umd-lhcb / UT-Aux-mapping

Python script to generate various UT auxiliary mappings
BSD 2-Clause "Simplified" License
1 stars 0 forks source link

Check the correctness of COMET internal mapping #3

Closed yipengsun closed 5 years ago

yipengsun commented 5 years ago

@manuelfs I have release a debug mapping (see 1) for COMET FPGA (IC3) -> COMET J1. Please help me check the correctness of the mapping (maybe with the help of some undergrads).

It might be useful to refer to the wiki (see 2) to see some workarounds that I implemented.

It is also worth noting that I rearranged the J6 pins in the comet_db.net so that they are ordered logically. (See umd-lhcb/lab-notes#16, umd-lhcb/lab-notes#17 for details).

yipengsun commented 5 years ago

v0.2.1 has also include J2 in COMET internal mapping. Please use the newer release to check connections.

yipengsun commented 5 years ago

The process of from COMET J6, then to COMET DB J6, then going back to COMET J6 is very hand-wavy. They seem to agree with manual tracing, but I can't explain line 354:

        another_comet_pin = ('J6', str(int(comet_db_pin[1])+1))

Previously I assumed COMET JD6, pin x <-> COMET DB JD6, pin x+1. Here we are converting a COMET DB pin to a COMET pin, so it makes sense to subtract 1, instead of add 1.

yipengsun commented 5 years ago

Actually, +1 does make sense:

This is because on schematic, the connector is flipped horizontally. This would produce the 2 rules listed above.

jorger99 commented 5 years ago

Firstly, After lots of arguing discussion about even/odd/left/right, we have determined a rule for matching pins for J6:

The even side of J6 on the COMET has Pin N corresponding to Pin N-1 on the DB’s J6. The odd side of J6 on the COMET has Pin N corresponding to Pin N+1 on the DB’s J6

There are also 6 pins that have their net name switched from N to P or vice versa which are located in the wiki.

Secondly, West COMET has IC3 -> J4 and IC3 -> J6 which connect to the DB's J1. East COMET has IC3 -> J4 and IC3 -> J6 which connect to the DB's J2. The important thing to know is that the J2 is shifted two pins down from J1.

So West J1-10 corresponds to East J2-8. Note that J1 and J2 are identical otherwise.

Thirdly,

A quick inspection shows a discrepancy: COMET J6-56 goes to DB J6-55, but the name goes from COMET 19-P to DB 21-N. COMET J6-74 goes to DB J6-73, but the name goes from COMET 21-P to DB 19-N. The nets have been swapped or something wacky is going on.

yipengsun commented 5 years ago

It is worth nothing that the J6 pin swapping scheme also applies to J4, after all. But because of the more regular pin out of J4, the result remains unchanged.

yipengsun commented 5 years ago

I have also updated this pin swapping scheme in the wiki.

yipengsun commented 5 years ago

@nhirsh and @bflaggs have done a great job in manual tracking. So far no violation has been found. (Ignore the ones that go to RJ45 connectors, they don't go to pathfinder).