Monday, September 19, 2016

Welcome to Mega Automotive

by Joanna Hill

Today is Arnold’s first day at Mega Automotive. He is fresh out of college with an electrical engineering degree and is eager to meet his new boss, Mr. Buttsworth.

Mr. Buttsworth: Hello Arnold welcome to Mega Automotive. I’m sure you’re going to like it here. It’s a fast-paced environment with lots to learn and loads of good people to work with. We have an exciting new position for you it’s called a Component Electromagnetic Compatibility Engineer. Your job is to prevent electromagnetic compatibility failures at final vehicle testing. We design the vehicle and many of the components. Some of the components are designed and manufactured by suppliers. After all the parts have been designed they are assembled into prototype vehicles. These vehicles are then tested to make sure they can meet all sorts of requirements; one of these requirements is electromagnetic compatibility.

First, you must understand the specific EMC tests we perform on a vehicle are considered a company intellectual property and you are not allowed to disclose these requirements outside of the company.

Arnold, your first assignment will be to test an engine controller for EMC as a component. You are to develop a component level test plan to catch the EMC issues so they can be fixed early in the programs rather then at the end of a program. It is very expensive to make any last minute changes.

This is what an Engine Control Unit looks like, we call it an ECU.



Figure 1 - An Engine Control Unit (ECU) from a 1996 Chevrolet Beretta courtesy of Delco Electronics and Wikipedia.

Arnold: Cool, I love working on cars. It’ll be great fun to work on them for a living. What exactly is EMC?

Mr. Buttsworth: Our vehicles are put into a lot of different electromagnetic fields created by a lot of different devices. Many of these devices are “self-certified’. In fact, automobiles are also “self-certified”. That means that Mega Automotive is obligated to make sure a vehicle will work correctly when exposed to the fields that it may encounter and not interfere with any other device. The problem is that not all of the companies making things are as diligent as we are here at Mega Automotive.

For example, farms can have a water sprinkler called a Pivot that is driven by a three phase, 480 volt, Variable Frequency Drive. The VFD takes in 60 Hz AC and makes whatever frequency you want at 480 volts and 800 amps. The variable frequency is used to drive the pump motor of the sprinkler. The VFD does this by rectifying the three phases with a bank of diodes and capacitors. Then this DC voltage is pulse width modulated to feed a three phase motor that drives the water pump. If any of the case ground connections in the wet and dirty farm environment gets rusty, a PN junction is formed in the ground path. This splatters RF noise all over the spectrum. To make matters worse, the pump drives water into a 16 foot high quarter mile long water pipe.

The farmer thinks of this as an agricultural sized water sprinkler, we know it as a folded monopole antenna driven by lots of high frequency PWM noise. The Farmer has no idea he is creating huge electromagnetic fields that can interfere with all sorts of devices including vehicles. And to make matters worse, occasionally this quarter mile long short wave antenna with taps every 50 feet made up of an A-frame with two tractor tires to elevate the water pipe is parallel to the highway! Never the less our vehicles driving by within 100 feet of this antenna are not allowed to misbehave.

Your job is to come up with a component level EMC test to make sure that when the initial prototype is assembled and tested there will not be any issues with EMC.

Arnold: Okay, can I see the harness of wires that will be connected to the ECU? 

Mr. Buttsworth: No, it has not been designed yet.

Arnold: Okay, can I know how many harnesses will be connected to the ECU? 

Mr. Buttsworth: No, it has not been designed yet.

Arnold: Are there things connected to this unknown harness?

Mr. Buttsworth: Yes, we know what kind of sensors and actuators will be connected to the ECU and we have the documents that describe their function, but we don’t actually have any of then yet. They’re all in design as we speak.

Arnold: Okay, can I see the schematic of the ECU? 

Mr. Buttsworth: No, it has not been designed yet.

Arnold: What can I know about this thing?

Mr. Buttsworth: We know the pinout and we have a description of the engine control system. After the harness has been designed and routed on the engine and when we find out how the engine group is going to ground the ECU, it will be assemble into a prototype vehicle and it has to pass the vehicle EMC testing.

Let me show you to your desk. We will have a computer for you shortly. In the mean time please read these documents. And here is your list of mandatory safety and company training classes.

Barney: Hello Arnold, welcome to the neighborhood. I hear you’re the new EMC component engineer. I do the vehicle level EMC testing.

Arnold: Oh boy am I glad to meet you. I am to develop an EMC test to prevent issues from showing up in your EMC testing. But I have to do this without knowing how the thing is going to be wired up. This is impossible.

Barney: Actually our job is to write test plans that specifies the EMC test that need to be run by EMC technicians. The EMC equipment itself is very expensive and getting chamber time is really hard. So we write an EMC test plan, it gets approved, and then it is performed by the technicians.

