How Ron affect our multiplexers design?

Dear sir,

I am interesting about high speed multiplexers. Now I have a question about Rds_on.

When turn on the FET, there are  Rds_on and parasitic capacitance on the signal trace. It will be a RC delay circuit.

There are a description from E2E --> We use a rough formula “Bandwidth = (3/2)xDataRate” which will provide sufficient margin. Does it have a limitation about PCB trace?

I have another question. There are three different Ron characteristics – Ron, Ron(p), ∆Ron. How this three characteristics affect our design?

Regards,

Ben

  • Hi Ben,

    The formula of bandwidth = 3/2 x DataRate is really a case by case for high speed protocal. For example, a 250MHz square wave actually consists of 250MHz sine wave(fundamental frequency), 500MHz(2nd order harmonics), 750MHz (3rd order harmonics) and so on.


    If the transmitter sends out 250MHz square wave and the receiver is able to detect the logic high and low only from 250MHz sine wave(fundamental frequency), then minimal bandwidth of 250MHz is required for the entire signal channel including the PCB traces and all the devices on the signal channel.


    As you know, some receiver is not as powerful as I imagined above. They require the signal waveform reaching at the receiver side still looks more square wave then sine wave so that they can correctly detect logic high or low. Then in this case, 750MHz bandwidth might be required.

    When we refer to the bandwidth, please be aware of the attenuation as well. It is very likely that bandwidth is wide enough but attenuation is too high. Becasue bandwidth is measured at the frequency where the attenuation is -3dB lower than the attenuation at DC. For example, if at DC, the attenuation is already -10dB and at the 100MHz, the attenuation is -13dB, then 100MHz will be considered as -3dB bandwidth. In this case, actual attenuation is -13dB at 100MHz!


    As for your Ron question, ron as you suggested is the resistance between input and output when the channel is conducting. It is measured at the frequency of DC. In other words, for high speed protocal, it is not much great of help.

    For Ron (flatness), since transistor is a non-ideal device, the Ron will vary as the input and voltage voltage change. For a NMOS type switch, when the input voltage is low, the Ron will be relatively small, while the Ron will increase as the input voltage increase.

    If the switch or mutiplexier have mutiple channels, ∆Ron is the Ron difference between channels.

    I do not recognize Ron(p). Would you please point me where you saw Ron(p)?

    Fan Wang

  • In reply to Fan Wang51:

    Hi Fan,

    Thanks for your reply. It's help me a lot.
    when I look into SN74LV4051A, there is a characteristics Ron(p).
    which is mean peak on-state resistance.

    -Ben
  • In reply to Ben Kuo:


    Fan,

     

    How do you choose device base on the on state resistance spec in real case performance, I can’t prevent this issue even when I know how this trace long I have.

    The only thing I can make sure is “longer trace you need lower on resistor”. Because it have to consider about capability of receiver.

     

    Do we have a roughly equation can figure out what value of on resistance performance I need in my switch?  What is the worst case Ron of a switch I can have in my system.

     

    Thank you,

    Adam

  • In reply to Adam Torma:

    Hi Adam,

    The data rate that a channel can carry is not only determined by Ron but also determined by the communication protocal. To be more specific, the data rate can be determined by the waveform, signal bandwidth, data pattern and the channel quality. Ron is just one factor of channel quality.

    For example, ethernet CAT-5 cable can carry 10Mbps, 100Mpbs and up to 1Gbps. The physcial Ron, in this case, does not change, but the data rate increases 100 times from 10Mpbs to 1Gbps.

    If you can share an example with me, we can discuss how to choose the Ron which best fits to your design.

    Fan Wang