summaryrefslogtreecommitdiffstats
Commit message (Collapse)AuthorAgeFilesLines
* switch to 32 bit sequence numbers. Will wrap in ~15 hours at max rateMatt Ettus2010-11-113-14/+16
|
* send message on eob to ack the end of transmissionMatt Ettus2010-11-111-1/+6
|
* typo which isn't caught by xilinxMatt Ettus2010-11-111-1/+1
|
* separated flow control and error reporting on tx path. should work with and ↵Matt Ettus2010-11-114-25/+43
| | | | without flow control
* go to the correct stateMatt Ettus2010-11-111-3/+3
|
* add a fifo to the end of the mux to help in timing.Matt Ettus2010-11-111-6/+13
|
* add trigger to makefileMatt Ettus2010-11-111-0/+1
|
* assign setting reg addressesMatt Ettus2010-11-111-2/+2
|
* declarationsMatt Ettus2010-11-111-2/+3
|
* checkpoint in flow control packet generationMatt Ettus2010-11-115-42/+147
|
* these got dropped during the rebaseMatt Ettus2010-11-114-31/+37
|
* Removed 'ifdef for second DCM that was a deign idea for external SRAM on ↵Ian Buckley2010-11-111-49/+4
| | | | | | u2plus. Hardcoded -90 degree clcok from first DCM as final solution
* 1) Created new FIFO IP in Coregen. 512x36 dual clcok FIFO with programable ↵Ian Buckley2010-11-1111-11/+555
| | | | | | full watermark 2) Put larger FIFO on output of u2plus extramfifo that is compatable with u2_rev3 EMI fixes.
* 1) u2p has added a new signal from the SRAM to the pinout, RAM_ZZIan Buckley2010-11-114-11/+17
| | | | | | | | | | | | | | | which allows the SRAM to be placed in a sleep mode. This pin was erroniously pulled high at the top level rendering the SRAM unusable. 2) Added declaration for extramfifo debug bus which had got deleted at some point in the past 3) Created a debug bundle of signals from extsramfifo to help diagnose problem 1) 4) u2p Rev1 PCB ommits control of any of the SRAM chip selects. Made a code change so that control logic does not rely on the presence of this pin and ensuring that the SRAM is always placed in READ mode in any idle cycles.
* Defaulted all SRAM pins to LVCMOS25 8mA FASTIan Buckley2010-11-111-67/+67
|
* Placed 2nd DCM into `ifdef DCM_FOR_RAMCLK which is dissabled by defaultIan Buckley2010-11-112-7/+23
| | | | | Derived RAMCLK from 270degree offset of principle core DCM giving theoretical 2.5nS timing advance on RAM_CLK relative to RAM_* signals.
* Added external RAM FIFO to u2plus.Ian Buckley2010-11-1120-100/+4498
| | | | | | | | | | Added code branch to ext_fifo.v using generate that instantiates different input and out fifo's and touched nobl_fifo code so that it works at 18 and 36bit widths. Added 2nd DCM to top level to generate off chip RAMCLK. Added explicit I/O instances to top level for tristate drivers and changed signals to core as needed. Creted new FIFO's in core gen to replace much larger FIFO's used on u2rev3
* revert unneeded changes and incorrect commentsMatt Ettus2010-11-112-34/+34
|
* reconnect GPIOs, remove debug pins, meets timing nowMatt Ettus2010-11-111-5/+3
|
* Modified phase shift of DCM1 to -64 which is intended to give more timing ↵Ian Buckley2010-11-111-1/+1
| | | | | | | margin on reads from the SRAM at the expense of Writes to the SRAM. Tested to be at least as stable as a phase shift of 12 and beter looking timing on the logic analyzer. Signals driven by the FPGA are observed changing on the SRAM pins about 4 nS after the rising edge of the RAM clock.
* Enabled phase offset adjustment on DCM_INST1 which drives the external Fast ↵Ian Buckley2010-11-111-12/+12
| | | | | | | | SRAM clock. Set phase shift to -12 after experimentation using logic analyzer to see results. This value gives near optimum 1.5nS setup times on the source sync signals FPGA -> SRAM under lab conditions.
* Added to DCM's and some BUFG's to align the internal 125MHz clock edge with ↵Ian Buckley2010-11-114-5/+100
| | | | | | | its presentation externally at the NoBL SRAM. Since we don't have a board level trace to use to estimate clock propigation delay we just loop through the I/O on the FPGA. This hasn't been verified as working on a USRP2 yet.
