Baldrick

[stiltman]

Just curious about the overheating CPUs, were these recommended upgrasdes to the box or outside the parameters for it?
or do you know?

I thought you were messing with Apache and PHP

Thanks for your newsgroup post

[pixel]

What's the newsgroup? Or is it just the irc channel (#vcdhelp)?

[Baldrick]

Outside the paramaters so I probably have to buy a new server case and new fans.

[stiltman]

What's the newsgroup? Or is it just the irc channel (#vcdhelp)?

I don't do IRC, well I haven't in probably 10yrs

The newsgroups were
alt.video.dvd.software
<22b6d3c1.0309060126.211f9308@posting.google.com >

rec.video.desktop
<22b6d3c1.0309060124.77528cfc@posting.google.com >

[The village idiot]

I never thought of checking those groups. Silly me. Now if it had been something like a.b......

[g_shocker182]

I thought newsgroups were for downloading illegal movies and such...

[tgm4883]

newsgroups are for just about everything

[stiltman]

I thought newsgroups were for downloading illegal movies and such...

I've been going to 1 group for about 7 years. There's only about 10 of us, but we all keep in touch in this small group.....very similar to what BBSs were used for back in the begining

[sterno]

I thought newsgroups were for downloading illegal movies and such...

A lot of people really do think things like that. A long time ago (you know, like way back in the early 90s, the DARK AGES man!) the web wasn't even remotely usable for anything practical. There definitely wasn't anything like a web bulletin board. People who only had access to BBS's used fidonet, people who had access to the internet (mostly colleges, scientific research, and military) used usenet. So newsgroups had a mixture of binaries (both legal and not), jokes and such, raging arguments, and serious discussions.

One amusing item from the history of usenet is the "Hitler rule". Usenet arguments (flame wars) are legendary for their vitriol. The Hitler rule was an informal guideline: as soon as anybody compared an opponent to Hitler all parties involved dropped the subject.

And before ICQ/AIM/etc people used IRC, finger, talk, and social MUDs....

And before massively-multiplayer online RPGs like Everquest, there were RPG MUDs....

[FOO]

I get hints here that someone has knowlege of the
reason this site was so down for so long. Will you
let us non-newsgroup members have a clue ?

[gmatov]

Baldrick,

How do you mean outside the parameters of the box?
The board will support the 2400's instead of the 1800's you had, so there should be enough power, shouldn't get a voltage drop to cause amperage to rise too much. The CPU's will draw X watts, volts go to low, amps have to rise to compensate.

I just bought a boxed 2400 today, the heat sink is probably an inch longer, 1/2 unch higher than supplied with my 2000. Can't be wider, of course, latches are fixed distance. Also has copper plug in base.

Haven't changed it yet, but think core is smaller than 2000 is, more concentrated heat source.

I think you need better heatsink, not box. Micro machines run high speed CPUs with no problem.

Are your new ones AMD Boxed, AMD fans/sinks?

George

[gmatov]

Foo,

Read above, and go back a couple days to when Baldrick posted the site had a planned outage to upgrade from 1800's to 2400MPU's on the servers.

And, again, above, and I think housepig's post where Baldrick said the new CPU's are getting too hot.

I haven't checked the price of 2400 MPU's for awhile but don't think you want to risk burning up 500 bucks worth of CPU's if you can find a fix.

Hey, I missed my daily fix, much as anyone, mebbe more.

I'd send a couple solid copper fan/sinks, but they're rated on the box only to 2200XP.

George

[The village idiot]

If it were me, I would put the old CPU's back in until I had better cooling. No sense burning up the new hardware.

[Baldrick]

Baldrick,

How do you mean outside the parameters of the box?
The board will support the 2400's instead of the 1800's you had, so there should be enough power, shouldn't get a voltage drop to cause amperage to rise too much. The CPU's will draw X watts, volts go to low, amps have to rise to compensate.

I just bought a boxed 2400 today, the heat sink is probably an inch longer, 1/2 unch higher than supplied with my 2000. Can't be wider, of course, latches are fixed distance. Also has copper plug in base.

Haven't changed it yet, but think core is smaller than 2000 is, more concentrated heat source.

