8
\$\begingroup\$

So I've got a 7805 Regulator and I've been looking it up and each website says different info. I am using an input of 12V DC 1A Wall Adapter as an input supply.

  • www.adafruit.com Says

    This regulator does not require capacitors for stability. We recommend at least 10 uF electrolytic capacitors on both input and output.

  • Datasheet Says

    0.33 uF into the input pin and a 0.1 uF into the output pin.

so I want to know which is better and if it matters. I am looking to make it stable to charge electronics and supply projects and such.

\$\endgroup\$
7
  • \$\begingroup\$ It will depend on load among other things. If you can afford 10 uF, add 10 uF. \$\endgroup\$
    – winny
    Feb 23, 2017 at 15:47
  • \$\begingroup\$ Who do you think knows best, the engineers who designed the 7805 and wrote the datasheet or some website which sells the same part ? In practice the 7805 should indeed be stable without the caps but do you want to spend time finding that out after having weird issues when you can simply prevent that by following the recommendation of the datasheet ? Just do what the datasheet says and be done with it. \$\endgroup\$ Feb 23, 2017 at 15:55
  • 2
    \$\begingroup\$ proximity is a factor: too much wiring inductance between regulator and capacitor can cause instability problems. Attach the recommended capacitors close to the regulator, especially if the regulator is fed its input from a remote source. \$\endgroup\$
    – glen_geek
    Feb 23, 2017 at 16:05
  • \$\begingroup\$ Doesn't the datasheet say 0.1µF at output? \$\endgroup\$
    – Oskar Skog
    Feb 23, 2017 at 16:22
  • \$\begingroup\$ yes it says 0.1uf \$\endgroup\$
    – Xios
    Feb 23, 2017 at 16:28

3 Answers 3

14
\$\begingroup\$

Obey the datasheet, but there is no harm in adding extra capacitors.

The 330nF and 100nF (non-electrolytic) capacitors are probably required to guarantee that the regulator is stable. They should be as close to the regulator as possible.

The 10µF electrolytics suggested on the website may be beneficial to the rest of the circuit. Eg. ripple smoothing on input and a "circuit-wide" decoupling on output.

So, I'd suggest combining the two. (Don't use just the electrolytics as they have strong parasitic properties.) If space is limited, I'd go with what the datasheet says only.

schematic

simulate this circuit – Schematic created using CircuitLab

\$\endgroup\$
5
  • 2
    \$\begingroup\$ "Obey the datasheet, but there is no harm in adding extra capacitors." It depends on the regulator. Some regulators may get unstable with higher capacitance. Mostly collector output regulators or LDOs. \$\endgroup\$
    – dannyf
    Feb 23, 2017 at 16:15
  • 2
    \$\begingroup\$ No it's emitter follower outputs with feedback. (-1/2) Only MOSFET types may be unstable with ultra-low dropout. \$\endgroup\$ Feb 23, 2017 at 16:27
  • \$\begingroup\$ what do you mean by "electrolytics have strong parasitic properties"? \$\endgroup\$
    – Xios
    Feb 23, 2017 at 16:37
  • 4
    \$\begingroup\$ Electrolytic capacitors have much higher series resistance, series inductance and leakage current compared to most forms of non-electrolytic capacitors. No capacitor is perfect, when the frequency gets high enough, the inductance in the capacitor will start dominating over the capacitance, and that high enough frequency may be quite low if you're using an electrolytic capacitor. The series resistance is also an issue, it will limit how much current can be drawn from the capacitor during a current spike. \$\endgroup\$
    – Oskar Skog
    Feb 23, 2017 at 16:46
  • 2
    \$\begingroup\$ Basically, the only good thing about electrolytic capacitors is that they have a high capacitance and voltage rating per volume, and high capacitance per $$$. But they have many many drawbacks that limits their use. \$\endgroup\$
    – Oskar Skog
    Feb 23, 2017 at 16:48
8
\$\begingroup\$

The datasheet does not recommend any specific capacitors, it simply mentions that the measurements were taken with 330nF and 100nF.

IMO this is a shortcoming of the datasheet.

If the regulator is stable without capacitor, then it should be written. If a specific value/ESR is mandatory, then it should also be written.

A cap at the input usually helps stability, as regulators tend to dislike inductive supplies. If the main supply caps are more than a few cm away, adding the 330nF cap mentioned in the datasheet would be a good idea, or any small value modern aluminium electrolytic.

Now, the output.

Considering the history of the 7805 regulator, I would pair it with a decoupling scheme matching its age, like a 10-100µF aluminium capacitor with ESR between 0.5 and 1 ohms, and a 100nF decoupling cap close to the load. I would avoid low-ESR caps.

If you guys are interested, I might be motivated enough to test one with the network analyzer.

\$\endgroup\$
6
  • \$\begingroup\$ What would be the consequence of having too low ESR on the output cap? \$\endgroup\$
    – Oskar Skog
    Feb 23, 2017 at 18:34
  • 1
    \$\begingroup\$ Depending on the regulator, load current, etc, it could either be very happy and have nicer transient response, be borderline stable, or oscillate like a bucking mad cow. LDOs are usually quite picky, hence we get nice datasheets with clear capacitor specs, and everyone's happy. There, no spec to be found... \$\endgroup\$
    – bobflux
    Feb 23, 2017 at 18:38
  • \$\begingroup\$ Please do ! (VNA) \$\endgroup\$
    – Mike
    Feb 23, 2017 at 19:04
  • \$\begingroup\$ Under application information (p7), it does show them. \$\endgroup\$ Feb 23, 2017 at 19:05
  • \$\begingroup\$ Nice catch! I has simply Ctrl-F "capa" and missed them. Still no word about ESR though, but we can assume a 100nF film or ceramic is fine, then. \$\endgroup\$
    – bobflux
    Feb 23, 2017 at 19:13
1
\$\begingroup\$

schematic

simulate this circuit – Schematic created using CircuitLab

If you can analyze a pulse load noise response, you can choose your own Cap.

The LDO is internally compensated to be unity gain stable with a capacitive load. But step load response may cause under-overshoot depending on rate of steps. So choose Cout based on Ic=Cdv/dt for dv/dt= ripple and output ESR of emitter follower of about 1 Ohm without feedback (depending on current rating) and with feedback Zout is reduced by OA gain at DC and is implied by load regulation error in datasheet as the R ratio. Assume OA BW is about 10kHZ.

Does this help you understand?

\$\endgroup\$
6
  • \$\begingroup\$ Your LDO equivalent circuit has positive feedback. \$\endgroup\$
    – The Photon
    Feb 23, 2017 at 16:32
  • \$\begingroup\$ "Does this help you understand?" lol nope, I don't understand what you even said. I only know basic electronics. \$\endgroup\$
    – Xios
    Feb 23, 2017 at 16:33
  • \$\begingroup\$ TY photon, ok Xios then what is your load? step like or continuous high frequency or DC? \$\endgroup\$ Feb 23, 2017 at 16:36
  • \$\begingroup\$ My load would be a continuous 5V DC Output, my Input is a 12V 1A Wall Adapter \$\endgroup\$
    – Xios
    Feb 23, 2017 at 16:40
  • 1
    \$\begingroup\$ Then any cap will do. But beware 7V drop * Iout = waste heat so you may need a big heatsink. and unregulated wall adapters are often 50% higher with no load. Poor choice of 5V source. \$\endgroup\$ Feb 23, 2017 at 17:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

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