• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 2 answers
  • Subscribers 16 subscribers
  • Views 3744 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Do I need digital multiplexers or analog switches?

Stefano Infante
Expert 2320 points
Other Parts Discussed in Thread: SN74CBTLV3251, SN74CB3T3257, TS3A44159, CD4512B

Hello!

I need a few signal switches, but I cannot figure out whether to use digital multiplexers or analog switches.

I need a couple of quad 2:1 muxes, a couple of single channel 2:1 muxes and a single 9:1 mux, which could be splitted in 8:1 + 2:1.

The worst case signals I have to pass are 8 MHz clocks, and my main task is to have the smallest possible input-output propagation delay (selection-output propagation delay does not matter). I found some parts (mainly SN74CBTLV3251 and SN74CB3T3257), but I am not sure if they are the right parts for me. And I don't understand why on the analog switches' data sheets there is no mention of input-output propagation delays.

Can you please help?

Thanks in advance.

Stefano Infante

  • 0
    TI__Mastermind 45575 points

    Stefano,

    Great news digital multiplexers and analog switches are two different names for the same devices.  Both of these devices are passive FET devices that can turn on and off signal paths and or route signal paths to different locations.  The I/Os of these switches and multiplexers are connected to the drain and source of these pass FETs so the propagation delays are very fast and hard to measure.  You are basically trying to measure how fast a charge will go from the drain to the source.  You can find more information on the switches and multiplexers FAQ page and this post below:

    To help you find a device with have a selection tool on TI.com that you can use to filter for 4-channel (quad) 1:2 devices and 1:8 devices.  Unfortunately we wont have all these devices on one searchable table until sometime at the end of March. You may have to look at both signal switches for 1:2 devices like TS3A44159 and analog mux/demux for 1:8 devices like SN74HC4851 sections to find both configurations.  All of these switches will have enough bandwidth to pass 8 MHz clock signals.

    Thank you,

    Adam

      

     

  • 0 in reply to Adam Torma
    Expert 2320 points
    Hi Adam,
    thanks for your reply.

    I found the post and the app note you linked very useful, and totally agree with the user on the other post about the opportunity to put this useful info on top of the e2e forum section, so that all users can get a quick answer to most of their problems.

    About propagation delays, my question was about the fact that the products marked as "multiplexers" do mention (then they measure) propagation delays, which are conveniently reported on data sheets. On the other hand, the products marked as "analog switches" do not mention propagation delays at all. And, by the way, this difference seems to be true also for non-TI products.

    But I can perhaps see an explanation somewhat hidden in your reply :)
    You say that "The I/Os of these switches and multiplexers are connected to the drain and source of these pass FETs so the propagation delays are very fast and hard to measure".
    A somewhat strange thing I noticed is that "multiplexers" have often higher rDSON than analog switches, sometimes even ten or twenty times higher. This makes me to think this way:
    - rDSON is the DC resistance that the charges have to "pass" to go from drain to source
    - rDSON is due to the FET channel that forms under the gate when applying a voltage higher than Vth to the gate itself
    - a higher rDSON means that the channel is longer (this is my guess)
    - when the channel is longer, it takes more time for the charges to pass from drain to source -> this time is measurable -> the situation of "multiplexers", with high rDSON and exposed tPD
    - when the channel is shorter, it takes less time for the charges to pass from drain to source -> this time is not measurable -> the situation of "analog switches", with low rDSON and tPD, which cannot be easily measured and then is not exposed

    What do you think?
    If this logic works, then I can simply consider the lowest rDSON switches as the fastest ones and definitely solve my problem...

    Thanks again.

    Stefano
  • 0 in reply to Stefano Infante
    TI__Mastermind 45575 points

    Stefano,

    Thank you for the feedback.  With more requests for this type of information will help me get resourcing to move these documents or write another.  

    About propagation delays, my question was about the fact that the products marked as "multiplexers" do mention (then they measure) propagation delays, which are conveniently reported on data sheets. On the other hand, the products marked as "analog switches" do not mention propagation delays at all. And, by the way, this difference seems to be true also for non-TI products.

    Unfortunately the naming conventions "multiplexers" "switches" "encoders" "digital multiplexers" are not standard company to company or within companies.  Even within TI there are multiplexers that do have buffered I/Os at TI like the CD4512B where the propagation delay would be note worthy.

    Another note, there are multiple different business within TI that produce switches and multiplexers.  Each business operates independently and characterizes the devices a little bit different. Some business decided to specify the propagation delays and others didn't for their own independent thought processes.  

     

    A somewhat strange thing I noticed is that "multiplexers" have often higher rDSON than analog switches, sometimes even ten or twenty times higher. This makes me to think this way:

    You have made a good "strange" observation.  There are certain trade offs in designing passive FET switches depending on the performance you are trying to optimize.  As you mentioned it is mostly due to the size of the FET.  Generally, to achieve a low Rdson you need a larger FET at the expense of increased capacitance and bandwidth and for higher bandwidth and lower capacitance a smaller FET.   

     

    I will need to double check on the relationship between Rdson and propogation delay.

     

    What are you working on that propagation delay is so critical.  I would say almost all switches and multiplexers are <5ns and most in the ps range.


    Adam

     What do you think?
    If this logic works, then I can simply consider the lowest rDSON switches as the fastest ones and definitely solve my problem...

     

  • 0 in reply to Adam Torma
    Expert 2320 points

    Adam Torma said:

    What are you working on that propagation delay is so critical.  I would say almost all switches and multiplexers are <5ns and most in the ps range. 

    Adam,

    I have to synchronize signals on different paths. So, propagation delay is critical, but maybe the knowledge about propagation delays is even more important: if I am confident with the fact that the maximum propagation delays of all multiplexers I use is far lower than my synchronization goal, I can be sure that they have no influence on the final result. Hence I need to choose propagation delays as low as possible, in order to have the lowest influence on my final result. And I have to use the same parts in all "forward and reverse" paths, so to have a greater confidence about their "nulling" themselves. I hope I have clarified... :)

    My goal is about 100 ns, but, as you know, one has often to go for the best possible design in order to have a chance to get what he needs in the final tests...

    Thanks again.

    Stefano