Bypass capacitor for a PWM

Discussion in 'General Electronics Chat' started by jasoto, Dec 11, 2014.

  1. jasoto

    Thread Starter New Member

    Mar 10, 2012
    11
    0
    I am using a PWM from Intersil (IS-1825ASRH-HCI), but this datasheet is not like for example the TI datasheet that recommend an specific bypass capacitor for the part. I contacted Intersil in order to understand if they recommend a specific value for the bypass capacitor for their PWM and this is the answer I get "
    As for the 1825, what is its operating frequency and what is the gate capacitance of the FET it is going to drive. This dictates how much charge the bypass capacitor must hold. "

    My Question is, based on the information provide from Intersil, is there is any formula to calculate this minimum value for bypass capacitor?
     
  2. crutschow

    Expert

    Mar 14, 2008
    13,032
    3,240
    Normally you don't use the "minimum value" for a bypass (power supply) capacitor, you just use a value that should cover all bases. Typically a 0.1 μF ceramic cap directly from the power pins to ground in parallel with a larger (1 μF or more) electrolytic is adequate for most applications.
     
  3. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
  4. jasoto

    Thread Starter New Member

    Mar 10, 2012
    11
    0
    I think I did not explain well..., what I need to know is based the information provided by Intersil "As for the 1825, what is its operating frequency and what is the gate capacitance of the FET it is going to drive. This dictates how much charge the bypass capacitor must hold. "
    how can I relate the the operating frequency and the FET gate capacitance with the charge the capacitor must hold?
     
  5. crutschow

    Expert

    Mar 14, 2008
    13,032
    3,240
    That's an expensive radiation hardened device. Is this a space application?
    Are you talking about the power supply bypass capacitor or a bootstrap capacitor?
     
  6. jasoto

    Thread Starter New Member

    Mar 10, 2012
    11
    0
    bypass Capacitor
     
  7. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
    What capacitor are you referring to?
    The charge on a capacitor is related to the voltage by the formula

    Q = CV

    The frequency is related to the reciprocal of the time constant given by the formula,

    tau = RC

    What else would you like to know?
     
  8. jasoto

    Thread Starter New Member

    Mar 10, 2012
    11
    0
    The Operating Frequency of the PWM and the Gate Capacitance of the FET that will be drive, will dictates how much Charge the bypass capacitor must hold.

    With this information..., How can I calculate the value of the Bypass Capacitor?
     
  9. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
    You got me there.

    I thought power was supplied by the power supply, not by the bypass capacitor.

    The bypass capacitor is there to bypass, no? Not to supply power.
     
  10. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,061
    3,826
    The bypass capacitor is not there for power. If it were, it would take away the benefits of PWM on brushed motors (Better low-speed control, ...) and filter out the pulses to the time constant of the capacitor.

    The 0.1uF capacitor is just there to prevent radio interference and soften the blow of the fast turn off of the inductive load on the motor and associated kick-back which saves the brushes to some degree.
     
  11. jasoto

    Thread Starter New Member

    Mar 10, 2012
    11
    0
    let see..., The Operating Frequency of the PWM and the Gate Capacitance of the FET that will be drive it by the PWM, will dictates how much Charge the bypass capacitor connected to the Vcc of the PWM must hold.

    With this information..., How can I calculate the value of the Bypass Capacitor connected to the Vcc of the PWM?
     
  12. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
    Let's see now.

    tau = RC

    f = 1/tau

    C = tau/R = 1/fR

    That should do it, no?
     
  13. jasoto

    Thread Starter New Member

    Mar 10, 2012
    11
    0
    What is the "R"? The ESR of the capacitor?
     
  14. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
    Oh. Forgot that part.

    R = V/I
     
  15. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,061
    3,826
    Assuming this is for the power supply-

    Something in the range of 2200 to 4700uF per amp on a 50 to 60 Hz supply with full bridge rectifier works well for most things.

    If you have battery power and you only want to eliminate power sag of each pulse of Pwm. And, assuming some normal Pwm rate of 8 to 20k Hz, then you can go much lower. About 10 to 100uf per amp. Even less if you have a battery intended for high discharge rates (lithium) - more capacitance is needed if you have a battery that cannot accept high discharge rates.
     
  16. wayneh

    Expert

    Sep 9, 2010
    12,148
    3,058
    This assumes an amount of ripple. It's still true that more C would give less, right? My point is that one needs to define the goal of using the bypass cap in the first place before it can be precisely specified.

    On the other hand, getting within an order of magnitude and then adding a bit more is probably close enough.
     
  17. Papabravo

    Expert

    Feb 24, 2006
    10,152
    1,793
    This thread might set a record for repeated statements by the TS/OP with no new information. If he repeats his post enough times we're sure to get the idea. Except I have no idea what he and intersil are talking about. In the early days of TTL logic it was not uncommon to have multiple layers of bypass capacitors for different purposes. The TS/OP and Intersil are implying that the 1825 chip has current demands based on its operation that require instantaneous current from bypass capacitors in much the same way that a TTL counter or shift register would behave (ca. 1971). Problem is that there is no generally known relationship between PWM operation frequency and the capacitance of the gate drive that we can use to estimate the current demand. So the answer to the original question is that we cannot calculate or estimate something for which we have no underlying theoretical basis.

    Empirically we learned what various TTL chips required in terms of bulk capacitance to avoid having Vcc sag below about 4.75 volts. Power supplies in those days were often +5V @ 35 Amperes with multiple 500,000 uFD capacitors in the supply. We employed a mix of .01, .1, and 1 uFD monlithic ceramics on the individual chips. Will this work for you? I have no idea.
     
    Last edited: Dec 11, 2014
    Sensacell likes this.
  18. MrChips

    Moderator

    Oct 2, 2009
    12,449
    3,363
    Nah! He quoted Intersil only four times. Did he get my tongue-in-cheek answer?

    Intersil uses 0.1μF capacitors.

    500,000μF caps should do the trick, no?
     
    Last edited: Dec 11, 2014
  19. Papabravo

    Expert

    Feb 24, 2006
    10,152
    1,793
    Those weren't bypass capacitor as they were in the power supply which was remote from the boards with the ICs. The usual mix of bypass caps was every chip got a .01 uFd, every 4th chip got a .1 uFd, and every row got a 1 uFd. Each row had power and GND traces separate from all other rows and they went straight to the supply.
     
  20. Midnite Jim

    New Member

    Dec 10, 2014
    2
    0
    I don't think there is. It depends a lot on power supply line impedence. The part can drive up to 1 amp. Do you not need to use that much? Anyway, I=C dv/dt but normally there is a series gate drive resistor that limits the current in/out of the gate. dv/dt (max) is determined by R(series) and C(gate) and you just assume the gate is initially uncharged so it's just I=V/R. A lot of people just use 10 ohms for the gate resistor and don't think about it much. What voltage are you applying to the drive? You want to get all of the energy to charge the gate from the bypass capacitor without the voltage on the supply line dropping by more than whatever you decide is the maximum ripple (a few hundred millivolts, maybe).
     
Loading...