Directory Hierarchy:
├── optohybrid_registers.xml -- XML Address Table ├── Hog -- Hog build system submodule ├── doc -- documentation/images ├── tools -- Python scripts ├── src -- Source code, constraints, IP cores │ ├── optohybrid_top.vhd │ ├── cluster_building -- Cluster building │ ├── control -- Control │ ├── gbt -- GBTx Tx/Rx │ ├── ip_cores -- Virtex-6 IP Cores │ ├── ip_cores_a7 -- Artix-7 IP Cores │ ├── mgt -- Transceivers │ ├── pkg -- VHDL Packages │ ├── trigger -- S-bits │ ├── ucf -- Constraints │ ├── utils -- Shared/misc utility modules │ └── wb -- Wishbone ├── Top -- Hog top levels │ ├── oh11 │ └── oh21 └── README.org -- This readme
This firmware is using the HOG framework as a build system:
- HOG Documentation: http://hog-user-docs.web.cern.ch
- HOG Source Code: https://gitlab.cern.ch/hog/Hog
A Makefile is provided that should make firmware building more straightforward. You can run for example:
make update # update from the XML file
make oh21-200 # builds the 200t version of the ge21 firmware
make oh21-75 # builds the 75t version of the ge21 firmware
make oh21 -j2 # builds both versions of the firmware in parallel with 2 jobs
make oh11 # builds long and short versions of the GE11 firmware
make all -j8 # update from the XML file and build all firmware flavors, do it all in parallel :)n.b. Support for GE1/1 through a planAhead wrapper is in development (but still experimental)
Alternatively, you can build the firmware manually, for example, the firmware can be built as:
git clone https://github.com/cms-gem-daq-project/OptoHybridv3
cd OptoHybridv3
git submodule update --init --recursive
Hog/CreateProject.sh oh21
Hog/LaunchSynthesis.sh oh21
Hog/LaunchImplementation.sh oh21The firmware can be updated from the address table by running make update from the root of the repository.
If you want to increment the version, just run make with an argument, e.g.
make update version=03.04.01
Releases are labeled as version_major.version_minor.version_patch.hardware_id
i.e. in HARDWARE_VERSION XY
x=chamber type (0=GE11 short, 1=GE11 long, 2=GE21) y=hardware generation
Specifically:
| Station | OH Version | Chamber Variant | Tag |
|---|---|---|---|
| GE1/1 | v3B | Short | 0B |
| GE1/1 | v3C | Short | 0C |
| GE1/1 | v3C | Long | 1C |
| GE2/1 | v1 | 2A | |
| GE2/1 | v2 | 2B |
TMR for different modules can be configured on and off to allow for customization between GE2/1 and GE1/1.
It is controlled by the platform specific hardware package, viz. hardware_pkg_ge11.vhd and hardware_pkg_ge21.vhd
For example:
constant EN_TMR : integer := 1;
constant EN_TMR_TRIG_FORMATTER : integer := 1*EN_TMR;
constant EN_TMR_GBT_DRU : integer := 1*EN_TMR;
constant EN_TMR_SBIT_DRU : integer := 1*EN_TMR;
constant EN_TMR_SOT_DRU : integer := 1*EN_TMR;
constant EN_TMR_CLUSTER_PACKER : integer := 0*EN_TMR;You can change a 1 to a 0 to to disable TMR for that module, or EN_TMR from 1 to 0 to disable TMR entirely.
| Optohybrid v3 LED Assignments | |
|---|---|
| led[15] = (rsvrd, always on) | led[7] = Cluster Rate >= 10,000,000 Hz |
| led[14] = Clock, divided | led[6] = Cluster Rate >= 1,000,000 Hz |
| led[13] = MGTS ready, blinking | led[5] = Cluster Rate >= 100,000 Hz |
| led[12] = GBT Ready, blinking | led[4] = Cluster Rate >= 10,000 Hz |
| led[11] = GBT Request flash | led[3] = Cluster Rate >= 1,000 Hz |
| led[10] = L1A Flash | led[2] = Cluster Rate >= 100 Hz |
| led[9] = Resync Flash | led[1] = Cluster Rate >= 10 Hz |
| led[8] = BC0 Flash | led[0] = Cluster Rate >= 1 Hz |
The “right side” LED indicators provide a logarithmic progress bar indicator which provides a convenient way to monitor the rate of incoming clusters received by the Optohybrid.
