My Corsair iCUE H115i RGB Pro XT randomly disconnects (Windows 11 Home 23H2, Ryzen 9 3900x)

[Hoover1979]

At least 3 times a day, my H115i RGB Pro XT goes silent and the event is accompanied by a Windows USB disconnect sound.  I have to immediately restart my PC, losing progress on any app or game in the process to prevent my CPU from overheating as the fans come to a complete stop, and it doesn't reconnect by itself.  This is bugging the $%#@ out of me!!!

I have included a Pastebin link containing my full DXDiag information, if that helps.
https://pastebin.com/kFH2W6m6

[c-attack]

Losing the usb connection should drop it into HW mode only, which other than changing lighting should not impact performance. If the fans and pump stop, that is a clear sign of power loss and it puts the speed issue into a different light. It appears the unit is experiencing some type of power delivery problem. There is no software mechanism to stop the pump and it should run at all times regardless of software state.
 

If you are still in your 5 year warranty period, contact Corsair Support to start the RMA process. If not, you’ll need to go shopping. Unless the SATA connector delivering power is only half connected, this is going to be a terminal failure pretty soon. 

[Hoover1979]

I can confirm the device is disconnecting completely.  I get the USB disconnect sound (without any following reconnect sound), the RGB goes dark and the fans grind to a complete halt.  I either have to restart my rig or restart the iCUE service in the iCUE settings, before my Ryzen 9 3900x overheats.  If I fill my USB ports (including USB 2.0 ports) the cooler disconnects every couple of minutes.  I am not using any USB 2.0 ports (due to a somewhat widespread issue affecting X570, B450, & B550 chipsets running 3rd gen Ryzen processors, causing random USB devices to disconnect and not reconnect until they are unplugged and plugged in again).  At the moment I have 1 keyboard (SteelSeries Apex 7), 1 mouse (Logitech G502 X Pro), 1 HDTV Tuner (Hauppauge WinTV HVR-935), and my receiver (Creative Sound Blaster X3), and I still have my cooler disconnected at least 1-2 times a day (3-4 times a day if I connect my Xbox Wireless adapter and 8BitDo wireless adapter for my arcade controller).  This can only be fixed by Microsoft, and we all know Microsoft doesn't fix things in their updates, Microsoft breaks things. 

I also had to disable fTPM in BIOS as I was hit with an extension of the fTPM stuttering problems affecting Ryzen users throughout 2022 and into 2023.  What happens for me is when Windows tries to engage TPM, the CPU freezes and never recovers, and I have no choice but to hard reset my rig.  I have to watch my PC like a hawk now as if I don't and the cooler disconnects, my CPU will fry for sure.

Edited March 25 by Hoover1979

[c-attack]

There's no device power on the USB connection.  That's the SATA, so if fans stop and lights go out that is more likely a device power drop, which then in turn triggers the Windows device disconnect sound because the thing just lost power.  This is not normal behavior or something that comes up very often.  Definitely follow up with support.  This is a serious hardware error.  Make sure you emphasize the RGB drops to dark and fans stop in the support ticket.  That's the red flag this is not one of the usual run of the mill things and hopefully the agent won't run you around the block a few times with elementary steps.  

[Hoover1979]

I am thinking of replacing my cooler with this one: https://www.corsair.com/us/en/p/cpu-coolers/cw-9060069-ww/icue-h115i-elite-capellix-xt-liquid-cpu-cooler

At the moment, I went into BIOS and changed the PCIe from "Auto" to "GEN-3", to see if that helps.  It hasn't disconnected yet since making the BIOS change, but I am expecting the issue to continue until I replace the cooler after the Easter weekend.

[c-attack]

That is the most like for like replacement at the moment.  It will come with the Commander Core fan controller (AIO power goes through this - required) but also gives you PWM control for 6 fans and RGB control for 6 Corsair fans.  Also be aware Corsair is moving toward the interconnected "one wire" setup like most manufacturers.  This is the new CUE Link gear.  There is a CUE Link H115i AIO that is controlled through the very small CUE Link hub and gets power via a PCI-e 6 pin.  I am not recommended one or the other, but be aware that new CUE Link stuff only works with CUE Link gear and the older dual wire stuff like we've had for the past 5-6 years only works with Commanders, Lighting Node Core/Pros, etc.  If you have a case full of current Corsair LL, ML, QL fans, then the Capellix model is an easy transition and everything connects to the one controller.  If you are thinking you might update all your case fans to RGB shortly, then it may be worthwhile considering the CUE Link model as most new gear will be configured to run with it.  

