Other Parts Discussed in Thread: EK-TM4C1294XL
Our product is using the TM4C1294NCPDT and software based on the enet_s2e reference example.
I have found that the data coming into our UART and then out the telnet port is bursty which has confused the desk top application connecting to our product by telnet. Our desk top poles several times a second with a protocol command ">04<CR>" That is a greater than sign, a zero a four and a carriage return which I will abbreviate as ">04". This is a request to our product for measurement information. Our product is to return a string.
The developer of the desk top application discovered that the response to the ">04" was what he called buffered up. Some times no response but after a subsequent ">04" was sent the instrument sent multiple responses.
I believe I have found the root cause in the enet_s2e reference design which displays such buffering and bursting of data through the telnet to UART connection.
With the reference enet_s2e project I wired PC6 to PN4 and PC7 to PN5 to connect the UARTs. Then I used RealTerminal to send repeatedly a text file containing only ">04" into one telnet port. RealTerm allows me to set inter-character delays and delays at end of line and delays before the tile to be sent is sent.
I captured and time stamped the data out the other telnet port with a Processing program I wrote which time stamps to milliseconds (since program started) . I see the ">04" command is passed through the two telnet pipes in bursts. In the screen shot I will provide there is a burst where the delay was 217 milliseconds and three ">04" commands were sent. See time stamp 2019 3 14 17 12 187874.
I would like to reduce or eliminate this burstyness.
I have begun to examine the enet_s2e code and see what I think as a 40 mS HOST_TMR_INTERVAL managing the telnet UART buffers and so I would have thought that the burstyness would be on the order of 40 mS not 200 mS.
Any improved insight would be appreciated.
I would like to understand the code enough to possibly flush the buffer when a <CR> or <LF> occures for example but am not sure that might violate the design in ways I can not anticipate.
I will add screen shots separate to this first post.