All you have to do is come up with an idealized harness with a load box to simulate the sensors and actuators at the end of the harness. Make the harness about half a wavelength long in the FM band and pass on the test equipment design to the test equipment group. Tell them to test for stuck at one and zero for all loads.

It’s not really that hard if you don’t get too picky about simulating the vehicle. We have no idea what it will look like at this point, but never the less it has to pass my vehicle level EMC test to be sold. And we cannot delay the first day of production of the vehicle if it fails the EMC testing. As a result we almost always end up working nights and weekends to get it out the door in time.

Good luck and welcome to automotive. It’s not for everyone, but I love to see my vehicles on the show room floor or on the road. You know you had a part of making it happen. We have lunch down in the cafeteria. I’ll come by around noon to show you where it is. By the way, the hardest part of the job is staying awake while you read all of the documentation. See you at noon.

Epilogue

After many years of trial and error, Arnold came up with a technique to test modules for EMC before the vehicle design was complete. At the same time others were also working to solve the same problem thoughout the industry. But as you can expect, their solutions were all different. Some demanded the load box be in plastic, where others wanted metal. All of these empirical techniques work, sort of. But still at the end of a vehicle program, it must pass an EMC test.

Disclaimer, the individuals named in this article are fictional, any reference to actual people or companies is unintended. 



Joanna Hill holds a MSEE from Georgia Institute of Technology and a BSEE from Florida Institute of Technology. She turns the magic of EMC into the technology of EMC with consulting and classes globally. Her classes have demystified EMC in China, Mexico, Germany and the United States.

She is a member of the IEEE, IEEE EMC Society, SWE, SAE EMI Task Force, ISO TC22/SC3/WG3 USTAG, and CISPR/D USTAG. And a member of the IEEE EMC Society board of directors. She has worked as an engineer for 39 years always with an interest in fields and waves. Her LinkedIn page URL is:www.linkedin.com/in/JoannaEMC. She may be contacted at: joannaemc@icloud.com.

Wednesday, August 31, 2016

DC-DC Converter Noise Evaluation

By Kenneth Wyatt, Wyatt Technical Services LLC

More of my clients are starting to use small third-party DC-DC converters to provide the multitude of voltages required for today’s processor and DDR RAM ICs. While these are convenient to drop onto a circuit board, they can be quite a source of radiated and conducted emissions – especially those that switch in the MHz range.

I recently published an article on how these converter circuits can generate harmonic noise all the way up to 1 GHz, and above, severely compromising RF receiver sensitivity in the wireless telephone bands [1]. Kevin Slattery and Harry Skinner called this “Platform Interference”, in their book, Platform Interference in Wireless Systems – Models, Measurements, and Mitigation [2].

One example of this type of “drop-in” DC-DC converter is manufactured by Murata and we’ll use their model UWE-24/3-Q12, which is an “Eighth Brick” power supply that can take 9 to 36V and convert it to 24V at 3A (Figure 1). My client was using three of these converters on a product and was measuring a high level of radiated emissions, as well as observing broadband noise throughout his system all the way through 150 MHz.

 Figure 1 – The Murata UWE-24/3-Q12 DC-DC converter. The manufacturer recommends specific capacitors on the input and output and I’ve tack-soldered these on as shown.

If you read the manufacturer’s specification sheet, you’ll generally find that to “pass” EMI will require “additional filtering”, and this converter is no different. In this case, the additional filtering required to meet EMI limits was not described. I decided to bring one of these back to the lab and try some experiments to attempt to quiet the EMI.

To do this required some instrumentation. I used a Siglent Technologies SPD3303C three output power supply, a Tekbox Technologies TBOH01 5uH LISN, A Tekbox Self-Powered Active Load, and a Siglent Technologies SSA3032X spectrum analyzer. All this gear is available from the U.S. distributor, Saelig Electronics [3]. The active load was really handy, because I could dial in the exact load current I wanted…in this case 0.5 amps, to avoid cooling issues. I connected a couple of Fluke DMMs to monitor the output voltage and current (Figure 2). The spectrum analyzer picked off the conducted emissions via the LISN.

 Figure 2 – The test setup for evaluating the conducted emissions from the Murata DC-DC converter.

After trying some inductors and common-mode chokes I had on hand, I determined that simply placing a 100uH inductor in series with the input terminal was enough to quiet the emissions rather drastically (Figure 3).


Figure 3 – A 100 uH inductor was all that was required to dramatically reduce the conducted emissions.

Figure 4 shows the result. The yellow trace was the ambient (baseline) signal level. The red trace was unfiltered (no inductor) and the blue trace was with the inductor added in series with the input voltage to the converter.

The addition of a single inductor nearly reduced the conducted emissions down to the noise floor of the measurement.