I think you need better heatsink, not box. Micro machines run high speed CPUs with no problem.

Are your new ones AMD Boxed, AMD fans/sinks?

George

I think that the cpu fans/heatsink are "outside the paramaters" so new fans maybe will be enough. The case is a small "2u case" in a rack mounted server and I don't know how it is with coooling. It is mushkin that hosts the server and they are going to check the temparature now and see what it needs.

[The village idiot]

A 2 RU case is only 3.5 inches tall. Not a lot of room for fans. Going to be expensive. And a new case that is rackmount is even worse. Rackmount stuff is a big rip-off. The cases don't need to cost that much.

You might be able to put some kind of case fan into the PCI card holder that will increase the airflow through the case. That will be the cheapest solution. Did they clean any dust out of the fans when they changed the processors? Oh wait, I bet it is a "clean" room so no dust

[Solarjetman]

Oh wait, I bet it is a "clean" room so no dust

[FOO]

The CPU's will draw X watts, volts go to low, amps have to rise to compensate.

Not true. Where did you get that ?

[gmatov]

Basic electricity. Any electrical device is designed to run at a given current draw, at a given voltage, result is total wattage.

Voltage falls, current rises. It's why a few years back, when o'clocking was the big thing, if your machine was unstable you were advised to bump the voltage up in increments. Usually worked. Go to an OC site, if you doubt it.

You can also try running a power tool at the end of a long extension, check the volts and current draw, then do it just plugged into the wall.

Just don't use a heavy extension, just a standard, cheap 16 guage homeowner type. 100 feet, you'll probably be about 95 volts and your drill or whatever will quickly overheat.

Cooling a rack mount case is outa my league. Don't believe I've ever even seen one up close.

[FOO]

You have no understanding of "Basic Electricity" at all.
I have no wish to start a war, but I have to complain
about erroneous information being presented as fact.

Any electrical device is designed to run at a given current draw, at a given voltage

Not true. You can't just claim that all electrical devices have the same
behavior. Comparing an electric drill to a microprocessor is ridiculous.

Voltage falls, current rises.

It is vey rare for an electrical device to exhibit such behaviour.
I assure you that a microprocessor does not behave like that.
Assuming that devices operate at constant power is wrong.

The reason that increasing the voltage allows a cmos processor to run faster,
is that the transistors can switch more current with a higher gate voltage
and charge up the gates of the next transistor faster. This WILL result
in higher total device current.

The reason that the drill (may) burn up at the end of a long cord is that
it is operating at lower than design speed due to lower voltage. Current draw to a motor is mostly determined by it's speed. If you were to load the motor down to the same speed at 120 V, it would burn up faster.

[The village idiot]

EDit, sorry didn't read all the way. You do understand.



Power=current*voltage
power in watts
current in amps
voltage in volts

[gmatov]

Foo,

I daresay you know more about the theory than I do.
However, if you should run "that drill" at the end of a long, light extension, it will run at rated speed, and it will draw more current, and it will overheat.

The same goes for a CPU. If the voltage falls, the current will rise, if the current rises, the heat production will rise. As the CPU heats, resistance rises, and the current will rise even more, thus the heat output will rise, again.

Voltage, per se, is nothing. Current is the thing. It's why the cross country transmission towers are carrying 765000 volts, at relatively low current, so as not to melt the wires from the towers. Actually, it's why we have an AC system, today, rather than Edison's favored DC.

There may be an overvoltage fuse/circuit breaker, but the vast majority are current sensitive/limiting devices. They are rated in AMPS, not volts nor watts. When the AMPERAGE gets to the set point they trip or blow, not when the volts/watts get to a point..

A CPU has, from some estimates, miles of conductor on the chip. The voltage drops the farther it flows. The heat rises. The CPU gets so hot from the resulting undervoltage that O'clockers boost the initial voltage so that the voltage "further down the line" does not cause a meltdown.

If you're an electronic engineer, show me where I am in error.

George

[Solarjetman]

Well, I do not know how a drill works, but it would be possible to make a item draw more current when the voltage drops, but it seems to me that would not be a good idea. However, you would have to add circuitry to accomplish that.

