A month or two ago, I got a linear, and immediately afterward started setting our hardwired smoke alaems off.
My first attempt was cheap inline split ferrites from Amazon. No luck.
Second, after some research, I thought it might be differential mode noise. So I made up delta networks of bypass caps, three caps per detector, and put them between the neutral, hot, and interconnect lines. No luck.
Finally, with TX5K running and 4 AM my only clear shot at working 'em on 40, I did some research. Turns out that the interconnect line between the detectors, the one that sets them screaming in unison, is only a 9V DC signal. With 1500w being thrown into the air and my house wiring about 20 meters long, it occurred to me that it could be common mode noise on the interconnect line.
My third attempt was therefore L networks on the interconnect line at every detector. I reused the 0.47 mfd, 1000v caps that I'd bought from Newark ELectronics in my second attempt. The caps went between the interconnect line and the ground wire. For the inductors, I took some big split ferrites, Type 31 mix, that I'd gotten from DX Engineering, and threaded six turns of hookup wire through them. These went between the cap and the detector, so any induced QRM would be steered through the cap and away from the detector. (For the prototype, I['d built a pi network, but this was so bulky that I left out the first cap for the others.)
The detector nearest the shack got a big ferrite around the power leads, just for good measure.
So far it's worked. Got TX5K on 40 tonight, which sweeps 80-10 in CW.
And not a peep out of the detectors. A really thorough test of the setup (key down for at least 30 seconds) will have to wait until the XYL is out of the house, or at least awake.
For those who are having the same issue, here is where I got the parts.
. They are quite bulky for the application; since I didn't go to the power lines, I'd use a physically smaller cap with a lower voltage rating if I was doing this from scratch.
Ferrite beads: http://www.dxengineering.com/parts/dxe-csb-275p
These are also pretty large, but I see no way of miniaturizing here. If you substitute this part, make sure to get one made from Type 31 ferrite mix. I won't mention the cheapies because they didn't do the job; I think they're likely made from a different mix that's only good at VHF.
I'd appreciate hearing from anyone else who has had, and successfully attacked, this problem. As I said, I don't want to declare total victory just yet.