Solved PCI USB expansion card causes CMOS error

[seankurth]

The USB expansion card is a Via Vectro vt6202 that I scavenged from an old build whose motherboard fried when the cooling fan suddenly stopped working (it was an OLD build, CPU couldn't shut off automatically when the temp got too high, and I know most boards have been able to do that for about 10 years).

It was tested working in another PC ( emachines T3124, its older than my little sister but it still works so I use it to test stuff, I upgraded it from 512mb to 768mb of RAM, which can comfortably run Windows 7 with aero turned off, although I do dual boot with Android x86 4.4 because why not), so I know the card isn't broken. The PC I'm trying to install it in is a Lenovo Thinkserver TS140 , which I upgraded to 8gb RAM (ECC of course, I know I don't need enterprise grade stuff but I hate blue screens and corrupted files) and put a 1TB Seagate Barracuda in.

When I put the USB expansion card into the PCI slot, and press the power button, it gives me a CMOS error (2 beeps), which means the card is incompatible with the BIOS. This doesn't make sense to me because PCI-e (the Lenovo) is supposed to be software backwards compatible with PCI (the expansion card). How do I get it to work?

EDIT: Its a full-size pcie, so its not like I'm trying to stick a big card into a tiny hole.

 

[TVeblen]

Ah... you've transported us all back to the bad old days of the IRQ error (Interrupt ReQuest).

At that time there were only 15 IRQs and it was quite common to need to manually change a new device's IRQ position so that it did not conflict with another device, freezing the system.

Modern PCs use the APIC system which greatly expands the number of IRQs and reduces the possibility of conflicts.
But this system relies on Plug-and-Play devices to automatically configure the IRQs. That is why the vast majority of people reading this will have never heard of it.

Can you get into your BIOS settings with this card installed?

You could try the card in a different PCI slot if you have one. That would sometimes change the IRQ in older systems.

You might also try doing a ClearCMOS procedure with the card plugged in. That may force the system to reassign IRQs for all the P&P devices and free up whatever IRQ that old card insists it has to have.

 

[seankurth]

I did what you suggested, and it didn't work. But I think I did find the problem: The computer has 8 usb ports (2 front 6 back), and I'm using a USB hub that expands the total to 14.

This might not seem like a problem at first. And it isn't, since USB ports only cause IRQ errors if a device that can issue IRQs and is recognizable by the BIOS is plugged in (or if there are more USB controllers than IRQs, but that's basically impossible on any board small enough to fit in a case). But, most of my devices are recognizable by the bios, and even more can issue IRQs.

Devices that can issue IRQs:

1.) Speaker
2.) Webcam
3.) Canon Lide 30 scanner
4.) HP Deskjet D1520

Recognizable by BIOS:

5.) Mouse
6.) Keyboard
7.) Seagate 500GB external drive
8.) SD card reader (yes, I put a microSD card in and the BIOS listed it under "Removable" in the boot menu)
9.) LG external DVD drive

Other IRQ devices:
10-11 .) DIMM slots (2 of the 4 are empty, so I'm ignoring them)
12.) SATA port 1 Seagate 1TB HDD
13.) SATA port 3 disc drive

If you include the integrated graphics, system bus, ACPI, clock, and all the empty expansion slots (mini pcie and one full size x16), that brings the grand total to 15, the maximum. When I inserted that USB expansion card into the full size slot, I ended up with 16 IRQ devices, which of course broke the motherboard's circuit and caused a CMOS error.

To test this theory, I started unplugging things. As soon as I unplugged the external HDD and dvd drive, the board POSTed just fine with the old usb expansion card inserted. Once Windows was up, I plugged them back in. Sure enough, instant blue screen, it even said IRQ error. I did a CMOS battery reseat just to see if it would fix things (reassign, like you said), and nope. Still gave me those 2 beeps until I unplugged something.

I actually hadn't thought of this until you suggested it, because I thought the same thing as you: That modern PCs can issue more IRQs and even if they couldn't, it didn't matter because PCIe is capable of IRQ sharing. Hence, the last time I worried about that was almost a decade ago. I guess server boards must be an exception.

 

[TVeblen]

Wow, nice work! Interesting results. I am as surprised as you.

I might check in the BIOS settings to see if there are any entries for ACPI. Maybe that board defaults to a legacy mode, or it got switched at some point?