Radiative Emissions 3

[ngaynor314]

I'm working on a book scanner product which contains several PCBs and a variety of cables and connectors. We recently submitted our product for ETL testing and failed the radiative emissions portion. I am a Mechanical Designer so my electrical knowledge is limited. We have incorporated shielded wires, enclosures for the boards, and verified proper grounding of all components. We're about 4.5dB over limits at 840Mhz and 2.0dB over limits at 660Mhz. Any suggestions to troubleshooting the EMI?

 

[MacGyverS2000]

No offense intended, but sending in a ME to resolve an EE problem is beating your head against a wall. Anything we suggest right now will be wide-dispersal shots on what to do, and in the end they will be complete guesses. Get an EE to properly investigate where the noise is coming from, and then resolving it becomes a significantly easier job.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[ngaynor314]

Believe me, I have been beating my head against a wall! I work for a pretty small company and we do not have an EE so it's been put in my lap. I understand resolving this quickly is out of my realm, so any suggestions on where we can find a consulting EE? We're located about 2.5 hours away from Minneapolis and the only services I could find online were pertaining to Civil Electrical Engineering.

Just a random question, I have been learning as much as I can about radiative emissions and found some information regarding component shielding which discusses "redundant shielding for noisy areas". Is this feasable?

Thanks for the feedback!

 

[MacGyverS2000]

A number of good engineers here do consulting... hopefully they'll chime in soon. biff44 owns MaguffinMicrowave.com, there's Higgler, VE1BLL... I believe these guys all work in the radiative comm field, so if they can't help, they may be able to point you to someone who can.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

MacGyverS2000 makes a good point, and I agree with him fully.

One should consider the cost of each test run (thousands?), plus the cost of delays (tens of thousands?), and balance that against the price of bringing in an expert to sort it out.

On the other hand, 4.5 and 2.0 dB are so close (on the log scale)! By way of comparison: on a different thread on a related forum, we had one poor fellow with a product that was as much as 26dB over the limit.

Are your enclosures shielded, or plastic? Do the shielded cables have correct 360° shield termination at the enclosure? Did the product pass any Conducted Emission testing?

 

[ngaynor314]

First of all, thanks for the prompt feedback MacGyverS2000 and VE1BLL. Myself and the owner of the company realize we need to figure out the most efficient way to resolve this, even if that means hiring a consultant, so that dollars are not wasted on delays and testing. That said, we are real close and that's why I'm scratching my head at this point.

The machine was originally designed by a contracted freelance designer, I've been trying to figure out some of his logic and correct problematic areas. Originally the machine had little shielding and improper grounding of all components. A few areas that raised flags to me were the main PCB layout, which had no shield, and the camera CCD (PCB), where the I/O power connection is right next to a data cable connection on the board. I added a sheet metal enclosure (grounded) to the PCB layout which contained a CPU board, servo controller board, and a camera board. We also added wrap around self adhesive wire shielding (grounded) to the power wires that run to the camera CCD (PCB) and a grounded sheet metal shield over the board. This all helped (we were about 15dB over @ 75Mhz on the first test) but we're still a little off at higher frequencies.

What bothers me is how the geometry of the machine may affect the far field test results (how do we analyze/manage this) and how much consideration for EMI/EMC testing was taken into account during the original design. I'm concerned we may have to redesign some of the electronic layout. Like I said, I'm not an EE but I'm not sure how much experience the original designer has had with EMI/EMC testing.

Any additional feedback would be great! Thanks again!

 

[ngaynor314]

Just to follow up...

The company I work for would consider hiring an EE that has experience with this type of work, meaning EMI/EMC product testing. I haven't had any luck finding someone that fits these qualifications at this point.

Also, we did pass the conductive emissions test (7.6dB below limit) after an inline filter was installed between the power cord connection on the back of the machine and the power supply.

I would really like to find an EE to work with that can analyze our test results along with the overall design and electrical component layout of the machine and provide some sound advice on how we should go about correcting these noisy areas so the next time we have it tested we will be confident in success.

 

[MacGyverS2000]

Motors (and control circuitry) can spew out a lot of EMI. Self-adhesive wrap ground screams in-the-field band-aid to me... use a properly shielded cable. Shields surrounding the entire board do little if you have huge gaping holes and wires coming out that acts as antennas.

