Trace Routing (IMPORTANT):
Regarding the Tip/Ring (T/R) traces:
All suggested spacing's are MINIMUM... i.e. METAL TO METAL,
in all directions - 360 degrees
1. Min trace width of T and R.... .010"
2. Min distance from metal of trace to ANYTHING outside of the
DAA.... 0.120"(3mm) or greater...
3. Min distance between Tip and Ring traces (within DAA
area)... 0.035" or greater
Conducted and Radiated Emissions:
Radiated Emissions
Your final product, with modem inside, will need to be tested
for Part 15(RFI emissions) There will be two types of
emissions tested... radiated and conducted.
We suggest the following to be placed in your mother board
layout to minimize possible radiated
emissions that can creep out the T/R lines.
In your artwork, place a .001uf/3kv from Tip to gnd and a
.001uf/3kv from Ring to ground. Ideally, tie them to your
metal front panel.
If that is not possible, find the strongest ground available.
This will help minimize radiation that may creep out the
T/R leads. The grounds should be as STRONG as possible...
these caps should be placed as close to the RJ-11 as possible.
Note: Add these caps in the artwork, but do not
load them. When you go to your Pt.15 testing, you
will then have the provisions, if needed, to add these caps.
If you don't need them to pass Pt.15, then do not load them in
production.... this is only a 'ace in the hole' that most
designs do not need.
If you need to load these caps, and you still have excessive
radiation coming from the T/R please give me a call to discuss
this... YOU NORMALLY DO NOT NEED TO GO HIGHER THAN .001uf, but
please contact me first.
Here are two physical versions of a .001uf/3KV caps.
SMT version:
Manufacture: NOVACAP
Part
Number: LS1808N102M302NT.
Through Hole version:
- rather large but you still might be interested.
Manufacture: CERAMITE
Part Number: 30TSD10.
These particular caps are suggestions, not mandatory.
If you have other sources, they would be fine also, just ensure
the voltage is at least 3KV.
Why 3KV min when the FCC Pt.68 Tests to 1500v PEAK?
- Actually you could get by with 1KV or 2KV caps, but to
keep a good margin of error, 3KV is recommended... 3KV are
physically about the same size and cost as 2KV.
Conducted emissions are RARELY caused by the modem.
Overwhelmingly, excessive conducted emissions are a result
of the hosts power supply.
Make sure your host power supply and power traces and planes are
substantive and well bypassed. You should add copious
amounts of HF (High frequency) bypass caps <to a strong
ground> on the power traces that feed the modem.
If
necessary, you may consider an inductor in series with the power
supply trace to the modem and add bypasses (to ground) on the
modem side.
There are usually three
fundamental reasons that cause excessive EMI/RFI... here they
are in order of priority.
- Grounding
- GROUNDING
- GROUNDING
and if that does not
help...
try more grounding
(get the point?)
IMPORTANT
- Do NOT put grounding plane (or Power Plane... or ANY
PLANE) in the areas of the DAA including traces and
components that are in the DAA area.... that is a
BARRIER AREA .... Components that must go into the DAA
and also require a ground, make the component straddle
the barrier... AGAIN, THIS IS IMPORTANT.
Still having emissions
problems?
If the radiation is propagating
out the RJ-11 cable and all else has failed, try adding a common
mode filter in series with Tip and Ring. The filter may be
added on your modem/mother board or actually in the RJ-11 cable
itself.
If you select the cable method,
you will need to ship the special cable<with filter> with every
product.
Here a link to a PCB mountable
Common Mode Filter which consists of 2 separate Inductors that
are to go one(1)each in series with Tip and and one in series
with Ring:
http://www.coilcraft.com/tttrf.cfm