Group Actions
Engagement
  • Author Author: Former Member
  • Date Created: Date Created
  • Views 2653 views
  • Likes 3 likes
  • Comments 14 comments
Related
Recommended

Spikes in the Ether – EMC and Compliance

Former Member

I was about to start talking about the design of the Pi but it seems that events have got ahead of me so instead I’ll talk about compliance.

 

It is common practice to build engineering samples/ dev boards without compliance testing simply to get programmers and developers up and running. In fact I cannot think of many companies who haven't done this.

 

It is normally not commercially sensible to start detailed certification of a design without first checking out that the feature set and functionality does not have gremlins. Even relatively minor design changes to support product level 'fixes' opens up the realistic possibility of having to re-certify again, possibly across multiple jurisdictions. The software system on Pi is sufficiently complex that a few 10's of boards are really not enough to prove the design, and as has been said elsewhere, most people buying the first boards were aware that there was still some development to do and they were very much helping us in that process. This was in line with the well trodden path of several other excellent community based developments.

 

The foundations decision to take on professional distribution partners has cause them and us considerable pain, but compared with the option of trying to raise a logistics operation out of the ground ourselves we would, even with the delays we would be much much further behind the curve.

 

Their compliance teams have been working diligently to try to formulate a response to the issues raised, and they have also had to take into account the diversity of Pi 'use cases' that they have. It had always been the foundations intention, and roadmap, to certify before the unit was put into the educational environment (both domestic and institutional), and this would also include a number of case options. Our partners have already helped us accelerate this activity.

 

Regrettably I am now minded to agree with their arguments based on sheer volume of interest and user demographics and some subtle arguments as to what constitutes a ‘finished’ product.

This now puts us all in the difficult position of having to complete the full certification in a short space of time, rather that the more leisurely plan we had in place for over the summer.  Anyone involved in compliance will know this is not a trivial issue and not something that we can absolutely guarantee to accomplish in a couple of weeks but we are going to give it our best shot. We have used solid ground planes, impedance matched traces and copious decoupling. A quick look around discussions on HDMI signal integrity and EMC will give pointers to the problems we were likely to face if you are interested.

 

Early pre-compliance work we conducted last week indicated that we were not simply going to be 'lucky' and everything is going to be perfect, but we could see quite clearly that the HDMI was a prime culprit, as expected. (Don’t you just hate being the lead engineer on a project with ‘issues’). However, work that Eben and myself did this week into the wee hours, (sorry Liz), brightened the horizon for me as we could see we were overdriving the HDMI and also had chosen a less than capable cable. (I'm not going to name and shame the supplier - should have bought it from e14!) The Broadcom chip is incredibly versatile and we have some definite options for ripping 10db or more off the key gremlin frequencies. Before you ask, yes, this should be enough.

 

We have assembled a talented team of engineers and supporters who are helping us work through the certification issues in association with our partners. We hope to have the critical results in by mid-week but that depends on a number of factors so in 'Pi time' could be towards the end of the week.

 

Taking the longer view, this needed to be done before we get these into schools so ultimately its all good and in the right direction.

 

Once we have the green light, rubber stamp or whatever you like to call it, our partners are primed and ready to do what we could never do – delivery of real volume. They have people working full time on resolving the world-wide logistics for both serious (and I mean serious) volume manufacture in double quick time and the subsequent distribution.

 

Cheers

 

Pete

 

image Slightly despondent – Yes 

image Gutted for you all - our supporters  – Definitely (the wait is agonising/agonizing)

image Belief in our goal - Unshakable.

Parents

  • Hi Pete, was the HDMI overdriving generating RF at unexpected frequencies because of clipping, or was it just that all the RF emissions from the HDMI connector were higher because of the higher output level?   I ask because I remember some pretty awful RF spectrum spreads that people got when they used custom firmware to boost the transmission power of their Linksys WRT54G access points - there it looked like clipping was generating a huge spray of RF at frequencies not seen when the amplifier was transmitting at the designed power.

    A good tight connector fit also seems to be pretty important to HDMI signal integrity...  (otherwise the impedance presented by the cable to the source is not as well matched?  I just hit the limit of my RF knowledge. :-)

Comment

  • Hi Pete, was the HDMI overdriving generating RF at unexpected frequencies because of clipping, or was it just that all the RF emissions from the HDMI connector were higher because of the higher output level?   I ask because I remember some pretty awful RF spectrum spreads that people got when they used custom firmware to boost the transmission power of their Linksys WRT54G access points - there it looked like clipping was generating a huge spray of RF at frequencies not seen when the amplifier was transmitting at the designed power.

    A good tight connector fit also seems to be pretty important to HDMI signal integrity...  (otherwise the impedance presented by the cable to the source is not as well matched?  I just hit the limit of my RF knowledge. :-)

Children
  • in reply to Former Member

    Jason - A good fit is important but people have seen more issues on the small form factor HDMI connectors (which I found surprising).  Cable quality is also an issue (the way the cores are screened and maintaining the characteristic impedance required.) Any discontinuity in the impedance can cause radiation.  One of the big issues is common mode noise that is not nulled out by the fact that HDMI is a differential signalling system - people have written whole books on this topic.  I'll talk about a PCB design trick we used to mitigate some of that later on in the blog when I get back on an even keel - right now I need to spend the rest of the day making sure the guys are all good to go for the next tests on Monday.

     

    Pete