V=IR (ohm's law)

R (the resistance) never changes. Thus if V (the voltage) drops then I (the current) would have to drop to compensate.

P=IV (equation for power)

If V is dropping, and P (power) stays the same as you say, then I (current) would have increase making ohm's law incorrect. And ohm's law is not going to be shown incorrect by a common household drill

HOWEVER! If the resistance is not a constant, then the current COULD rise as voltage drops. Thus is is possible to make a device like gamtov is describing. I do not know if drills do that though.

because...
I=V/R (ohm's law re-written)
P=(V/R)*V = (V^2)/R

So to make P remain constant while V decreased, then R would have to decrease.

Q.E.D.

And as it is 3:00 in the morning here, that was WAY too much thinking for my brain to handle at this hour, and I think it just overheated from too much current or something. So, I am going to bed.

The Jetman has spoken

[gmatov]

Solar, if you're still awake,

ALL electrical devces work that way.. Why the hell do you think we have gone from the 3.5 volts of the P2s, K5s etc. Because the traces in the CPU have gone down to .13 microns, and now down to.09 microns. Finer wires. The finer the wire, the less current, note, current, it can draw without burning out.

It can take 27000 volts, as the super thin traces in a TV. Oh, by the way, those traces vapor deposited into the channels in a CPU are aluminum, and aluminum in a piss poor conductor. Heat builds up, because of the inherent resistance of aluminum. Everyone is talking about Copper interconnects, but that is not copper deposition, it's copper for the 460, or 478, or for the newer CPUs to come, 700+ and 900+ pin connections..

The hell of it is, the finer, and closer the traces are on a CPU, the lower the voltage has to be, as it will jump to the next trace. Short!! Burnout!! Shot CPU!!

Ohm's law does work, here.

I just went back to your post..

OK, if you have 500 feet of X guage wire wound into a "drill", it will run at , say, 4.5 amps, at 120 volts, that is 540 watts, that is nearly 3/4 horsepower (US=746 watts per HP, I think Europe says 756 watts).
I have charts that have the impedance/resistance of nearly every metal, given in ohms per foot/ It is simple mathematics, with a little electrical theory thrown in. 500 feet of a given guage wire will cause an impedance of X Ohms, which will cause a draw of X Amps, at X volts. Double the voltage, as any computer box you buy, dual voltage, halve the current.

Simple. Electronic engineers, please explain why Ohm's law is irrelevant to electronics. Not counting vapochill, as we have been trying from the time of Willie Shorts to arrive at a room temp super conductor..

Ah, well,

George

[gmatov]

Solar,

I might have missed something, there..
Actually, it's E=I over R, or any combination. You can find volts, amps, resistance, depending on your inputs and the order.
E -= Electromotive Force, ie, volts.

Innyhoo, a drill, or any other motor for that matter, has X feet of wire, at a given, and calculated diameter, resistance per foot, wound onto the armature, the rotating member, in a set number of poles. 2 pole at 60 cycle, US, is 3600 RPM. You have to have at least 2 poles..

This is one of the big, big points that Shill is missing.. The traces n MoBos are also getting closer and closer together. As should be evidenced from the problems Intel had with their RD RAM. Crosstalk. Not enough insulation to keep the 2 or 3 RAM chips from crosstalking, interfering with the neighbor chip.. Chaos. All f'd up.

Hey, go read up on it.

George

[sterno]

The basic error is that you're taking a few principles of high school electronics, muddling them together, and then applying them where they don't belong. The voltage/current relationships you talk about only really apply to resistive elements and certain electromagnetic components (like transformers). Every electrical device you see isn't that simple, if it were there wouldn't be much need for 4- and 5-year college programs in electrical engineering, plus specializations within the field, plus ongoing advanced research, etc.

Pick up an introductory electrical engineering text and you'll probably find one chapter that covers basics like V=IR and P=(I^2)R and another 15 or 20 chapters describing the equations and techniques you use to model other stuff as resistors and voltage/current sources. Semiconductors have their own rules, and a few types of IC usually get a special focus because they're really common. Inductors and capacitors throw AC voltage and current out of phase, which means your circuit sees less power than your source is delivering - correcting that is a big deal in power supplies.

