• State TI Thinks Resolved
  • Locked Locked
  • Replies 8 replies
  • Answers 1 answer
  • Subscribers 34 subscribers
  • Views 1257 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

XIO2001: XIO2001 not detected by PEX8732 (PCIe Gen3)

Niels Koll
Expert 1950 points

Part Number: XIO2001
Other Parts Discussed in Thread: DS80PCI402

Hello,

we have a problem with the XIO2001 not detected by a PEX8732 Gen3 PCIe Switch.
We have two different board designs that both show the same behaviour.
We found that the XIO2001 continuously generates PCI resets (approx. every 60ms) in this case.

Other PCIe-to-PCI bridges, like Pericom PI7C9X111SL work fine with the PEX8732.

With other (Gen2) PCIe-Switches, our XIO2001 based designs work fine.

Are you aware of any compatibility issues between XIO2001 and PEX8732?

Are there any known issues with the XIO2001 and Gen3 link partners?

Any idea how to get the XIO2001 working with the PEX8732?

Looking forward to your answers,
Niels

  • 0
    TI__Guru* 93570 points

    Hi Niels,

    I checked XIO2001 known errata. The only thing that could be related to what you are observing perhaps is CLKREQ# not functioning properly in Gen3 power managed states. The thought is that CLKREQ# could be causing this continuous PERST# generation. Maybe the workaround for this could take care of the issue you are seeing. 

    Regards,, Nasser

  • 0 in reply to Nasser Mohammadi
    Expert 1950 points

    Hello Nasser,

    unfortunately, it is the PCI-Reset# which is continiously generated by the XIO2001.
    (PCI, this old, traditional, parallel interface)

    The PCI-Express Reset Signal PERST# behaves as expected. A short "low" period during power-up, and then it transitions to a stable "high".

    Here is a closer view on the pulsing PCI-Reset#:

    PCIe-REFCLK and PCI-Clk are both running with correct frequencies, even during PCI-Reset# transitions:

    Do you still think that
    "Set the FORCE_CLKREQ bit to disable CLKREQ support on the pci-express bus."
    could be the solution?

    By the way, we have an EEPROM connected to the XIO2001, but it is empty.
    Does TI provide any tools to program the EEPROM and a recommended EEPROM content?

