Image of modification schematic (100k)
- Open the module by removing the screw between
the legs of the incoming power.
- Change C14 (2E225,blue) from 2.2uF/250V to 1uf/400V. Fix this one with silicon or epoxy glue.
- Change C10 (2E224, blue) from 0.22uF/250V to 0.1uF/400V.
- Change R25 from 22 Ohm 1/2W to 47 Ohm 1W.
- Change R12 from 100k 1/2W to 220k 1/2W.
- Change the VDR from 130V MOV to a MOV of 250V RMS. 230 and 240V users should use a 275V MOV.
- To improve reliability change the 68V/1W Zener diode to a 3W type. You could perhaps also use two 33V, 2 or 3W zeners in series. This diode fails often, even on 110V.
Location of components shown in picture: Credits: Alex Kurz, personal correspondence:
Full spec of TW523 and schematics (290k pdf file) source: X-10 web site
Modification to improve reliability
The following modification of the TW523 helps it comply
with X10's OWN PROTOCOL.
If you are:
1. Using the TW523 with the Lynx10 interface, Aegis HAI or any other
'smart' controller that tries to sense collisions and retransmits if it does
and
2. Are experiencing either delays in transmission (typical of HAI) or
Lynx10 returns E4 Timeout Error or Retransmit Failure errors -Then,
The problem is likely that the TW523 power supply is having a hard time supplying when
receiving its transmitted commands. The TW523 detects
commands it transmits, thusly helping the controller to verify that the transmission was
actually sent out. This happens on EVERY transmission. The
problem above reportedly happens on newer units that went through some 'value
engineering'. The manifestation is that the controllers don't receive
the commands they've sent, retry, fail again and thus the delays and errors.The situation
may be marginal and depends on the capacitive load of the
powerline.
HAI went out with a solution using two TW523 and a splitter cable, with one used for
transmission only, the other for receiving only. The problem is more acute with
controllers that comply fully with the protocol and transmit THREE times per half cycle to
cover all phases in a three-phase system. Another proposed solution was NOT to comply with
the X10 protocol and have the controller send out only once during a half cycle. This
controller would, of course, not work on three phase systems
- Open the TW523 case by loosening single screw between the two prongs and carefully pushing the two tabs at the bottom (RJ11 plug) side.
- The TW523 is a two PCB device, that is connected by a stiff Grey ribbon connector. Take care not to damage it. The area of interest is the PCB where the red LED is soldered. Carefully pull it out of the half-case. Two wires go through notches on the side of the PCB to the LED.
- CAREFULLY clip one wire going to the LED (any one). Make the cut such that it will be easy to solder back in case you'll need the LED indicator again in the future. One can also solder wires to the two cut ends and fit an external switch. I haven't - but I'm sure it can be done.
- Close the case, starting by carefully inserting the modified
PCB to the case, aligning the two halves from the tabs at the bottom and finishing with
tightening the screw.
Alternative
(source: Mark A. Day)
: If you need the LED, than you can do the modification by replacing the power supply
circuit filtering capacitor C12 (refer to the schematic)
from 100uF/25v to 470uF. This capacitor will store more charge necessary to complete the
operation. Note: I have not tried this. Source:
Now the TW523 poorly-engineered power supply have more of a
margin for doing some real work instead of blinking.