• State
  • Locked Locked
  • Replies 6 replies
  • Subscribers 18 subscribers
  • Views 929 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

CD74HCT4067 what looks like charge injection noise

Steve Clynes
Mastermind 43875 points
Other Parts Discussed in Thread: CD74HCT4067, LMV324

I am working on a design which uses the CD74HCT4067 to switch 16 high impedance sources (~2K ohms) to the input of an ADC.

Whenever the source switches I see what looks like charge injection back to the newly selected analog input. This then disrupts the analog sample captured by the ADC.

This is a switch issue since I see exactly the same issue even if I lift the common output pin of the CD74HCT4067 so that it is not connected to the ADC.

I modified my FPGA controller code so that the CD74HCT4067 was disabled whilst switching inputs, then re-enabled and I see the glitches at the instant of re-enabling the CD74HCT4067.

It seems to be at least partially dependent on what the previous input voltage level was, almost as if there is a large capacitance on the common node, but as I mentioned I see the same result even if the CD74HCT4067 common pin is lifted (i.e. no ADC or PCB capacitance).

I see about 1V drop of a 2V input for close to 1uS before it recovers (seems to depend heavily on the source impedance and on the other input channel voltage levels.

Any ideas why I would see this noise being injected back to the inputs with a 2K ohm source impedance?

BR,

Steve

  • 0
    TI__Guru* 96681 points

    Steve,

    That sounds like an interesting problem.  Can you post a schematic and scope shots?

  • 0 in reply to Emrys Maier
    TI__Mastermind 43875 points

    Emrys,

    Sorry for the delay in getting back to you.

    Please find attached the schematics and scope captures.

    In the first scope capture the top signal is injected as "C_CC1",the second trace signal is captured at TP221.

    This arrangement gives a source impedance of about 50K and results in a 6uS signal drop of about 2V

    Reducing the source series resistance to 2K8 gives scope capture 2 with a 500ns drop again of about 2V.

    ADC.pdf

    BR,

    Steve

  • 0 in reply to Steve Clynes
    TI__Guru* 96681 points

    Steve,

    I believe what you are seeing is feedthrough noise from the channel select.  We found a standard feedthrough test setup that indicates a waveform very similar to what you are seeing:

    Control-to-Switch Feedthrough Noise Test Circuit

    Can you reduce the input impedance to 600 ohms to see if it alleviates the problem?

  • 0 in reply to Emrys Maier
    TI__Mastermind 43875 points

    This looks like the same issue. Reducing the impedance from ~50K to ~2K does reduce the issue, so I expect reducing further will too.

    I can reduce the impedance as a test, but not in the application.

    Do you have any suggestions for an analog switch which does not have the noise and can more cleanly switch ~5V 4K ohm signals?

    BR,

    Steve

  • 0 in reply to Steve Clynes
    TI__Guru* 96681 points

    Steve,

    In order to get a low r_on for a switch, the FET's gate must be rather large.  Unfortunately, the large gate also produces a large capacitance and I believe that capacitance is what is causing the trouble.  I doubt any switch would be better.

    Would it be possible to replace the switch with a buffer or transceiver?  Either of those will eliminate input impedance problems and provide a more stable drive for the output.

  • 0 in reply to Emrys Maier
    TI__Mastermind 43875 points
    Hmm. OK. Seems like there is a LOT of capacitance though.

    Even with a directly buffered source (LMV324) there is a 280mV 100nS disturbance. This is OK for my application (other than requiring the buffers) but still seems excessive.

    I will need to look at some of the high speed switches which hopefully suffer less from this issue.