1
\$\begingroup\$

I have a circuit that switches a 50W Sunbeam heating pad via a relay (for regulating temperature). My circuit is driven by a 5v AC/DC switching converter. The AC input to the power converter is shared with the heating pad.

When the heating pad is switched off, I notice a lot of noise in the 5v DC output of my power supply. I also notice a lot of noise in the same DC output as I change the heating pad thermal output (from off, to low, to medium, to high, and back). Every "transition" on the heating pad seems to generate a lot of power noise to the microcontroller in my circuit (see o-scope images).

I've tried a simple RC snubber on the AC input to my supply (0.033uF + 120Ohm in series), and also a filter on the 5V DC output (2 electrolytics in parallel... 470uF and 150uF, separated by a 0.47uH choke). Nothing I've tried seems to affect the noise. The AC/DC converter I'm using is the XP Power ECE05

Any ideas? These power spikes are causing my controller to reset.

http://imgur.com/Bqe10KR

http://imgur.com/GzQOgZr

The switch in the schematic isn't there in the real circuit. The base of the transistor is controlled by the GPIO pin of an Atmel processor, but this illustrates the point.

http://imgur.com/Bb2i5sO

NOTE: I get identical noise when everything after the power supply filters is replaced with a 50 Ohm 10W resistor. The oscilloscope captures above were all done with just that 50 Ohm resistor as the load so I could eliminate the rest of the circuit as the problem.

Some data:

  • Connecting the heater to a different AC circuit than the power supply and o-scope did make the noise go away.
  • Connecting the heater and power supply to the same AC circuit but putting the o-scope on a different circuit did not make the noise go away.
  • Adding a 0.1 uF mylar cap in parallel (and after) the existing filter caps added some attenuation to the noise (a volt or two). Other (smaller) ceramic caps I had also did little.
  • I noticed that switching the heating pad temp setting a rate of about once a second didn't produce any noise. Once every 1/2 second produced a few noise spikes every couple seconds. Rapid switching caused noise everywhere on the o-scope trace.

I added an EMI filter (EMI103T http://www.bourns.com/data/global/pdfs/emi_t_series.pdf) to the power supply output and I also tried it at the AC input. Neither configuration had the attenuation I would have expected.

Close up of the noise signal. 8-10MHz range. This is why I thought an EMI filter with 15db at 8MHz would have a positive effect... I'm confused why it didn't have more than a 2-4V attenuation effect on the noise on the DC output.

http://i.imgur.com/11qLtnv.png

\$\endgroup\$
  • 2
    \$\begingroup\$ What is the relay and driver schematic? \$\endgroup\$ – Spehro Pefhany Apr 20 '14 at 6:18
  • 1
    \$\begingroup\$ What Spehro said, plus some details on how you are viewing this waveform. The screencap says Picoscope, which suggests that also involved is a desktop or laptop computer. It would be good to detail that, and whether it is grounded to power ground, or powered through an isolated power brick, or completely floating by operating on battery. \$\endgroup\$ – gwideman Apr 20 '14 at 6:31
  • \$\begingroup\$ Also, you might try adding 0.1u (or so) ceramic or other non-elecrolytic caps parallel to your electrolytic filter caps, as these have better frequency response. It would be useful to know what result you get from that. \$\endgroup\$ – gwideman Apr 20 '14 at 6:33
  • \$\begingroup\$ And have you tried powering your controller circuit from an entirely separate AC circuit, and what happens under that situation? \$\endgroup\$ – gwideman Apr 20 '14 at 6:49
  • \$\begingroup\$ @gwideman It's a Picoscope 2204A connected to a Windows 7 box. I hope the edits to my question provide enough answers. Thx. \$\endgroup\$ – tjd Apr 22 '14 at 3:12
2
\$\begingroup\$
  1. Disconnect the heater. If the glitch persists, try a capacitor across the coil driver. If long distance, add a line filter to the coil wires with a common mode choke and twisted pair.
  2. Connect the heater. If a disconnect fixes the problem and if the glitch persists, add a snubber without 120 ohm on AC. Or an LC filter using X rated capacitors.
  3. Use low-ESR capacitors near the relay power draw.
\$\endgroup\$
  • \$\begingroup\$ Thx Joe. Any particular guidance on using x rated caps or what types of dielectrics/materials to look for? From what I've read, I assume "no". Also, since it's 5V DC output, should I also look at a TVS or some sort of diode across Vcc and Gnd? \$\endgroup\$ – tjd Apr 25 '14 at 20:34
1
\$\begingroup\$

Judging by your new and old question details, it looks like the source of the problem is erratic load behavior when switching the heating pad load either with the relay, or with the heating pad's controller. This is evidently generating transient voltage spikes on the power line, which are being carried through the ECE05, or confusing the ECE05's regulation circuitry so it generates noise by mistake.

See http://www.samlexamerica.com/support/faqs/faq18.aspx item 3.

The erratic load behavior could be relay contact bounce, or bounce of the thermostat in the heating pad controller if it's simple electomechanical. Or if it has an electronic controller (LCD display, fancy functions), it might do some funky behavior when first turned on, and especially if switched rapidly on and off... not a condition it would normally see in the wild :-).

I don't personally know details about inrush currents and other anomalies of heating blankets, but this thread suggests that as loads they are not as simple as might be assumed: http://forums.whirlpool.net.au/archive/1898784.

If the blanket switches a substantial amount of current on and off, that will inevitably show up as a voltage spike on a device sharing the same supply -- that is separated from the upstream AC source by the same resistance (in this case the same power cord and AC wiring circuit). 05.

You might try an EMI/RFI filter on either the blanket or the input to the ECE05 or both. (http://www.jameco.com/webapp/wcs/stores/servlet/Product_10001_10001_1172613_-1 looks like a cheap experiment, other units available with higher current capability and without the IEC receptacle.)

You might find that an ordinary transformer with linear regulator would be less vulnerable to this noise. I suggest experimenting with swapping out the ECE05 for a wall wart -- 5V if you have one, or a higher voltage one plus a 7805 or similar regulator.

\$\endgroup\$
  • \$\begingroup\$ Thx! PCB through-hole EMI filter ordered from Digikey. I will update when tested. \$\endgroup\$ – tjd Apr 22 '14 at 17:48
  • \$\begingroup\$ See update #2 above. The EMI filter you suggested is much too big for the final packaging I want to use. I need a relatively small solution for what I'm doing. \$\endgroup\$ – tjd Apr 25 '14 at 20:09
  • \$\begingroup\$ Regarding update #2: We still don't really know the noise's signal chain. If the noise you see at 'probe' is the XP Power ECE05 reacting badly to a glitch at its input, note that ECE05's normal freq of operation is 130kHz, so that neighborhood is a frequency it may be able to produce noise at. The filter you chose has no effect at that frequency. Anyhow, it would be useful to see a scope plot of sample noise bursts expanded to see if there's a discernible frequency. \$\endgroup\$ – gwideman Apr 25 '14 at 21:44
  • \$\begingroup\$ See update #3 and why I thought the EMI filter I bought should have worked. \$\endgroup\$ – tjd Apr 29 '14 at 2:54

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.