We have made a design change to the T2 board to improve the timing of the TTC clock vs data. The extracted TTC clock is sent on the MicroTCA clock network (received on FCLKA on an AMC card), while the data is sent as DC-coupled LVDS on fabric B (port 3 on an AMC card). The data is simply un-encoded 80 MB/s serial data with the TTC A/B channels alternating.

The original idea was that the timing would be well enough aligned at the source that a simple DDR register could receive it without any skew adjustment. It turned on in the prototype AMC13 (v1 and XG) that there is a large delay range in the M-LVDS driver chips used to launch the clock on the backplane. This results in troublesome timing adjustment required on the receiving end to reliably capture the TTC signal.

We have redesigned the clock fanout hardware slightly as described in this sketch: TTC_Clock_Skew_Adjust_Scheme.pdf.

This new topology allows for an individual phase tune of the fabric B data to match the measured delay in each M-LVDS chip (the channel-to-channel delay is small).

We have just now (March 2014) received a run of prototype boards with the new scheme.

-- EricHazen - 11 Mar 2014

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r2 - 11 Mar 2014 - EricHazen
 
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback