Custom Board XLINKS

[Bianco]

The suffix A denotes the silicon revision. The 01 is the number of cores in the chip.
There is no real difference between an XS1-L01A and a XS1-L1 right now (only the first being more specific) since there is only one silicon revision shipped.

[MaxFlashrom]

Well, there's the problem. My datasheet is wrong. This is the one I have been using for ages:
http://www.xmos.com/system/files/xs1-l1 ... ete-me.pdf

This is most regrettable! The datasheet has been revised several times and it's noted that the labelling of links C and D was fixed in the revision history. It's helpful to subscribe on xmos.com to document update notifications.

This datasheet error was also responsible for XMOS' issues with their XTAG2 interfacing to the XK1, to enable their xscope implementation, which uses links.

See https://www.xcore.com/forum/viewtopic.p ... g&start=20

Max.

[huw]

This is indeed regrettable and apologies for the problem it has caused. The error was revised in version 2.2 of the datasheet issued in December 2011, since when there have been two further reissues of the datasheet.

We currently deliver all documents from a /published/* path which issues the latest version of a document. Previously, however, documents have been available from /systemfiles/* urls, which are now only used for images. We've instructed Google to remove all /system/files references from their indexes and blocked the indexing of such paths in future - it'll take a couple of days for this to take effect. I'll also make sure that our distributors all link to the correct path.

Going forward, please subscribe to any XMOS datasheets you're using so we can immediately inform you if there are any changes to the documents.

Huw

[MaxFlashrom]

This is indeed regrettable and apologies for the problem it has caused. The error was revised in version 2.2 of the datasheet issued in December 2011, since when there have been two further reissues of the datasheet.

Huw

I think that should be November 2010. The sheet naming was also changed to a new format.

Rev. X1154D–11/10
1. Updated pin table to show correct direction of links X0LC and X0LD.

Version 2.2, as it was labelled then, said:

2010-11-23 2.2 Direction of XMOS Links X0LC and X0LD (X0D49 to X0D70)
updated. Previous versions of datasheet showed the direction
inverted in/out and out/in Pin Table

Either that, or XMOS are really bringing us technology back from the future :)

Max.

[huw]

Thanks for correction, Max.

2010 it was.

[rp181]

Thank you Huw!

[rp181]

So am I right if I change to the following:

Code: Select all

X0D64: OXD1
X0D65: OXD0
X0D66: OXU0
X0D67: OXU1

From the old version of:

Code: Select all

X0D64: OXU1
X0D65: OXU0
X0D66: OXD0
X0D67: OXD1

[Bianco]

That should fix it!
You might want to consider adding series termination too while you're at it.

[MaxFlashrom]

That should fix it!
You might want to consider adding series termination too while you're at it.

Place series inline resistors as close as possible to the L1-128 for X0D17/LB0out pin 110, X0D16/LB1out pin 113 , X0D65/LD0out pin122, X0D64/LD1Out pin124
The resistor values are chosen depending on the Tx line characteristic. You can experiment, later if necessary. The Tx line impedance will depend on how thick you make your track and the distance to the ground plane on the layer below. You want the output impedance of the L1 driver+the series resistance to equal the Tx-Line impedance. 8 mil tracks give about 50Ohms Tx-Line impedance on a typical stack-up.

Keep the lines far enough apart to prevent crosstalk;these are not a differential pair.

Here's a nice calculator:
http://www.mantaro.com/resources/impeda ... _impedance

Use stripline calculator for embedded lines.

Note you only do the out lines as close as possible to the driving point, not the in lines.
Max.