3
\$\begingroup\$

Looking at this module to convert a supply voltage into 3.3 V for boards with the ESP8266 chip, I don't understand why there are two capacitors on both input and output side:

enter image description here

As far as I can tell, there is nothing about that in the datasheet: http://www.advanced-monolithic.com/pdf/ds1117.pdf

I don't have this module, but I have an AMS1117 and would like to use it to convert 5 V from a USB power supply into 3.3 V for my ESP-01S board. Do I need two capacitors on both sides, too? What's the reason for this? I'm trying to keep things as simple as possible.

\$\endgroup\$
5
  • \$\begingroup\$ Cout(s) is clearly in the datasheet. Cin(s) is most likely for decoupling. Is your question "Why are the two (on each side)?" or "Why are they there (at all)?" \$\endgroup\$
    – Tyler
    Commented Oct 31, 2021 at 14:33
  • \$\begingroup\$ @Tyler Both I guess? ;) \$\endgroup\$
    – finefoot
    Commented Oct 31, 2021 at 14:35
  • \$\begingroup\$ The requirements are quite similar between all 1117-type regulators. Sure, some datasheets might say slightly different values based on capacitor type. If some datasheet does not mention capacitor requirements, change to a better datasheet. The AMS1117 datasheet does say why output capacitance is required and how the values are chosen. \$\endgroup\$
    – Justme
    Commented Oct 31, 2021 at 14:51
  • \$\begingroup\$ real capacitors have a frequency response different from ideal capacitors (because they have inductance and resistance in addition to capacitance). Using different sizes in parallel helps to mitigate that fact. \$\endgroup\$
    – vicatcu
    Commented Oct 31, 2021 at 15:21
  • \$\begingroup\$ The values of the capacitors is written above them (106 and 104). That means you have a 100nF part in parallel with 10uF. This is intended to give you a wider frequency response than either one alone. The 10uF output part is probably also required for stability. \$\endgroup\$
    – Peter Smith
    Commented Oct 31, 2021 at 15:42

1 Answer 1

Reset to default
1
\$\begingroup\$

For an ESP8266 application with a decent quality 5V source, you can get by with just a single 47-1000 μF cap on the output side, just place it physically as close to the ESP power pin as possible.

But what you can get by with is, of course, not the same as what is recommended. Capacitors serve the important functions of smoothing out voltage fluctuations and also filtering out errant signals (high/low frequencies), both of which can cause erroneous operation of a microcontroller and the voltage regulator itself.

For this reason, ideally both the input and output sides of your AMS1117 linear voltage regulator will have 2 capacitors in parallel -- one with a high capacitance (1 to 100 μF) for filtering low-frequency noise and one with a low capacitance (0.01 to 0.1 μF) for filtering out high-frequency noise.

Extra benefit can be achieved by using the proper capacitor chemistry for each application -- ceramic for filtering high-frequency and tantalum for filtering low-frequency. Standard can-style polarized electrolytic capacitors are not ideal. This is why AMS1117 modules like the one OP is asking about exist; they provide the best possible characteristics for the widest range of applications without any of the guesswork.

enter image description here See: Arduino Forum discussion of AMS1117 application with ESP8266

See also: this StackExchange question

\$\endgroup\$
5
  • \$\begingroup\$ Is something about a capacitor adding a pole in the frequency response of the feedback loop used by the regulator and thus certain values of capacitor must not be used to prevent oscillation? \$\endgroup\$
    – quantum231
    Commented May 2 at 17:18
  • 2
    \$\begingroup\$ "This is why AMS1117 modules like the on OP is asking about exist; they provide the best possible characteristics for the widest range of applications without any of the guesswork" If anything most of the chinese modules that are not proper dev kits add an insane amount of guesswork... and are very far from "best possible" \$\endgroup\$
    – Wesley Lee
    Commented May 2 at 17:39
  • \$\begingroup\$ I have to agree. Those chinese modules do not use best possible but worst possible capacitors for that regulator. They have very low ESR ceramic caps and they are generally good or best but the problem is that regulator requires a tantalum capacitor with higher ESR for stability. That is not to say that it happens or can be made to work with ceramics, but rarely there is much thought given about it. Also it does not make really sense to use tantalum caps these days either, while good in ther own right, they are not the cheapest or best option for many reasons today any more. \$\endgroup\$
    – Justme
    Commented May 2 at 19:08
  • \$\begingroup\$ True, not all LDO modules are created equal, just as not all caps are created equal. The point here is that using a module is far superior to no capacitors or just a single capacitor on the output. We can be critical of the inferior components on the specific model depicted in the OP but in reality even cheap ones like that work just fine for this application. \$\endgroup\$
    – Jamesfo
    Commented May 2 at 20:52
  • \$\begingroup\$ "using a module is far superior to no capacitors" - not if the module oscillates :) (I'm not trying to be annoying but I think these modules are overestimated by a lot - it doesnt even guarantee that you have the actual part(fake ICs), in which case you can't even be sure how much ESR or capacitance the IC is specd to work well with, not to mention the other specs) \$\endgroup\$
    – Wesley Lee
    Commented May 2 at 21:55

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