TMS320F280049: FSI daisy-chain node limitation

Hi Jordan,

Jordan Zhou said:

For FSI, if daisy-chain is used, what will be the limitation for the node devices? My point of view, there's no limitation.

What sort of limitation are you asking about specifically? Are you meaning the number of possible nodes, or something else? Note that increasing the number of nodes in a daisy-chain will increase the overall communication latency, since each node will have to forward the FSI packet (frame) to the next device in the chain. See SPRACM3 app note below for daisy-chain topology details.

https://www.ti.com/lit/spracm3

Jordan Zhou said:

If 100 bytes are needed to be transmited to each device, 7 times frame transmit activities are needed, as each time max data is 16 words. Is that right?

That's correct, however each FSI word is two bytes so it can be 4 frame transmits even.

Jordan Zhou said:

With 100Mbps, what will be the total latency for 100 bytes transmit, from the first frame transmit, to the 7th frame received at node device No.50 (just an example)?

I will provide an example estimate based on the table below from the SPRACM3 App Note. The actual time would have some additional dependencies such as differential drivers/receivers used (if any), transmission line length, or any additional frame processing time needed. The example below assumes the FSI frame is immediately forwarded along to the next device in the chain upon being received, with no additional frame processing or manipulation.

Let's use the last row (DMA control) of the table for the example, which is 100mbps (i.e. 50 MHz with 1 data line) and using the max 16 word frame size. The time for the frame to be received and forwarded to the next device is 4.0us, therefore it will take approximately 50 * (4.0us) = 200us for a frame to be forwarded to the 50th node.

For the case where 7 consecutive frames are transmitted we would just add in the time it takes for the frames to be transmitted out from the host device. For the example let's say a frame is transmitted every 10us (could be more or less depending on how TX data is being loaded). This would be 7 * (10us) = 70us to transmit all 7 frames. In total we have 200us + 70us = 270us for the 7th frame to be received at the 50th node.

Best,

Kevin

Thanks Kevin.

It's very clear.

Br, Jordan

You're welcome Jordan. Let me know if you have any further questions I can help with about FSI.

Best,

Kevin