This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Can TSW1400 support 2 TSW30SH84 EVMs

Hello all,

Does TSW1400 support to 2 TSW30SH84 EVMs? I want to connect 2 TSW30SH84 EVM to have 4 channels RF. Are the four RF channels aligned?

Or how about TSW14J56

Thanks

Rick

  • Rick,

    The TSW1400 only has one transmit connector so it can only support one TSW30SH84. The TSW14J56 has a different connector that only supports the JESD204B products. You could try to connect two sets of TSW1400/TSW30SH84 and synchronize the two TSW1400's so that they are both sending the same data at the same time. See the HSDC Pro GUI software guide for more information on this mode of operation.

    Regards,

    Jim

  • Thanks, Jim.

    One more question, it looks to me TSW30SH84 works with TSW3100 and TSW3100 has network port. I am a bit concerned with the waveform download speed. I am wondering if waveform download speed from the host to TSW3100 is much faster than the host to TSW1400. Is this true? Do you know for example how long it takes for a 400k samples to TSW1400 and to TSW3100?

    In general how do you choose between TSW1400 vs TSW3100?

    Thanks,

    Rick  

  • Rick,

    I cannot answer the speed question but the TSW3100 will soon become obsolete. The TSW1400 is the replacement for this. How fast does this download speed need to be for your application?

    Regards,

    Jim 

  • Jim

    I thought TSW3100 is newer as the number is bigger -}. As for the speed the faster the better. We had some platform taking us 30 second to download a waveform.

    If there is a way to manipulate the data inside FPGA such as times a coefficients for each sample that will be great. that way we do not need to download every time we time each sample with some coeff. Do you think you can add this function to TSW1400 GUI? or does TI open the code to customers so the customers can add their own functions?

    Thanks,

    Rick

  • Rick,

    I am not 100% sure what you are asking, but one method might be to replace the lsb in your pattern and use it as a control bit. This has been done before by other customers. Are you trying to identify when the pattern repeats, kind of like a trigger or time stamp?

    Regards,

    Jim

  • Jim,

    Sorry I did not make it clear. We have a test cases where the waveform samples will change by a constant factor, so instead of download the new waveform again if the FPGA can multiple the constant that will avoid download the waveform.

    How did the customer manage to use the lsb as the control bit? Does he/she need to modify the vhdl codes?

    Thanks,

    Ruikang

  • Ruikang,

    We are looking into this request. Below is an answer to your data transfer time from our software team. 

    So theoretically, for 400K Samples (6400kbits) it will take ‘218ms + SW latency’. We can roughly consider the Software latency to be from 200ms to 700ms (just an approximate figure). So the transfer time for 400K samples from PC to TSW1400 would be around 418ms to 918ms roughly. This can still vary based on the Host PC configuration.

  • Jim

    wow, I am definitely happy with under 1 second to do the download.

    Really appreciate your help and insight. Thanks lot!

    Best Regards,

    Ruikang 

  • Good Morning Jim,

    Some follow up questions. From the HSDC Pro GUI manual I learned you can synchronize 2 TSW1400 in transmit. Let's say I have 2 sets of TSW1400+TSW30SH84, I use one as the master and a cable connects SYNC1 from the master to EXT_TRG_INPUT of the slave.  I have two questions here: first what is the delay between Master's RF channels and the slave's RF channel if the cable delay is ignored? This is to refer to the hardware delay of internal software/fireware delay. The second question is if the delay is fixed and in what accuracy, pico seconds?

    This is to help me to understand how can I synchronize 4 RF channels from 2 sets of TSWs and if this is a delay how accurate I can compensate.

    Thanks,

    Ruikang  

  • Ruikang,

    To setup for triggering, you will need two cables that are the same length. One cable will go from SYNC1 of board #1 to EXT_TRG of board #1. The second cable will go from SYNC2 of board #1 to EXT_TRG of board #2. Both boards will use EXT_TRG to start the trigger sequence. The signal is clocked inside the FPGA using a 166MHz clock. The delay will be anywhere from 0-6 ns (since both 166MHz clocks are not synced) plus the delay difference between the cables.

    Regards,

    Jim

  • JIm,

    1. is there a way to sync the two 166MHz clocks?

    2. If I understand correct, every time I start to play the waveforms, the delay is different since this is depending on when SYNC1 goes to high. Is this right? In other word the SYNC can come anytime between 0-6 ns.

    3. I assume the two 166MHz clocks are phase stable, in other word once the boards are powered on the two clocks will keep same phase relationship. If this is the case may I ask the trigger signal being first aligned with the master 166MHz clock before being sent to SYNC1? That way, once boards are powered on the delay is fixed so I can do proper compensation. 

    I am wiling to customize the boards if need to.

    Thanks,

    Ruikang

  • Ruikang,

    1. No.

    2. No. It depends on when EXT_TRG is registered by the 166MHZ clock.

    3. The two oscillators have ppm that may be different, thus causing a drift over time. The SYNC signal is asynchronos to the 166MHz clock.

    Regards,

    Jim

     

  • Jim,

    A bit confusion on your answer on question 2.

    Let's say I download waveform 1 to TSWs and start to play the waveform, due to asynchronous SYNC the delay between the two boards is 2ns. Now I have a new waveform and download to TSW and play again, does the delay this time remain same as 2 ns or it is going to be different, ignore the draft due to different ppm?

    On my question 3, I think we can ignore the draft. Then can you modify such that the SYNC signal is first aligned with 166MHz before be sending out?

    We can also do the modification under your guidance if you prefer. I imagine SYNC probably coming from FPGA. So only FPGA code change is needed if luck enough that there is no wire changing. 

    Thanks,

    Ruikang

  • Ruikang,

    We do not have the bandwidth at this time to do any modifications to the firmware. If you would like, we can send you the source code and you can use this as a starting point to do whatever you would like with it.

    Regards,

    Jim

  • JIm,

    Ok, great please send us the source codes and whatever help documents to help us to set it up.

    Could you please also comment my question below?

    Let's say I download waveform 1 to TSWs and start to play the waveform, due to asynchronous SYNC the delay between the two boards is 2ns. Now I have a new waveform and download to TSW and play again, does the delay this time remain same as 2 ns or it is going to be different, ignore the draft due to different ppm?

    Thanks,

    Ruikang

  • Ruikang,

    The firmware can be downloaded from the following link:

    https://txn.box.com/s/6qb9kica5szgn90j1rx1

    3757.TSW1400_doc_dac.pdf

    I have also attached a firmware description document.

    Regards,

    Jim

  • Jim,

    Can I use the USB cable that comes with TSW1400 to download the modified FPGA codes or I need a special cable for that. Is there any switch/jumps I have to set in order to let Quartus II to program the board?

    Thanks,

    Ruikang