Dell Precision 7540 and 7740 Owner's Thread

This is the only thing that I can think of...

Open the legacy "Sound" control panel.
Go to the "Communications" tab.
Set it to "do nothing" rather than "reduce the volume of other sounds by 80%".

If this fixes your problem then there is some background task running, doing something to make Windows think you are on an audio/video conference and that it needs to reduce the volume level.

 

Thank you. I will try it. It seems that in some moment laptop switches between speakers and in that time I can hear "glitching" noise.

P.S. No, the problem is still here. Also I have tried connecting my wired earphones and had no success. Laptop doesn't recognize them and plays aduio with built-in speakers.

 

I've had this with the Precision 7730. Please check using Aida64 or some similar tool what is the temperature of your PCH. It must be specifically the PCH. I know for sure Aida64 shows this temperature. If it's more than 70 Celsius on idle, it means that your PCH is being overloaded with temperature and does a ramp-down, causing this.

Quick fix is to NOT use the NVMe slot that is near the RAM. So just clear that slot (put your NVMe drive in another slot). Another fix is to set the PCIExpress Power Management to "Low Power Mode". This will force the NVMe drives to go to sleep when unused. DO NOT, I repeat, DO NOT use that NVMe slot to hold the Windows installation as it will be constantly used.

Rationale: If the slot near the RAM contains NVMe drive, laptop will pump electricity to it constantly, which will keep it awake, so it will run hot. On the other side of the system board, exactly on the same area there is this small chip (PCH) that runs hot whenever there is I/O (including sound) so PCH hot, NVMe hot, there's no place to blow the heat (PCH is not connected to the heatsink) and thus these things appear.

If these didn't help, but your PCH controller is still running at 70 Celsius or more in IDLE, then it might have melted down partly and you need to call Dell for a replacement. If you get a replacement, you'll need to either use the Power Management setting I mentioned earlier and set PCIe to low power mode, or simply not use the slot near the RAM at all. PCH temperature should be 50 Celsius in IDLE or LESS when booting up. In time it will run to about 55 Celsius, but on cold boot, that should not go over 50 in IDLE.

If the PCH is fine, but you still have the sound issues, then you will need to switch the sound driver to the one from Microsoft. Please see here a guide on how to do this. Reboot computer after this.

I had 3 x Precision 7730 full laptop replacements because of this and about 14 system board replacements in these 3 laptops just because of this PCH temperature issue (my PCH was running at 103 Celsius in IDLE). I started noticing this because I had exactly what you describe. It's a flaw by design, so the fix is for Dell to design a proper system board with PCH someplace else where it won't melt.

 

If you are having temperature issues with the PCH, please make sure that PCIe link state power management is set to Maximum power savings in Windows power settings / advanced area. It is basically required to use the Precision 7X30 systems this way to keep its temperature under control.

 

I'd say this should be the default settting on all notebooks—I haven't seen a single notebook where the PCH chip was cooled at all, either passively by a simple fin assembly, or actively by being under the heatpipe and fan system.

In fact, there should be no real reason to use any other PCIe ASPM mode at all except Maximum Power Savings, as modern OSes do a very good job of firing up the links when necessary. Furthermore, keeping the PCH controller (and the PCIe links attached to it) at high power all the time would be a huge power drain by wasting power maintaining a link down which no I/O is happening. Furthermore, on a convectively cooled PCH chip, temperatures above 80 °C would just mean a less stable system.

The negligible increase in latency is entirely worth massively improved system stability and improved power management.

I get a feeling @Soromeister kept his ASPM power state 'Off'—otherwise, there should have been little reason to burn through three notebooks, all with the same problem.

 

For anyone out there in need of a long battery life Linux distro, I have found that Manjaro (KDE Plasma edition) with the latest 5.6 kernel, -0.110v undervolt and default TLP optimisations can achieve nearly 12 hours of actual (light) usage. I suspect it must be due to new chipset optimisations on the new kernel as previously (with Linux Mint Kernel 5.4) I was getting closer to 7-8 hour mark, often less than I would get with Windows 10. If I turn off WIFI and use it just for reading docs then consumption averages around 6w, which translates to estimates of 15 hours battery life. This is on a 7540 with a 9980HK 128GB RAM Intel GPU 4K Screen.
Also, on a slightly related note, avoid Clear Linux if you want battery life, for some reason its atrociously bad at around 3-4 hours of use.

 

Can you please detail this? How do you undervolt in Manajaro and what TLP optimisations are done? Is this some package you install or how is this carried out?

I have Precision 7740 Intel + Nvidia with Optimus (4k display + 6-cell 97Wh Lithium Ion battery) and I couldn't make it last more than 3 hours with Windows 10 with light usage and everything on max power savings mode. If I just leave it as it is (IDLE) then I get 4 hours, sometimes 4 and a half, so I really find this unbelievable when people say they achieve battery time of 6 or 8 hours with light usage. Power draw oscilates between 22w and 60w, depending on what I'm doing.

NVidia GPU will never stay in sleep mode so if I start a movie, even though everything is set to use the Intel GPU, the Nvidia GPU will still wake from sleep and not go back.

I've tried different settings on freshly installed Windows 10 and concluded that getting more than 4 hours on this laptop is pure fantasy so I really wish to test this in Manjaro.

 

There are some exceptions, e.g. Alienware m15 R2 (a poorly designed laptop otherwise due to soldered RAM). My AW m15 R1 is set to moderate PCIE power savings and temps never breach 70C under load. I've actually added a custom mod which cooled the PCH down a few C which made absolutely no difference to any benchmark results - system stability has been great throughout. Looks like these Precisions might be more prone to PCH-related issues for some reason.

 

Really nice figures you've got there. I'm running Arch + default kernel 5.5.5. I also have a dGPU, so I've set up PRIME offloading and RTD3 power management. However, the latter doesn't appear to work and I've set my battery to stop charging at 80%, so I get around 4.5 hours there. Idle power draw is at around 13-14 W, and nvidia-smi reports that Xorg is drawing power, and only the P8 state has been reached by the GPU.

Have you tried the 'hack' I mentioned, by keeping a 3D application (Paint 3D should work) in idle in the background?

 

When I pulled the 7740 apart the other day I noticed that a thermal pad was placed on the PCH so it could dissipate heat to the palm rest. My PCH temp tends to be between 45C-60C depending on the load.