Post Go back to editing

ADM2582E and EVAL-ADM2582EEBZ EMI problem

Category: Hardware
Product Number: ADM2582E


We designed a PCB with an ADM2582E onboard as the RS485 transceiver for our customer, but got feedback w.r.t EMI issues of our design. As we don't have a lot of experiences on this matter, we read the application notes AN-0972 and AN-1349, which helped quite a lot. We also bought an ADM2582E EMI evaluation board EVAL-ADM2582EEBZ as the reference for our design revision and tests.


However, when we were doing EMI tests with the EVAL-ADM2582EEBZ board in the last few days, we couldn’t reproduce similar results shown in the EVAL-ADM2582EEBZ user guide (UG-044). The settings we used for the board are as follow:

  • 9V battery for the onboard regulator ADP667, and 3.3V VDD for ADM2582E
  • ADM2582E’s driver and receiver are both enabled
  • 8MHz TxD signal from onboard oscillator LTC6900 (to do this, we connected the Pin4 ‘DIV’ of LTC6900 to ground and replaced the resistor between Pin1 ‘V+’ and Pin3 ‘SET’ with a 25kΩ one to get 8MHz signal out of LTC6900)
  • Y, Z are connected to A,B respectively (Jumpers LK5, LK7 are shorted)
  • Two 120Ω termination resistors are fitted at RT and RT1 footprints

EMI tests were carried out in a 3-meter semi anechoic chamber, and the results we got is shown below, where we can see the emission levels near 180MHz are still quite high, and lots of small peaks across the whole 30MHz to 1GHz frequency range which we suspected come from the 8MHz oscillator.

The result shown in the UG-044 Figure 6 is much much better, especially at near 180MHz range where there is almost no very obvious high emissions. (Though it seems that there is a problem with Figure 6. As Table 6 says it's 37dBuV/m at around 358MHz, in the Figure 6 the peak point at this frequency is lower than 30dBuV/m.)

We’re tring to figure out why we cannot get similar results before we go on to revise our own PCB. Is there something we missed or something we’re not doing right? Could you help? Thank you!

  • Hello,

       We've been internally discussing your questions and have some comments.

    1. First, there are multiple versions of this evaluation board which were developed and want to confirm that the board you are testing matches the board pictured in the user guide.   If, for some reason you ended up purchasing an older revision, then it will be missing some of the board level EMC suppression techniques and so would have higher emissions.

    2. I'm assuming that your board is entirely powered by the 9V battery and there is no additional/external power source/connections.  Is the 9V battery connection made with a very short cable?   Unfiltered cable connections can act as an antenna which can significantly impact the results.  The longer the cable, the better the antenna.

    3. In Table 6, the 37dB uV/m value are the EN55022 Class B limits, not the measured value.  Also, the detectors used for the scan in Figure 6 and the values in Table 6 are different.  The scan is performed using a peak detector while the readings for Table 6 are made with a quasi-peak detector.  Thus they will be different.

    4. Note that our emissions data was made in a 10m chamber while your data is being collected in a 3m chamber.  Due the difference in the sizes of the setups, an approximate difference of about 10db would be expected.


  • Hi Eric,


    Thank you very much for the reply.

    With regard to the questions in your comments,

    1. The board we bought looks the same as the one pictured in the user guide.The first picture below is the board we bought and the second one is from UG-044 Rev D.  
    1. The board is entirely powered by a 9V battery and there is no additional connections, as shown in the picture below. The 9V battery is placed in a battery holder. The cable is about 2cm long and during the test the cable is wrapped by copper foil(we already took the copper foil off when we were taking this picture). The thin blue wire is used to shorted the Pin4 ‘DIV’ of LTC6900 to ground. Also we replaced the resistor between Pin1 ‘V+’ and Pin3 ‘SET’ with a 25kΩ one to get 8MHz signal out of LTC6900 as mentioned before. Other than the battery holder and the blue wire, we didn’t add anything else for the EMI tests.
    1. In Table 6, the QP level reading is 32.30 dBuV/m at 357.152MHz(I made a mistake before about the number), and in the Figure 6 the peak level is less than 30dBuV/m. However, if the scan is performed with a peak detector and the readings in the table are made with a quasi-peak detector, as I understand it, the readings from the QP detector would be less than or equal to (no greater than) the results got from the peak detector, right? (I don’t have much experience in this field so correct me if I’m wrong)
    2. We understand that there will be about 10dB difference due to the 3-meter chamber.

    Because the emissions levels near 360MHz range look fine in the test results, we already took many EMC suppression techniques from your application notes to revise the bus side of our PCB design.

    I suppose the main problem we’re trying to figure out now is how to reduce the emissions levels near the 180MHz range. As we can’t get similar results shown in Figure 6, in which the emissions levels near the 180MHz range are controlled very well, we think there must be something we missed or aren’t doing right, and we’re trying to figure it out so that we can revise our design properly. We plan to do some more EMI tests but in the meantime any suggestions are welcomed. Thank you very much!



  • Hi Rentao,

       Some additional comments following more internal discussions.

    1. The results from the figure and the table are made under different test conditions for the board and antenna set up:

    The figure's results were measured with the board flat on an insulated table ~0.8-1m off of the floor of the chamber and the antenna at the same elevation as the board.  The top edge of the board (as shown in the above pictures) is facing the antenna.  This setup scanned all of the frequencies and identified the "worst case" frequencies for further study in the 2nd test setup.

    For the table results a 2nd setup was used where the board was rotated 360 degrees (the table the board is sitting on is a turntable) and at each rotation position, the antenna height/elevation was adjusted between 1-4m.  The measurements at the frequency of interest were collected for each combination of antenna elevation and board rotation and the largest values are what are reported in the table.

    Thus, the QP values reported in the table are very likely from a positional combination which is worse than the initial configuration use in the figure.  If the peak value results from that positional combination were available, I would expect them to be the same or larger than the QP value.

    2. Was the orientation of the battery pack during the testing the same as in your figure?  If so, because there is no common mode filtering, the battery pack is effectively increasing the size of the dipole radiator which will increase common mode emissions. 

    Placing the battery pack under the board and rotating it so that the long side of the battery pack is parallel with the short side of the board should improve the emissions.  In this orientation, it will no longer be increasing the effective area/size of the dipole because it will be within the area of the existing ground plane.  You will want to make sure that the battery pack is NOT under/crossing the isolation barrier.

    Note that adding the copper foil around the power leads will likely not be effective if the emissions are common mode which I believe is likely.

    3. some additional tests you might perform for additional debugging information might include repeating the test but with the transceiver's input pulled continuously high and then low (i.e. not switching).  This will help separate the emissions due to the isolated power supply and the data transmission.