HP Support Forums
Join in the conversation.
I've been working in laptop thermal repairs for a good part of my PC career... boredom being what it is, and money being as scarce as it is, I often spend hours, even days, researching and Googling the inner workings of these systems, to solve the problem itself, not just replacing the parts and hoping for a better outcome (you know what they say about "insanity"...). I know that replacing the heat sink won't solve the problem, replacing the fan just mutes the symptoms (I haven't yet had a reason to replace the fan itself, BTW - laptop fans are nearly indestructible).
Here are my findings with my overheating dv4-1225dx:
- The overheating is caused by a too-high TDP processor paired with a lackluster heat pipe design. This is a raw, inexcusable marketing/design flaw by HP. Amusingly, the CPU itself isn't even advertised on the PC case. It's like this just fell through the cracks of the design/marketing guys.
- CPU is an AMD Turion 64 x2 (Athlon 64 x2 - don't understand AMD's Turion/Athlon thing), RM-72, "Lion" core. Claims a TDP (thermal design power) of 35 watts.
- The 780G (Radeon HD 3200 IGP) chipset, paired with the same heat pipe as the 35w CPU, is found to have about a 15W TDP under load (THAT was a pain in the arse to find, AMD seems to have the TDP specs pretty well guarded). But here's the kicker: the 780G chipset is a DESKTOP chipset - commonly used in MiniATX boards and PCs! Most results in searching for 780G refer to their desktop counterparts.
- This equates to about a 50-watt thermal load on the tiny heat pipe under full load. Let's go on and see why that's a problem. - Some heat pipes in HP laptops I've seen from the Pentium 4 era are able to, somewhat uncomfortably, exhaust the heat from a 125W 3.8GHz Pentium 4 with Hyperthreading. How? Extensive, thick-built heat pipes, and multiple channels of dissipation from the source (CPU). That's only a little over twice the heat produced by this chipset+CPU combo under full load. This laptop, though? It gets a tiny stick of a heat pipe.
- Heat pipes are miniature phase-change heat exchangers. Inside the heat pipe, there is a somewhat-complex system of spongy material and refrigerant, which, when it becomes heated, will transition to the inner channel of the pipe, carrying the heat with it, until a point where it becomes cooled and flows along the outer edge again to get re-heated. It's an amusingly efficient system, but it does involve physical movement within the pipe (much to many peoples' surprise). This also adds a little bit of complexity...
- If the heat pipe is kinked or damaged in just the slightest way, or even improperly angled or oriented, it significantly restricts the flow of this cooling material and causes a "hot spot" on one side of the obstruction. Many instances of first-hand experience to back this up... however, it can sometimes be repaired by padding the edges and "nudging" the edges of the pipe to get the dent out. I've had luck with this on my main PC, where the bottom panel got banged and slightly dented the pipe - frustratingly causing my Mobile i5 CPU to go 85C+ under load, without turbo (thermally throttled). After "massaging" the dent out and slightly expanding the height (into more of a square) in the process, my temperature is down to about 72-75 C under full turbo. Night and day difference.
- However, at the same time, temperature variations can cause "partitioning" as well. I've found that heat pipes can be rather picky when it comes to their thermal distribution - cool off a part too fast and it will "lock up", it'll stop moving heat! Except for the copper conductivity, but that's very limited. The further you push the heat pipe's thermal conductivity, the more likely you are to "crash" the pipe, causing it to stop (or significantly slow) moving heat. In the case of the 50W TDP design of the dv4-1225dx, this "heat pipe overload" is nearly inevitable, and most people have it happen without even noticing (other than the roaring fan and throttled performance). - Now let's consider the dismal heatsink attached to the end of the pipe. It's only connected on one face of the heat pipe, leaving the top side of the heat pipe completely un-conducted. It's blocked by the LCD hinge, and most of its exhaust facing is pointed downward in a half-ellipse shape. That also means that much of the air moved by the fan is dropped through the shortest point - the bottom. It both relies on, and is hindered by, the fact that the notebook sits on a table. Without sitting on a table, the air escapes the "short route" of the heat sink (downward) - but if it's sitting on a table, the LCD hinge typically obstructs most of the airflow. There's only one specific angle where the LCD obstructs the airflow the least, and that is with the LCD at a 90* angle to the keyboard.