[Hoover1979]

I opened Device Manager and expanded the "Universal Serial Bus controllers" There were 2 instances of "USB Root Hub (USB3.0)", 2 instances of "Generic USB Hub", and 2 instances of "AMD USB 3.1 eXtensible Host Controller - 1.10 (Microsoft)".

I started at the bottom of the list. I uninstalled the 1st instance of "USB Root Hub (USB 3.0)" and chose not to restart. I then uninstalled the 2nd instance of "USB Root Hub (USB 3.0)" and again, chose not to restart. For me, the "Generic USB Hub" instances were then gone. I then uninstalled one instance of "AMD USB 3.10 eXtensible Host Controller - 1.10 (Microsoft)", choosing not to restart. Finally, I uninstalled the last instance of "AMD USB 3.1 eXtensible Host Controller - 1.10 (Microsoft)", and chose to restart. Then I let Windows 11 reinstall the USB hubs and controllers.

After that, I went into BIOS and changed both PCIE settings from "Auto" to "GEN 3". I have all my USB ports (both USB 2 and USB 3) filled up now, and I am yet to see any peripherals or my cooler disconnect, and it's been 24 hours without issue, so far. If it keeps working after a week, or so I will mark this post as the solution. If the cooler disconnects again, it's time for a new cooler, but it's gone from 2-5 disconnects a day to no disconnects in 22 hours or so.

Here is a pic of what I uninstalled. I started at the bottom of the list and worked my way to the top, and chose to restart later, until you uninstall the last device, then restart.
https://pixeldrain.com/api/file/5tKiRRKa

 

Edited March 26 by Hoover1979

[Hoover1979]

It's been nearly 48 hours since I reinstalled the USB hubs/controllers via Device Manager, and set the PCIE from "auto" to "GEN 3", and my cooler hasn't disconnected once since, nor have my other USB peripherals.

Edited March 28 by Hoover1979

[c-attack]

Switching from PCI-e Gen 4 to Gen 3 was the old solution for the X570/B550 motherboards.  Those first generation PCI-e 4.0 boards along with the Intel Z690 counterpart had a tough time keeping steady USB voltage at the socket. Most users chose a powered USB hub to keep the devices connected and many needed one anyway to handle more than 2 connections and side stepping the need to set it to PCI-e gen 3.  

 

The bothersome aspect here is there is no way you went X number of years with this thing and then only now does it suddenly decide the USB connection and motherboard pathways are insufficient.  Maybe if you have been on a very old BIOS all this time and then recently updated.  This was the problem X470 owners ran into.  They were just fine for a year, but when new CPU compatible BIOS versions rolled out for the next generation, they got pulled into this as well.  Even then, the normal behavior for a cooler that looses it's USB connection is to drop into hardware mode -- that's HW Lighting and the fan curve you were using is automatically written to the device.  The fans should not stop.  Even if they did, it should take 5-10 min before the zero fan state could elevate your CPU temps to warning zone.   If the pump stops, then you have 30 seconds at most.  I am not sure which is the case here.  A possible but unintentional way the fans might stop when the AIO drops into HW Mode is if you have the fan curve set to a value it cannot obtain without the software -- like CPU temp.  They should be running from the coolant temp and that is a value the AIO can read even with no software installed.  Still, what should happen with a null value variable is the fans max out rather than stop, so I still have some concerns.  I suppose all you can do now is move forward and keep an eye on things.  

[Hoover1979]

I had USB issues for a year, and I don't flash BIOS myself (I worry the power might go out as I have bad wiring in my public housing apartment and don't want to brick my  MoBo, if I get a blackout, as often all the power cuts off for 5 seconds before coming back on).  The issues started with my Gigabyte Aorus X570 Pro WiFi (Rev. 1.0). The same issue continued with my Aorus B550 Elite AX v2 (Rev. 1.3), with the same behavior.  I also have to disable fTMP in BIOS or every single time Windows 11 tries to invoke TPM, the CPU completely freezes, and never recovers, forcing me to reset my PC by holding the power button down, until the system switches off, often having to use DISM to allow an elevated command prompt to repair the OS (whenever I hard reset, when I run "SFC /Scannow" it always says it found corruption but is unable to repair it, requiring me to use DISM first).  Since disabling fTPM, it takes far, far longer to boot up, sign out, shut down, or restart, but at least my system isn't freezing, often during startup.

