[RFI] ADSL2+ modem crashing solved using power line filter and choke

[Cortland Richmond]

On 6/2/2016 10:37 PM, Don Kirk wrote:
> I then decided to try some filtering on the power line cord that feeds the power supply for the new ADSL2+ modem, and to my surprise my new modem so far has been functioning fine

In a firm far away and long ago, I was the R&D EMC design engineer.  We 
were selling subscriber line cards to telco's all over the United States 
– our line cards were touted as handling 90% of all telephone calls in 
the country.  And we had come up with a dual-purpose line card that 
would add ADSL service without requiring any further equipment upgrades 
at the central office; it looked like it would be a really hot seller.

We got through regulatory compliance tests just fine, but the customer 
telco had problems keeping them online at longer reach distances (feet 
of wire to the receiver) and I had to go with the team that visited that 
customer.

I took along a spectrum analyzer, a *personal *loop antenna, and an 
oscilloscope because I had a hunch what was going wrong.

Sure enough, when we set up in the labs own central office (for testing 
only) I noticed that disconnects happened when the 20-Hertz ringing 
voltage relays were energizing, and connected the loop antenna (20Hz-5 
MHz) directly to the scope to see what was going on.

Bingo!  There was a huge magnetic field spike every time we lost sync, 
coinciding with line card 20 Hz relays switching.  It wasn't our 
fault.   When they moved the customer modems out of the central office 
and away from all of those busy relays, synchronization was preserved.

However… this was a hint there would be problems later on, and so there 
were.  When you got to 17,000 feet (of wire) from the central office 
there was very little tolerance left for broadband noise at the modem, 
which would drop sync with his little as one or two micro bolts of RF noise.

I prevailed on the modem vendor to send us a couple of his units so I 
could see how they might be fixed, and I found that *adding a better 
common-mode filter between the board electronics and the wall-wart power 
connection* would do the trick.

I'm afraid this rather impressed my manager,  who found me in our own 
test lab central office holding the modem board in one hand while I 
operated the spectrum analyzer with the other; "Cortland!"  he yelled, 
"Cortland!".

"What?"

"Cortland!"

The modem board was on fire in my hand; a solder bridge on my makeshift 
fix drew enough current to ignite the phenolic. Luckily, I didn't set 
anything else on fire, it was a two-layer board, and I was able to 
remove the charred portion with a coping saw...

"Hot" seller, indeed.   It was replaced with a more noise resistant ADSL 
technology.

Cortland Richmond