Analog multiplexer - resistor at input

  1. Hello PF!

    I've been quite busy on my faculty, doing my academic final work.

    Anyway I need somebodies expertise which involves analog multiplexers.

    I am using MPC506

    Datasheet
    http://www.ti.com/lit/gpn/mpc506

    I tested it and gives nice result at 5 MHz sine wave, which is more than good for my project, which is Reconfigurable active analog filter.

    I am using it to multiplex different resistors and capacitor at the output. Now this might be late because my work is almost done(theoretical), but in practice i had some problems.

    Here is the schematic:

    [​IMG]

    Does anybody know will this work in theory? I believe I will get problems because they are in parallel in some sense.

    I need thoughts of experts, I prey that this works :( because I wouldn't be able to finish my project. I did simulations but, they were fine but still...

    Thank you!
     
  2. jcsd
  3. It should work fine without any problems. But just to be safe I would add a diode to each branch so you can be assured no problems will exist.

    BiP
     
  4. phinds

    phinds 8,482
    Gold Member

    "parallel in some sense" does not make any sense. Either things are in parallel or they are not. These are not.

    EDIT: I can't see that a diode would do any good at all but it WILL cause your results to be off.
     
  5. jim hardy

    jim hardy 4,630
    Science Advisor
    Gold Member
    2014 Award

    Make sure I understand ---

    Sounds very clever - multiplexing in reverse ?
    You cause the resistor or capacitor that's on the input to appear at the output, for purpose of tuning an active filter over there? Wish i'd thought of that !


    My worry would be the internal resistance of your multiplexer- ~1500 ohms , per "RON: On Resistance" line in datasheet. That's in series and is not very predictable.

    That is an amazing multiplexer though, with its overvoltage capability.

    There exist multiplexers with lower 'on resistance' - eg
    http://www.vishay.com/docs/72091/72091.pdf

    and http://www.intersil.com/content/dam/Intersil/documents/fn60/fn6095.pdf
    but that one's just 3 volt supply so wont handle much signal voltage.

    and an interesting article...
    http://www.digikey.com/us/en/techzo...log-switches-and-muxes-for-sensor-arrays.html

    Good to hear from you again.
    Have fun !
     
  6. Hmm its good to hear!

    I will be using these multiplexers in combination 2-1 multiplexers to achieve this circuit:

    [​IMG]

    I did the theory behind it, calculated cutoff frequency and from the theoretical part it looks promising. But I needed a confirmation from some experts that the resistances and capacitors will in fact show up at output. Of course, I will compensate or better said take into considering multiplexers internal characteristics like on resistance, off resistance capacitance etc.

    If this works well, I will be some kind of pioneer in this because I haven't found anybody on internet that did anything similar which is using the multiplexers to achieve variable impedance.

    Thank you all for your responses!
     
  7. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    Similar CMOS "analog switches" have been around for a long time (20 or 30 years). I've used them for routing audio signals with no problems (and no audible distortion). I haven't seem this multiplexer version before but it looks like a bunch of the single switches with one common terminal (plus the logic to only have switch on at once, or course).

    You need to remember the "on" resistance is not zero, so if you are using capacitors in a filter you need to design the circuit to work with a 1k resistor in series with the cap (and that probably means you can't use most standard filter circuits straight from of a textbook).

    The other issue is that you might get some "interesting" transients when you change channels. IIRC the single channel switches I used could be switched without clicks and pops by ramping the control voltage up and down over a few milliseconds (or longer). You don't have that option with a digital multiplexor

    Ignore the comment about diodes. For the CMOS switches I used, "off" meant the same as for a mechanical switch - any signal leakage was unmeasurably small.
     
  8. Yes yes, precisely what i thought!

    I will model the filter using resistor+capacitor in series.

    But I had an idea!

    Using relatively small capacitors, i can get relatively high impedances on capacitor, and then i can neglect the resistor to some extent, you agree?
     
    Last edited: Apr 21, 2013
  9. AlephZero

    AlephZero 7,300
    Science Advisor
    Homework Helper

    Hmm.... a 30 pf capacitor has an impedance of about 1k at 5 MHz.

    I think you will have a hard time using caps small enough for that idea to work!
     
  10. Yes, my bad i forgot about the frequency range I was working with.

    Guess I will have to compensate some other way.

    Thanks!
     
  11. This is a very cool circuit, and good work for a school project!

    I must let you know, however, that a somewhat similar circuit appears in the analog baseband path of virtually every cell phone or cell tower. Almost all continuous-time analog filters are programmed in a similar way because the cutoff frequency of the filter depends on the absolute accuracy of the passive components. Since the value of an integrated resistor or capacitor may vary +/- 20% or more across process variation, voltage, and temperature, continuous-time analog filters must be tuned. They are usually tuned using multiplexers (although there are other techniques). So you had a very good idea! Keep at it! Analog electronics is a really fun field!
     
  12. Thank you!

    Yes I do love more analog electronics than digital. This is actually my BSc final work. So it has to be precise to very last word!

    I will keep this thread updated.

    Thank you all, again!
     
  13. Hello, a little update:

    I did manage to get the switching between HP and LP filters(I think, I will confirm it tomorrow), but I have major troubles with accuracy.

    Those 16-1 multiplexers are bringing in a constant 1.6k of resistance. Now with 16 resistors its not a problem, but capacitors are giving me headaches.

    The transfer changes from this:

    [itex]NF1=\frac{- \mathrm{G1}\, \mathrm{G3}}{\left(\mathrm{C2}\, \mathrm{C5}\right)\, s^2 + \left(\mathrm{C5}\, \mathrm{G1} + \mathrm{C5}\, \mathrm{G3} + \mathrm{C5}\, \mathrm{G4}\right)\, s + \mathrm{G3}\, \mathrm{G4}}[/itex]

    to this:

    [itex]NF2=\frac{- \mathrm{G1}\, \mathrm{G3}}{\left(\mathrm{C2}\, \mathrm{C5}\right)\, s^2 + \left(\mathrm{C5}\, \mathrm{G1} + \mathrm{C5}\, \mathrm{G3} + \mathrm{C5}\, \mathrm{G4} + \mathrm{C2}\, \mathrm{Gp} + \mathrm{C5}\, \mathrm{Gp}\right)\, s + \left({\mathrm{Gp}}^2 + \mathrm{G3}\, \mathrm{G4} + \mathrm{G1}\, \mathrm{Gp} + \mathrm{G3}\, \mathrm{Gp} + \mathrm{G4}\, \mathrm{Gp}\right)}[/itex]

    So I have to model capacitor in series with resistance.

    This is a NIGHTMARE!

    The idea was this:

    To give Q, K and w0 and construct an algorithm that selects Y1 Y2 Y3 Y4 Y5 that best suits those parameters.

    The way I did that was find functions of Q K and w0, and take 2 resistors and see what I get.

    So i would then do this for all 16 available resistors(256 combinations) and get the best capacitors that suits my needs.

    This is a major problem with that resistance in series, if i model the Y as Y=C+Gp, where Gp is 1/Rp, and Rp=1.6k.



    Can I get any advice here, how do I compensate these 1.6k?
     
Know someone interested in this topic? Share a link to this question via email, Google+, Twitter, or Facebook

Have something to add?

0
Draft saved Draft deleted