Figure 4 – The results with the 100 uH inductor installed (blue trace) looking from 150 kHz to 150 MHz. The red trace is the unfiltered noise and the yellow trace was the ambient baseline noise. The display line is the approximate emissions limit.

Conclusion

Many EMI tests may be conducted right at the bench top. Evaluating various vendor products, such as DC-DC power supply converters, is always wise, prior to committing to a PC board design. It’s also wise to verify EMI performance as well as reading the “fine print” within the product specification sheet. Very often claimed EMI performance will require additional components.

References


[2] Slattery and Skinner,  Platform Interference in Wireless Systems – Models, Measurements, and Mitigation, https://www.amazon.com/Platform-Interference-Wireless-Systems-Measurement/dp/0323281451/ref=sr_1_2?ie=UTF8&qid=1450492921&sr=8-2&keywords=Platform+interference


[3] Saelig Electronics, http://www.saelig.com

Wednesday, August 24, 2016

EMC & Its Role In Reliability


EMC and Its Role In Reliability

by Ken Javor, EMC Compliance

Totally anecdotal, but this just happened to me.

My wife uses a powered wheelchair. A very high end, high quality chair. We were on vacation driving north and had just crossed the 45th parallel in the state of Michigan. The 45th parallel anywhere is halfway between the equator and the North Pole. In Michigan it’s also halfway to nowhere, and it was there my wife’s chair broke down.

The chair weighs close to 250 pounds empty, and is not much fun to move around less the electric motors, especially say up and down the handicap van’s ramp. The chair has enough smarts to give you a code when something fails. Maybe not the correct code, but a code. In this case, it told us the joystick module had gone bad, so we called a repair place and got one ordered to be sent to the place we were headed. I mentioned we were in the middle of nowhere, but we had pulled over in a cell phone-serviced spot just off I-75 because I was supposed to support a telecon with a customer and I had been looking for just such an oasis.

While looking at the chair, I saw a cable had been abraded and the sheathing was damaged and so were internal wires. I repaired those, and although that cable had nothing to do with the joystick module, once I got it repaired the fault code changed to bad power module cable. We haven’t priced the joystick module, because essentially price was no object with the chair stalled at the side of the road north of the 45th parallel, but I know it will be well north of $500 US. Seeing a new code unrelated to the joystick module did not lighten my mood.

Several days later after cooling our heels at what is called the “tip of the mitt”, the northern most point of Michigan’s lower peninsula, the new joystick module arrived and I installed it and got the same cable fault code as before and no life from the chair. So I removed all the chair plastic covers and traced the cable from the joystick module back to its other end, and I ended up at two plastic connectors with various wires emanating, and ferrite cores around them.

Now we are finally getting near the end of the shaggy dog story. Or maybe not the end, but the beginning of the end. Or maybe just the end of the beginning. But I digress.

One of those connectors or a wire/pin connection within had loosened and when I pushed on it, the chair woke up and worked. So clearly an intermittent. The chair is several years old, so this was not some sort of infant mortality problem, nor something to be caught by QA before shipping the chair. The ferrite cores I mentioned earlier are not the typical snap-on type, but they are actual toroids around which one could wind a choke or transformer. But they weren't so wound; they were just encircling the cables. Anyway, pushing on one of the two cores caused the intermittent to change states, so clearly over time that heavy core bouncing around the cable had caused something to open up.

So the lesson here is that adding EMI fixes can actually decrease reliability. Stating that in some circles where EMC is held to be a reliability issue is nigh on to heresy and can literally cause a violent argument. I could tell that story, but I won’t.

But it is in fact the case that adding parts decreases reliability – it’s just probability theory. Snapping a core around a cable at the last minute may be attractive and relatively risk-free on a piece of portable electronics, but when it is a vehicle and/or people’s lives depend on its functionality, then the modification must be subjected to review.

Related to this problem, I once worked a spacecraft that had a long RS-232 cable interface, for which they had chosen some very fast logic family, and the cable was electrically long relative to the transition time. There was considerable ringing on leading and trailing edges, and there was an expressed concern that such ringing could look like an extra bit. The fix here was to slow down the rise time (the data rate was OTOO 100 kbits/sec), but no one wanted to change anything, so I suggested one of these press-in connector filters (ee-seal by Quell), and that was demonstrated to work. But the ancient program manager, confronted with this news, asked the question: “Do capacitors fail open or closed?” The answer was “closed.” The program manager, the recipient of no telling how many whippings by Murphy, responded, “Not on my spacecraft,” and that was that.

Intelligence guided by experience.

Mr. Javor has worked in the EMC industry for over 30 years. He is a consultant to government and industry, runs a pre-compliance EMI testing facility, and curates the Museum of EMC Antiquities, a collection of radios and instruments that were important in the development of the discipline, as well as a library of important documentation. He can be reached at ken.javor@emccompliance.com.