    Regards, Niels

  • 0 in reply to Niels Koll
    TI__Guru* 93570 points

    Fullscreen 8240104C.txt Download 
    'WinROM 3 EEPROM Data File
'DEVICE ID: 8240104C	SILICON REV: 00
'Created 1/15/2009 4:38:21 PM
000	00	00000000
001	25	00100101
002	00	00000000
003	00	00000000
004	00	00000000
005	00	00000000
006	5F	01011111
007	02	00000010
008	00	00000000
009	86	10000110
00A	00	00000000
00B	00	00000000
00C	00	00000000
00D	40	01000000
00E	00	00000000
00F	01	00000001
010	00	00000000
011	00	00000000
012	00	00000000
013	08	00001000
014	01	00000001
015	12	00010010
016	00	00000000
017	80	10000000
018	20	00100000
019	14	00010100
01A	32	00110010
01B	00	00000000
01C	00	00000000
01D	00	00000000
01E	00	00000000
01F	00	00000000
020	43	01000011
021	04	00000100
022	08	00001000
023	7F	01111111
024	00	00000000
025	C0	11000000
026	00	00000000
027	80	10000000
028	FF	11111111
029	FF	11111111
02A	FF	11111111
02B	FF	11111111
02C	FF	11111111
02D	FF	11111111
02E	FF	11111111
02F	FF	11111111
030	FF	11111111
031	FF	11111111
032	FF	11111111
033	FF	11111111
034	FF	11111111
035	FF	11111111
036	FF	11111111
037	FF	11111111
038	FF	11111111
039	FF	11111111
03A	FF	11111111
03B	FF	11111111
03C	FF	11111111
03D	FF	11111111
03E	FF	11111111
03F	FF	11111111
040	FF	11111111
041	FF	11111111
042	FF	11111111
043	FF	11111111
044	FF	11111111
045	FF	11111111
046	FF	11111111
047	FF	11111111
048	FF	11111111
049	FF	11111111
04A	FF	11111111
04B	FF	11111111
04C	FF	11111111
04D	FF	11111111
04E	FF	11111111
04F	FF	11111111
050	FF	11111111
051	FF	11111111
052	FF	11111111
053	FF	11111111
054	FF	11111111
055	FF	11111111
056	FF	11111111
057	FF	11111111
058	FF	11111111
059	FF	11111111
05A	FF	11111111
05B	FF	11111111
05C	FF	11111111
05D	FF	11111111
05E	FF	11111111
05F	FF	11111111
060	FF	11111111
061	FF	11111111
062	FF	11111111
063	FF	11111111
064	FF	11111111
065	FF	11111111
066	FF	11111111
067	FF	11111111
068	FF	11111111
069	FF	11111111
06A	FF	11111111
06B	FF	11111111
06C	FF	11111111
06D	FF	11111111
06E	FF	11111111
06F	FF	11111111
070	FF	11111111
071	FF	11111111
072	FF	11111111
073	FF	11111111
074	FF	11111111
075	FF	11111111
076	FF	11111111
077	FF	11111111
078	FF	11111111
079	FF	11111111
07A	FF	11111111
07B	FF	11111111
07C	FF	11111111
07D	FF	11111111
07E	FF	11111111
07F	FF	11111111
080	FF	11111111
081	FF	11111111
082	FF	11111111
083	FF	11111111
084	FF	11111111
085	FF	11111111
086	FF	11111111
087	FF	11111111
088	FF	11111111
089	FF	11111111
08A	FF	11111111
08B	FF	11111111
08C	FF	11111111
08D	FF	11111111
08E	FF	11111111
08F	FF	11111111
090	FF	11111111
091	FF	11111111
092	FF	11111111
093	FF	11111111
094	FF	11111111
095	FF	11111111
096	FF	11111111
097	FF	11111111
098	FF	11111111
099	FF	11111111
09A	FF	11111111
09B	FF	11111111
09C	FF	11111111
09D	FF	11111111
09E	FF	11111111
09F	FF	11111111
0A0	FF	11111111
0A1	FF	11111111
0A2	FF	11111111
0A3	FF	11111111
0A4	FF	11111111
0A5	FF	11111111
0A6	FF	11111111
0A7	FF	11111111
0A8	FF	11111111
0A9	FF	11111111
0AA	FF	11111111
0AB	FF	11111111
0AC	FF	11111111
0AD	FF	11111111
0AE	FF	11111111
0AF	FF	11111111
0B0	FF	11111111
0B1	FF	11111111
0B2	FF	11111111
0B3	FF	11111111
0B4	FF	11111111
0B5	FF	11111111
0B6	FF	11111111
0B7	FF	11111111
0B8	FF	11111111
0B9	FF	11111111
0BA	FF	11111111
0BB	FF	11111111
0BC	FF	11111111
0BD	FF	11111111
0BE	FF	11111111
0BF	FF	11111111
0C0	FF	11111111
0C1	FF	11111111
0C2	FF	11111111
0C3	FF	11111111
0C4	FF	11111111
0C5	FF	11111111
0C6	FF	11111111
0C7	FF	11111111
0C8	FF	11111111
0C9	FF	11111111
0CA	FF	11111111
0CB	FF	11111111
0CC	FF	11111111
0CD	FF	11111111
0CE	FF	11111111
0CF	FF	11111111
0D0	FF	11111111
0D1	FF	11111111
0D2	FF	11111111
0D3	FF	11111111
0D4	FF	11111111
0D5	FF	11111111
0D6	FF	11111111
0D7	FF	11111111
0D8	FF	11111111
0D9	FF	11111111
0DA	FF	11111111
0DB	FF	11111111
0DC	FF	11111111
0DD	FF	11111111
0DE	FF	11111111
0DF	FF	11111111
0E0	FF	11111111
0E1	FF	11111111
0E2	FF	11111111
0E3	FF	11111111
0E4	FF	11111111
0E5	FF	11111111
0E6	FF	11111111
0E7	FF	11111111
0E8	FF	11111111
0E9	FF	11111111
0EA	FF	11111111
0EB	FF	11111111
0EC	FF	11111111
0ED	FF	11111111
0EE	FF	11111111
0EF	FF	11111111
0F0	FF	11111111
0F1	FF	11111111
0F2	FF	11111111
0F3	FF	11111111
0F4	FF	11111111
0F5	FF	11111111
0F6	FF	11111111
0F7	FF	11111111
0F8	FF	11111111
0F9	FF	11111111
0FA	FF	11111111
0FB	FF	11111111
0FC	FF	11111111
0FD	FF	11111111
0FE	FF	11111111
0FF	FF	11111111

