The second production run of the PUPE boards was completed and the
boards have had basic functional testing performed. The TMS
pre-series system has been assembled and basic testing with the TMS
test signal generator has been performed. We have installed and
performed basic testing with the TMS system on the actual PS machine
at CERN as part of the pre-series installation visit.
The following work has been performed:
The second production run of the PUPE boards has been
The PUPE boards have been functionally tested to a basic
level. Full testing and ADC characterisation tests have yet to be
performed. There may be a production issue with the cPCI termination
The FPGA firmware has been ported to the PUPE board and has
been tested with the full TMS system software using the TMS test
signal generator. There are a few minor issues to be resolved.
The TMS software has been tested and further developed using
the PUPE board, FPGA Firmware and TMS test signal generator.
3.CERN Pre-Series Visit
We visited CERN on the 6/6/2007 through to 8/6/2007 to install the
TMS pre-series system and perform initial testing with the CERN PS
machine. The work done during this visit included:
The system was set-up and tested stand-alone with the TMS
test signal generator to make sure all components were OK after
transportation to CERN. All parts functioned fine.
With CERN's help, we installed the TMS rack components into
one of CERNS main racks and again performed tests with the TMS test
signal generator to make sure all components were OK. There was a
problem at this stage due to a loose bolt finding its way into the
power supply causing a warning light to be on. We have brought this
power supply back for testing.
We then demonstrated the TMS system's software to CERN's
staff and showed the system working with the TMS test signal
The system was then connected to the CERN PS machine using
the CYCLE_START, FREF, INJECTION and Sigma signals. The 10MHz
SYS_CLOCK was provided by the TMS test signal generator and the
CYCLE_STOP signal was supplied by software. Once the
pllInitialFrequency and frefPhaseDelay0 was setup to match the PS
machine the TMS PLL was seen to lock and the system was seen to
It was noticed that the PLL took some time to lock. One
reason for this was that the PLL was running at 62.5MHz rather than
125MHz as per CERN's original example code. Andrew updated the
firmware for this and we later worked with the 125MHz PLL.
It was noticed that the ADC data had some glitches on the
DeltaX and DeltaY inputs. Derek had a look at this and Andrew
modified the firmware to latch the ADC data on the opposite clock
edge to fix this issue. This needs further investigation.
It was noticed that the final data from the system looks a
little noisy. This needs some further investigation. The actual
analogue input signals were quite low. It may be better to set the
TMS system to use +-1 Volt peak-peak input signals rather than the
current +-2 Volt peak-peak setting.
Derek worked on the ADC PLL clock synchronisation chip during
the visit as well as looked at the ADC data latch timing issue..
We helped CERN get the TMS system running on their network
and accessible from their remote computers.
We gave basic training to CERN's staff on how to use and test
We had a meeting with some of CERN's software support staff
to help them with integrating the system into the CERN main control
The pre-series visit and installation went very well. We were able
to connect the system to the CERN PS and capture and process live
signals from the system. The diagnostics and test systems worked
well. There are a few issues too investigate and fix.
A full characterisation test on the PUPE ADC's needs to be
performed. Alpha Data and CERN will carry out this work. This is
probably best done using Derek's ADC test FPGA code.
We need to look at the noise in the final data to see where
this is coming from. This could be ADC capture noise or a problem in
the FPGA algorithm possibly the BLR algorithm.
The ADC clock's PLL needs the option to be able to be
synchronised with the 10MHz system clock. We need to setup the PLL
chip and test this functionality.
There are some issues with writing to some registers on the
PUPE board. This is causing the pllPhaseDelay register to not be set
occasionally and the diagnostics capture to occasionally fail.
We need to check the ADC data latch time in the FPGA to make
sure it is in phase with the ADC data.
There is some problem where the FPGA bit file will not run
after the first load from a cold boot.
There is some problem with the FPGA internal test data from
SDRAM system in the current firmware release.
The TmsControlGui test application could be improved to add
abilities to setup the State/Phase tables.
The TmsControlGui test application could be improved to store
data files in user defined locations.
We will provide the source code for the TMS API to CERN's
system software staff and assist them in porting this to their
LynxOS based platform.
Full series system work including testing with multiple PUPE
boards, Module Controllers and client applications.
The following TMS hardware has been delivered to CERN:
The pre-series system was delivered slightly behind the original
schedule. The detailed project schedule is at:
Although we are a little behind with shipping the pre-series
system we can deliver the full system on time assuming we get an
early agreement to full system production from CERN. However, we
believe it would be much better for both parties to spend more time
with the pre-series system performing testing prior to agreeing to
the final system production. This we believe would yield a better
system for CERN. We understand that there are no real project
pressures for the August 21st date. The main project
pressure is to have a functioning system prior to the end of year PS
shutdown leaving sufficient time for CERN to test the system on real
BEAM signals this year. A date of the 28th of September
was thought to be mutually acceptable. We need to confirm this.
The basic schedule information is:
Pre-Series Unit Complete
CERN Agrees to full system production
Full System Installation
September 28th ?