TDC bunch counter errors (10/24/98):
Frank Chlebana observed synchronization problems with the bunch counters on TDCs in the Cosmic ray test stand. Over the past 2 weeks we have been exploring these problems with the TDC Full Crate test.
General observations:
- If the TRACER module is near the end of the crate (slot 1
or 21) the TDCs near the TRACER exhibit bunch counter errors. The
TDCs far from the TRACER stay synchronized.
- The more TDCs installed in crate the worse the
synchronization gets.
- If the TRACER is in the middle of the crate the
synchronization is much better.
A few example tests are outlined.
Configuration:
- Slot 1 - MVME 167
- Slot 4-17 some number of TDCs
- Slot 12 or 19 TRACER
- Slot 21 TESTCLK V7
Tests performed:
- TRACER in slot 12; 12 TDCs in slots 4-10, 13-17; 35 L1 Rejects followed by L1A.
Readout 9Bx (155) from all bunch counters, there may be a small
failure rate (<5% of events) this needs to verified more systematically.
- TRACER in slot 19; 12 TDCs in slots 4-10, 13-17; 35 L1 Rejects followed by L1A.
Readout 9Bx (155) from bunch counter in slots 4-10. Slots 13-16
give bunch counter results varying from 91-9Cx depending on slot and
event. Slot 17 readouts out 10-21 depending on event.
- TRACER in slot 19; 12 TDCs in slots 4,10,17; 35 L1 Rejects followed by L1A.
Readout 9Bx (155) from bunch counter in slots 4,10.
Slot 17 readouts out 9B,9B or 9D depending on event.
- TRACER in slot 19; 12 TDCs in slots 4-10, 17; 35 L1 Rejects
followed by L1A.
Readout 9B (155) from bunch counter in slots 4-10.
Slot 17 readouts out 8Ax or CAx depending on event, could be dropping a bit.
- TRACER in slot 19; 12 TDCs in slots 4-10, 17; 37 L1
Rejects followed by L1A.
Readout 9Dx (158) from bunch counter in slots 4-10.
Slot 17 readouts out D0x on every event.
Note that in the last test we incremented the number of L1R by 2 and add 2
to the bunch counter for slots 4-10. This is the expected behavior. However
in slot 17 it increments by 3. Adding or subtracting more L1R continues with
this picture. This would lead one to believe that extra clock cycles are being generated in Slot 17.
On examination of the PC board, we discovered the PECL clock receiver
is about 10" from the P2 pins. This is true of both the Rev B and Rev C
boards.
Postscript files of scope pictures of Clock signal after receiver on TDC boards in crate.
In all cases lower trace (blue/purple) is from Slot 17 TDC and upper trace
(magenta) is from slot 3. Four configurations:
There is a clear problem in cases 1 and 2. When the TRACER is in the center
of the crate the signals look about the same as if there are only 2 TDCs
installed. These are consistent with reflections off the stubs on the TDC
boards causing extra edges.
There are a couple of other puzzles about the bunch counter.
One puzzle is the offset. There are 35 L1 rejects after
the B0 has been received but the bunch counter is reading out 155.
The other is that the boards that are synchronized have the bunch counter roll
over from 158 to 00. The board which is out of synch does not roll over (evidenced by getting a value CA).
Added Nov 20, 1998: More Postscript files of scope pictures of Clock signal after receiver on
TDC boards in crate.
In all cases Blue trace is from Slot 4 TDC and Green trace
is from slot 3 and the Red trace is an ADMEM in slot 16.
Four configurations:
- TDCs in slots 4 and 17, ADMEMs in 10-16, TSCRs in 5-9, TRACER in slot 19
- TDCs in slots 4 and 17, ADMEMs in 5-10&16, TSCRs in 11-15, TRACER in slot 19
- TDCs in slots 4 and 17, ADMEMs in 5-7&16, TRACER in slot 19
- TDCs in slots 4 and 17, ADMEMs in 16, TRACER in slot 19
Note following on clock receiver chips on the different boards:
- TDC uses Synergy SY10H842 about 8-10" from P2 pins
- ADMEM uses Motorola MC10H350 about 1-1.5" from P2 pins
- TSCR uses Motorola MC10H350 about 3-3.5" from P2 pins
Created by P. Wilson (pjw@fnal.gov).
(last updated Nov 20, 1998)