As a part of a circuit, I want to power an opamp LM311 with a 5V DC single supply. The whole circuit will be supplied by a 24V DC supply. To avoid another power supply I want to use this 24V to 5V DC-DC converter to power the opamp. I guess this converter doesn't require any heat sink. The opamp will draw around 5mA from this converter. Here is the data-sheet of the converter.

Below is the ripple of 5V output of this converter in AC coupling without loading. The ripple is around 1.7kHz with 40mV peak to peak:

enter image description here

And here below is the ripple in again in AC coupling without loading but this time I used a 1000uF 25V electrolytic capacitor. In this case the ripple is around 80Hz with around 10mV peak to peak:

enter image description here


I used a 1000uF capacitor in parallel with the converter and reduced the ripple as you see above in the scope outputs. But still there is ripple. Is this ripple acceptable? What is the criteria? I mean for a 5V supply is 10mV pk-pk ripple by using the 1000uF cap fine for this application(powering an opamp)? And if I use 5000uF or more I guess the ripple reduces more. But what is the limit here for the capacitor value? Is there a methodology I can summarize as a rule of thumb?

  • \$\begingroup\$ You'll get more meaningful results if you test using a realistic load. Many DC-DC converters switch their operating modes when lightly loaded, sacrificing ripple for efficiency. It looks like yours might be going into a pulse-skipping mode. If your only application load is the single comparator, it might actually be worth your while to add a resistive dummy load in order to get better performance. \$\endgroup\$ – Dave Tweed May 14 '17 at 22:05
  • \$\begingroup\$ I just don't want to upset LM311 output which is used as a comparator. I dont know what ripple is acceptable for the opamp supply. \$\endgroup\$ – user1245 May 14 '17 at 22:08
  • \$\begingroup\$ Why don't you try it and see? Without knowing anything about your actual circuit, all we can recommend is reading the datasheet. \$\endgroup\$ – Dave Tweed May 14 '17 at 22:12
  • \$\begingroup\$ I mean if I use 1000uF I get different ripple than I use no cap. Im wondering and want to learn in general what ripple is acceptable. How to quantify the cap approximately. \$\endgroup\$ – user1245 May 14 '17 at 22:14
  • 3
    \$\begingroup\$ Which part of "it depends on your application" is unclear to you? Unless you tell us something about your circuit (a diagram, perhaps?), there's nothing else we can tell you! \$\endgroup\$ – Dave Tweed May 14 '17 at 22:47

The problem is this cheapo buck regulator runs at 2kHz and Zcap is much higher than 200kHz, ...


  • due to low current loading it appears to run in hysteric mode. Adding shunt capacitance with unknown ESR reduces ripple but also relaxation frequency making C harder to reduce ripple, thus limited by cap ESR.

  • this mode is essential for this class to improve phase margin under light loads. , which I leave for you to understand later. Some design dictate range of load caps inside feedback loop and ESR in order not to suppress f ripple feedback and lean towards open loop) This is why my second schematic shows the RC "outside" the feedback loop as a LPF with higher series R.

You can understand the behaviour by examining the Cap specs.

Dissipation Factor D.F. = tan δ of General Purpose (G.P.) electrolytic caps. are standardized for 120Hz bridge rectifiers and in general implies not ultra-low ESR. But if ESR is specified then it can be special ultra-low ESR needed to suppress (low frequency) ripple sawtooth.

Compare using a GP 1000 uF cap with 350mΩ and one tantalum ultralow ESR 1000uF cap with 10 mΩ effect on ripple attenuation with estimated buck regulator source ESR = 100 mΩ .

  • If Load regulation for 500mA on 5V is 1% then Vdc drops 1% or 50mV /500mA=100mΩ ESR


simulate this circuit – Schematic created using CircuitLab

So what would you prefer 10~20mΩ ultralow ESR 1mF cap or add a few Ω series ESR for your 5mA load to reduce ripple.

N.B. First define your specs what you need

  • % ripple or +/-mVpp max
  • % load regulation error (ok)
  • % tolerance or +/-mV max

    • Then design impedance ratio to achieve it.
    • Then choose parts

    • *learn to remember range of ESR * C = Ts for different cap chemistry, just like batteries.*

p.s. Although I estimated ESR of Buck regulator near 100mΩ , it is more complex like Op Amp with negative feedback reducing with rising f, Zout rises with f which includes ESR and is also affected by duty factor and RdsOn of Switch.

  • But then with Zcap(f) on regulator output, impedance drops with f, so net effect can be fairly flat. (hand waving argument)

So what cap part number did you choose? and what was the ESR?

| improve this answer | |
  • \$\begingroup\$ I missed the change in sawtooth f with 1mF load. So ignore the estimates in ESR and follow the flow of impedance (f) analysis and examine Cap ESR, D.F. differences. Define specs 1st not an afterthought. \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 15 '17 at 2:01
  • \$\begingroup\$ Buck regulator runs at 2kHz? But data-sheet says the Switching Frequency at Vin = 24V is 570 KHz typ. docs-europe.electrocomponents.com/webdocs/10a2/… \$\endgroup\$ – user1245 May 15 '17 at 7:27
  • \$\begingroup\$ It appears to operatein hiccup mode at low current loads and thus the ripple frequency drops with detected ripple voltage as a relaxation oscillator from 2kHz no 1mF C load (with unknown ESR) to 80 Hz with 10mV ripple . look again. reread my answer \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 May 15 '17 at 14:12

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.