4

I have a problem with TL431 voltage reference circuit:

enter image description here

There is no load connected to VREF25.
When I remove C8, voltage VREF25 is perfectly stable 2.5V.
But when I put C8 back into the circuit, voltage VREF25 is oscillating and I can't figure out why:
enter image description here

It is irony because C8 was meant to stabilize VREF25.

UPDATE:
I tried to use capacitor with higher capacitance (470nF). It did not help.
I tried to use resistor (R10) with lower resistance (330Ω) to increase current (from 1mA to 2mA). It did not help.

UPDATE2:
3.3V line is this: enter image description here I use genuine TL431 from NXP.
There's no 139kHz signal in my circuit. Actually I disconnected this voltage reference from the rest of circuitry.
I use SMD components, capacitors and resistors are 0805.

Chupacabras
  • 5,394
  • 2
  • 23
  • 50
  • Mighty strange. Is anything else in the circuit operating at 139 kHz? Is the 3.3 V stable? – winny Sep 30 '17 at 20:20
  • @winny 3.3V is stable, actually that is the first thing I checked. Actually it's a li-ion battery and MCP1700 LDO. No, there is nothing with 139kHz. Actually I detached all other circuitry from this voltage reference. So it was connected only to 3.3V power, and that is clean. – Chupacabras Sep 30 '17 at 20:27
  • Please add stable 3.3 V reading and no 139 kHz anywhere else into your question. Are you using some strange super high ESL capacitor? Off-brand TL431? – winny Sep 30 '17 at 20:56
  • @winny I updated my question. I am using cheap 0805 MLCC capacitors. I doubt there is too high ESL. – Chupacabras Sep 30 '17 at 21:14

1 Answers1

4

According to the datasheet provided, 0.1uF is right at the boundary condition for oscillation (ref. figure 23). Use either a larger or a much smaller capacitance.

Ignacio Vazquez-Abrams
  • 48,282
  • 4
  • 73
  • 102
  • I tried to use bigger capacitor, it did not help. I have updated question. – Chupacabras Sep 30 '17 at 19:32
  • @Chupacabras According to TI's datasheet, 470nF is within region of instability. Try placing a 2.2-uF cap. By the way the datasheet says that the TLV431 is designed to be stable without caps. – next-hack Sep 30 '17 at 19:41
  • 1
    Yes, I see it in the datasheet, but apparently it doesn't work for me. 470nF is in stable region, but in reality it is unstable for me :( I tried 10uF cap now and oscillations are gone. I'm going to remove that cap. It really seems it is stable without that cap. Thank you. – Chupacabras Sep 30 '17 at 19:53
  • Capacitors are usually inserted to cut noise, not for stability. To help deciding right values they give instability regions. – andrea Sep 30 '17 at 20:27
  • The cap increases HF PSRR and gives better transient response if varying current is pulled from the TL431. 10µF is OK, you can use a much larger electrolytic like 470µF with ESR below 0R1 if you want best PSRR and lowesr noise. – bobflux Sep 30 '17 at 21:22
  • What is stable region for MCP1700? – analogsystemsrf Oct 01 '17 at 04:32