. . .

"  It originally came with a 1000uF cap, which I have replaced with a TCS KA2, but all this appears to have done is let the sound continue for longer.   "

And this is why it is working in this manner. The cap you added is only driving the sound chip wired in this manner.

Browse this web link to figure out where to connect your cap so it is before any motor drive or sound related chip.

Needs to be wired to the bridge rectifier circuitry in most cases.

Marc

. . .

Maybe if I stuffed the link it would help you : -)

http://www.members.optusnet.com.au/mainnorth/alive.htm

BLI keep alive.

The capacitor you replaced is ONLY for the sound.  It has a very low reserve so anytime track power is interupted, the sound only lasts a fraction of a second.   You NEED to find the rectifier diodes by probing with a multimeter set to 20V DC.  look for 12V- 16V (track voltage) and note which end is positve and negative. This is where you can solder a "keep alive" so it will power the loco during a power interuption.  I will upload a photo of my BLI SW1500 that I did this to.

So stay tuned

Solder points Photo

In this photo I have highlighted the solder points that I used.  I found these by locating the four (4) rectifier diodes that convert the DCC track power to DC power that the decoder uses.  The yellow arrow points to the "negative" pole and the red arrow points to the positive pole.  The blue arrows highlight the other 2 diodes.  These 4 diodes make up the bridge rectifier for the decoder.  This is where I soldered my KA.  (Do this only if you are comfortable with electronics and wiring otherwise seek expert help local to you). The ONLY issue I have run into is during programming.  Blast programming a bunch of CV's will fail because the capacitors charge while programming.  Best to do all your programming BEFORE you commit to this circuit.  

Hi Nelson.  Your picture

Hi Nelson.  Your picture isn't showing for me (is there another link you can provide)?

As noted in my post

As I noted in my original post, I was reasonably confident that the + and - of the original cap pins (which I used for the KA2) were direct shorts  (e.g. directly connected) to the V+ and 'ground' rectifier diode ends, but it would certainly explain what I'm seeing if they are not! 

I have no issue with the soldering, but to confirm - you're KA keeps your Paragon2 equipped loco rolling in the absence of the DCC signal?  In other words it's possible to do what I'm looking to do, not an issue with the decoder software deciding to cut the motor when it loses the signal?

Thanks!

Clarifications

Dear SAR, Nelson,

Nelson, the link was a "share" URL, it did not end in a ".JPG" or other clear filename.extension path.

SAR, in order of appearance:

Anytime there's an polarised capacitor soldered to a PCB, there will be at-least some indication of the "+" terminal (sometimes a square solder pad rather than a circle). This however cannot be presumed to directly connect to the "+ve Supply" and "GND" supply-lines of the decoder.

If the decooder does not explicitly state as being "Keep Alive Equipped", then it's a default-safe-bet than the factory-connected caps will only keep the sound portion alive. (in the same way that, unless explicitly stated, always work on the basis that Resistors will need to be added for each function output...   ).

BLI would certainly suggest it's possible...

http://www.broadway-limited2.com/support/GoPacks%20and%20Keep%20Alives.pdf

Happy Modelling,
Aim to Improve,
Prof Klyzlr 

Thanks all, and especially

Thanks all, and especially thanks for that BLI link which I did not find previously.

For some reason, when I probed the board, it looks like the original cap pads (which I was reusing for the KA2) were directly connected (no impedance, direct connection) to the DC side of the bridge rectifier, but clearly they weren't as both Marc and Nelson said, and I must be touching something else that is so connected accidentally on the board while probing.

I am inferring this error on my part from 1. The BLI instructions didn't say 'remove the cap and solder your keep alive there' as that's presumably a lot less work and risky and 2. I moved the KA2 to the DC side of the bridge rectifier diodes and it works (which I have done in numerous other locos).   Somehow I'm missing some sort of gating circuitry between the sound and motor controls with my probing, but it's clearly there!

There's apparently still some cleverness going on with the decoder as it drops speed noticeably when it loses the DCC signal and then proceeds for now only a few seconds at a constant (but much lower speed) before exhausting the KA, and then the sound dies out about a second later.  It seems to still move at a satisfactory speed when moving slowly, so hopefully the KA will now do the expected job and let the loco transition any momentary dead spots!

I appreciate all the help!

BLI decoders form factor

BLI decoders tend to use a specific form factor (DCC boards) depending on the model.  Your NW2 MAY have a different board layout than the SW1500.  (Meaning components are in slightly diffent locations) but the process is the same.    

Nelson your picture share isn't working

Computer crashed during upload!

Joe I have to start over

Hopefully later will get it done.

Keep Alive on P2/3 board:

Here is where I soldered the "Go Pack" to an older Paragon board.

Minus wire to the minus side of the capacitor at the end of the board and the positive side to the second leg of the motor control IC as shown. I have also used Soundtraxx and TCS keep alive caps successfully, too.

Hope that helps, Ed

Clarification RE "gating" of power-sources/distribution

Dear SAR,

Just some minor clarification...

There is unlikely to be "active gating" of the power, rather a basic case of:
- different power requirements by different parts of the decoder
- localised power/volt regulation for different parts of the decoder

and thus, different "drop-out" minimum Volt/Current conditions for various parts of the decoder
- Core microprocessor
- Motor-control H-bridge
- Lighting power-transistor stages
- Audio Power amplifier
- etc etc

...typical of a KA with an operating voltage which is lower than the designed-operating-voltage of the decoders Vcc/GND rails...

(EG if a given decoder has +15VDC designed Vcc rails or local Volt-regulation stage,
but a given KA has a 12VDC protection Zener diode,
the KA will only charge-to, and deliver-when-called-on, 12VDC...)

Capacity of the KA, VS in-the-moment/situation power-draw configuration of the decoder
(mute sound, turn off some lighting, pull-less-train/less-hill,
all will change the "current/amperage demand" of the decoder in that moment,
and thus change the "KA uptime" observed)

KA is working, but with (presumed) lower-voltage being delivered to the Motor Control H-bridge transistors,
the RMS Voltage available at the "Motor Out" terminals is less than when "rail-powered",
despite the PWM drive signal/duty-cycle/timing not changing...

makes sense, KA Power is being fed almost-directly to the Motor H-bridge transistors.
once it drops below required Power (Volts x Amps), the motor can no-longer turn against the load

...because the decoder Power Amplifier/PWM "Audio Out" is kept alive that little bit longer by the Factory-installed Capacitor you were originally looking-at as the "keep alive position"...
(IE that Factory-installed capacitor is being fed by the 3rd-party KA under "no rail power" conditions,
its a "cascaded power distribution" situation...).

Given the evident VCC-supply-voltage difference between "rail powered" and "KA powered" conditions,
I'd suggest there likely is a _small_ difference in speed when transitioning Rail-power--> KA-Power under "crawling" conditions,

it's just that the difference between the 2 "speeds" is relatively tiny
(possibly not-evident to naked uncalibrated Mk1 eyeball)

as compared to a "full speed" Rail-power--> KA-power transition event...

Assuming you're not wilfully contravening the very nature of rail-borne power/control by trying to get the loco to jump multi-inch-long "dead frogs" and similar,
a KA which gives "a few seconds" uptime which be fine for the "fractions of a second" intermittent-pickup issues commonly reported by most modellers.
(Confirming 100% properly-working All-Wheel-Pickups,
track/wheel clean with Mineral Spirits,
and post-clean Graphite treatment,
you should be golden,
and the KA can relax into the role it should forfil, a "belt and braces" backup).

I hope this helps,

Happy Modelling,
Aim to Improve,
Prof Klyzlr