4
\$\begingroup\$

I have a 230VAC inverter that runs off 12V (battery). But it seems to cause a lot of noise.

The noise is causing problems with my MCU and other circuits. It seems the problem is measurable in the +12V supply.

Diagram

schematic

simulate this circuit – Schematic created using CircuitLab

Oscilloscope

I'm measuring the +12V bus, which is connected to battery positive, and with the probe ground connected to the negative bus, which is connected to battery negative terminal, via 30mm2 cables of ~1m length.

Measuring directly at the positive terminal gives the same result. Except when the probe ground is connected to battery negative. (This leads me to suspect the problem might be with the inverter input)

enter image description here enter image description here enter image description here enter image description here

However, when probe ground is connected to battery negative terminal, the noise disappears. The scope looks the same when the inverter is turned off.

enter image description here

Attempts to reduce noise

  1. I've tried to use an extension cord to move it ~4 meters away: no effect

  2. Turning off lights: no effect

  3. Holding the inverter in my hand: no effect

  4. Connecting oscilloscope probe to battery negative terminal: causes noise to disappear

  5. Connecting a wire from negative battery terminal to oscilloscope ground: no effect

Context:

  • Inverter pulls 0.6A
  • Scope ground is connected to negative bus, which is connected to battery negative terminal
  • No natural earth is connected. "Earth" is connected to negative bus, which is connected to vehicle chassis (motorhome installation with external mains cable intake, mains cable lacks earth wire).
  • Battery bank is LiFePO4
  • No VAC output is connected
  • Oscilloscope and computer is powered from mains cable during testing
  • No other appliances are connected except for a 12V LED strip (no driver).

Question

What is the cause of this noise?

Do I need a beefy input capacitor? The inverter already has some beefy input capacitors, but maybe they're not enough?

What else might resolve the problem?

\$\endgroup\$
  • 1
    \$\begingroup\$ "However, when probe ground is connected to battery negative terminal, the noise disappears. " So where is the probe ground when you see this noise? \$\endgroup\$ – JonRB Jan 30 '17 at 1:10
  • \$\begingroup\$ Thanks for asking to clarify. The probe ground was connected to the negative bus, which is connected the battery negative terminal. In other words, despite relatively thick cables, the problem might be at the bus, but not at the battery terminals. (capacitance? inductance?) \$\endgroup\$ – user95482301 Jan 30 '17 at 1:13
  • \$\begingroup\$ it will be inductive. due to the switching of your devices causing some current to flow. Be careful where you connect your scope probe (ie tie to EARTH), especially when you get a load on the output of the inverter ... \$\endgroup\$ – JonRB Jan 30 '17 at 1:15
  • \$\begingroup\$ I'm aware of the dangers of the probe ground, which is why I'm trying to leave it connected to the battery negative bus (via earth wire). I was hoping that it would serve as an absolute reference. \$\endgroup\$ – user95482301 Jan 30 '17 at 1:18
  • \$\begingroup\$ Does this mean I would need a beefy capacitor to negate the induction? I assume the cable-inductance is the reason why the LiFePO4 aren't able to maintain the voltage level (despite very low internal resistance) ? \$\endgroup\$ – user95482301 Jan 30 '17 at 1:20
2
\$\begingroup\$

Noise can be killed by shielding and filtering. Of course, it would be nice, if no noise were generated anywhere, but that's a dream as long as there exixt computers, logic circuits, modern power supplies and radio transmitters. Shielding stops the fields in the air. Filtering prevents the noise propagating further along the Wires.

Oscilloscope input GND should be there, where your interesting signal is. Any detour creates a risk to measure the signal + the noise catched by the detour wire.

The detour wire works as antenna that collects noise from the air. If it has a current, then all sharp changes in the current cause varying voltage difference that also look out as additional noise.

Check, how noisy is your DC input in the device that uses the current. (no detour for oscilloscope probe wires). That matters.

Even a serious noise is easily hidden if the oscilloscope is at DC position, because the DC forces to use so high volts/div that even one volt of noise is unnoticed. The noisefullness of a DC should be checked by having the probe in AC position.

ADDENDUM for the system diagram Only the inverter without a load and the noise is harmful for your other gear in the same room - No doubt, your inverter is a radio transmitter. There are some fast signal state changes in the inverter and that's enough, if no countermeasures are done. If your inverter has a metal case (=earth) and it's connected to only one DC input wire, then you even have the antenna and the earth as different entities - just like in the radio stations before VHF, UHF and microwaves.

The following image is a normal approach to keep the rf noise inside the metal:

RF filtering

Inside the housig there should be symmetric LC rf filter for input and output. The filters must be designed for the used voltages and currents. The filter coils must be well away from converter's coils. The only allowed contacts between the circuits and the housing are X and Y and the wires from them to the filters must be short, for example 2 cm. All input and output wires must be decoupled with a few nF feedthrough capacitors. The protective earth wire can be connected to the housing. More filtering can be achieved by wounding the input and output cables few turns over ferrite rings that are positioned outside the metal housing, but as near as possible.

Good parts are available from old PC power supplies.

\$\endgroup\$
  • \$\begingroup\$ Given that I moved the inverter 4 meters away, wouldn't that exclude the possibility of airborne noise? \$\endgroup\$ – user95482301 Jan 30 '17 at 9:28
  • \$\begingroup\$ @user95 You should provide to us a block diagram that shows all the involved devices and their interconnections. Configuration = the minimum, where the noise is harmful. Accuracy=power or signal or grounding, is power =DC or AC, Voltage? How you have decided that it's the inverter, not a high current device that drains sharp pulses from DC? Do you know the difference between common mode noise and signal noise in wires? Do there already exist LC filters or ferrites to tame the noise? \$\endgroup\$ – user287001 Jan 30 '17 at 10:43
  • \$\begingroup\$ I've updated the question with a diagram. No other devices are connected, so it should only be the inverter affecting the DC. I realize now from what you're saying, that the noise I'm trying to identify must be common mode noise, not signal noise. Initially I thought it might be airborne noise, but it doesn't seem to be the case. Or perhaps that will result in another question later (I assume "proper" ground plane shielding of the inverter box is the main way to approach that problem?) \$\endgroup\$ – user95482301 Jan 30 '17 at 11:03
  • \$\begingroup\$ @user95 The suggestion for fix is available. Nothing special, quite the same as in PC power supplies. I have bought some low-cost power supplies from an international webshop. All filters were left out. No problem to keep the price incredibly low. Simply unusable stuff, altough voltages were ok. \$\endgroup\$ – user287001 Jan 31 '17 at 21:11
0
\$\begingroup\$

Is that a 50Hz inverter? The noise has a 50Hz pattern and the ringing can be caused by zero-crossing. If your battery discharges considerably fast you might have a fuge in your electrical network.

Where did you place the probe's ground in the first 4 graphs?

If you wanted to measure the battery voltage, the fifth graph is the right one: with GND connected to negative terminal of battery.

\$\endgroup\$
  • \$\begingroup\$ 50Hz pure-sine inverter, 230VAC. Battery is 12V 300Ah LiFePO4 and is not affected in any noticable way. A fuge? Or fuse? I've updated the question (see revision for specifics). The reason for placing the probe at the bus, is that that's where the problem is (causing problems for my MCU). \$\endgroup\$ – user95482301 Jan 30 '17 at 11:53

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.