You are not logged in.
Hello, I just got my WARP board v2. I have some naive questions.
1. Board v2 should be able to work with Xilinx design tools v11.1, since it uses Virtex 4 FPGA, right?
2. I connect the board with my PC through USB, and I verified the USB connection with IDCODE looping, it's working. I then use Xilinx iMPACT v11.1 to download bitstream file - "WARPLab_2x2_v05_01_FPGAv2.bit" by using "Assign New Configuration File" option to the board. It shows "Added Device xc4vfx100 successfully". But then I did not see any LED light on board on (its DIP is being set to "0000" ), neither I can ping the board (IP 10.0.0.1) from PC (I did set its IP to 10.0.0.200).
Any suggestions will be highly appreciated.
Offline
After assigning the configuration file, did you run the Program step (right-click on the FPGA's icon; Program is the first option). This step finalizes the process. If the programming succeeds, a green LED will light up (D8, labeled "FGPA Done" in this photo).
Offline
Thanks a lot! I did forgot "program" it, lol, I did it now. On one board, it's been working, D8 is on, can ping it, everything is fine. On the other one, however, D8 is on, but no the LED light, and does not work at all. Could that be the problem of the board? Anything else I should check?
Offline
OK, just fixed it. Now it's fine.
Turns out just the clock board got off from the slot a little bit.
Offline
A few suggestions:
-Power cycle the board. When you plug it in, D8 (the green DONE LED) should be off. After programming via iMPACT, D8 should light up.
-Double check that the clock board is fully seated. If it's lifted from its socket (which happens occasionally when the boards are handled), the FPGA will configure successfully (i.e. D8 will light up), but the design won't boot, as it won't be receiving a valid clock. You can slightly loosen the hex nut the clock board is bolted into to make sure the connectors are fully seated.
Offline
I was following OFDM Reference Design - Wireless/Wired Bridge
I've downloaded bitstream "download-nomac.bit" (get it from the website, and did not do any modifications) in "OFDM_ReferenceDesign_FPGAv2_v14.0_public" to WARP. Then both D8 and LED light up. I then connect each WARP to a PC through a different switch, and assign IP 10.0.0.1 and 10.0.0.2 to PCs respectively. But I still can't ping through each other from any of the PCs. Did I miss something?
Thanks a lot for your help !
Offline
The reference design assumes you have a gigabit Ethernet connection to the board. Make sure the switches can communicate at 1000Mbps. The way to verify the speed that the board and switch have auto-negotiated is to look at the LEDs next to the Ethernet connector. Link 1000 LED (D2) will be on if the switch can maintain a gigabit link. If not, try connecting the laptop directly to the board.
If the Ethernet link is 1000Mbps and it still does not work, try hard-coding the ARP entry: http://warp.rice.edu/forums/viewtopic.php?id=245.
Offline
I 've just tried connected a PC to a WARP board directly with a CAT5 cross-over cable, and it seems that the WARP is very strict with the cable. CAT 5 does not work. Did you use CAT5e, or CAT6 cross-over cable, or a gigabit switch with regular CAT5, 5e or 6 cables? Thanks!
Last edited by Jettank (2009-Nov-19 12:43:19)
Offline
The FPGA board's Ethernet interface can do 10, 100 and 1000 Mbps Ethernet. 10/100 should work with any Ethernet cable. 1Gbps requires a cable with all 4 pairs, with tight enough tolerance to sustain the faster signaling. We've used CAT 5e and 6 cables for 1Gbps connections successfully.
The Ethernet PHY (the Marvell chip near the Ethernet jack) auto-negotiates a link at 10/100/1000 (and lights up the corresponding LED). But the OFDM Reference Design hard-codes a 1000 link between the TEMAC and the Ethernet PHY. You can change this (it's the ENET_LINK_SPEED macro in warpmac.c). The TEMAC also supports auto-negotiation, but we haven't implemented the required software hooks in the reference design yet.
Offline
I've changed "#define ENET_LINK_SPEED 1000" to 100. It's been working now.
Thank you so much!
Offline
Hi,
I know this is an old thread but I am having the same issues. Even with both pc's at 1000 I cant ping either of them (ive set static ip's, and turned off firewalls, and tried the arp hardcode) so I was gonna try changing the hard code to 100. When I change it then save the warpmac.c file, I 'update bitstream' and get this error:
ERROR:Xst:1484 - A core is unlicensed !
ERROR:MDT - Aborting XST flow execution!
INFO:MDT - Refer to
C:\WARP_Tutorials\Demo\OFDM_ReferenceDesign_FPGAv2_v14.1_public\synthesis\tri
mode_mac_gmii_wrapper_xst.srp for details
Running NGCBUILD ...
Rebuilding cache ...
ERROR:MDT - platgen failed with errors!
make: *** [implementation/system.bmm] Error 2
Im still pretty new to FPGAs so its totally possible that I did something wrong.
Any suggestions?
Offline
The license for the Trimode Ethernet MAC is not available by default but can be downloaded for free. Here is a previous thread on how to download the license.
Offline
Thank you, I was able to download the license and update the bitstream, however still no connection.
I have used 4 different pc's all with either Windows XP or Windows 7 installed on them, have turned off all firewalls, set static ip's to the 10.0.0.x subnet, right now Im just using a SMA cable to connect the 2 WARP boards (connecting board1_radio2 to board2_radio2), manually negotiated all pc's to 100MB Full duplex, and tried a few different cables incase they were the problem. Generally if pinging from the W7 machines I would see 'Reply from 10.0.0.1: Destination host unreachable', where 10.0.0.1 was the machine I was pinging from; and if pinging from the WXP machines I would see 'Request timed out.' Even tried the arp thing suggested before.
