I have a CM11A and I don't have any lock up problems - I have written in the past about layout of the X10 network for better reliability - but the bottom line is that the X10 protocol has very simple redundancy - it sends the data twice, and uses a very simple error check - but there are no two way communications that the message has been received - it is not really upto TCP kind of expectations - but hey it was conceived in the 1970's with the objective of cheap receivers.
I have seen occasions when the message hasn't gotten through - even in a well designed network - I think maybe about 2 to 3% of the time. Those things that I care about - I send the message twice a few minutes apart. Pool pump off, 2 minutes, pool pump off. For those things I can live with - garage lights off - sometimes (<2%) they just stay on until the next time around to turn them off for inactivity. I have my irrigation system running off of X10 with a home built module to receive the messages - once it gets a turn on signal - it sets a hardware safety timer to turn off after 80 minutes unless it receives another valid X10 command - because I can't afford a flood!
If you are controlling life and death things for your fish I would do one of two things:
1) Equip Misterhosue with some kind of sensor to know if a pump / light  is on or not - you can use other than X10 for feedback loop, or
2) Use some isolated relay board, or theatrical lighting system that is directly connected to the PC to control the vital things. (Since they are all close together you really don't need to send X10 protocol to remote places in your house).
If you have others - such as lights and what not - that if they stay on longer than programmed - and the fish just get tired because the day is long - maybe X10 is ok - still I would use dual send off, or  on commands on top of a reliable network.
On your electrical circuit - it sounds fine - check to see if the power bars have surge suppressors that can interfere. Also I would try to put the sendor (CM11A) in the middle of the receivers - rather than at one end. You could write a module that turns something on and off every 5 seconds for say 10 minutes - grab a beer and watch it for a missing pulse - that would give you an idea of reliability - however you might need to repeat with each hardware module. The hosue code that the module is on is irrelevant but different modules have their detector circuits tuned not that accurately so you might want to verify the important modules you need to trust.
Jim Serack
From: [] On Behalf Of Joey French
Sent: March 28, 2006 1:08 PM
Subject: Re: [mh] So, tell me about the CM11A...

Okay, so on the test system at home, I am running the cm11a with one lm14a and a full capacity of eleven three prong appliance modules. All of these modules are on one "bank" type power strip, 48" long with 12 vertically aligned outlets I got specifically to use the modules with. I have none of the two way modules (yet) sending anything back to the cm11a. I have entertained the notion of putting the MH box (soon to be an old headless hp evectra series mini-itx running linux) on a battery back up.  Could this configuration be made more foolproof? Maybe a noise filter on the "outlet bank" strip? I am really thinking that I may try to use the cm11a to control the one lm14a for the all-important % function, and use the cm17 for the other modules (opinions?). 
Keep in mind all of my modules are within four feet of each other, on the same circuit. Maybe I could attempt to isolate this circuit in some way? It seems like it will take some trial and error. Thanks again for the advice, everyone. If anyone has any more experiences, good or bad, I'd love to hear them.

Joey French