Respectfully, you must have something wrong, it definitely works as designed.
(OK - have not personally tested on a Rocket, but the circuit is the same as other Triumphs - I bench-tested original design with a simulator and it has been proven on a Bonneville which has same circuit)
First question - do you actually
have HIDs?
If not, the first solution is much simpler & perfectly adequate.
But if you do, then go through all again and ensure that you have followed implicitly.
1) Ensure the capacitor is connected with correct polarity, as is also case with the diodes of course. Note that the capacitor is marked for the
negative lead, as opposed to the positive being ID'd. It is imperative the diodes and capacitor are connected correctly.
All three diode cathodes (end with the white stripe) and the capacitor positive are all connected at the common node of the 86 terminal on the new relay.
2) Ensure you have all the terminals correctly ID'd on your new relay - what model of relay are you using?
Let's run through the operation of this system again:
1) ignition off, OEM starter relay is off - no input power at '3' so nothing connected through N/C terminal '4' to new relay; and new relay is also off; so no power out to headlight.
2) Ignition on - Starter relay remains off, but power is now connected through the N/C contact '4' to the common input '30' of the new relay; however new relay is 'off', so still no power out to headlight.
3) Pull Clutch, press 'Start' - Starter relay switches state, removes power from N/C '4' and transfers to N/O '5', the connection to the solenoid - starter runs & at the same time turns the new relay on, via the diode connection to '86'; however the power is now disconnected from the input (common)'30' so the headlights still remain off. At this time the capacitor is also being charged.
4) Release start button - the OEM starter relay de-energizes and switches state back to provide power through the N/C contact '4' out to the common on the new relay '30'; without the capacitor, the new relay would turn off at this time and even with power now on its input, it would be a 'race' to see if it could couple the power through to the output, to set the latch via the output coupled diode back into the coil; however the stored energy in the capacitor will delay the turn-off of the new relay - the relay is already on from stage 3, so the delay time that the capacitor would take to discharge through the relay coil, would enable the power through to the N/O contact, where it would then latch via the diode back to its coil.
While testing, you should be able to 'feel' the new relay click on, while you are starting (stage 3 above) - and then it should not drop when you release the button.
Most likely candidate for me is suspect the capacitor is hooked up backwards. Also ensure that the 85 terminal is connected to a good common 'ground' point (another system 'black' wire typically - do NOT connect to the frame! - the black wire on the Turn Signal Module which is directly next to the Starter Relay is a good convenient candidate source)
Good Luck, it should come together as laid out.
p.s. Here it is executed on a 675 Daytona - should be the same for the Rocket -
http://www.triumph675.net/forum/showpost.php?p=728568&postcount=12