I really dont know what to try next.
Offline
Just to clarify, did you set the DIPSW (SW5 on FPGAv2) on the boards to 0 and 1? Try setting the IP address of the computer to 10.0.0.1 for the board with DIPSW 0 and 10.0.0.2 for the board with DIPSW 1.
Offline
Also, you should always use at least 30dB attenuation when connecting two radio boards via a cable. The transmitted power from one is sufficient to damage the receiver on another without attenuation.
Offline
Sgupta, thanks for the clarification. I originally started with SW5 set to 0 and 1 for the respective computers, but I think in a fit of frustration I changed them to both 0. Anyways It now works at 100 and 1000. I havent looked at any more of the code yet, will do soon, but is it difficult to set this up for a MIMO configuration? Thanks for all the help.
Offline
So I found this thread, http://warp.rice.edu/forums/viewtopic.php?id=589, that discusses using the OFDM Ref. design in a 2x2 setup. But changing 'warpphy_setAntennaMode(TX_ANTMODE_MULTPLX, RX_ANTMODE_MULTPLX);' didnt seem to work, radio2 was still the only operating radio board. Is there anything else that needs to be done so I can perform the OFDM ref. deisgn with mimo?
Offline
How were you gauging whether one or two radios were active after switching to multiplexing mode?
Also, keep in mind the 2x2 multiplexing mode will have significantly worse PER performance than SISO or Alamouti. It may not work at all if there are strong line-of-sight paths between Tx and Rx antennas (spatial multiplexing relies on lots of scattering in the channel). The 2-antenna Alamouti mode is a more robust MIMO mode.
If you're using v14, there are a few other parameters you can change that might help with performance. The energy threshold for packet detection defaults to 4000, which is fairly low. Raising this (to ~6000, as it was in v12) will help reject false-positive detection events. This is set with P/p on the keyboard in csmamac.c.
In v14, coarse carrier frequency offset correction is disabled by default. On some kits, enabling this will improve the PER. This is set with Q/Q on the keyboard. If you enable coarse CFO, I also suggest raising the value of mimo_ofdmRx_setCFOMaxDiff() in warpphy.c to 255 (i.e. change line 338 to mimo_ofdmRx_setCFOMaxDiff(255);). This parameter controls the maximum CFO the coarse estimator is allowed to apply; we've found the default of 38 is probably too small for some kits.
Offline
well, on the radio boards there is a group of three leds by the edge. When I download the bitstream to the board one of the leds turns yellow (i think the middle one does, cant remember). I assumed this was a status led meaning that radio board was active. When I modified the project for multiplexing only the radio2 board led turned yellow, and not the radio3 board led. Also, as I connected an antenna to both radio2 and 3 and streamed a video, and individually disconnected antennas to see how the quality degraded. With radio2 antennas connected and radio3 antennas removed there was no change in quality. But when radio2 ant were removed and radio3 ant left on the video stopped. If I'm wrong about the led or anything else just let me know.
I am using v14, and will look into the other parameters, thanks.
Offline
Ive been attempting all of this with just the NoMAC project, should I try to use the csmaMAC? Also I got both radios to be active, probably did something wrong when I generated the bit file. However it is very difficult for me to even receive a response back from a ping nevermind see a video.
Offline
If you're getting any packets through, your code is probably working fine. Keep in mind multiplexing mode is not very robust. The receiver implements zero-forcing, which will only function well in rich scattering environments. Line-of-sight between Tx and Rx antennas will degrade performance significantly. The PHY'a Alamouti mode should perform significantly better.
Offline
ok. I understand that the scattering needs to be rich. I separated the 2 nodes by a significant distance (20-30 meters, non-LOS) with desks, racks, filing cabinets in between and got nothing. Only when it was between 5-10 meters away with non-LOS can I get anything but its not consistant and as godd as not getting anything at all. I will try the alamouti mode.
Offline
I changed the antenna mode to "warpphy_setAntennaMode(TX_ANTMODE_ALAMOUTI_2ANT, RX_ANTMODE_ALAMOUTI_ANTA);" but why is it that the Rx node only activates radio2 and not both. Again I am going by the bright yellow LED (assuming this is a status LED) and I also disconnected the radio3 antenna with degrade in quality but video stopped with ant3 in and ant2 out.
Last edited by whitehat09 (2010-Apr-05 14:35:02)
Offline
Alamouti is a 2x1 scheme. RX_ANTMODE_ALAMOUTI_ANTA uses just antenna A for Rx (radio in slot 2 in the reference design). RX_ANTMODE_ALAMOUTI_ANTB will use radio 3. RX_ANTMODE_ALAMOUTI_SELDIV will listen on both antennas, and select the one with a stronger signal per-packet (a scheme sometimes called "selection diversity").
For a full list of antenna modes, see the PHY docs.
Offline
murphpo wrote:
Alamouti is a 2x1 scheme. RX_ANTMODE_ALAMOUTI_ANTA uses just antenna A for Rx (radio in slot 2 in the reference design). RX_ANTMODE_ALAMOUTI_ANTB will use radio 3. RX_ANTMODE_ALAMOUTI_SELDIV will listen on both antennas, and select the one with a stronger signal per-packet (a scheme sometimes called "selection diversity").
For a full list of antenna modes, see the PHY docs.
oh man I totally messed that up, I meant the Tx node only has radio2 active and not both
Offline
