search

sinara-hw / Sayma_AMC

AMC FPGA carrier board with FMC, dual SFP and RTM
8 stars 6 forks source link

WR FF to FPGA connection #139

Open sbourdeauducq opened 4 years ago

sbourdeauducq commented 4 years ago

Why are there both single-ended SSTL18 connections and differential SSTL15? Which one is populated on the boards?

Also this indication of "LVDS" is misleading: image

@gkasprow

gkasprow commented 4 years ago

They are just in case. There is a note below that explains this. LVDS is just a routing rule. At the moment the single-ended SSTL18 is populated, so do not use the differential ones. It is a source terminated.

sbourdeauducq commented 4 years ago

Currently I cannot get any signal out of those FFs. The FPGA constraints I am using are:

Subsignal("rec_clk", Pins("F12"), IOStandard("SSTL18_I")),
Subsignal("main_xo", Pins("L9"), IOStandard("SSTL18_I"))

Are they correct?

gkasprow commented 4 years ago

They look fine. Please, look with fast scope at the pins 14, 15 and 16, 17 of IC19. If you don't have such fast scopes, measure DC voltages instead.

gkasprow commented 4 years ago

@marmeladapk did you check this circuit?

marmeladapk commented 4 years ago

@sbourdeauducq We checked DDMTD_OUT_MAIN_CLK_SSTL18. And the signal that we got out of it looked fine. We used the same IO standard as you. Do you have both main and helper clock signals?

With CDR we had a problem to get clock out of MGT. It didn't work on the first few tries, so we moved to other issues.

sbourdeauducq commented 4 years ago

I have both working now (since today). The GTH needs this: https://github.com/m-labs/artiq/commit/307a6ca140b1cea8cf0c08072fb733312de6fd0a

sbourdeauducq commented 4 years ago

Another issue was this: https://github.com/m-labs/artiq/commit/14e09582b6ea141624d3f50d85498d4590f55c7c

marmeladapk commented 4 years ago

@sbourdeauducq Can we close this?

sbourdeauducq commented 4 years ago

Just two issues remain:

  1. While I am getting consistent DDMTD data from the main XO FF, the DDMTD data from the GTH FF is unexplained. I do not know where the problem is coming from. FWIW, I am also getting inconsistent data with the RTM implementation that does not use FFs, so it is perhaps not the FFs.
  2. If we don't use the differential signal, it should perhaps be removed.
jbqubit commented 4 years ago

I created separate issue about the misleading LVDS label. https://github.com/sinara-hw/Sayma_AMC/issues/143

jbqubit commented 4 years ago

@sbourdeauducq Can this be closed?

sbourdeauducq commented 4 years ago

No, there should still be a decision about which signal (SE or differential) to keep, and then remove the other.