TDP / power consumption: how does it relate to actual heat?

[niels007]

How does it work?

Say a harddisk is specced at 7W idle and 8.6W seek (Samsung Spinpoint 1614). If it actually produced 7W and 8.6W in HEAT, wouldn't it have no energy left to spin the platters and power the circuit board? Am I seeing this wrong? I mean my CRT is now using 150W yet the casing is only lukewarm. Surely a lot of the 150W goes effectively into displaying the image?

It would seem, to me - which doesn't have to say much - that you need to know the 'efficiency' of the device? I.e. perhaps the FDB motor has ~60% efficiency so if it needs 4W it will output 40% and thus 1.6W into heat?


The reason I am wondering is that I'm doing some simulations with flow software and at 10W I need loads of airflow to keep the virtual harddisk cool. 10W is too much, but isn't the value in the specs too much as well?

[Rusty075]

It's all about The Law of Conservation of Energy. However much energy comes into the closed system (like a HDD or a monitor) has to come back out. In a HDD most of that energy is used to spin the platter, but that's not the end of it. The energy used to spin the platter gets transfered to the air inside the HDD and to the bearings via friction, and is converted to heat. Same with the energy used to move the heads. The power used for the circuit board gets converted to heat within the IC chips. A tiny fraction of the energy is used to create physical changes in the drive, such as bearing wear, but the rest does come back out as heat.

With a monitor a portion of the 150w gets converted to light energy and transmitted out the tube, but most gets wasted as heat. The reason your 150w monitor feels only warm, while a 150w CPU would be too hot to even touch is all a matter of surface area. On the monitor that heat is being radiated from an area a thousand times as large.

[Straker]

Think of it this way - most drives use 20-30W when spinning up, and almost all of that power goes into getting the drive up to 7200rpm (or whatever). Now, the drive's spinning at 7200rpm, you're still putting energy into it but it's not speeding up, so almost all of the 5-8W a drive uses at that point has to be turned into heat.
If you're cooling it in a vacuum or something then yeah, it'll take a lot of air, but 7W is a trivial amount of power to dissipate through a big metal block, even via convection.

Better example - take a car with a 300hp engine, floor it, once you reach a constant velocity you're generating 300hp of heat via friction.

[wisdomtooth]

How does it work?

Say a harddisk is specced at 7W idle and 8.6W seek (Samsung Spinpoint 1614). If it actually produced 7W and 8.6W in HEAT, wouldn't it have no energy left to spin the platters and power the circuit board?

Newton's Laws of Motion. The HDD needs most of that 30W of power to overcome the inertia and static friction of a stationary platter to get it spinning on bootup. Once the platter is spinning (in motion), it requires only very little energy input to keep it spinning at a constant speed (it only has to overcome dynamic friction).

As far as heat generation is concerned, there are three sources of waste heat from an HDD: The mechanical friction of the spinning platter passing through air (air resistance), the mechanical friction of the HDD's motor spindle bearing, and the electrical resistance of its circuit board and electric motor. Most of that 7 to 8.6W of waste heat is generated by electrical resistance FYI, NOT through the platter and motor's mechanical friction.

And no, a car's engine running at 300 horse power is NOT generating 300HP of friction heat. The power output from the engine is also being used to power electrical systems, and if all of it is being used to overcome friction, you wouldn't be able to listen to your pimped up stereo system with 1000W speakers or have air conditioning/heating or power for that GPS navigation system or have headlights to avoid a deer staring at you from the middle of the road. On top of that, most of the heat generated by a car is from the chemical combustion of fuel, which requires dissipation via the radiator/liquid cooling system and the engine exhaust (about 70% in the most fuel-efficient cars, in fact).

[Straker]

And no, a car's engine running at 300 horse power is NOT generating 300HP of friction heat. The power output from the engine is also being used to power electrical systems, and if all of it is being used to overcome friction, you wouldn't be able to listen to your pimped up stereo system with 1000W speakers or have air conditioning/heating or power for that GPS navigation system or have headlights to avoid a deer staring at you from the middle of the road.

I found a primary source for that in Annals of the Obvious, but didn't bother citing.

