ESD, FCC Certification

Go To Last Post
15 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am working on a design that will need to undergo FCC Certification testing (first time certifying)

 

I only have 4 pins leaving my box: +12V, GND, and two I2C lines.

I plan on putting low capacitance unidirectional TVS diodes on the SCL and SDA line, and a bidirectional TVS on the +12V.

 

Am I missing anything critical?  Any recommendations?

Attachment(s): 

Science is not consensus. Science is numbers.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

FCC does not care diddly-squat about ESD or any other susceptibility. It is ONLY concerned about emissions, both radiated and conducted. CE, on the other hand, is concerned with all this.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Right.  Thanks for pointing that out --- I should have been more specific in my title.

 

 

ESD/FCC are the two things I am worried about in this design.  I know enough about ESD to be [somewhat] confident about my tvs usage.  Still not sure if I should do anything else to avoid radiation/FCC issues...

Science is not consensus. Science is numbers.

Last Edited: Fri. Jun 5, 2015 - 05:11 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I/O lines tend to be less critical for radiation. That is because the logic signals tend to be intermittent and low effective bit rate. 

 

What is more important is that power supply rail noise gets connected into those IO lines in the MCU and THAT can get radiated. Thus, it helps to have a well bypassed supply bus.

 

The power input is often the dominant radiator because of the high current spikes. I've seen spikes above 50mA with widths of a few ns. That really radiates broadband. Good low inductance bypass caps and ferrite beads help. But, those caps have to be properly placed with short trace lengths between the caps and the MCUs Vcc and ground pins.

 

I've been through more than 10 FCC certs.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
The power input is often the dominant radiator because of the high current spikes.

...

In testing, the device was indistinguishable from background noise when not attached to USB. When it was attached to USB, I ran into one issue which almost broke the threshold, but thankfully did not. I think this signal was caused by the fact that I had no ferrite beads on the USB power lines. This document gives an excellent description of how to avoid this problem:

...

by crwper

https://www.avrfreaks.net/comment/931857#comment-931857 in https://www.avrfreaks.net/forum/designing-pass-emi-testing-requirements

"Dare to be naïve." - Buckminster Fuller

Last Edited: Fri. Jun 5, 2015 - 06:21 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Why a bidirectional tvs on the 12V? A unidirectional would be more appropriate methinks.

It appears you have 12V on the same connector as i2c. You might want to add some protection for the 5V tvs in case 12V finds its way onto the i2c.

As for i/o not radiating - it is the edge rate that gets you rather than the bit rate. I once had a lcd that radiated at around 100MHz - the bit rate was nowhere near that.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is the edge rate that gets you rather than the bit rate

Not just a pretty face are you Kartman

 

I know I am a rare one but I think most people know that a perfect square wave has (beautifully distributed) infinite harmonics but its very educational to see how if we add a rise time to that square wave and make it a trapeze (like the real world) how it shifts the harmonics around, adding power to the harmonics that are related to those rise times

 

I will get my coat!

 

 

Last Edited: Fri. Jun 5, 2015 - 11:31 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

BOTH edge rate and bit rate get you.

 

Edge rate determines how high the harmonics go (and how strong). The more high harmonics there are (above, say 50MHz), the more problems you have with resonances on the traces and such.

 

For amplitude, a sort of RMS measure is used (they call it "quasi-peak"). You can have stronger spikes that are infrequent and get away with it. And, you can get blasted by lower amplitude spikes that happen every CPU clock cycle. 

 

Jim

 

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What piece of equipment is used for the test? A high end spectrum analyzer? So you can look for bad stuff before the Real Test?

Imagecraft compiler user

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The real test is a spectrum analyser with calibrated antennas in a shielded room.
You can use a less sophisticated setup to do a prescan to give an idea if there is anything particularly nasty.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, THE tool is a spectrum analyzer. Needs to be able to go from under 1MHz to 1GHz (or is it 3GHz, don't remember). Because no one antenna will do that full range, in the cert tests, it is broken into 3 or 4 sub-scans. For the stuff I worked on (8051-driven), the most challenging was the low end and from 25MHz to 1GHz. 

