"Keep Alive" Capacitor Installation

[BCR751]

I think I remember seeing a thread on installing an SMD "keep alive" capacitor in an N-Scale loco whose decoder doesn't already have one.  I looked here on TRW but didn't see it.  Maybe it was on another forum.  Anyone else remember it?  I want to do this to my locos that have the decoders sans the cap.

Doug

[peteski]

AN excellent subject for the DCC/Electronics forum.    Maybe that is where the original thread it.

I don't recall that  thread but I have taken apart the TCS version of that circuit.  It is more than just a super-cap. It has a simple diode/resistor circuit in series with the cap to allow for gradual charge and quicker discharge.  without that the charging current could overload the decoder's rectifier.

It needs to be hooked up to the output of the bridge rectifier. The easy part is the positive side of the cap circuit: It hooks up the the blue decoder lead (positive common for the function outputs). The tougher part is the negative end. Some decoders have the negative (or ground) lead. Most don't.  You need to find that circuit on your decoder. It will be the negative output side from the bridge rectifier.  There usually is not universal place on the decoder to find for that circuit. It will be on the anode side of the 2 rectifier diodes. Possibly on the negative side of the larges filter cap on the decoder.

[craigolio1]

I know you said "to a decoder that doesn't have one" but this blog is very detailed and helpful for choosing the caps. The fellow if also very knowledgable.

Craig

http://www.trainboard.com/grapevine/entry.php?2289-Install-the-Digitrax-SDN144K1E-sound-decoder-into-your-N-scale-FVM-GEVO-part-3

[jdcolombo]

There is a bit of discussion of keep-alive caps in my thread on sound decoder installation here:

https://www.therailwire.net/forum/index.php?topic=31981.0

A more thorough discussion of keep alives in general is here:

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

And you can find more information by doing a general web search on "DCC keep alive"

I use 220uf, 16v SMD tantalum caps for keep alive in my sound installations.   These are available from Digikey or Mouser.  Here's one example:

http://www.digikey.com/product-detail/en/F931C227KNC/478-8257-1-ND/4005713

However, I'm very careful to keep my track voltage at 11.5v, using the N-scale setting on my Digitrax boosters.   If you don't know what the track voltage is on your system, or you'll be using the engines with your keep alive caps on other layouts that may have a higher track voltage, a 20v cap (instead of 16v) would be a better choice.  If  you overload a tantalum cap, it will explode (literally).  It doesn't melt, or give up with a puff of smoke.  It EXPLODES.  A small explosion, admittedly, but still dangerous.

You can get a 150uf 20v cap that is roughly the same size as a 220uf 16v cap.  If you wire caps in parallel, you double the capacitance.   So two 150uf caps wired in parallel gives you a total of 300uf, which should cure momentary sound dropouts.   If you want something that will allow you to coast six feet, the TCS KA's would be a good choice, but you'll have to find room for them.

John C.

[BCR751]

AN excellent subject for the DCC/Electronics forum.    Maybe that is where the original thread it.

/quote]


I'm sure Gary Hinshaw will move this if it's in the wrong place

There is a bit of discussion of keep-alive caps in my thread on sound decoder installation here:

https://www.therailwire.net/forum/index.php?topic=31981.0

quote]

That's where I must have seen it.  Thanks, John.

I have six or seven locos that have TCS M4 decoders but they have no "keep alive" feature.  Although I haven't experienced any issues with momentary drop-outs, I thought it would be pragmatic to try and retrofit the M4's with the circuit addition(s).  I don't have a schematic of the M4 so it would take some optical sleuthing to figure out where those connect points would be on the board.  Also, as Pete mentions, there ain't a lot of room left under the shell for more parts.  Back to the drawing board.

Doug

[M.C. Fujiwara]

Timely thread, as I'm about to tackle installing these on some HO brass geared steam that have sound decoders in them.
There's no space inside, so I will be disguising the KAs as boxes, barrels, tarped stuff, whatever.
Much more important to have the KA feature than the locos recycling sound startup every 3-second hiccup.

Just wonder what the difference is between the KA2 and the KA1.
Or, more specifically, will the smaller KA1 keep the HO sound decoder going for a couple seconds.