Electric motors are a little complicated. The only thing that keeps the coils in a motor from burning out with normal power is that the shaft is turning fast enough that individual coils don't heat up too much. If you don't give it enough power (for instance, if you put in a big resistor like a cheap cord) it bogs down under less of a load and heats up faster.

I'm not sure quite where crosstalk comes in to this, that's generally from the electromagnetic field around a wire/trace inducing an electron flow in a nearby wire/trace that interferes with its proper signal. It's been a problem in cheap data cables for years, and the faster you push data around the stronger the effect will be - inductance is partly a function of how fast the field is changing.

There's a whole lot more to electronics and power systems than Ohm's law. Knowing that gasoline burns might be a big part of understanding how a car engine works, but you couldn't build a Ferrari with only that knowledge.

[argiris]

In semiconductor chips, power dissipation depends on voltage, current and switching frequency:

Total power dissipation= DC power + Switching losses + Capacitive losses
P=Pdc + Psw +Pc

Pdc=voltage x current
Psw=Voltage difference x drive current x # of gates switcing
Pc= Capacitance x Voltage^2 x frequency

Processors get hotter because the frequency and number of gates increase (by user demand). Manufacturers try to keep voltage low because Pc increases exponentially with core voltage. The construction of the core gets smaller (less microns) to try to keep the capacitance low.

Major contribution to power budget on modern cpus comes from frequency and number of gates increase!!

And I agree with sterno on that engine example! Exactly on the point!

In short: No plain Ohm's law here!

[Baldrick]

Thanks for the electronic education....I could send this to my webprovider so he understands the heating process....

[stiltman]

Thanks for the electronic education....I could send this to my webprovider so he understands the heating process....

All this just because your boxen has a small fan in it

[FOO]

Baldrick,
I wouldn't send this thread to anyone. Profound ignorance is embarassing.

If you're an electronic engineer, show me where I am in error.

I am an electrical engineer. Other people have shown you where you are in error but you ignored them. There is a long list.

1. Ohms law applies only to resistors . A transistor is not a resistor
2. If you use wire resistance only to calculate the current in an AC motor
you will be drastically wrong, A drill is also not a resistor. Did you know
that you can run your drill at full power on about 35 Volts DC ? A drill
is a compound wound DC motor. It has around 10 poles.
3. A two pole AC induction motor runs a 3600/2 or 1800 rpm
4. You gotta get this outta your head: ELECTRICAL DEVICES DO NOT
RUN AT CONSTANT POWER !!!! If the voltage goes up, the current
goes up. (almost always)
5. The reason CPU voltages are going down is purely for speed. It takes longer to charge a capacitor to 5 volts than to 1.8. And the power goes down too. BECAUSE THE CURRENT GOES DOWN !

[Solarjetman]

What a convoluted mess we have here. I am having a hard time making out who is disagreeing and who is agreeing with what.

Sterno, argiris, (and maybe FOO): I THINK you guy are going against something I said. However, this is entirely uncalled for. I never said all electric circuits are pure resistive circuits. And I never said that I knew how a drill works (in fact I stated I did not). The entire case of my post was to show how if Voltage dropped, then Current could increase. And as resistive circuits and ohm’s law are easier to understand, I used it as an example.

Even though I do not have a degree in Electrical engineering, I do have the same credits as an electrical engineer (a story I will not get into). So I know first had about textbooks only having one chapter on ohms law I have had several college level electronics courses so I have had experience with Capacitance, Impedance, Power Factors, True / Reactive / Apparent power, the importance of Impedance matching and the like. But I am not debating any of those as I have not used them in two years. I do not want to get my facts mixed up. And as any person who knows electricity can tell you, when you throw alternating-current, capacitance, and inductance into the mix, it complicates matters over a simple resistive DC circuit.

In short, I am not trying to make an entire car from knowing gas explodes, only showing that it is possible for current to rise as voltage drops by using a simple example. I am not relating this to any particular implementation.

So please, do not take my post out of context. If you find something iccorrect about what I said as it relates to a purely resistive circuit, I would like to hear it.