Prior to the receipt of the first S-bit (after a reset or resync), the progress bar will show a strobing “cylon” pattern to indicate an idle state.
The Optohybrid has an HDMI connector connected to differential pairs of the FPGA.
The functionality is programmable by changing firmware, but in the current configuration they provide a programmable way to output S-bits for use in the cosmic ray test stand.
Each conductor can be individually programmed into one of 4 modes:
| Mode | Description |
|---|---|
| Mode 0 | Each signal was a single VFAT (set by selN) |
| Mode 1 | Each signal was the OR of three VFATs in an ieta row (row is set by selN) |
| Mode 2 | Each signal was the OR of four VFATs in an iphi half column |
| (e.g. 0-3, 4-7, 8-11, 12 15, 16-19, 20-23) (phi half is set by selN) | |
| Mode 3 | Disabled (all outputs driven to zero) |
The registers you will want to use to control the HDMI output are:
- GEM_AMC.OH.OH{X}.FPGA.CONTROL.HDMI.SBIT_SEL{0-7}
- GEM_AMC.OH.OH{X}.FPGA.CONTROL.HDMI.SBIT_MODE{0-7}
The assignments of these signals in the HDMI cable are:
NET "ext_sbits_o[0]" LOC = H17; # tmds clk p
NET "ext_sbits_o[1]" LOC = G17; # tmds clk n
NET "ext_sbits_o[2]" LOC = J16; # tmds d2 p
NET "ext_sbits_o[3]" LOC = J17; # tmds d2 n
NET "ext_sbits_o[4]" LOC = L14; # tmds d1 p
NET "ext_sbits_o[5]" LOC = L15; # tmds d1 n
NET "ext_sbits_o[6]" LOC = M17; # tmds d0 p
NET "ext_sbits_o[7]" LOC = M18; # tmds d0 nIn the version 3 firmware, all Optohybrid registers are accessible through a Wishbone bus with an address table defined in single XML file, optohybrid_registers.xml (accessible in Github at ./optohybrid_registers.xml):
This file can be used in the same way as the backend (CTP-7) registers file and with the same tools, e.g. rw_reg.py. The contents of this file should be embedded into the CTP-7 firmware at a sub-address which is allocated to the Optohybrid.
16 bits are allocated for the Optohybrid address space. Within this 16-bit address space, the bits are subdivided according to the scheme:
– highest 4 are used as the module ID (wishbone slave #)
– lowest 12 bits are used by the wishbone splitters as individual register addresses
Thus we can have up to 16 slaves, and 4096 endpoints per slave. Presently we use only 6 slaves with a firmware that is nearly finalized, so it is expected that 16 slaves will not be a limiting factor.
The Optohybrid has a local accumulator which increments bunch crossing number (bxn) and produces a locally generated bx0 flag.
In order for the two flags, (1) locally generated and (2) received from GBTx, to be in sync a programmable parameter (bxn offset) is provided which allows you to set an “offset” which is the value which will be assumed by the bxn counter after ttc resync.
The synchronization of these two flags can be monitored by checking the status of bxn_sync_error in the OH status register (n.b. that bx0_sync_error is NOT suitable for this, as it is a 1bx wide pulse useful for internal counters when the bx0 is received, while bxn_sync_error will persist for the entire orbit).
When the Optohybrid is correctly timed in to the TTC system, bxn_sync_error should be 0. A software routine in the CTP-7 to find the correct bxn offset should be easily designed.
A counter in the counters module counts the number of bx0 sync errors since the last resync or hard-reset.
The anticipated synchronization process is:
- From the backend electronics, send a BC0 is to all Optohybrids
- Inside of the Optohybrid, adjust the BXN offset until bxn_sync_error is read to be 0
- This indicates that the local OH bxn counter matches the remote counter
- Inside of the backend electronics, adjust the delays of incoming trigger data until the 0x50 (BC0) frame markers are aligned between all Optohybrids