Very good article!

Couldn’t resist showing or re-showing you all a very easy way to find the grounding spot and hot spot on an adequately sized, inexpensive soldering pad.  It’s much cheaper to make a soldering mistake here than on a many dollar decoder.  It’s actually just another easy soldering joint minimizing the possibility of a mistake!!!

Nick

Missing zener diode info

Zener diodes

A regular diode blocks current flow in one direction, but a Zener diode (above a certain voltage) will also conduct the other direction too. In effect, the Zener diode acts as a pressure valve for the capacitors, protecting them from over voltage.

If the capacitor charging voltage is too high then the Zener Diode will start conducting (clamping the voltage down to its rated voltage). The extra voltage will be dissipated as heat across R1. A 1.5W rating for the Zener diode is quite sufficient.

As for the specific part number, just search for 13V 1.5V Zener diode (specs on the circuit drawing) in Google and you will get pages and pages of options. The specific part is not critical as long as it meets the specs and the part isn't too large or costly.

Here's an axial lead Zener with the right specs 1N5928B -- from Mouser: $4.48 each, or $36 for 10.

Here's an SMD Zener with the right specs 1SMA5928BT3G -- from Digikey: $0.47 each, or $3.56 for 10.

As for looks, it varies. Sometimes the diode case is transparent glass, other times it's black plastic. Axial diodes look as you would expect, and SMD ones look like little black squares with metal tabs on the ends. The negative end (cathode) typically has a band on the case marking it.

Very interesting BUT

What kind of performance can we expect from these two keep a lives?

Bob Harris

Says in the article ...

It says in the article.

And ...

If you want to get some idea of how the 5-F version will perform, the zen stay alive demo uses a similar circuit, and the video really helps you see what it will do for you because they put tape on the rails all over the place. Check out the Zen stay alive demo, That's what the 5-F one will do. It's impressive:

Capacitors In Series

The schematic for the "Build your own stay alive" article shows the 1.0 F caps. in series, not parallel.  That would produce an equivalent capacitor of 200,000 uF.

Oops

Actually they're 1F caps with 2.7v, so it's not even that good. They're actually 200,000 uF total. Still that's enough to give over 10 seconds of total run time with sound+lights+motor. They need to be connected this way to get the 13.5v rating. Connected in parallel would burn them out since at 2.5V they couldn't take 12v from the rails.

We'll correct the article and upload (red face).

How about a happy medium?

I don’t want 10 seconds duration — that’s way too much time to get in trouble, but 1 to 1.75 seconds (presumably less under a load) seems too short.  How about 3-5 seconds?

Middle-of-the-road

Well we know 4700 uF gives 1.75 seconds and 200,000 uF gives 10+ seconds, so it seems like somewhere around 125,000 uF would be what you're after.

Now to go look up components that fit that ...

But do watch the zen stay alive video. It shows what ~10 seconds can do -- and it's darned impressive.

thanks

Thanks for pointing out  that I missed that information in the article. But I'm with Jeff, could we put two in series for about 3 seconds?

Bob Harris 

Yes and no

Not exactly. You need 13.5 - 15v, so two caps would give only 5.4v tolerance before they burn out. So no, that won't work. Think in terms of 125,000 uF and a total of 13.5-15v (at least) for the sum total of the voltages.

I found these caps: DGH504Q5R5 (just Google the #) ... they're 500,000 uF 5v and if you string 4 of them together in series you get 125,000 uF as desired, and they'll handle 20v so no problem there. They're available for ~2.50 each from Mouser.

That should give you ~5 seconds at best, maybe a little less. But be aware these caps are a bit larger and almost 3x more expensive than the ones in the article. Everything's a trade off.

Nice, but...

If someone has a removable bridge and expects to kill the loco before reaching where the bridge was removed (via dead track protecting the bridge area), then no bueno having keep-alive circuits in your locos. I can see 1-2 seconds possibly helping immensely. 10 seconds seems like absolute overkill - clean your track for crying out loud!