- Of course, in traditional HP style, the cooling fan only has one air inlet: directly through the bottom of the unit. If that's blocked, airflow go bye-bye! No alternative airflow channels as in most PCs (my primary notebook, an Asus, pulls in air from the whole underside and is completely unimpeded by being covered by a bed or lap). It's pretty much entirely desk-bound due to this design. The cooling fan is 2-sided internally, but 75% of the area is blocked by the left speaker, while the rest is blocked by the fact that there's no inlet channels in the keyboard. The hole in the bottom is absolutely *IT* for cooling.
HP is 100% to blame for this terrible design, as it could have happened no other way. These are not defective one-off products; every single owner of a similarly-configured dv4-1225dx or related models will run into this thermal issue if they know what to look (or listen) for. Of course, being that it's a configuration problem and not a product-line problem (like the dv6000/9000/tx1000 nVidia northbridge/GPU failure), it's almost entirely unlikely that HP will do anything about it. One more, HP will probably not even issue any official forum response here either, as admission of fault nor denial of fault is actually possible in a business sense. Admission of fault would mean HP would be obligated to repair the issue or reimburse users. Denial of fault is pretty much made impossible by the proof posted by myself and others. Sooooo... we're pretty much screwed!
Now for a few solutions.
There aren't many options for making it work better - due to the thermal conductivity rating of the heat pipe in this model, it can *never* match the TDP of the CPU and chipset, it's simply impossible to exhaust the heat this CPU produces in a reliable manner. However, there are a few options for reducing the maximum TDP of the processor and making it run within specification of the shoddy heat pipe.
- Replace the CPU with one of a lower TDP. As previously linked, cpu-world.com is an amazingly useful resource for finding and comparing CPUs that will work in the restricted notebook environment. Socket 1 is nororiously unforgiving, so you will have to be sure to match up S1G2 with another S1G2 processor (contrary to some sources, I've first-hand found that S1 or "S1" does not work in the dv4's S1G2 board). Through a brief comparison, a suitable replacement may be the AMD Turion 64 X2 RM-70 (2.0GHz), It has a 31W TDP but a lower idle state (500MHz) which should help reduce heat. This is somewhat of an untested crap-shoot, but it's almost guaranteed to plug-n-play work in the dv4, so someone might want to give it a try...
- Reduce the maximum state of the CPU. This is a pretty drastic step, as the RM-72 only has 3 clock states: low, medium, and high (in layman's terms). Low state is 520MHz, 0.8v, 5W, reported to Windows as "28% Maximum Frequency". Medium state is 1050MHz, 0.95v, 11.4W, reported to Windows as "51% Maximum Frequency" (bouncing between 50% and 51%). High is, of course, 2100MHz, 1.125v, 31.50W, and reported as "100% Maximum Frequency".
- So, with this information, you can modify your Balanced (or "HP Recommended", big fat "[sic]" there) power scheme, and set your Processor Power Management -> Maximum Processor State -> "Plugged in", to 51%, and set "On battery" to 28%. This is what I'm using now. I typically recommend setting On Battery to 50% to keep a CPU at its lowest state on battery (that's what the throttling is there for, IMO!), but apparently this CPU uses a <50% state as its lowest clock. Also BE SURE to change "Minimum Processor State" to 5% for both - there's absolutely no reason to prevent the CPU from downclocking, as it will always instantly speed up on demand when the need arises (there is zero performance hit for allowing it to clock down).
Hopefully these tips help someone, as I sure did spend a good amount of time typing this out 
And hopefully HP's learned after all these years: use good, thick heat pipes, and exhaust to an unobstructed side, like a side vent! You'd think they learned with the dv6000's, the "first time"!
Oh, and HP, yes, I would absolutely love to work there. 