Edited April 2 by Hoover1979

[Hoover1979]

The cooler started disconnecting again as of this morning (7:15am Australian Central Daylight-Savings Time).  I flashed my BIOS this morning, to see if I could use fTPM, but for that, no dice.  If I have TPM enabled, my CPU freezes whenever Windows tries to invoke it, forcing me to hard reset my PC.  Is it possible that disabling fTPM could be partly, or wholly respponsible for my H115i shutting off completely, and not coming back on, unless I either immedialely restart my system or restart the iCUE service in iCUE's options, before my CPU fries, as whenver it happens the fands grind to a complete halt.  I normally have the fans maxed, so the fans shutting off are noticeable.

 

These are the 2 events dictationg the fTPM error in Event Viewer, they always appear on top of each other, indicating 2 simultaneous fTPM errors :

 

EVENT 1:

SCEP Certificate enrollment initialization for WORKGROUP\DESKTOP-VRTDR64$ via https://AMD-KeyId-907d65e9b562315997dd5ad086b2b7598957b92c.microsoftaik.azure.net/templates/Aik/scep failed:

GetCACaps
GetCACaps: Not Found
{"Message":"The authority \"amd-keyid-907d65e9b562315997dd5ad086b2b7598957b92c.microsoftaik.azure.net\" does not exist."}
HTTP/1.1 404 Not Found
Date: Tue, 02 Apr 2024 20:33:05 GMT
Content-Length: 121
Content-Type: application/json; charset=utf-8
X-Content-Type-Options: nosniff
Strict-Transport-Security: max-age=31536000;includeSubDomains
x-ms-request-id: dd013d89-822e-45b7-b289-aed8c9bd74d4

Method: GET(250ms)
Stage: GetCACaps
Not found (404). 0x80190194 (-2145844844 HTTP_E_STATUS_NOT_FOUND)

 

EVENT 2:

SCEP Certificate enrollment initialization for Local system via https://AMD-KeyId-907d65e9b562315997dd5ad086b2b7598957b92c.microsoftaik.azure.net/templates/Aik/scep failed:

GetCACaps
GetCACaps: Not Found
{"Message":"The authority \"amd-keyid-907d65e9b562315997dd5ad086b2b7598957b92c.microsoftaik.azure.net\" does not exist."}
HTTP/1.1 404 Not Found
Date: Tue, 02 Apr 2024 20:33:05 GMT
Content-Length: 121
Content-Type: application/json; charset=utf-8
X-Content-Type-Options: nosniff
Strict-Transport-Security: max-age=31536000;includeSubDomains
x-ms-request-id: d42644e6-8adf-4347-9aab-2ab7d53f18e4

Method: GET(562ms)
Stage: GetCACaps
Not found (404). 0x80190194 (-2145844844 HTTP_E_STATUS_NOT_FOUND)

 

[Hoover1979]

I am trying uninstalling iCUE (selecting delete settings for all users), reinstalling iCUE, setting it up again, and restarting.

[DevBiker]

If you are in your warranty period, I'd start the RMA process.

I had a similar issue with the Pro series some time ago. Mine would, however, come back on. Never fixed it; had to RMA the cooler.

And I can't see how the TPM would have anything at all to do with the issue. I mean ... perhaps .. but I've run iCUE (and that cooler, among others) on systems with and without TPM ... no differences.

[Hoover1979]

I was referring to fTPM as a cause of my entire rig freezing, whenever Windows tried to invoke TPM.  Also, I found after disabling fTPM, it takes a lot longer for the system to boot up, as well as to sign out, shut down, and restart, but it's better than having the entire rig freeze, forcing a hard reset, potentially corrupting system files.  I doubt fTPM could play any part in hardware disconnecting.

My warranty has expired.  I got it in early 2019, so it's 5 years old.

[DevBiker]

Looking at your specs, you're running Windows 11, which requires a TPM (at least for installation) - which probably explains why the system is taking longer to boot. I don't think that your issues would be related to that though - iCUE doesn't require a TPM for any operations and that's not something that is accessed by the cooler at all (it can't).

I would be concerned about the freezing 'whenever it accesses the TPM' - but how do you know that it's from accessing the TPM? If, for example, you have Windows Hello or a login PIN set up, that will access the TPM. And the TPM gets cleared out with a BIOS update (ugh ... hate this) - I'm not sure if that's relevant or not. But if you've not updated your BIOS lately, perhaps look at that? I may be wrong but I don't think that any of this will have anything at all to do with the cooler. If the coolers had a problem with the TPM, considering the broad adoption of the TPM forced by Windows 11, we would most certainly have heard loud cries across the Interwebs. But ... there may be other software in the picture that does use the TPM and is causing the freeze.