EFT Phenomenon
Recently I tested a medical device for EFT susceptibility. The requirement is that the medical device must be immune to EFT bursts generated per IEC 61000-4-4. Per this standard EFT bursts are applied in four steps. For level 4, steps 1 and 2 are +/-2kV at 5kHz, and steps 3 and 4 are +/-2kV at 100kHz. For each step, these bursts are applied at 300ms interval for 60 seconds.
I took the DUT to a testing house for evaluation. EFT bursts were applied onto the AC line, specifically Line plus Neutral to ground. The behavior of the DUT was as follows:
- Physiological signals displayed were distorted making it difficult to read the data.
- The screen was flickering.
- LEDs on the front panel were turning on and off.
This behavior was repeatable with each EFT test and for each EFT test step.
Troubleshooting EFT
In order to troubleshoot the above described behavior of the DUT to EFT it was necessary to have an EFT test set-up. Troubleshooting in a test house is very expensive as the cost of immunity tests are in the $250-$300 per hour. In addition to the cost, one has to bring troubleshooting equipment with them to the test house.
A better solution is to rent an EFT/Surge generator if you don’t own one. There are many vendors that carry EFT/Surge generators and can lease them on a week to week or a month to month basis. My preferred location for getting test equipment very quickly is www.TheEMCShop.com. Through this shop I rented the Haefely PSURGE 8000/PIM 110 Ring Wave Impulse Generator for $875/month. It arrived at our lab within two days of ordering. The equipment came with detailed manual and it was straight forward to set up.
When troubleshooting EMC issues, it is best to start with a troubleshooting plan. Taking a moment and studying the schematic of the problem areas and understanding the circuit is very important. In my case I was dealing with bursts applied to the AC line. I came up with a diagram of the power distribution system. The diagram below shows an example.
Create a Plan
Creating a plan in a way creates an approach. Having an approach requires a starting point. A good way to start is to have a hypothesis. Once you start troubleshooting you either prove the hypothesis or you reject it.
I based my approach for EFT troubleshooting on the fact that bursts were injected directly into the line. Looking at the diagram above I could see how bursts could get through isolation barriers via stray capacitance. I thought at first that bursts were getting into the physiological signal processing circuits. So I decided to take the approach of problem area to source path investigation. Based on this approach I came up with the following hypothesis:
- EFT bursts are capacitively coupling through the power supply chain and affecting physiological signals, processors, etc.
Instead of looking at the above diagram as the power distribution block diagram I looked at it as energy travel diagram. Starting from the problem areas, I studied the path of energy travel in order to find the point of distribution. The solution or the fix would be implemented at the distribution point.
Using an oscilloscope I started probing voltage rails that supplied power to the problem areas observed during EFT tests. Keep in mind that because of high dV/dt created by EFT bursts oscilloscopes can be affected and give erroneous results. I would recommend using a floating oscilloscope (battery powered). With the EFT/Surge generator set to +2kV at 5kHz (with EFT burst duration of 15ms) and applying bursts on LN (Line Neutral) to ground I monitored voltages on all physiological signal processing circuits and I saw that the voltages were dropping by about 70% for 15ms at the rate of EFT bursts. Going down the power supply chain showed that VBUS was also dropping. Finally, the output of the AC/DC open frame switching power supply was also dropping for 15ms. That explained why the DUT was affected by EFT. I rejected the hypothesis. The mystery was the AC/DC power supply. Why was the output voltage dropping?
The Root Cause
EFT bursts applied on the AC line, which is the primary of the AC/DC transformer were common mode coupling onto the secondary due to the capacitance between the primary winding of the transformer and the secondary winding and due to stray capacitance on the board. When bursts got onto the secondary side, one of the effects they had was a ground bounce. Given that the power supply is a switching power supply, the microcontroller driving a MOSFET on the DC output was getting into an unknown state when its reference point was experiencing a large dV/dt due to the ground bounce. As a result all pins of the chip were becoming High Z and the MOSFET stopped switching on and the output voltage dropped.
Choosing an Off the Shelf Power Supply
I looked through the specification documentation for the power supply, which was an off the shelf module built by a company in China. The manufacturer claimed compliance to the IEC 60601-1-2 edition 4, which requires level 4 of EFT testing. Not believing the manufacturer’s claim I reviewed their EMC test report for the power supply. The manufacturer tested the power supply by itself with a resistive load. The power supply was attached to an EFT/Surge generator and the output voltage was monitored. When I reviewed the test set-up I found that the power supply output voltage was monitored using a hand held Digital Millimeter (DMM). That was very strange to me. DMMs do not have the ability to capture fast transient events. Sure enough when I repeated the power supply standalone tests per the manufacturer’s EMC report, the output of the power supply remained at correct voltage when monitored with the DMM. When I used an oscilloscope I saw the voltage drop. So the power supply is not compliant.
Suggestions on Choosing an off the Shelf Power Supply
- Do not trust any claims that manufacturers make.
- Always request the EMC test plan, protocol, and report that shows compliance.
- Carefully review test set-ups and methods.
- Test the power supply for EFT and other AC line tests prior to commitment.
