|
DCT/DCZ Latency Test Feb/24-27/2004
Scope tests for TSF latencies. A single clock-4 tick pattern is loaded
into TSF input memories and setting all wires on. Channel-1 is connected
to the high-neighbor-out connector on the TSFi as the trigger, while
channel-2 is connecting to the BLT/PTD/ZPD output on TSFi. BLT/PTD
outputs are looked at from the TSFi output plug themselves, while the
ZPD output is spied from the resistor leg just before entering into the
TSFi LVDS channel-link. The TSF address pattern 0xffff has LUT wt=1,
stretch=3 and delay=4. For the old TSF test, the weight was poked
to 3 to force the output for PTD.
1. Old TSF -> PTD test
2. new TSF -> ZPD test
3. Old TSF -> BLT test
4. new TSF -> BLT test
5. Old TSF + BLT chain test
Scope test of TSF-BLT chain latency. Similar to TSF latency test
above with X2,X3,Y1 simultaneousply play the same single tick all
FFFF wire pattern. Some other neighboring BLTi TSF input sockets
are also connected to live new TSFi/TSF (open sockets have all TSF
signal constantly on). Still triggering on Channel-1 TSF hi-neighbor
out data. Channel-2 is connected to the BLT A output plug for GLT.
6. Old vs new TSF GLINK
Scope test comparing the old nad new TSF latency at the GLINK section.
Running DCH-DCT calibration, spy on exactly the same pins of the
old and new TSFi high neighbor out connector for the old and new
TSF-X3. Channel 1 is old TSF; Channel 2 is new TSF.
7. Old vs new TSF->BLT latency
The was done on Mar/5/04 during Marc's SLAC trip just before any change
of TSF firmware to follow for the BLT output. DCH-DCT calibration
patterns simultaneously feeding both old and new TSFs. Signals at
TSFi BLT output are compared for the same output pin. Ch1=Old TSF;
Ch2=New TSF:
For the TSF firmware version 2.3 with BLT data coming out correctly,
this latency difference grew another clock4 to ~1 microsec.
The new TSF firmware 2.3 was also tested (Mar/10/04) for single tick4 all
wire on pattern playback (TSF LUT delay set to 7 for all SL). CH1=TSF high
neighbor out; CH2=TSFi BLT output (this can be compared to test 4.
above for older version of TSF firmware):
and the new TSF->ZPD playback test with CH1=TSF hi neighbor out; CH2=TSFi
ZPD data just before going into LVDS channel link. Comparing to test 2,
this verifies that V2.3 indeed reduced the TSF ZPD output latency by one
clock4:
|
