AD8155 (XStream Switch) HOT, does not pass data

I'm using the AD8155 to split upstream and downstream USB 3.0 data for use as a USB 3.0 protocol analyzer analog front-end. I'm unable to get this to work. The current draw on the 3.3V rail is 1.1A @ 3.3V so the IC gets extremely hot. This is without any data passed through the device. It also does not pass data. I have reviewed schematics and replaced the AD8155 without luck. All power supplies are 3.3V. Resistance between the 3.3V rail and GND is about 6 ohm when measured with power off. The AD8155 is hard-strapped to do BICAST C => A/B.

Please see attached for schematics. Are there any gotchas I've missed?

Thanks,

John.

p.s.

Please note that the StdA_SSRX and StdB_SSRX inputs should be swapped (incorrect on schematics but fixed on actual prototype board). The idea is to route upstream USB 3.0 data from one connector, split in the AD8155 and pass on via the 2nd connector. Same for downstream data. Separate copies of the upstream/downstream data is then sent to USB 3.0 PHYs for protocol analysis.

AD8155Schematics.pdf
  • EDIT: I can now pass data through the switch but the power consumption is still very high. The switch didn't pass data since being in reset. Can the high power consumption be due to the unused terminations having VCC = 3.3V and the inputs being grounded? Per my calculations, this would be approximately 100 mA per termination pair. For the four unused inputs, which are grounded (IX_A0, IX_A1, IX_B0, IX_B1) the total dissipation around 400mA * 3.3V = 1.3W. I will try to cut VTTI, VTTO of unused inputs and outputs to see what effect this has.

  • When cutting the traces to VTTI_PA, VTTI_PB and VTTO_PC, the load resistance went up from 6 to 8.7 ohm. This indicates a current draw of (V/R) 380mA. Power dissipation would be (VI) 1.25W, which is quite large for such a small package.

    I'm considering moving to a single 1.8V supply instead (next board rev). How will this affect the I2C pins? The microcontroller in the system uses 3.3V I/O so the I2C lines are pulled up to 3.3V. Will the AD8155 handle 3.3V I2C levels when using VCC and DVCC 1.8V? Or is the correct design to use DVCC=3.3V, VCC=1.8V, VTTI_PC=1.8V, VTTI_PA=1.8V, VTTO_PB=1.8V? Is DVCC used for all digital I/O pins?

    Thanks,

    /John.

  • 0
    •  Analog Employees 
    on Apr 28, 2016 7:59 PM

    Hi John,

    Apologies for the late reply.

    The high-power consumption you are seeing is a result of grounding the unused inputs. There is a protection diode between the IPx/INx pins to VEE that you are effectively shorting out and creating the high current consumption. This can damage the part in the long run.  If you cut the connection to ground across the inputs you should see the current come down to nominal levels and see the part cool down.

    For the redesign:

    For any unused input it is recommended that you pull it up to VTTI supply.

    For any unused output you can either leave it floating or pull-it up to the VTTO supply.

    Please keep in mind, if you decide to float the unused inputs or outputs they can be properly disabled by writing to the internal registers via the I2C communication bus.  This will reduce the overall power consumption of the part and prevent unwanted noise from coupling into the inputs and propagating to the outputs.

    With regards to the supply voltage:

    You can configure VTTO=VTTI= VCC to 1.8V supply rail.

    The DVCC supply can operate independently and be set to the 3.3V rail to maintain level compliance with the I2C bus. The DVCC supply sets the level threshold for all of the control pins.

    Hope this provides some clarity. Please let us know if you have additional questions.

    Regards,

    Jarrod

  • Hi Jarrod,

    Thank you for your response.

    I can confirm that the package temperature went down from over 75 degrees C to about 65 degrees C when the inputs were left floating. The temperature is still rather high but I will change the VTTI/VTTO to 1.8V per your suggestion in the next board spin, which should bring it down additionally.

    Regards,

    /John.