* Enhanced test bench to be more like real world applicationIan Buckley2010-11-112-7/+14
|
* hangedddddddextrnal fifo size to use full NoBL SRAMianb2010-11-111-1/+1
|
* Corrected extfifo code so that all registers that are on SRAM signals are ↵ianb2010-11-115-46/+59
| | | | | | | | packed into IOBs Explcit drives and skews added to GPIO pins Corrected minor error in FIFO logic that showed data avail internally incorrectly
* capacity logic fixMatt Ettus2010-11-111-1/+1
|
* Added capacity to the module pinoutIan Buckley2010-11-111-3/+4
|
* Added a bunch of debug signals.Ian Buckley2010-11-114-9/+19
|
* Regenerated FIFO with lower trigger level for almost full flag to reflect ↵Ian Buckley2010-11-118-236/+113
| | | | | | | | | | | | | logic removed from nobl_fifo. Improved ext_fifo_tb further, try to simulate more combinations of decomation rates and packet arrival patterns. Strip out the logic in nobl_fifo that made it look like a Xilinx fall-through FIFO...it is now very simple logic but a propriatory interface that exposes the high inetrnal latency of reads. Allow the USED size of the external FIFO to be parameterized from the core level. Currently set at only 256 Corrected a bug in vita_tx_deframer.v that can write to a FIFO when its full causing illegal state. Made further edits that are currently commented becuase simulation indicates they cause problems, however suspect a further bug is in this code.
* Regenerated FIFO's for extfifo.Ian Buckley2010-11-1111-726/+15
| | | | | | | | There are problems with certain configurations it seems. It is important that the fifo_xlnx_512x36_2clk_18to36 is generated with the "almost_full" pin even though it is not used in the application. if this pin is omitted the FPGA image doesn't work correctly
* Edited FIFO instance to delete port that was not regenerated after ↵Ian Buckley2010-11-111-1/+0
| | | | reconfiguration
* Adding in files that probably didn;t exist in the ISE10.1 version of coregenIan Buckley2010-11-115-0/+808
|
* Bringing all coregen files checked in into syncIan Buckley2010-11-1110-137/+60
|
* Found bug due to not accounting for the correct number of possible in flight ↵Ian Buckley2010-11-117-52/+110
| | | | | | | | | READ operations that can be in the extfifo pipeline. Regenerated fifo_xlnx_512x36_2clk_18to36 to include prog_full output triggered at 1017 so that there are 6 empty spaces to accept in flight read data upon completion. Had to generate the FIFO using Coregen from ISE12.1 due to 10.1 verion not working correctly in FPGA Still have to tackle making this simulate in Icarus
* checkin of generated coregen filesMatt Ettus2010-11-1118-8/+556
|
* External FIFO tested in simulation and on USRP2 from decimation 64->8 with ↵Ian Buckley2010-11-1118-236/+7297
| | | | | | | | current head UHD code. Apparently operation is "flawless" but more regression and corner case regression could and should be done. Tristate drivers have been added at the top level of the hierarchy for the SRAM databus as is considered good practice for both Xilinx and ASIC design flows and so both top level and core fils have been touched.
* Checkpoint checkin.Ian Buckley2010-11-1113-0/+1507
| | | | | Loopback is running via the external ZBT SRAM...HOWEVER, its not running well, its stable but the data is corrupted sometimes. Not clear if its a logic or AC timing/SI issue yet.
* get it to buildMatt Ettus2010-11-115-5/+309
|
* moved forward from the old branchMatt Ettus2010-11-118-4/+876
|
* reverting part of the reversion of the spi settings.Matt Ettus2010-11-101-2/+2
|
* u2p needs the bigger regs for some reasonMatt Ettus2010-11-101-4/+4
|
* need to enable both 16 and 32 bit spi interfaces -- 16 used in u1e, 32 in u2 ↵Matt Ettus2010-11-101-0/+1
| | | | and u2p
* occ needs to be 2 bits wide on a 36 bit fifo interface.Matt Ettus2010-11-101-1/+2
|
* Merge branch 'u1e' into merge_u1eMatt Ettus2010-11-1064-215/+3325
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | * u1e: (130 commits) invert led signals because they are active low duh allow for CS to rise before, at the same time, or after OE better debug pins watch the ethernet chip select on our debug bus fix timing issue on DAC outputs with rev 2. This puts the whole system on a 90 degree phase shift send all gpmc signals to mictor updated pins to match rev2, removed dip switch, etc. seems to compile ok. pins are different on rev2 fixed makefile to compile with our new system add register to tell host about compatibility level and which image we are using move declaration to make loopback compile no need for protocol headers since we're not doing ethernet match the signal names in this design debug pins cleanup properly integrate the new tx chain catch up with tx_policy attach run_tx and run_rx to leds connect atr delay the q channel to make the channels line up on the AD9862 ... Conflicts: usrp2/control_lib/Makefile.srcs
| * invert led signals because they are active lowMatt Ettus2010-11-091-1/+1
| |
| * duhMatt Ettus2010-11-041-1/+1
| |
| * allow for CS to rise before, at the same time, or after OEMatt Ettus2010-09-241-7/+6
| |
| * better debug pinsMatt Ettus2010-09-232-7/+11
| |
| * watch the ethernet chip select on our debug busMatt Ettus2010-09-233-6/+8
| |
| * fix timing issue on DAC outputs with rev 2. This puts the whole system on a ↵Matt Ettus2010-09-212-50/+25
| | | | | | | | 90 degree phase shift
generated by cgit v1.2.3 (git 2.39.1) at 2024-11-16 09:52:12 +0000