On top of that, most of the heat generated by a car is from the chemical combustion of fuel, which requires dissipation via the radiator/liquid cooling system and the engine exhaust (about 70% in the most fuel-efficient cars, in fact).

yeah, I had originally typed "heat and friction heat" but thought even that was superfluous.
point being that nearly the entirety of the power going into the drive during idle is being turned into heat just to keep it spinning at the same rate.

[wisdomtooth]

...yeah, I had originally typed "heat and friction heat" but thought even that was superfluous...
point being that nearly the entirety of the power going into the drive during idle is being turned into heat just to keep it spinning at the same rate.

Nope...

You are thinking the majority of waste heat from an HDD is coming from the motion/friction of the platter and motor. It is NOT.

When an HDD spins up using that initial spike of 30W of power, it is NOT converting all of that into heat. It is converting about 20W of that power into KINETIC ENERGY (moving a stationary mass against inertia to 7800RPMs), NOT HEAT. The remaining 10W of that spike is being dissipated as heat, the VAST majority of it due to ELECTRICAL RESISTANCE in the IC's and the electric motor, and only a tiny percentage from mechanical friction.

And once the platter has spun up to its operating RPM, it required VERY LITTLE additional energy to keep it moving (to overcome dynamic friction). That's why the HDD's draw drops down to around 10W at idle, and again almost all of that 10W being dissipated as heat is coming from ELECTRICAL RESISTANCE, rather than MECHANICAL FRICTION.

The majority of waste heat from an HDD comes from electrical resistance in the circuits, and will remain so, until someone invents a practical room-temperature electrical superconductor.

See, in this case, your "Annals of the Obvious" won't help much... But consulting a high school physics textbook might.

HTH.

[Devonavar]

All energy eventually ends up as heat.

The power that is used to spin up the drive only remains as kinetic energy as long as the drive is spinning. Once the drive spins down, all that kinetic energy turns into heat via friction.

A very miniscule amount of energy is turned into sound energy, which eventually turns into heat after it leaves the drive. This is the only place where energy leaves the drive when it is not heat.

Hmm, I suppose that there's also a small amount of energy that leaves as kinetic energy because the drive vibrates. But we all know that heat = vibration, so maybe that counts as heat too...

[wisdomtooth]

And since the KE in a spinning platter is dissipated ONLY after you power it down, that KE is NOT part of that 8-10W of heat being dissipated by the drive while it's under power (which is what the OP is asking about).

The original poster is wondering how could the drive still be spinning or the circuit board still be working if the amount of heat the HDD is putting out is almost as much as the power going in.

The answer is that the platter requires only a very small amount of energy input to keep spinning once it has been spun up to its operating RPM-- There is very little heat loss from mechanical friction in the platter.

The vast majority of heat being produced by a powered HDD is NOT coming from mechanical friction as everyone else here seems to be thinking. Rather, it is coming from electrical resistance in the circuit board and electric motor.

HTH.

[Devonavar]

Ok, time for some actual facts:

From the review of the Hitachi 7K250 hard drive, which happens to have a low RPM mode:

Normal Idle (7,200 RPM): 6.9 W
Low Power Idle (7,200 RPM, heads unloaded): 4.8 W
Low RPM Standby (4,200 RPM, heads unloaded): 2.6 W

As you can see, the vast majority of the power comes from friction, since simply unloading the head, which flies a few nm above the platter, causes a 2W drop. Dropping the RPM to ~60% drops it by ~4W (almost a 60% drop ... could this proportionate response be a coincidence?)

Now, take a look at the typical power consumption of a notebook drive, which uses a similar controller, but a smaller, less powerful motor and smaller, less friction-prone platters: It's typically around 0.9W for an ATA controller, and slightly more for a SATA controller.

Bottom line: The vast majority of heat comes from friction. Notebook drives use smaller and fewer platters and less powerful motors spinning more slowly, and therefore consume an order of magnitude lower power than desktop drives. If you poke around on the SATA web site, you should be able to find power specs that put chipset power consumption around 0.7-0.9W. With power saving features enabled, this can be dropped to ~0.4W, roughly on par with the ATA interface.

You're being rude and condescending, and you don't have your facts straight. Cut it out.