Embedded Intel Atom® Processors
Technological Conversations about Intel Atom® Hardware, Software, Firmware, Graphics
1149 Discussions

Atom Z520 hangs when accessing EHCI

idata
Employee
1,993 Views

I'm working on Atom Z520 & US15W.

Board hangs immediately when EHCI PCI configuration access by BIOS

I've checked COMP resistor which is fine.

Anyone has any idea what is causing it ?

0 Kudos
11 Replies
Natalie_Z_Intel
Employee
548 Views

Hello, Naresh! We are still working on your question but here is the feedback so far from an engineer:

"I do not see any known issues with this anomaly, however, what is provided by the customer is not enough info to help debug it. It is probably a BIOS issue or could be a board design issue. I will check with another technical support team to see if they have encountered this before."

0 Kudos
idata
Employee
548 Views

Hi,

Among 4 boards we developed for prototyping. one board is absolutely booting fine (EHCI PCI Configuration space access is success).

Remaining 3 boards hangs instantly when PCI configuration space of EHCI is accessed.

If bios is configured to ignore EHCI device completely & proceed ahead with further initialization.

Boot to OS is successful but USB doesn't works at all.

In that way I felt that this issue is not due to BIOS (Software). Its due to hardware.

I suspected it to be with chipset US15W . so in that board I've replaced the chipset. & tested.

Same EHCI issue is found after US15W replacement.

Now I feel that neither BIOS nor Chipset are faulty. It should be among other things connected to Chipset like Power supply noise or clocks etc.

So I verified

1. Clocks going to chipset & found it absolutely fine as in expected (133MHz, 14.318MHz)

2. USB rcomp resister value was verified & was as per reference schematic.

3. Chipset Power line noise was measured & found it to be 20mV peak to peak max which is acceptable.

4. RTC SUSCLK is fine.

Now I'm confused with what to suspect for this issue.

Please advice what additional debug info you need for the problem

0 Kudos
Natalie_Z_Intel
Employee
548 Views

Thank you for the added information, Naresh. Upon speaking with the engineer he said, "I would agree that if one board boots it is probably not a BIOS issue. This is not a known issue on the chipset so it would lead me to believe it is a board/timing issue." Naresh, can you verify the rest and power sequencing on your platforms?

0 Kudos
idata
Employee
548 Views

Hi,

1. Power up sequencing is completely controlled by Embedded Controller & its common for all board.

2. The power up sequencing as mentioned in SCH EDS is verified.

State of each & every net of US15W which are involved in power up sequence till CPU reset happens are as per SCH EDS.

3. As I mentioned before only USB EHCI is not working. And skipping accesses to EHCI & going for further initialization, boots to OS is happening without USB support.

4. All PCI devices except EHCI are very well accessible & configurable as well.

5. USB port are utilized on board for only full speed devices. No high speed usb device is connected to US15W on board.

Should I suspect USB devices connected to US15W on board ???

I assumed that USB EHCI functionally will not be active till it is initialized by bios.

Is there any chance that USB EHCI hang is due to usb full speed devices connected on board??

Please advice...

0 Kudos
idata
Employee
548 Views

Also,

All the pins are at 0V ( D+, D-) of all 8 ports though full speed device is connected.

Where as in the board that is working, some ports measure 3.08V at D+ line.

Can this have any relation with the issue ?

0 Kudos
idata
Employee
548 Views

Hi,

I've checked the board with all USB port left disconnected.( i.e., no USB device connected to US15W )

Still the board state is the same.

0 Kudos
Natalie_Z_Intel
Employee
548 Views

Hello, Naresh!

Another engineer reviewed this thread and said that since you mentioned that 1/4 of the prototype boards do not have this issue, he says that it is a really rare chance that it is caused by software/BIOS. He suggests that you double check the USB part design and layout to see if it meets the requirements of the PDG. After that check the components and trace around the chipset as it could be caused by some signal integrity issue. At this time we are not aware of any known issues on the US15W chipset, but we will continue to investigate this.

Please try these things and let us know the results. Thanks! LynnZ.

0 Kudos
idata
Employee
548 Views

Hi Lynnz,

1. USB parts contains USBD/P 0 to 7 port. Comparing working & Non working board shows that

Working board measures resistance of 2M ohm between USBD/N to ground.

Non working boards measure resistance of 0.3M ohm between USBD/N to ground.

2. USB_OC 0 to 7 resistance to ground is identical in both the boards,

3. USBRBIASN/P resistance to ground is identical in both the boards,

4. All voltages ( 3.3V, 1.5V ) going to USB section, resistance to ground is identical in both the boards,

5. For USB part, pull up resistor for USB_OC & RBIAS resistance to ground is used which is identical in both the boards.

6. Signal Integrity comes into picture only when we are transferring data on usb port but it always reads 0V in CRO for not working board for all ports.

7. Can you please tell me why the D+/D- lines measure low resistance to Ground. Which power line it is using to drive D+/D- lines.

8. In the beginning thinking that it is Chipset issue we had replaced US15W. After replacement, same USB EHCI issue was found. But when I examined the Removed chipset, the resistance between USBP0 pin & nearby VSS pin measure 0.3M ohm. Is the Resistance between USB P/N to GND, 0.3M ohm normal ?

Note: Resistance measure is in complete off condition of board.

0 Kudos
WILLIAM_C_Intel
Employee
548 Views

Hello, Naresh

I carefully read it again, the crux of the issue is the Resistance between USB P/N to GND.

You have a board which is absolutely booting fine, could you please measure twice more?

1. The resistance between USB P/N to GND of this "successful" chipset.

2. The resistance between USB P/N to GND of an absolutely new chipset.

William

0 Kudos
idata
Employee
548 Views

Hi All,

I've examined all the extra chipsets I have & found that the resistance to GND is ~0.4M ohm. ~2M ohm

But when I check impedance on prototype made by another team a few years back, there it measures ~2M ohm on all the board.

I feel that, after mounting chipset, something is causing the resistance of USB D/N line to gnd to go down to ~0.3M ohm

0 Kudos
WILLIAM_C_Intel
Employee
548 Views

Hello, Naresh

 

May I have a question? Did you take off the chipset inwhich board is absolutely booting fine, then please measure the Resistance between USB P/N to GND.

(Note: Resistance measure is in complete off condition of board.)

 

William

 

0 Kudos
Reply