|
|
|
# FAQ
|
|
|
|
|
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
|
|
## I start looking a bit more in detail to the possible association of the FMCADC100M14b4cha with a SPEC carrier, and when I look at the connector used on this FMC, I see that it is a HPC (High Pin-Count connector) whereas on the SPEC there is a LPC (Low Pin-Count connector)?
|
|
|
|
|
|
|
|
The FMCADC100M14b4cha actually has a HPC connector mounted (indeed that
|
|
|
|
is in the BOM). However, none of the HPC signals is used in any current
|
|
|
|
applications, so actually an LPC could have been mounted instead.
|
|
|
|
|
|
|
|
The clock for the ADC is generated on the mezzanine card itself (by a
|
|
|
|
programmable Si570 oscillator) and this clock is output from the
|
|
|
|
mezzanine to the carrier on an LPC pin. We had foreseen that some
|
|
|
|
applications may need a clock coming from the carrier going into the
|
|
|
|
mezzanine. In this case one must use a HPC connector as in the FMC
|
|
|
|
standard there are no dedicated clock pins in that direction on an LPC
|
|
|
|
connector (we believe this is a weak point in the FMC standard). In this
|
|
|
|
case one has to remove two capacitors (C22, C23) and move them to the
|
|
|
|
locations C24 and C25 (see [schematics
|
|
|
|
V5-0](https://edms.cern.ch/file/1165457/1/EDA-02063-V5-0_sch.pdf), page
|
|
|
|
5). We have never used this and commercial FMCADC100M14b4cha mezzanine
|
|
|
|
are all sold with the clock generated on the FMC mezzanine. Furthermore
|
|
|
|
indeed the SPEC carrier cannot generate a clock to the mezzanine as it
|
|
|
|
uses an LPC connector
|
|
|
|
|
|
|
|
For [White Rabbit](https://www.ohwr.org/project/white-rabbit/wiki),
|
|
|
|
where we actually do generate a clock on the carrier and need to sample
|
|
|
|
to that very precise clock, we actually program the Si570 oscillator
|
|
|
|
(see [schematics
|
|
|
|
V5-0](https://edms.cern.ch/file/1165457/1/EDA-02063-V5-0_sch.pdf), page
|
|
|
|
4) so that it follows and locks to the White Rabbit clock on the
|
|
|
|
carrier. We started in October 2012 the [ADC FMC board clock
|
|
|
|
synchronization
|
|
|
|
mechanism](https://www.ohwr.org/project/fmc-adc-100m14b4cha/wikis/Si570-clk-sync-funcspec)
|
|
|
|
task that will do precisely this. There will some code running in the
|
|
|
|
Xilinx that will basically implement a Phase-locked loop that controls
|
|
|
|
continuously the Si570 to track the White Rabbit clock.
|
|
|
|
|
|
|
|
The above should also explain the somewhat cryptic text in the
|
|
|
|
specification: *"FMC to carrier interface FMC high pin count connector
|
|
|
|
(HPC only used if external clock is
|
|
|
|
selected)"*
|
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
|
|
## I supposed that the external clock was entering the FMC directly with a Lemo00 connector?
|
|
|
|
|
|
|
|
That is wrong. It is not foreseen to have an external clock input on the
|
|
|
|
mezzanine card. There is only place for five Lemo00 connectors and these
|
|
|
|
are used as four signal inputs and one external trigger input.
|
|
|
|
|
|
|
|
-----
|
|
|
|
|
|
|
|
Erik van der Bij - October 2012
|
|
|
|
|