1
\$\begingroup\$

I'm going to use MIC5205 voltage regulator.
According to the datasheet and many sites and discussions I must use tantalum capacitors for the output. This LDO is unstable with ceramic capacitors (low-ESR caps in general). So I'm pretty aware of that.
But I'm not sure if I can use ceramic capacitors for bypassing in my circuit.

This is my LDO:
enter image description here

This is one of ICs powered by that LDO:
enter image description here

Here it's C26 that concerns me (that is just a sample, there are more ICs that are powered from "+3V3AN" rail). It is ceramic cap and I'm not sure if it will or will not cause a instability in my LDO.

I would assume it depends on the distance between LDO and that ceramic cap.

If this hypothesis is correct I have no idea what distance would be sufficient.

UPDATE:
According to comments, it is a problem.
One solution would be to add a resistor (R7):
enter image description here

Unfortunately that would deteriorate line regulation.

\$\endgroup\$
9
  • \$\begingroup\$ To be sure you should ask Microchip but what's wrong with using the MIC5245 as they suggested? I can remember at least once seeing someone use a non-ceramic stable regulator like this the correct output caps but plenty of ceramic decoupling at other chips. He eventually figured out that was the source of some supply noise we were always seeing. Anecdotal evidence though I guess. \$\endgroup\$
    – Some Hardware Guy
    Commented Jan 2, 2018 at 20:51
  • \$\begingroup\$ Not distance - resistance. The equivalent series resistance of the capacitors has to be high enough. Check the ESR of your decoupling caps, figure the ESR for all of them in parallel. Add a low value resistor in series if the ESR for the whole circuit is too low. \$\endgroup\$
    – JRE
    Commented Jan 2, 2018 at 20:55
  • \$\begingroup\$ @SomeHardwareGuy MIC5245 has inferior performance. Much worse line and load regulation, much worse PSRR. Higher output noise. And maximum input is only 6V vs. 16V. I'm actually planning to use another MIC5202 to regulate 12V rail to 5V (I need 5V and 3.3V rails) \$\endgroup\$
    – Chupacabras
    Commented Jan 2, 2018 at 21:19
  • \$\begingroup\$ @JRE So, I assume that those ceramic bypass caps are a problem. You suggest to add resistor in series with C26? Or to the output of LDO after C62? \$\endgroup\$
    – Chupacabras
    Commented Jan 2, 2018 at 21:40
  • \$\begingroup\$ Output of LDO, after C62. It shouldn't take but a few ohms. \$\endgroup\$
    – JRE
    Commented Jan 2, 2018 at 21:43

3 Answers 3

Reset to default
2
\$\begingroup\$

The distance of the ceramic cap from the LDO will not matter much. It will add some nH of inductance which will only matter at frequencies much higher than the bandwidth of the LDO. The ceramic capacitor however will be in parallel with the tantalum capacitors and could degrade the phase margin. There are many LDOs you could use instead that are stable with ceramic capacitors. This link might be helpful:

AN-1482

\$\endgroup\$
3
  • 1
    \$\begingroup\$ I know there are many LDOs compatible with ceramic capacitors, but unfortunately none with such parameters. But I don't want to turn this question to the shopping one. My question was not to find replacement. My question was whether it is or is not a problem when ceramic caps are in circuit. \$\endgroup\$
    – Chupacabras
    Commented Jan 2, 2018 at 21:37
  • \$\begingroup\$ You are wrong about there not being a replacement, check out LT1762. It took me a few minutes to find this and I am sure I could find others. Insisting to not change the part is not a good approach IMHO. Whether the existing part will work depends on the characteristics of all the capacitors connected to it. A loop stability analysis might tell you if it would work. \$\endgroup\$
    – EE_socal
    Commented Jan 5, 2018 at 18:24
  • \$\begingroup\$ I'm not insisting on anything. My question was whether bypass ceramic capacitors in the circuit can cause instability or not. According to comments it is a problem. So, of course I will use different part in the end and change my circuit. \$\endgroup\$
    – Chupacabras
    Commented Jan 5, 2018 at 20:04
1
\$\begingroup\$

Now I tested this circuit:

schematic

simulate this circuit – Schematic created using CircuitLab

Output voltage was oscillating. Unknown ceramic capacitors were used.

Then I changed C4 to tantalum capacitor (C3 remains in the circuit), and oscillations were gone.
Then I populated bypass ceramic capacitors to the rest of my PCB, like 2-4cm away from LDO. 6x 100nF, 5x 10uF. Oscillations did not come back.

So, when this LDO has a tantalum cap at its output, it can be stable even if circuit contains many ceramic bypass capacitors.
Of course it could be just a coincidence or luck. But I will definitely try it in the future again.

\$\endgroup\$
2
  • \$\begingroup\$ You need to take care even with tantalums; there are low esr series parts out there. \$\endgroup\$
    – Peter Smith
    Commented Jan 17, 2018 at 14:19
  • \$\begingroup\$ @PeterSmith yes they are. But the point is that ceramic-incompatible LDO can be stable even with ceramics in the circuit. That was surprise to me that I changed only 1 capacitor on the rail and that stopped oscillating. \$\endgroup\$
    – Chupacabras
    Commented Jan 17, 2018 at 17:44
0
\$\begingroup\$

I had the problem with noise generated from MIC5205 on cerammic caps. Solved by adding 0.3 Ohm resistor in series with the rest of the circuit. BTW: take care to never put higher than 470 pF cap. on the by-pass (BP) pin. Causes oscillations.

\$\endgroup\$

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