Best Of Retrofitting a Kato light board with a LokPilot 5 Micro DCC

[GaryHinshaw]

I have a sizeable fleet of Kato diesels with TCS decoders and a small fleet of LokSound equipped locos (Kato and Scale Trains) and I've been much more impressed with the motor control on the Lok units, so I've been looking into converting the TCS units.  At the same time, I am still underwhelmed by N scale sound, so I'm not really looking to convert to LokSound en masse (yet).  Enter the sound-free LokPilot -- there are no drop-in-place models available for Kato, but the LokPilot 5 Micro is tiny and inexpensive (~$30 USD), so I decided to try a Kato/LokPilot conversion.

@Scottl suggested trying to use the guts of a Kato light board as the basis for a drop-in-place conversion, so here is my implementation of that idea.  I'm converting a Gevo with a 17632 mechanism and light board, but the Kato light board layouts are all pretty similar, so the same approach should work for a variety of models.  For the decoder, I'm starting with a LokPilot 5 Micro DCC 59820 which has an NEM connector on it, but any style of connector will do, since I cut it off and direct wire to the board.  (ESU does not sell a LokPilot Micro without some form of connector.)  The conversion is slightly involved, but pretty straightforward, and the result is one of the smoothest-running and quietest units I have. 

1. The first step is to unsolder the motor clips, the front LED, and the rectifying diode next to the front LED:



It's not strictly necessary to remove the motor clips, but it does make it easier to cut the traces in the next step.  Note that when I removed the LED and diode, a short bit of the trace came off with them (the white segment in the photo).  I had to bridge this with a bit of wire when I re-installed the front LED.

2. The next step is to cut the traces in 8 places, indicated by the red segments in the photo above.  I made these cuts with a Dremel and a cutting wheel.  The purpose here is to isolate the motor and light outputs from the track power.   A few of the cuts are only needed to isolate the segments indicated by the black arrows from any of the decoder outputs.  These segments can touch one or the other frame sections and if they are not isolated you could inadvertently apply track power to one or more of the decoder outputs and fry it.  (Ask me how I know...)

3. The next step is to re-install the motor leads and the front LED, and to identify which decoder wire goes to which trace on the board.  Note that the polarity of the front LED had to be reversed to take advantage of the common-anode wiring of the decoder light outputs.  I did this by simply flipping it over and soldering it to the top of the board.  (You can also see the wire bridge I made to bridge the missing segment of trace.)



4. The next step is to wire the decoder per the scheme shown above.   I placed the decoder under the rear LED leads and ran the wires up between the leads, but other arrangements are possible since the decoder is so small (It comes wrapped in shrink wrap, so no worries about shorting on the LED leads.):



I don't have an immediate use for the remaining wires, but I didn't want to cut them off, so I routed them forward and wrapped them with Kapton tape to keep them out of harm's way:



The result is a drop-in-place decoder board that requires no frame or shell mods:



The stock LEDs are golden white and give pretty good illumination of the light tubes, but I am looking into replacing them with sunny whites if I can find a suitable bulk supply of them.

-gfh

[samusi01]

Good tips, thanks. The first ones I did this to are the SMD boards (Kato p/n 3066 and 17681) and I’d not considered repurposing the SMD resistor like you did, instead removing all the SMD resistors on the board and adding in-line resistors between the board and decoder. Unfortunately I took this mindset of removing everything onto the p/n 17632 board when I did those conversions. I’ll have to change my notes for that board to adopt your method.

As an addendum, at the recent NMRA national show I did bring this up to the ESU folks and the German guy seemed interested in the concept of a Kato form factor LokPilot but I don’t know if it was enough of a bug in his ear to actually generate anything. He did say that it’d be fairly straightforward to do.

[carlso]

Thanks for sharing Gary, very nice work.

I did the same install with the ESU 59820 in a KATO SD-90 using the original light board. Very easy, clean and professional job. That SD-90 is one of my best performing diesels bar none. I have not looked lately to see what Bryan is getting for that decoder today, but IMHO they are the best buy in decoders.

[jagged ben]

...  For the decoder, I'm starting with a LokPilot 5 Micro DCC 59820 which has an NEM connector on it, but any style of connector will do, since I cut it off and direct wire to the board.  (ESU does not sell a LokPilot Micro without some form of connector.) ...

-gfh

I was wondering about this, thanks for confirming. 

My feelings about switching to ESU are the same.  Starting with some 6-pin FVM locos but planning to do similar stuff as what you've shown here with most everything else.

