1 2 Previous Next 15 Replies Latest reply on Aug 11, 2012 6:25 AM by user_14586677

    Sampling data


       Hello ALL


      I was working onUART and just started with SPI....


      In all the communication protocols, the data is sampledi.e  in UART it is either 8x or 16x.


      i want to know


      1. How does the component decide wheteher the received data is 1 or 0( is there something like out of 8 samples maximum occurance of high voltage is decided as 1)


      2. What if due to noise interference during  sampling wrong values are sampled.


      3. the devices are far apart and connecting the common ground is not possible...and differential data is sent.....how will the device at the other end point take this voltage level .... what will the scenario at the input of diff amp if present for amplifying data and sending to sampler..

        • 1. Re: Sampling data

          In every datasheet for a component is a short description how it is working and how it has to be used. I2C is not designed to transmit over longer distances or not using an open collector output.


          Longer distances are bridged easily with RS232 or if all fails with a TTY current-loop interface.


          Sampling of signals is usually done at a clock edge to give the signal time to settle.





          • 2. Re: Sampling data

            RS232 is NOT a long distance transmission scheme. See URL summary of serial schemes. Excluding LVDS.








            There is an excellent chart in this reference showing capabilities of various serial PHY solutions.




            As far as oversampling algorithm, my curiosity is up as to how many samples, what sequence,


            how many bit samples are used to determine a bits validity. I have a question into Exar on


            this, they more or less are the industry leaders in this area.




            Last, regarding differential receivers, as long as all common mode limitations are met the receiver


            only cares about the differntial voltage. Said another way, they do not respond to DC or common


            mode ground potential differences, within their spec range. So ohmic drops are solved, ignored,


            by a differential receiver.




            Regards, Dana.

            • 3. Re: Sampling data

              Here is the response I got from Exar -




              The only bit that is validated is the start bit.  The RX pin should idle high.  When there is transition low, then the internal state


              machine starts.  To validate the start bit, the RX pin must be low on the 4th baud clock (middle of the start bit).  Once it has


              been validated, then the data will be sampled after every 8 clocks until the stop bit has been sampled.  Ideally, each bit will be


              sampled during the middle of the bit.  Once the stop bit has been sampled, then the UART receiver will wait until there's another


              start bit (or transition to low). 
              If the RX pin is high on the 4th baud clock, then the UART assumes it was noise on the RX input and waits for a valid start bit.




              Regards, Dana.

              • 4. Re: Sampling data

                Dana: are you referring to RS 232 or SPI with regards to sampling of data? This is interesting, I was not aware that either used a clock that is so much faster than the data rate...but I guess it makes sense to do it that way to allow adequate set-up times.

                • 5. Re: Sampling data

                  Current loops can achive long distances, but speed severly compromised


                  due to L and C. And complience of the I generator key limitation.








                  Regards, Dana.

                  • 6. Re: Sampling data

                    For longer distant, better use differential signal type such as RS-485 or RS-422.

                    • 7. Re: Sampling data

                      I was referring to RS232 UARTs.




                      Regards, Dana.

                      • 8. Re: Sampling data

                        @porcine, I was referring to UARTs in general,. typo for me to say RS232 UART


                        as RS232 is a PHY specification, RS232 does not care if a UART is driving it or


                        a watermelon.




                        Regards, Dana.

                        • 9. Re: Sampling data

                          Hey i have another question ....


                          for long distance transmisson (say 2 lines are used for communication - CLK and data) due to wire capacitance the signal rise and fall slowly(assume no amplifiers in between and reproducers) ,,,, if such is the case how would be the data sampling...


                          how is the data considered...

                          • 10. Re: Sampling data

                            "Long" is a relatively uncertain expression, when I compared I2C and RS232 capabilities Dana insisted that the latter is not designed for "Long" distances and gave a link to an article where RS232 could span 12m, which I would consider to be "very long". Besides getting knowledge about data transmission the question: is: Hari, do you have a special problem in mind you'd like to solve? What is the distance you are talking about, what are the data-rates needed (eather direction), what are the cable restrictions etc.


                            And for more knowledge: ask Mrs. Google (or Wikipedia) for the "OSI model" which will give you something at hand to classify your problems and finding solutions.






                            BTW: Hari, I cannot identify your picture, try to ger a better one with better resolution. Is it a dog?

                            • 11. Re: Sampling data

                              Hey BOB


                              I am not particular about about the length neither i am in any project(i completed the project...i sent you right!!!!),  out of curiosity i am asking these question.... You are big guys...you will be knowing......


                              Google gives lots of stuff....which can be both true and not true ...since you guys have hands on experience you guys are the right men to clarify doubts..

                              • 12. Re: Sampling data

                                And you want to change my profile pic....


                                i will do that so that i can be seen......


                                And why did you say that ...was my profile picture was a dog ????


                                hey i will not copy you man.....


                                i will have my own identity.....not to be worried....

                                • 13. Re: Sampling data



                                  As a matter of fact, my forum-picture is a cat!



                                  • 14. Re: Sampling data

                                    It is largley because of cable capacitance and ohmic drop, skews (like in LVDS), that challenge transceivers


                                    to recover degraded digital signals. This is a subject that fills many ap notes and comm books.




                                    Yes, RS232 is "long" if being compared to SPI, LVDS, I2C, "short" compared to RS485, fiber, wireless......


                                    In fact RS232 can transmit .0000001 baud over incredibly "long" distances. So can a telegraph key.




                                    Regards, Dana.

                                    1 2 Previous Next