then the power supply started working normal again at least
for a somewhat lighter load (didnt have resistors to test fully).
So all said and done it's really not that hard to do this. Phillips screwdriver, soldering iron,
solder sucker, solder, that's about all the tools needed. Each cap only has two leads too
so they can probably be removed without damaging the board without using a solder sucker.
Traces are a little close so it's a good idea to carefully check for solder bridges of course.
LEDs vs Bulbs, LEDs are winning.

User avatar
CeaSaR
Posts: 1691
Joined: Sat Nov 08, 2003 1:01 am
Location: Phoenixville, PA USA
Contact:

Re: Power Supply Voltage Tolerances and Measurements

Post by CeaSaR » Fri Oct 16, 2009 9:45 am

MrAl,

Check over at All Electronics for power resistors. I have the summer catalog in front of me and
the list is on page 61 at the bottom. Pricing is by power rating:
1-5 watts - 3 for $1.00
6-10 watts - 2 for $1.00
11-30 watts - $1.00 each
31-50 watts - $2.00 each
These come in ranges from .5 ohm to 2.2k ohm.

They also have 2 others, a 0.015 ohm 50W Dale resistor RH-50 (aluminum clad) for $2.00 each
and a 18 ohm 225W wire-wound reisistor for $5.85.

You might find what you need (want) there.

CeaSaR
Hey, what do I know?

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Sat Oct 17, 2009 8:26 am

Hi Ceasar,


Thanks, i'll take a look there.
LEDs vs Bulbs, LEDs are winning.

User avatar
MrAl
Posts: 3862
Joined: Fri Jan 11, 2002 1:01 am
Location: NewJersey
Contact:

Re: Power Supply Voltage Tolerances and Measuremen

Post by MrAl » Sat Oct 17, 2009 1:33 pm

Hi again,


I almost forgot to mention that after looking at the power supplies and the one schematic i see now
that it would not be extremely difficult to modify a power supply to put out maybe +2 to +12v instead
of the usual constant +12v. That would make it a variable voltage supply. Going over that (say 2 to 14v)
would be a little more difficult due to the overvoltage circuit, but most likely doable also.
Going over 16v would probably not work without changing the output cap(s).
LEDs vs Bulbs, LEDs are winning.

Post Reply

Who is online

Users browsing this forum: No registered users and 3 guests