[peteski]

Pretty nifty conversion Gary, however I would not recommend reusing that Kato's 560 ohm resistor.  Kato uses such a low value resistor on DC locos because they rarely get run at full 12V (since that would make the high-geared locos run at over 200 scale MPH).  Kato probably estimates most modelers will run their models at 9V or less.

Customarily a 1k resistors is used in series with white LEDs in DCC installs.   At 12V (which is usually at the "blue" positive output on decoders) the LEDs current will be (12-3)/560= 0.016A (16mA). I suspect the LED will is very bright.  The T-1 size LEDs Kato uses can withstand such currents

Also, there are no diodes  on the Kato board, except for those 2 LEDs.  That component  you mentioned is a small multi-layer ceramic capacitor which is there to prevent the LED in the opposite direction of travel from flickering.  When a DC loco momentarily looses contact with rails (at medium to high seeds) while the motor is still coasting, the motor produces voltage (BEMF) which is reverse polarity to the track voltage.  That reverse voltage supplied to the very sensitive white LED, causing it to flicker.  That capacitor shunts those voltage spikes, so the LED does not flicker.

[GaryHinshaw]

Thanks for the clarification about the capacitor @peteski, that makes more sense. 

Regarding the resistor, I'm going to stick with it.  I believe these LEDs are rated up to 20 mA and the LokPilot blue lead is only 11 V, so (11-3)/560 gives only 14 mA.  In fact, I wish they were a bit brighter!  ESU recommends 470 Ohms for LEDs.

Here is another wiring example, using a fried decoder (the one alluded to above) and a scrap board with those awful amber LEDs.  I turned the decoder over and taped it to the rear LED leads, snipped off the pink and teal leads, and routed only the central red, black, white, and yellow wires between the LED leads.  This gives a slightly neater package.  (Still not sure if I'm going to keep the green and violet wires, going forward.)







In theory there is plenty of room for a TCS keep alive (KA-N1) in front of the decoder.  I've ordered a few to test.  If I understand correctly, these can be hooked up to the blue lead and the ground pad.  Alas, the LokPilot ground pad does not have a lead soldered to it, and I have been singularly unsuccessful soldering a wire to it, no matter what I try!  I suspect the decoder has a coating (Kryptonite?) on it to protect the other leads and that this is interfering with soldering to the ground pad.  Has anyone successfully connected anything to this ground pad?  (It's the pad below the violet wire in the shot below.)



-gfh

[peteski]

Regarding the resistor, I'm going to stick with it.  I believe these LEDs are rated up to 20 mA and the LokPilot blue lead is only 11 V, so (11-3)/560 gives only 14 mA.  In fact, I wish they were a bit brighter!  ESU recommends 470 Ohms for LEDs.
. . .
Alas, the LokPilot ground pad does not have a lead soldered to it, and I have been singularly unsuccessful soldering a wire to it, no matter what I try!  I suspect the decoder has a coating (Kryptonite?) on it to protect the other leads and that this is interfering with soldering to the ground pad.  Has anyone successfully connected anything to this ground pad?  (It's the pad below the violet wire in the shot below.)
-gfh

Yes, ESU states "470 ohm to 2.2 k). Yes, with 470 ihm resistor and 12V blue lead voltage  white LED will conduct around 19mA.  As you mentioned, that is within specs for the larger T-1 (3mm) size LEDs.  The smaller SMD LEDs have lower current ratings. Some as low as 5mA.

As for the ground pad, it is not Kryptonite.  The explanation is much more mundane.    ESU decoders use multilayer PC boards. The tiny ground pad is internally connected to the decoder's ground plane, which is one of the inner layers in the PC board.  It basically is a large copper area,  and it acts as a heat sink when trying to solder to the tiny pad.  Plus the lead-free solder used in electronic circuits has a higher melting temperature than the old lead-based electronic solder.

If your iron has a clean tip tinned with solder, and you use tiny bit of extra flux (it acts as a heat conductor, you should be able to solder to that pad.  You will have to heat it a bit longer than you are used to, so the pad heats up even with that internal heat sink connected to it.  I set my iron for 700 deg. F and I have successfully soldered wires to the ground pad.  Heating the entire decoder with a hair dryer will reduce the temperature differential and should make soldering easier (but I have not found that necessary).

What I do is to first remove the original lead-free solder blob from the pad. I take some solder wick braid, dip the end in paste flux, then using the iron (again, with a clean tip) melt and wick the solder.  Then I apply some standard electronic solder - the  63/37 Sn-Pb type. With a lower melting point, soldering a wire will be easier. Again, add some paste flux to  the solder on the pad before soldering the wire.  Even with rosin core solder, the extra flux is very helpful. Did I also mention using clean and tinned iron's tip? 