[randgust]

I had absolutely no success with a recent TCS decoder installation, and tech support bailed on me because 'this installation is too small for DCC'; mostly the problem was erratic performance at low speed (surging/hiccupping).  TCS said the decoder was constantly resetting DC/DCC due to erratic contact.  Would a 'keep alive' cap help this kind of problem?

[jdcolombo]

I had absolutely no success with a recent TCS decoder installation, and tech support bailed on me because 'this installation is too small for DCC'; mostly the problem was erratic performance at low speed (surging/hiccupping).  TCS said the decoder was constantly resetting DC/DCC due to erratic contact.  Would a 'keep alive' cap help this kind of problem?

What locomotive were you installing this in and what TCS decoder were you using?  If it really is an electrical conductivity issue, then yes, a keep alive would help, but it would probably need to be more than one or two 150uf caps.   But I honestly doubt this was the problem.  Erratic slow speed performance indicates a problem with the BEMF circuit, similar to what happened with the TCS decoder in the new Kato FEF (they have since fixed their FEF decoder).  Does this engine have a coreless motor?

For hard-wire installations, I've learned the hard way to stick with the Europeans.  ESU's LokPilot micro, Lenz Silver mini, Zimo MX621 all have superior motor control circuits.  My current favorite is the Zimo.

John C.

[peteski]

Keep Alive cap (as normally installed) is simply acting like a very small battery for the  primary internal power supply of the decoder.  Primary power supply gets its power by rectifying the DCC signal.  Most small decoders do not have any keep alive capacitor.  That supply then directly feeds the motor driving circuitry and a secondary power supply for the microcontroller (miniature computer) which is the brain of the decoder.  That secondary supply is usually designed to supply constant 3V to the microcontroller and all the related circuitry.

It is a very simple supply using a Zener (as a voltage regulator) and a resistor. That supply (even in small decoders) has a small filter cap.  You can call that cap a "keep alive" for the microcontroller. The cap's capacity is relatively small (usually between 22 and 100 uF) but that is sufficient enough since the microcontroller consumes very little current.   Still, it only keeps the microprocessor's supply voltage high enough for less than a second.    But the microcontroller itself can also be programmed to hut down the instant it stops seeing the DCC packets (even if it still has enough power to function).  That could be the reason for jerky movement f the contact with rails is intermittent.

I thing an interesting experiment would be to see whether the problem is BEMF or intermittent pickup.  An easy way to do this would be to get a pair of Kato caboose or passenger trucks. and make a dummy electric pickup car only. Attache them to an underframe made of plastic, add some weight and electrically hook up the truck together for electric pickup. Then feed the pickup from those trucks to the locomotive under test (to greatly improve its electric pickup). See if that changes the behavior of the decoder.

[randgust]

What locomotive were you installing this in and what TCS decoder were you using?  If it really is an electrical conductivity issue, then yes, a keep alive would help, but it would probably need to be more than one or two 150uf caps.   But I honestly doubt this was the problem.  Erratic slow speed performance indicates a problem with the BEMF circuit, similar to what happened with the TCS decoder in the new Kato FEF (they have since fixed their FEF decoder).  Does this engine have a coreless motor?

For hard-wire installations, I've learned the hard way to stick with the Europeans.  ESU's LokPilot micro, Lenz Silver mini, Zimo MX621 all have superior motor control circuits.  My current favorite is the Zimo.

John C.

These installations have been in Kato 11-105 chassis; basically a motor similar to the Bachmann 44-tonner only single-shafted, and about the same weight when complete.   After I failed with it I sent it down to Lee Weldon to analyze it, he got just as frustrated, TCS replaced it under warranty, we repeated the entire exercise again, we both failed again.   I still have the decoder, but my customer was not very happy.   I have just enough DCC to test and install but I'm a Luddite in this stuff, which is why I enlisted Lee's help.  It would seem to me that a 'keep alive' would help, but I sure don't have any enthusiasm for going through this again.

This was a TCS Z2, primarily chosen for size, remember I'm trying to do this in a Climax A.  http://www.tcsdcc.com/Customer_Content/Products/Decoders/Z-Scale/Z%20Series.html

If there's a similar-sized hard-wire decoder out there without this problem, I'm all ears.

