12 Replies Latest reply: Nov 11, 2013 10:56 AM by NTipping RSS

    ADUM3402 and ADM3078 versus ADM2682

    NTipping

      I have an existing RS485 half duplex design using the ADUM3402, the ADM3078 running at 3.3V and a DC-DC convertor. I'm looking to save some space (as always) and am considering using the ADM2682. However, I've been trying to interpret the supply current specifications, and however I look at it, the single device solution apprears to require about twice the current under the same conditions (60R load). There doesn't appear to be a spec. for the power conversion efficiency, so am I misinterpreting the specs? I need a low-power solution.

       

      On the subject of the ADM3078, are there any known issues with the fail-safe detection features? I appear to get false start characters from the receiver on a low-capacitance bus (<600pF approx, 2 x 120R terminators) when the the transmitter turns off, with a baud rate of 240k. Once the bus capacitance rises above about 600pF, the voltage collapse appears to be slow enough for this not to happen.

        • 1. Re: ADUM3402 and ADM3078 versus ADM2682
          RAnslow

          Hi NTipping,

           

          What DC-DC convertor are you using? Is it an isoPower solution, such as the ADuM5000?

           

          The power conversion efficiency is not specified for the ADM2682E, it is restricted by the small size of the internal transformer and this is the reason for the high supply current.

           

          In order to achieve maximum integration (lowest board space), but a lower power solution than the ADM2682E, i can suggest using the ADM2485. The ADM2485 has integrated an RS-485 transceiver, iCoupler signal isolation, and a transformer driver into a single package. The ADM2485 will operate at lower power than the ADM2682E.

          Please see the ADM2485 datasheet pp16 Figure 34, which shows an applications diagram.

           

          For your second question can you please provide some more information?

          Are you refering to a Transmitter and Receiver on the same ADM3078 transceiver? Or on seperate bus nodes?

          Is there a Pull-Up resistor on the RO?

           

          Regards,

           

          Richard Anslow

          • 2. Re: ADUM3402 and ADM3078 versus ADM2682
            NTipping

            Hi Richard,

             

            I'm using a discrete DC-DC converter - a TRACO 1W unregulated device (5V to 5V then post-regulated to 3V3). I've now looked at the ADM2485 and although the power consumption is similar to the discrete solution, the increased footprint of the external transformer plus rectification versus a dc-dc convertor pushes the real estate requirements up to about the same too. Ah well, I guess you can't have everything.

             

            I am referring to the transmitter/receiver on the same ADM3078, and there is a pull up on RO (although it is 100k).

             

            Best Regards,

             

            Nick Tipping

            • 3. Re: ADUM3402 and ADM3078 versus ADM2682
              RAnslow

              Hi Nick,

               

              If you remove the 100K pull-up resistor on the RO pin, then this will remove your false start/glitch issue on the receiver output for when the transmitter turns off. The RO pin will stay Fail-safe HIGH.

               

              Regards,

               

              Richard Anslow.

              • 4. Re: ADUM3402 and ADM3078 versus ADM2682
                NTipping

                Hi Richard,

                 

                Really? IF A/Y finish > B/Z  at the end of a character, RO is fail-safe high, and is pulled high with a resistor, why would it glitch low when the Transmitter is disabled if a pull up is present (A/Y-B/Z should never be < -50mV)? The Receiver is disabled during transmission, so I need a pull-up to keep RO high.

                 

                Best Regards,

                 

                Nick Tipping

                • 5. Re: ADUM3402 and ADM3078 versus ADM2682
                  RAnslow

                  Hi Nick,

                   

                  Can you please share your schematic for the ADM3078E?

                   

                  Can you please provide an oscilloscope plot showing the glitch issue that you are seeing for the RO pin? Can you please also show the DE pin, and /RE pin on the oscilloscope plot?

                   

                  Can you then please temporarily remove the pull-up resistor on the RO pin and repeat the oscilloscope plots?

                   

                  Regards,

                   

                  Richard

                  • 6. Re: ADUM3402 and ADM3078 versus ADM2682
                    NTipping

                    Hi Richard,

                     

                    I can and I will, but unfortunately I don't have access to a system at the moment. Please bear with me and I'll post as soon as I can (but it could be a few days).

                     

                    Best Regards,

                     

                    Nick Tipping

                    • 7. Re: ADUM3402 and ADM3078 versus ADM2682
                      NTipping

                      Hi Richard,

                      I don't quite have it all together yet, but here are my initial findings from a test set up. I still need to attach a circuit which is representative of what I have on the bench, but a description follows:

                       

                      I am connected to the ADM3078 via a ADUM3402 for isolation. Both are powered from a 3.3V linear regulator which is fed from a 5V 1W unregulated isolated DC-DC convertor. Suitable decoupling is present. DE and nRE are tied together for half duplex operation. A/B run about 5mm to a screw terminal connector. A complete ground plane doesn't exist, but power and ground are distributed through realatively large copper areas. There are no other components connected to A/B on the board.

