Forum Discussion
Honored ContributorGated transceiver operation on Transceiver SI Development Kit Stratix V GX board
Hi,
I have been trying to use the transceivers on the Transceiver SI Development Kit Stratix V GX board and managed to successfully use them in continuous mode. However I would like to be able to make them work in a gated mode where bits will be sent only during specific interval. I have implemented a binary counter which I would like to use as a reference for gating the output data. The operation that I want to implement is basically to have as an output a signal that is a result of an AND operation between the 10G continuous data and the binary counter (~400ns period). I have tried by using the reset signal, but that apparently messes up with the transceiver. I have tried to play with the assignments and allocate the output to two different transceivers (1:2 demultiplexing) but I always get some physical limitations warnings and the fitter doesn't compile successfully. If someone has a suggestion on how to implement this or has already done it, I would appreciate your help :) I have instantiated two .qsys systems, one as a 10G clock for triggering purposes and another one as a data channel. The top level file is the a2ch_TDM_clock.v file which I adapted from the transceiver examples.14 Replies
Altera_ForumHi vaka,
The SV transceiver is consisted of each individual and dedicated channel. So you would not be able to perform 1:2 demultiplexing. Probably you could try to send some dummy pattern ie idle pattern dring the interval where you do not want to send any data.- Altera_Forum
To stop the data transmission at the serial output, you can tri-state the TX serial output by controlling the tx_forceelecidle input port if you are using the Custom PHY or tx_std_elecidle if you are using Native PHY. You may refer to the transceiver PHY IP user guide for further details by looking up these ports.
- Altera_Forum
Hi vaka,
Agree with tiny007, the output tri-state is something to try with. Regarding the reset, transceiver channel would require a specific reset sequence to work. Therefore, every time you press the reset, it will go through the reset sequence where all the blocks need to be retrained. You would not be able to gate the data transmission as per your requirement. - Altera_Forum
Hi Vaka,
I would suggest that you use the mux at the core logic instead of resetting your transceiver. if you reset your transceiver, it would take time for them to restart again. - Altera_Forum
Hi Vaka,
As mentioned by ah_zhi02, you can try to mux the parallel data fed into the TX parallel input. However, you will still need to switch between your data pattern and another dummy pattern. Like what you have mentioned, the transceiver will work continuously. Note that if you have the 8b10b block enabled, even if you are sending 0x00, you will still see toggling at the TX serial output due to the 10b encoded data. - Altera_Forum--- Quote Start --- To stop the data transmission at the serial output, you can tri-state the TX serial output by controlling the tx_forceelecidle input port if you are using the Custom PHY or tx_std_elecidle if you are using Native PHY. You may refer to the transceiver PHY IP user guide for further details by looking up these ports. --- Quote End --- I was actually using the Low Latency PHY, however maybe I can try to replace the PHY and see if it will work. Thanks :)
- Altera_Forum
So, I have tried to use the Custom PHY, however I get some weird data at the output. Attached is the eye diagram. The line is overlapping with the eye diagram as a result of the fact that the bit clock and the signal that puts the transceiver in idle state are not locked, so they appear overlapped. However, I would have expected that that line coincides with the low level ('0'). I don't know if it matters that I terminate one of the differential outputs and use it as a single-ended to connect it to equipment...Also, could you tell me what exactly does the tx_forceelecidle do? I have read that the transceiver buffer is put in an idle state, but does that mean exactly? Is there no data transmission at all, are '0' sent or what...?
- Altera_Forum
Hi Vaka,
Regarding the tx_forceelecidle, for your information, it will tri-state the TX output buffer. During tri-state, there is no valid signal output from the buffer. The differential voltage output (Vp-Vn) should be 0V. Note that 0V is not equivalent to logic 0 here. - Altera_Forum
Hi bfkstimchan,
Thanks for the reply. I need the output during that interval to be equivalent to logic zero, is that possible somehow? The reason is that I want to use two different signals for performing optical modulation and multiplex them optically afterwards. - Altera_Forum
Hi vaka,
If I understand it correctly, you are trying to switch between specific data pattern and logic zero. If yes, then you could mux the parallel data input (mux between your data pattern and zero's) to your low latency PHY TX ie the tx_parallel_data. You would not need the tx_forceelecidle since you are still transmitting valid data (logic zeros is considered valid data as well). It is recommended for you to try your design with Modelsim simulation as well to verify the functionality before testing on hardware.