The most annoying part was that it ran smooth as silk under normal DC (same track, same locomotive), but the decoder (either under analog mode or DCC) the low-speed stop/go/stop/go hiccupping at minimum speed wouldn't stop.   I could pile additional weight and it wouldn't stop, so TCS's 'its too small' response held no value to me.   I'm not sure what Lee was using but I had a Digitrax Zephyr, and we had the same problem.  I may be a DCC idiot, but Lee isn't.

[victor miranda]

you had pick-up issues with a kato 11-105 chassis?

that seems strange.

how are you getting the power from the wheels?

I've been experimenting with pick-ups and I have a few conclusions.
they need more testing for me to become confident.

weight helps a lot. the problem is that it has to be applied with care.
one of the things I am still surprised about is the effects of bulb grease in the axle point cups.
I add it to my atlas diesels and every time I get far better slow speed performance.

I have not tried neo-lube yet.
I suspect the two together will be dramatic.

victor

[jdcolombo]

These installations have been in Kato 11-105 chassis; basically a motor similar to the Bachmann 44-tonner only single-shafted, and about the same weight when complete.   After I failed with it I sent it down to Lee Weldon to analyze it, he got just as frustrated, TCS replaced it under warranty, we repeated the entire exercise again, we both failed again.   I still have the decoder, but my customer was not very happy.   I have just enough DCC to test and install but I'm a Luddite in this stuff, which is why I enlisted Lee's help.  It would seem to me that a 'keep alive' would help, but I sure don't have any enthusiasm for going through this again.

This was a TCS Z2, primarily chosen for size, remember I'm trying to do this in a Climax A.  http://www.tcsdcc.com/Customer_Content/Products/Decoders/Z-Scale/Z%20Series.html

If there's a similar-sized hard-wire decoder out there without this problem, I'm all ears.

The most annoying part was that it ran smooth as silk under normal DC (same track, same locomotive), but the decoder (either under analog mode or DCC) the low-speed stop/go/stop/go hiccupping at minimum speed wouldn't stop.   I could pile additional weight and it wouldn't stop, so TCS's 'its too small' response held no value to me.   I'm not sure what Lee was using but I had a Digitrax Zephyr, and we had the same problem.  I may be a DCC idiot, but Lee isn't.

If it ran smoothly under normal DC, it's not an electrical pickup issue.  Ten bucks says its a problem with the TCS BEMF circuits interfacing with the motor.   If you want to try again, I'd strongly recommend the Zimo MX621.   The MX621 is a bit wider than the Z2, but shorter and thinner:  12mm long, by 8.5mm wide, by 2.2mm high (for the MX) vs. 12.95mm. X 6.65mm. X 2.79mm for the Z2.  The question would be whether you have the width.

I'd be willing to bet you'll have success with the Zimo.   Streamlined Backshop is a good Zimo source, and I could give you some pointers on how to set up the BEMF CV's for even better results than stock.

John C.

[Ken Rice]

It's easy to check if it's something related to the TCS BEMF code - just turn off BEMF (CV61).
There are also bugs in a number of decoders of different brands where if you set the top speed CV down too low (CV5), or get certain combinations of  CV 2, 6, and 5, there are obvious overflow errors in the arithemtic the decoder is doing to determine the motor speed - the result is if you step up slowing one speed step at a time in 128 speed step mode you'll see the loco speed up progressively, and then slow down when you hit one more step up - in short instead of the speed step to speed curve being smooth, it turns into a bit of an ascending sawtooth.  I most recently saw this with a TCS AMD4, but I've also seen it with some NCE decoders, and at least one other brand which I can't remember offhand.

So, set CV2,6,5 to 0 (the default), turn off BEMF, and make sure you're not using a custom speed table.  If it runs smoothly then, you know where the problem is.  If not, well, at least you know a few places the problem isn't 

[victor miranda]

mmmm...
there is a lot to a decoder install.

On the few kato mikados I have been asked to 'fix'
some one soldered wires to the truck thumbs and removed the strips in the tender.

If you think that is a good way to do it, you may want to ask why those locos
ended up at my place.

While back emf settings may fix the issue (and how to turn it off is the only question.)
that does not exclude that there may well be a pick-up issue.

my experience points to pick-up issue.
I have a z2 at home for a mikado install .... and I have a couple of old 11-105 chassis.

I can try to replicate the problem.

since I do not know how randgust installed.... we may have to live with my best guess...

victor

[victor miranda]

I like the direction, Ken.

now to see how my weekend goes.