                       

                      The bench tests used to generate the attached images include no external wiring, and consist of software which normally holds the transmitter disabled and the receiver enabled until there is data to send:

                      DE-nRE and 1 A.jpg

                       

                      At the end of transmission plus a few uS, the bus is then turned around ( see "DE-nRE and A 2.jpg" for an expanded view), with the bus idling high (note that 'A' is with respect to RS485 common, not with respect to 'B'):

                      DE-nRE and A 2.jpg

                       

                      I have not yet run the circuit with no pull up on RO, but I have run tests with a 100k pull up and a 10k pull up to 3.3V. The following image shows how RO behaves with a 100k pull up when the receiver is disabled and the bus is terminated with 120R:

                      DE-nRE and RO with 100k pull up and 120R terminator 1.jpg

                      Presumably, leakage is causing the decay to about 1V. The ADUM3402 appears to interpret this as a logic 1, so the decay is having no effect on the ADUM3402 output

                       

                      The following image is an expanded view and highlights the problem:

                      DE-nRE and RO with 100k pull up and 120R terminator 2.jpg

                      As DE/nRE goes low and re-enables the receiver, there is a 200nS period where RO is driven high (as expected), but then pulses low for 500nS. I am currently running at a baud rate of 230400, and while 500nS should be too short to trigger a start bit, I'm not sure this is being filtered out by the PIC UART that the ADUM3402 is connected to. The following image shows the effect of using a 10k pull up on RO:

                      DE-nRE and RO with 10k pull up and 120R terminator.jpg

                      The only difference is that RO doesn't decay while the receiver is disabled. It still pulses low for 500nS about 200nS after the receiver is re-enabled.

                       

                      The following image shows the effect of removing the terminator or adding some bus (cable?) capacitance of 1nF:

                      DE-nRE and RO with 100k pull up and no terminator or 120R in parallel with 1nF capacitor.jpg

                      There is no 500nS pulse. With the 10k pull up on DE/nRE, there is no decay, and RO stays permanently high. Presumably, with no terminator or a higher bus capacitance, the A/B bus voltage doesn't collapse as quickly from the idle high state.

                       

                      I have tried terminating a network of a few metres length with 120R in parallel with 1nF, and the problem disappears until a node is a certain distance from a terminator, at which point the problem reoccurs.

                       

                      Is this a device issue, a layout issue, an issue with the RS485 common? Any suggestions would be appreciated.

                       

                      Thanks in advance.

                       

                      Nick Tipping

                      • 8. Re: ADUM3402 and ADM3078 versus ADM2682
                        RAnslow

                        Hi Nick,

                         

                        Your first 5 images, and the details you have given, correspond to what we have seen in the lab before when testing the ADM3078E. Have tested according to the following concept diagram? With a 100K or 10K pull-up on RO?

                        Removing the Pull-Up on RO can solve the glitch issue.

                         

                        Engineering Zone.png

                        For your 6th (last) image you have stated that when there extra bus capacitance added (1nf) and there is a pull-up on DE and /RE there is no decay on the RO signal - it stays high constantly.

                        Is this correct? -- Have you now added a pull-up to DE and /RE instead?

                         

                        For your network, with two 120 Ohm termination resistors, and bus capacitance, you shouldn't see any issues provided that you don't have a pull-up on RO.

                         

                        Best Regards,

                         

                        Richard Anslow

                        • 9. Re: ADUM3402 and ADM3078 versus ADM2682
                          NTipping

                          Hi Richard,

                           

                          Yes, I have tested in accordance with your concept diagram.

                          My mistake re: Image 6, I meant to say that there is no decay and no glitch with a 10k pull up on RO (not DE/nRE) with either no terminator or a 120R/1n terminator. Image 6 shows the decay on RO with a 100k pull up, but RO stays high with a 10k pull up (Looks like image 6 without the decay - no image posted).

                          I will try removing the pull up all together. Will that not potentially give me an issue with a floating RO driving an input on the ADUM3402? Could that not potentially float to a low enough level to be interpreted as a '0'?

                           

                          Best Regards,

                           

                          Nick Tipping

                          • 10. Re: ADUM3402 and ADM3078 versus ADM2682
                            NTipping

                            Hi Richard,

                            I have further investigated the unit I was experimenting on and found that the 100k pull up was not actually connected. So, read 'no pull up' for '100k pull up' on RO in the images above. The results with a 100k pull up look the same as with a 10k pull up. Therefore, the 500nS glitch is present, whether I have a pull up on RO or not.

                             

                            Best Regards,

                             

                            Nick Tipping

                            • 11. Re: ADUM3402 and ADM3078 versus ADM2682
                              RAnslow

                              Hi Nick,

                              Please see the Receiver Fail-Safe section of the ADM3078E datasheet.
                              http://www.analog.com/static/imported-files/data_sheets/ADM3070E_3071E_3072E_3073E_3074E_3075E_3076E_3077E_3078E.pdf


                              If you terminate the transmission line, and all the transmitters are disabled, the receiver differential input voltage is pulled to 0 V by the termination.
                              With the receiver thresholds of the ADM307xE family, this results in a logic high [on the RO pin] with a 50 mV minimum noise margin.

                              On adding the transmission line termination impedences to your network does this solve your problem?
                              We have seen on previous cases that a bus capacitance that exceeds 600pF, in addition to two 120 Ohm bus line terminations, solves the 'glitch low' issue on the receiver output RO.


                              Regards,


                              Richard

                              • 12. Re: ADUM3402 and ADM3078 versus ADM2682
                                NTipping

                                Hi Richard,

                                 

                                600pF was the value we arrived at by experimentation, so we gave it a margin and opted for 1nF in parallel with each terminator. However, it would now appear that as the bus length grows, the lumped capacitance might not work for units some distance from the terminators. I will need to do a bigger system test to confirm this.

                                 

                                Best Regards,

                                 

                                Nick Tipping