On 09/22/2014 02:20 PM, Slichter, Daniel H. wrote:
>> The receive parallel clock is reconstructed from the received data
>> and the latency is constant once locked.
> 
> Yes, but if this timing relationship is different each time the
> system is rebooted (I have seen on datasheets that for some of the
> gigabit systems e.g TI TLK2711A or TLK2521, it can vary from chip to
> chip and relock to relock, up to ~half a parallel UI) then you will
> have different skew across deserializer chips from power-up to
> power-up.

This is not a new problem. In The Old System there is a variable
alignment of SYNCLK to the other signals from reset to reset. To get
accurate timing with the current hardware you will have to add a return
channel for SYNCLK anyway.

> The other issue is that any SerDes-type solution means we lose the
> advantages of fine timing adjustment provided by the RTIO core, which
> would let us perform deskewing of TTL pulses, for example, via some
> calibrations saved in the parameter list.  A compromise might be that
> we could have a few direct channels to the RTIO core (in/out) over a
> reduced-conductor parallel cable for signals (like FUD, DDS SNYC_CLK,
> PMT counts) where we might care about timing to a high degree, and
> then send the rest (DDS programming, most TTL outputs) over a high
> speed serial link where we don't care about the timing resolution as
> much.

Yes.

Robert.
_______________________________________________
ARTIQ mailing list
https://ssl.serverraum.org/lists/listinfo/artiq

Reply via email to