[GaryHinshaw]

Good info about the ground pad Peteski, thanks.  I was very hesitant to apply heat for too long, but knowing that the pad is sunk to the whole backplane makes me much less nervous about it.  I still wish they had added a lead to it though...

Clean tips ordered.  I've done a fair bit of soldering lately and my last tip is getting pretty shot.  I thought I had a stash of replacement tips, but apparently I ran out.   

[Jim Starbuck]

Thanks for the clarification about the capacitor @peteski, that makes more sense. 

Regarding the resistor, I'm going to stick with it.  I believe these LEDs are rated up to 20 mA and the LokPilot blue lead is only 11 V, so (11-3)/560 gives only 14 mA.  In fact, I wish they were a bit brighter!  ESU recommends 470 Ohms for LEDs.

Here is another wiring example, using a fried decoder (the one alluded to above) and a scrap board with those awful amber LEDs.  I turned the decoder over and taped it to the rear LED leads, snipped off the pink and teal leads, and routed only the central red, black, white, and yellow wires between the LED leads.  This gives a slightly neater package.  (Still not sure if I'm going to keep the green and violet wires, going forward.)

In theory there is plenty of room for a TCS keep alive (KA-N1) in front of the decoder.  I've ordered a few to test.  If I understand correctly, these can be hooked up to the blue lead and the ground pad.  Alas, the LokPilot ground pad does not have a lead soldered to it, and I have been singularly unsuccessful soldering a wire to it, no matter what I try!  I suspect the decoder has a coating (Kryptonite?) on it to protect the other leads and that this is interfering with soldering to the ground pad.  Has anyone successfully connected anything to this ground pad?

-gfh

The Lokpilot/ TCS KA-N1 is a great combination. I use it often.
I do the ground wire differently and had good success with it. I had trouble keeping the tiny joint from breaking when the wire was soldered in the same direction as the stock wires. I add the ground wire from the opposite direction which keeps the tip of the iron away from components on the decoder.

[ Guests cannot view attachments ]

Depending on the space available, I will pot the ground wire with epoxy or hot glue to take strain off the fragile solder joint. It’s easy to route the ground wire back in the opposite direction if that’s where the keep alive is located.
I have also glued the keep alive directly to the decoder trapping the ground wire. This makes a nice compact unit again depending on space availability.
I also desolder any unused function wires as well.
[ Guests cannot view attachments ]

[ Guests cannot view attachments ]



Jim

[GaryHinshaw]

Thanks Jim, that really cool.  I was picturing something similar for a constrained mechanical installation, though probably side by side on a plank of some kind, since I have more horizontal space than vertical.  RE the ground pad, I'll just keep at it.

Quick question: I note that you've directly connected the blue wires between the devices, but that you have not snipped the white & yellow wires.  How are you handling +V for the lighting?

-gfh

[Jim Starbuck]

On an install with low vertical space I’ll still solder the ground wire like that but instead of gluing the keep alive on top I’ll secure the wire by gluing a thin piece of styrene over the top to provide strain relief. This doesn’t add much thickness and still allows the keep alive to be located even at the opposite end of the chassis as space allows.
I typically use SMD LEDs with magnet wire leads. I solder a small SMD resistor to each of the function wires then the negative magnet wire solders to the resistor. The positive blue wire pad on the keep alive is a convenient place to attach the positive  magnet wires.

[Scottl]

Great work @GaryHinshaw! Your method and this discussion point the way forward for these installs.  Aloha!

[peteski]

I wouldn't call that TCS KA-N1 device a real "keep-alive". It is just a tantalum capacitor (IIRC, less than a 1000 uF). To me a keep-alive device will keep the entire loco moving for 1 second or more.   This device stores enough energy for milliseconds worth or time. But it will improve running during those very brief power drops.  Similar to what adding few individual tantalum capacitors accomplishes.  But I guess the additional cost of the KA-1N over individual components is worth it for someone who does not want to deal with the hassle of adding discrete capacitors.

[jagged ben]

What do the other components on the KA-N1 do?

[peteski]

What do the other components on the KA-N1 do?

While I haven't actually taken a detailed look into that one, those components are likely for protecting the sensitive tantalum cap from over-voltage, and limit the charge and discharge current.  When adding caps, when you go over certain value additional components should be added.  This has been discussed in the past.  ZIMO sound decoder manual has several examples of such installations and recommended components.