Sorry to be a stick-in-the-mud, but somebody has to mention this - it might as well be me.

YMMV,

Joe W.

Something wrong with the math

Others have already pointed out the series vs. parallel, 5F vs. 200,000uF, error, but something still doesn't add up. If 4700uF gives 1 second of stay-alive power, then 200,000uF should give about 43 seconds, so either the 1 second or 10 second figure is wrong. I suspect it's the 1 second figure, because a 4700uF cap being discharged at, say 100mA, would drop at about 21 Volts per second, whereas a 200,000uF cap with the same load on it would lose only 1/2 a Volt per second.

It's not linear ...

You can debate the math all day long but the results in the field shows the stay alive isn't linear to the farads. Not everything in nature is linear -- some things are exponential, parabolic, and so on.

Stay alive and removable bridges

Stopping KA locos without physical barriers...

No barriers, 1 second, linear

So, what you’re saying prof, is that keep alive locos cannot in fact be stopped without having physical barriers It is too bad not everyone implements asymmetric braking, it’s kind of a cool feature.

In HO scale, 1 second at 1 scale MPH moves 0.2”.  That’s a pretty decent crawl, and plenty of distance to get over a dirty spot.  So if all you’re trying to cover for is the occasional dirty spot on the track, 1 second ought to be plenty.  If you’re trying to avoid powering frogs and avoid putting in point jumpers, 5 or 10 seconds is probably in order.

I suppose the stay alive not being linear to the farads makes sense - capacitor discharge curves aren’t linear.

Another capacitor in parallel??

If a second 4700 uf capacitor was added in parallel to the first, would that about double the time?  They would both be 16 volt or would that change as well?  This would not be very small, but may give additional time

Bill

ten seconds

I too am mystified by why anyone would want 10 seconds of keep alive. I looked at the video, it seems to me that running on for two meters without the ability to send a stop signal (because the loco isn't receiving any DCC signals) is just a really bad idea - you're asking for a rear-end wreck if not an off-the-table experience. It seems like a 'because we can' thing more than a 'because we need to' thing.

Is the assumption that a loco will have only one pair of pickup wheels, that it will be crawling at speed 1 over a multi-inch long dead frog and it will take more than 3 seconds for the pickup wheels cross the dead spot?

I just don't quite understand the parameters of the problem that 10 seconds of KA is required to fix. Could someone give a fer-instance? I'm assuming that most of us can keep gremlins from sneaking into the layout room when we're asleep and putting down six feet of tape on our rails.

What real-layout track problem would still stall your loco if it only has 2 seconds worth of KA?

Yard Throat

Decoder Time-outs

I'm not sure about the settings in other decoders but Tsunami 2's have a setting where if the decoder hasn't received a signal from the system then it stops the engine.  The soundtraxx YouTube channel has a video on it.

But it is linear ...

Sorry, electricity doesn't work that way. The time to discharge a cap with a constant current source (or a constant resistance) from one specific voltage (say 12V) to another (say 3V when the motor stops) is a linear function of the capacitance. I'd like to see a demo of a loco staying alive for 1 second on a 4700uF cap.

EDIT: That said, even a fraction of a second is probably long enough to get over a frog or dead spot.

Reality vs. Theory

If reality and theory don't match, it's pretty rare that reality is the one in the wrong 

There are complicating factors, I don't know exactly what effect they have but I can easily believe some combination of them would explain Joe's practical real world observation.  First, we're not completely discharging the cap, we're just discharging it down to whatever the min voltage the decoder/loco will run on.  And second, the load is not constant current or resistance.  It's a bit of electronics running off an on board regulator that probably pulls more current as the input voltage decreases, and a spinning motor whose effective resistance changes as the speed changes.  Some decoders with bemf will compensate for input voltage variation to keep the speed the same, which will have the effect of increasing the input current as the voltage decreases.  Lots of variables.

well,well

As much as I like to save money and build things myself this tread is doing a wonderful job of selling me a few TCS KA4's. That way  I get the performance and small size I need.

Thanks

Bob Harris