1
\$\begingroup\$

I put a simple current source on a circuit to drive a UV LED with a constant current.

Schematics Constant Current Source

According to the LTSpice simulation setting 0 V at the signal input results in 0 A through the LED. But when I build up this circuit I get a non-zero current for 0 V. Ergo the LED is glowing. The inputs of the AD8616 are both almost zero (the feedback is a maybe 3 mV higher because of the small current going through the fb resistor). However the AD8616 drives 600 mV at the output. The noise floor on GND is quite high in amplitude (approx. 2 mV).

enter image description here

Can anybody explain this behavior?

EDIT: I replaced the opamp. Now I get a true 0. I can imagine, that if you leave the supply voltage connected, and you set a jumper to a new position (to switch to another input signal), the opamp drives much more than it should and dies.

\$\endgroup\$
  • \$\begingroup\$ First of all, any simulation is of perfect products, real ones have things like offset voltages and leakage currents. Then also, have you measured what current is actually flowing? It may be consistent with the noise and everything else you have. If you really want to hard shutdown something, you better go quite a bit below the "off" threshold and not hover around it. \$\endgroup\$ – PlasmaHH Jun 28 '17 at 8:48
  • \$\begingroup\$ What do you mean by "off" threshold? \$\endgroup\$ – m3x1m0m Jun 28 '17 at 8:59
  • \$\begingroup\$ In your case you seem to have it at 0V. Anything above that should make a (proportional) current flow, anything below that should make no current flow. \$\endgroup\$ – PlasmaHH Jun 28 '17 at 9:00
  • \$\begingroup\$ So you mean you go below 0V with the input signal? \$\endgroup\$ – m3x1m0m Jun 28 '17 at 9:04
  • \$\begingroup\$ either that or you add an extra switch \$\endgroup\$ – PlasmaHH Jun 28 '17 at 9:06
0
\$\begingroup\$

Your schematic shows the second op amp in the package as floating. This will lead to oscillations within the IC that can also cause power dissipation problems. You should set the unused amp to unity gain by shorting the output to the negative input and grounding the positive input.

\$\endgroup\$
  • \$\begingroup\$ Thank you. The other opamp I am using though. The problem is described in my EDIT. \$\endgroup\$ – m3x1m0m Jun 28 '17 at 14:13
  • \$\begingroup\$ What is the other op amp used for? It could be the source of the failure. \$\endgroup\$ – Glenn W9IQ Jun 28 '17 at 16:00
0
\$\begingroup\$

Process variation is causing different results when using different opamps. Actually, 0V differential at any amplifier's inputs, is roughly undefined. They either see it as roughly 0.001V, meaning the output would tend to go high, or -0.001V, meaning the output would tend to go low.

Have a look fig. 27 (COMMON-MODE VOLTAGE) in this datasheet: AD8616 datasheet - Rev. G

That shows the process variation of the offset voltage, based on measurements performed on over 5 wafer lots. You'll see that at 0V common mode voltage, the input offset is somewhere in between +/- 100uV.

To guarantee you always switch the LED off with your circuit, you'd have to pull pin Signal to below -100uV, and that would only apply to room temperature.

\$\endgroup\$
0
\$\begingroup\$

You should consider stabilizing your amplifier loop, or ensure that your simulation shows that it is stable.

Have a look at this post: Stability of Constant Current circuit

\$\endgroup\$

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.