0
\$\begingroup\$

I'm going to apologize in advance as I don't have oscilloscope screenshots for the problems I'm about to describe, so words will have to suffice.

I've put together a non-isolated SMPS using the LNK304 as seen below (sorry for LTSpice). Perhaps a better laid out schematic of the power supply can be found on page 4 of the LNK302/304/306 datasheet.

System Schematic

The problem I'm seeing is that when the universal motor (14A) on the right hand-side of the schematic is triggered a noticeable dip repeating at exactly 60 Hz is seen in the LNK302 output, the dip gets progressively larger as the motor is driven at higher and higher currents. At its peak this dip is around 5V. The dip gets less prominent as more capacitance is added to the output but I'm seeing diminishing returns as I add more capacitance (247uF so far).

Additionally, the 5V regulator output sees a dip of the same magnitude on its output.

I did look at the drain input of the LNK302 with an oscilloscope and saw a negative ramp wave, the amplitude of which increased as the motor current draw increased.

I strongly suspect that this issue is related to the neutral line getting yanked around by the motor but I find the repetitive nature of the issue troubling, especially since it continues at steady state. If this is the problem I'm not at all sure how to solve it other than by limiting the current to the motor or designing a fully isolated power supply, unfortunately, neither of these are an option.

\$\endgroup\$
  • 1
    \$\begingroup\$ What is happening to the voltage across C6 while this is going on? Have you tried increasing the size of C5 and/or C6? \$\endgroup\$ – Dwayne Reid Aug 9 '16 at 14:24
  • \$\begingroup\$ I've tried bumping up the capacitance of C5 and C6 to 10uF, as well as using a full bridge rectifier as an input, both had no effect. Theoretically C5, C6, and L1 are only a pi filter to limit EMI and the system can work fine without them at all, so there values shouldn't have an impact at all! \$\endgroup\$ – nreath Aug 9 '16 at 14:34
  • 1
    \$\begingroup\$ C5 and C6 have to supply your circuit for an entire AC cycle (given your half-wave rectifier) so the value is important. But they don't appear to be too low- the input voltage might drop 10 or 15 or 20V each cycle due to the capacitors discharging- unless your circuit is already marginal. \$\endgroup\$ – Spehro Pefhany Aug 9 '16 at 14:43
  • 1
    \$\begingroup\$ I have a feeling you need to radically increase C9, which appears to be the main internal reservoir. Start by doubling it, and keep going until there's little enough ripple on the 7805 input that it never drops out. (Then add a little more....) \$\endgroup\$ – Brian Drummond Aug 9 '16 at 15:19
  • \$\begingroup\$ I increased it from the 100uF shown in the schematic to 247uF which decreased the drop in the supply voltage to around 1V. At this point I could probably keep adding capacitance until it's acceptable but I'd rather understand the underlying cause, especially if it's going to give me headaches later. Since it's not a high frequency signal I can't really understand why the LM7805 passes this through and doesn't regulate it to a stable 5V... \$\endgroup\$ – nreath Aug 9 '16 at 15:27
2
\$\begingroup\$

First, that is a very nice schematic, especially coming from LTSpice.

I don't see any reason for interaction between the motor and the power supply, except possibly because your mains voltage is dropping.

However, presumably those LEDs illuminate at higher motor power so the current draw progressively increases with increasing motor speed and there is also the opto drive current. I would suggest testing the power supply at the full circuit current draw from the 12V (add 20% for margin) and input mains at nominal voltage less 20% and see if the output 12V remains in regulation.

I don't know how much your pressure sensor draws, but you might be getting close to the maximum capability of the LNK302, even if you are using the recommended inductor etc.

Edit: Here is a guide from Tektronix on safely making mains measurements with an oscilloscope. Using a grounded input oscilloscope with reference connected to the neutral can cause inaccurate readings and can damage the oscilloscope as some of the motor current will flow through the scope probe and circuit.

\$\endgroup\$
  • 1
    \$\begingroup\$ Re 1N4005- okay. Do you have a variac to reduce the supply voltage deliberately? Do you know the actual current draw form the 12V? Have you looked at the mains waveform with that motor being switched? If you are operating through an isolation transformer for safety that might be introducing some issues. \$\endgroup\$ – Spehro Pefhany Aug 9 '16 at 14:35
  • 1
    \$\begingroup\$ So you've got -10V (RMS?) due to the motor, -15 (say) due to the input capacitors discharging. If your power supply is marginal that's probably what's happening. Test the power supply by itself- make sure it meets the maximum current at the minimum mains voltage (including low mains and IR drop due to the motor) withe some margin to spare. Divide and conquer. \$\endgroup\$ – Spehro Pefhany Aug 9 '16 at 14:46
  • 1
    \$\begingroup\$ @nreath With full load on the power supply? \$\endgroup\$ – Spehro Pefhany Aug 9 '16 at 15:24
  • 1
    \$\begingroup\$ P.S. Are you sure this is a real effect and not an artifact of measurement? I use an isolated-input oscilloscope rather than one with the low side tied to ground, if you don't have such an instrument you can try two probes in differential mode. \$\endgroup\$ – Spehro Pefhany Aug 9 '16 at 15:26
  • 1
    \$\begingroup\$ Because you could be measuring between a DC output and the ground in the wall on another outlet, so voltage that doesn't affect your circuit operation appears to be there (is subtracted from the output)--- but keep in mind I can't see your setup. \$\endgroup\$ – Spehro Pefhany Aug 9 '16 at 15:36

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.