'DATA FILE FORMAT: <HEX ADDRESS><TAB><HEX DATA>. EVERYTHING AFTER <DATA> IS IGNORED.
'USE <'> FOR IN-LINE COMMENTING.
'DATA FILE PARSING STOPS AT FIRST BLANK LINE.
    Hi Neil,

    1). You can try this - set the Force_CLKREQ bit. I haven't been able to find an errata pointing to the issue you are seeing and it is worth a shot.

    2). Also, I am attaching EEPROM file or content to this email. Perhaps the initialization is different and it may help us here. Please note extension of the file is WR3 but since i was not able to attached to this case with WR3 extension i had to change this to txt.

    Regards,, Nasser

  • 0 in reply to Nasser Mohammadi
    Expert 1950 points

    Hello Nasser,

    thanks for posting the EEPROM content.
    We programmed our XIO2001 with your content, but the error stayed unchanged.

    As an additional test, we interposed a Cabled-PCIe connection between XIO2001 and PEX8732, and everything was fine.
    Sounds like an signal integrity problem.
    The cabled-PCIe "interposer" consisted of redrivers only. The signal chain during this test was as follows:
    PEX8732 ---> DS80PCI402 ---> PI2EQX5804 ---> DS80PCI402 --> XIO2001

    Any idea why interposing a redriver changes the behaviour?

    Regards, Niels

  • 0 in reply to Niels Koll
    TI__Guru* 93570 points

    Hi Niels,

    Yes this seems to be a signal integrity issue. 

    1). Do you see the problem if you just take out DS80PCI402 before the XIO part and leave the other one? 

    2). DS80PCI402: Are you using evaluation board for this part or this is part of the interposer? 

    Regards,,, Nasser

  • 0 in reply to Nasser Mohammadi
    Expert 1950 points

    Hi Nasser,

    as "interposer", we used a cabled-PCIe extension. It consists of:

    • TPCE278
    • OSS-XMC-HIB25-X8
    • OSS-PCIe-CBL-x8-xM
    • OSS-PCIe-ECA-x8

    Due to the different form factors of the boards, I am not able to "snip out" one of the redrivers.

    As you can see fron the listing above, the DS80PCI402 is located on two different boards of different vendors.

    Are there any options to "tune" the XIO2001's receiver equaliser or anything similar?

    Regards, Niels

  • 0 in reply to Niels Koll
    TI__Guru* 93570 points

    Hi Niels,

    Checked and there is not a register for signal conditioning. One thought crossed my mind, is it possible to force certain preset on this switch ? I am wondering there could be a setting that would open up the eye more - to some extend do the job of equalization but on the TX side.

    Regards,, Nasser

  • 0 in reply to Nasser Mohammadi
    Expert 1950 points

    Nasser,

    tuning the switch's transmitter is an alluring ude. But the board holding the switch is not from us, so I can't say if it is possible to tune the transmitter or not. I will try to find out on monday.

    Regards, Niels