Strange USB Behavior

[Dricht1]

Alright, I've never met a personal computer issue I can't work out, but this one may be it. I've spent a good month working on it to no avail.

The issue: USB connectivity drops when using USB 3.0 hub, powered or otherwise. Seems to occur regardless of which port is used, and has some connection with load. Rarely occurs during normal use but happens almost immediately when games are booted. Typical symptoms are brief disconnects and reconnects, getting progressively longer until the USB devices stop responding entirely. Sometimes unplugging and plugging back in will restore that specific usb device temporarily. Does not seem to affect all hub connected usb devices equally/simultaneously. Does occur when few/low power devices are connected.

The System: Custom build, z170-a Asus motherboard with a 6700k chip. Trident Z RAM rated for 3200. EVGA 1080 Hybrid card, corsair h115i cooler, EVGA Platinum PS1000 psu. Currently booting from a 960 Evo m.2 drive with a 950 evo ssd for extra storage. Everything is new and seems to be fully functional in an identical build used by my gf.

The Hubs: sabrent powered 7-port usb 3.0 with dual fast charging ports. Also seems to affect the two port usb hub built into the monitor.

USB devices affected (confirmed working as normal when not connected to any hub) are a razer mamba mouse (affected by issue whether used wireless or wired), razer blackwidow chroma keyboard, creative Soundblaster katana soundbar, and corsair void wireless headset. Charging ports on the powered hub do not show any issues and do not influence occurrence of the problem for the other ports.

Everything is plugged into a 1000w APC ups that I've never seen rise above 65% or so utilization. I normally run an overclock but have tried stock with the same effects.

Steps Tried So Far: At first I wondered if it was the hub, but I've tried three of them with identical behavior as well as the monitor hub. I thought it could be the PSU which I was thinking of upgrading anyway, so I swapped the original 750 Platinum for the current 1000. I thought it could be the motherboard or overclock so I have tested with stock settings and updated the motherboard to the latest firmware. I have also reinstalled all drivers I could under the USB section of device manager and also made sure power management and selective suspending options that might turn off the ports were not enabled. I just finished reinstalling windows and plan to test again when I get home, but am wondering if anyone else has experienced this issue. Nothing I can find seems to be an exact match to this issue.

Any thoughts? I believe the affected devices are USB 2.0, which screams driver issue to me but reinstalling everything I could find didn't fix anything. It's especially puzzling that the identical system is not exhibiting any of these problems.

Thanks in advance.

 

[Lutfij]

From what I'm reading if the issue comes from the HUB's why not have them disconnected in the first place and see if the issue persists if you use each port individually as a stand alone components port? It's quite possible to be an issue with the USB drivers for which it's a good idea to reinstall the chipset and USB drivers found off of Intel although the latest should be found on Asus's product suuport site.

I would ofc like to ask, if you've recreated the USB installer for Windows 10 in order to rule out the installer being corrupt. I'm assuming Windows 10 since Skylake would have more issues on Windows 7 than on 10.

Lastly, might I ask, why the need for a hub? Aren't the ports on your board and the front panel enough for all peripheral connections? There was a documented case here on Tom'sHardware where one user found a 5v power being fed to the system and causing an issue with bootups(this came to me after you mentioned your hub can power devices too). which is why I'm asking to remove the hub entirely since it's brought about the issue.

 

[Dricht1]

Looks like I forgot to mention (I thought I had) that all devices seem to behave normally if plugged into USB ports rather than a hub. However, due to the arrangement of my desk using the ports is not ideal. I also don't have enough to plug in all peripherals and device chargers.

Working through the drivers again now that windows is reset.

I have been getting by without the hub but can't accept that a modern build would be unable to use hubs at all. It happens with both the powered hub and the monitor hub.

 

[Lutfij]

The glitch is in Windows 10 trying to constantly download drivers it thinks is right for your system. For which reason I merely say, keep your devices that need charging off the system and onto a power block/adapter for purely delivering power. For all the other devices that are necessary for communication like wireless adapters, keyboard's, mice, webcams and even older USB printers there is the USB ports but in this day and age you have ports coming off the back and front of the case.

The USB hubs as of late(since early 2016) have been wonky on Windows 10 and Skylake in general.

 

[Dricht1]

I am gradually rebuilding the system in terms of software, drivers and settings in hopes of determining a solution. The identical PC is still operating as expected, and until it came time to enable XMP for the 3200 memory sticks everything seemed to be working fine on this one. I will post if I can isolate the exact point of failure.

 

[Dricht1]

I'm not ready to call this solved just yet, but modifying the RAM settings in bios seems to be the key. The default XMP settings at the advertised timings and speed cause the issue to occur, even with raised or lowered voltage. Utilizing a 2133 speed and auto voltage has eliminated the problem for the moment. Further testing will have to happen to see if this is the actual and only cause...

 

[Lutfij]

This is why we ask for a full system's specs listed out including their SKU numbers. Often times ram kits need slightly higher volts but since they conform to JEDEC's specifications, they should not be going over 1.3v(for DDR4 at least). When you speak of AUTO Voltages, I'm sure in BIOS you can see what sort of power the AUTO voltage sets for your ram...that way you can work your way down from there and see if the issue crops up.

Is the BIOS for your motherboard up to date? On the Z170 platform you should be able to hit 3200MHz with the X.M.P profile. With 2133Mhz that would mean you are way below what you paid for. Further testing on your end will reveal more but if it comes to reinstalling your OS again, make sure all your peripherals except keyboard and mouse is disconnected. Gradually adding peripherals after you've installed your chipset, USB and MEI drivers.

 

[Dricht1]

As I mentioned in the OP, the mobo firmware is the latest version. It sets the RAM to the 1.35 voltage advertised by the RAM manufacturer, but moving the voltage up or down in stages (min tested 1.25 max 1.4) does not seem to have any noticeable effect other than reduced overall system stability at low volts. Usb hub still cuts out on anything that is stable.

I've been pretty methodical about testing each piece of software and peripheral one at a time during the rebuild to try to minimize variables. Thus far absolutely everything has worked flawlessly aside from the XMP settings.

Once I have time I'm going to see what else (if anything) the mobo changes when xmp is enabled, and hope the RAM or mobo doesn't need a return or RMA.