I highly suggest getting that EE in sooner rather than later as you guys are throwing darts at a board and you can't be sure where the board is even located. A guy worth his salt would probably be able to calm most EMI by many dB just by looking at how the thing is currently designed and suggesting changes, no test equipment necessary.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[ngaynor314]

Thanks again for the feedback. You're correct, some of the things I have applied so far are basically quick fixes that have been a shot in the dark. Seeing as though my expertise is not EE work I was hoping a little common sense and basic logic would prevail. Unfortunately we fell just short and passing the EMI/EMC tests has proven far too complex for my basic approach.

Again, any help pointing me in the right direction to find a consulting EE with this kind of expertise would be extremely helpful.

Thanks!

 

[VEBill]

I found this gentleman using some carefully-selected Googly keywords:

http://www.linkedin.com/pub/david-zimmerman/10/531/535

David Zimmerman
EMC Consultant at Spectrum EMC Consulting
Location Greater Minneapolis-St. Paul Area
Industry Aviation & Aerospace

(I know nothing about him.)

 

[VEBill]

Here's another one:

HOOLIHAN EMC CONSULTING
Address: 32515 NOTTINGHAM CT, LINDSTROM, MN 55045-9102
Telephone: (651) 213-0966

 

[ngaynor314]

Thanks much. I'll look into this and see if they may be able to help us out.

 

[twalter]

Side note on EMI issues, we tell our customers "test often, and test frequenty" as Ill often first here "We are at a test facility and failing.... ". Sigh.

Quick Example... those darn little mag jacks (combined magnetic + RJ45 connectors) drive me nuts! There area some pretty good ones, but there area ton of junk ones (they are a few cents cheaper for a reason). If you have the room using a seperate magnetic and jack does help the EMI issues.

Other one I ran into was, when reviewing a schematic, was no ferrite beads were installed. When asked about it, the Engineer said they told him to save 0.02c per board as they aren't needed. Hmm, small volume product (5K boards), so I was curious if the $100 savings offset the board spin to add them back on.

Other favorite one, when I was designing a dashboard controller, was to use a 21.8Mhz XTAL. Later someone that that was stupid, as 20Mhz was a cheaper XTAL. I about died laughing about a purchasing agent saving a few cents.... Sigh.

 

[2dye4]

As a cheap shot in the dark get some clamp around ferrites from Digi Key or Newark and have them for the next test. You may be able to add them and achieve a pass.

Also look at the metal pieces with either wires or components mounted to them. These metal pieces should be bonded together with the entire chassis. If painted parts are screwed together the metal may be isolated and forming a good antenna.

Wire routing. Wires should be routed near well grounded metal surfaces not floating in the air.

When contracting electronic design for consumer sales always include an EMC clause in there so the original designer has to keep this in mind throughout the process.

Good luck.

 

[DHambley]

The solution I come up with for most EMI problems is to focus on the source of the noise instead of adding band-aids down stream (like huge shield beads). It is almost always related to a problem of the PCB layout or, gross mistakes by the EE like not designing in a common mode filter.
A re-layout of the board sometimes may not make economic sense. Once I helped a large computer manufacturer solve an EMI problem. Within 1 hour of troubleshooting it became apparent that the circuit layout in the power supply was causing the problem. My client however, liked that PS manufacturer and didn't want to force them to deliver a product which meets FCC specs. The solution was huge ferrite beads on the power line. The new "small" product became bigger.

 

[ngaynor314]

Based on what I witnessed during testing the areas that produced the most noise were around the PCBs (the two that seemed the worst pertain to the camera). The test engineer used a "sniffer probe" and those areas seemed to produce the highest levels on the spectrum analyzer. Problem is the boards we're using are a standard component that we purchase from a image system company, so redesigning the board is not an option. I'm pretty sure the original engineer that developed this product did not incorporate a common mode filter in the design. Is this something we could consider that would be a proper solution rather than getting carried away with shielding and containment?

Thanks for all the feedback!

 

[MacGyverS2000]

The CCD board will have some high-speed digital logic to shift out the photon readings... that can be quite noisy. It may make sense to shield that entire piece as one and make sure any lines to/from are properly shielded.

Dan - Owner

http://www.Hi-TecDesigns.com