0
\$\begingroup\$

I am using LTC3624-3.3 regulator, with components as in the datasheet app notes - 12 V Vin, burst mode, 10 uF at in, 47 uF at out, 2u2 inductor rated at 2.3 A...

At no load, voltmeter says 3.34 V. Good. But with 47 ohm resistor across output (i.e. 70 mA load), Vout drops to 3.20, and at 10 ohm - 330 mA, to 3.05 V. This indicates practically no regulation. I changed output capacitor from tantalum to ceramic X5R, inductor from 2u2 to 10u, and neither changed anything.

Has anyone had no-regulation with LTC3624? Could part arrangement cause this? I believe I satisfied all design considerations from the datasheet, except that Vin trace runs for a 5-6 mm parallel to SW trace with 2 mm gap and with ground flooding on the bottom PCB layer...

\$\endgroup\$
1
  • \$\begingroup\$ Done, tnx for advice. \$\endgroup\$
    – Zigmunad
    Nov 20, 2018 at 22:47

2 Answers 2

1
\$\begingroup\$

Because it's working on no (or very light) loads it can be assumed that your problem lies in one of these areas (despite what you may think you have done): -

  • Input supply (12 volts) is not capable of supplying the current that the load needs - measure the input voltage when adding loads.
  • Inductor chosen for the design has too much series resistance - you do need to choose a decent inductor for switching applications.
  • Inductor core is saturating i.e. it cannot handle the current demanded by the load and is losing energy as heat.
  • Load resistor chosen is too low in ohms i.e. you have misinterpreted the load resistance value.
\$\endgroup\$
5
  • \$\begingroup\$ Thanks Andy for thinking along. Besides the wall adapters (labels say 12 V / 1 A, and I know they sometimes over-promise), I also tried 12 V / 2.9 Ah sealed lead-acid battery, all with the same result - battery measures 13,15 V with no load, and 13.10 V when 10 ohm across regulator output; A-meter shows 290 mA through resistor, and 2.9 V across it. (Since the load resistors came out of my old parts bins, I checked all three with ohmmeter, they are indeed 47, 10, and 2 ohm). \$\endgroup\$
    – Zigmunad
    Nov 19, 2018 at 7:02
  • \$\begingroup\$ Inductor is Taiyo Yuden NR6028T2R2N, ferrite core, 3.7A, 26 mOhm. To make things really confusing, I got the same volt drop reading with Murata I33115 switcher (3.3V, 1.5A, drop-in replacement for 78xx). I'm doing something stupid, am I? \$\endgroup\$
    – Zigmunad
    Nov 19, 2018 at 7:18
  • \$\begingroup\$ @Zigmunad hmmm sounds odd. You are saying all the right things and this is good. Maybe post a schematic and a picture of your prototype. \$\endgroup\$
    – Andy aka
    Nov 19, 2018 at 7:51
  • 1
    \$\begingroup\$ SOLVED: Bad pin crimping job on a hookup wire gave few ohms of additional resistance, which is a big deal at 10 ohms load. Unfortunately, I was measuring voltage at the resistor, since the regulator is very small. Cut off badly crimped pins off Made-in-China hookup wires, installed new pins myself, and all peachy now! \$\endgroup\$
    – Zigmunad
    Nov 19, 2018 at 19:37
  • \$\begingroup\$ Excellent news. \$\endgroup\$
    – Andy aka
    Nov 19, 2018 at 20:54
0
\$\begingroup\$

While moving things on the workbench with regulator loaded, I noticed that the voltage across resistor is swinging few hundred millivolts (I noticed that before, but I thought it was regulator instability). Checking the hookup wires, it turned out one had insufficiently crimped pins, which sporadically added few ohms to the circuit, more than enough on 10 ohm load. Unfortunately I was clipping voltmeter leads over the load, and not over the regulator output (regulator is so tiny, and voltage should be the same, except for the hookup wires...), so there was always 3.3 V at the regulator, but between 2.9 and 3.3 over the load. Fixed hookup wires, and LTC2624 regulates like a champ!

\$\endgroup\$

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.