DAC glitch, rc filter and glitch energy

Discussion in 'General Electronics Chat' started by andrewmm, May 8, 2015.

  1. andrewmm

    Thread Starter Active Member

    Feb 25, 2011
    General question :

    A DAC can give out a glitch at major code change.

    one option to 'minimize' this on the output is to use a R/C low pass filter.

    Debate raging here, does the filter 'remove' the glitch , or smear it over time ?

    One side here says the glitch is attenuated,
    effectively the glitch is dissipated as 'heat' in the capacitor.

    One side says the glitch is attenuated ,
    the impedance of the DAC output forms a frequency dependent voltage divider,

    One side here says the glitch is attenuated,
    but the energy before and after the filter is the same, its just spread out in time,

    Anyone got any thoughts that I can hook together ,
  2. MrChips


    Oct 2, 2009
    Is this a homework or class question?
    What is your school level/grade/age?
  3. andrewmm

    Thread Starter Active Member

    Feb 25, 2011
    Hi ,

    none of the above.

    I'm 58 years old,
    most of the others I work with , apart from the youngsters, are about the same age,
    we have a professor of Latin, a few of maths, and the rest of us are just doctors or Degree level .

    it came up over our afternoon Tee break,
    we were reading some articles , and the lively conversation ensued.

    We had all sorts of thoughts from Lapalce transforms and FFT's , down to making a big circuit and putting lots of power out to see if the capacitor got hot as the frequency changed. ( that was the fun option )

    joy of being at a 'traditional' British establishment !!

    but we could not decide,

    so needless to say, I'm doing a simulation this weekend, some one else I know is doing a pile of maths,
    but its 'got' to be easy !

    I thought
  4. DickCappels


    Aug 21, 2008
    The filter will smear the glitch over time, but DAC manufacturers strive to minimize the glitch energy so that the remaining artifact after is minimal.
  5. Tesla23

    Senior Member

    May 10, 2009
    For a voltage output DAC followed by an RC filter, there is in fact a spike in energy dissipated in the resistor. The capacitor voltage can't follow the glitch, so most of the glitch voltage appears across the resistor. The challenge is to remove as much of the glitch energy without removing the signal, but tiny smeared glitches will remain. The glitch energy is probably small compared to the other higher frequency sampling aliases also suppressed.

    The effect on the DAC output depends on the signal. At one extreme, if the glitches occur randomly, then the effect is that the residual smeared glitches add white noise. On the other hand, if you are producing a periodic signal then typically the glitches have some periodicities, which produce spurious frequency outputs (probably at a low level if the glitch is small and the RC filter effective).
  6. MrChips


    Oct 2, 2009
    There are two approaches and you ought to look at both, time domain and frequency domain.
    An RC filter is just that, it filters or removes something from the signal.
    With regards to "a major code change" on a DAC, let us consider the "glitch" as a "step function", i.e. an abrupt change from one voltage level to another.

    Let us look at this in the frequency domain.
    A sharp step function has infinite frequencies, from DC to infinity. A low pass filter will pass the low frequencies and attenuate the highs. The energy in the signal is lowered. Where does it go? It is lost as heat in both the resistor and the capacitor. Both resistor and capacitor have "reactance" measured in ohms. The equations to solve for this reactance uses "impedance" which is a complex quantity using the operator j =√-1.


    Frequency response curve of RC low pass filter:


    Now let us look at it in the time domain.
    The step transition is slowed down by the nature of the RC circuit. Instead of a sharp transition, we get a damped rise or fall, an inverse exponential function. Hence you do get a "smearing", a damping or smoothing of the signal. Again, this is a result of removing the high frequencies and preventing the signal from changing too rapidly.


    So in summary, the glitch is smeared or smoothed because the high frequencies have been attenuated. Some energy has been removed as heat (hardly measurable) in both the resistor and capacitor.
    Last edited: May 9, 2015
    Sinus23 likes this.
  7. andrewmm

    Thread Starter Active Member

    Feb 25, 2011
    Interesting . Thank you. Monday Tee could be interesting.

    Having thought with your input about this some more
    I'm thinking

    With the removal of the high frequencies, then the energy / voltage / power of the glitch is reduced,
    not smeared out


    The low frequency part of the glitch is still present, ( as one would expect )
    which is why one sees the 'hump' out the back of the rc filter.
    its the low frequencies in the time domain.

    Thanks for all the other stuff about DAC's,
    Been around the block a few times over the decades,
    started with Transistor driven R2R networks, with a DTL register,
    am now using Ghz things that are just amazing,
    Exciting times,

    You know the biggest problems is still the same in DACs,
    The analog output, you try driving DC to even 500 MHz ,

    thank you guys,