 

In the test lab, it is done in an anechoic chamber (shielded room) with calibrated antennas. Really big $$. The procedure was to collect lists of "peaks" where the detected peak exceeded some value based on a simple signal strength measure that ignores the duty cycle of the signal. Then, you go back to those peaks with the spectrum analyzer set for "quasi-peak". I gather that this was some sort of signal-processed measurement that was not part of a normal spectrum analyzer. I think that it gets added as some sort of "software module". This does what appears to be an RMS measurement over some time interval (thus, "averaged") with the spectrum analyzer set to be very narrow bandwidth and tuned to the signal in question.

 

For units that failed, we would use a "sniffer probe" with the spectrum analyzer to try to locate where the signal was originating. In those pre-smt days, the signals often came from loops in the circuit board, usually involving the power supply pins at diagonally opposite corners of the (8051) MCU (really hard to properly bypass) and loops in switch-mode power supplies. 

 

Nowadays, most of the modern SMPS chips come with a recommended layout to minimize such signals. It REALLY pays to pay attention to those recommended layouts. Likewise, Atmel has an app note on layouts and EMI. Read it and do what it says!

 

An FCC certification will cost you several thousand dollars, MINIMUM. It will cost more if you have problems. Most labs have a base cost with an additional per-hour if it takes diagnostic time and rescanning. You save big dollars if it makes it through the first time.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Another instrument that can be operated "before the Real Test" is a Mixed Domain Oscilloscope (MDO).

Practical tips for troubleshooting EMI issues

by Varun Merchant (Mainstream Technical Marketing Manager for Tektronix)

June 03, 2015

page 3 of 5

http://www.embedded.com/design/prototyping-and-development/4439605/3/Practical-tips-for-troubleshooting-EMI-issues

...

Examining the coincidence of EMI problems with electrical events is arguably the most time consuming process in EMI troubleshooting.

In the past, it has been very difficult to correlate information from spectrum analyzers and oscilloscopes in a synchronized way.

However, the introduction of the mixed domain oscilloscope (MDO) has changed that by providing synchronized time-correlated views and measurements.

...

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the test equipment suggestions... As a former analog engineer, I am know my way around a Spec An.  We usually went whole hog, and got the 40GHz models...

 

As far as connectors, the +12V/gnd and I2c will actually not be on the same one -- even though they are proximate on the schematic.

 

@Kartmann: why do you state that a unidirectional would be more appropriate for the +12V?  I wanted unidirectional on the I2C lines, because I could get a lot lower capacitance on those.

Science is not consensus. Science is numbers.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Unidirectional on +12V so it could not go negative.

 

Unidirectional is a bit  of a misnomer. Bidirectional means symmetrical around 0V for TVS. Unidirectional is more like a traditional zener diode. A higher voltage clamp in one direction and a low-voltage one in the other.

 

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Last Edited: Sat. Jun 6, 2015 - 10:26 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ka7ehk wrote:
The power input is often the dominant radiator because of the high current spikes. I've seen spikes above 50mA with widths of a few ns.
IIRC, XMEGA i/o risetime is 4ns; some of the 32-bit MCUs are a few ns.

Atmel Corporation

AVR040: EMC Design Considerations

http://www.atmel.com/images/doc1619.pdf (108kB)

...

(page 9)

4.6 Power Supply, Power Routing and Decoupling Capacitors
...

The current is drawn in very short spikes on the clock edges, and if I/O lines are switching, the spikes will be even higher.
The current pulses on the power supply lines can be several hundred mA if all eight I/O lines of an I/O port changes value at the same time.

If the I/O lines are not loaded, the pulse will only be a few ns.
...

"Dare to be naïve." - Buckminster Fuller