I’ve been replacing the Sierra SoundTraxx gel betteries with supercaps. I’ve found that three 2.7 volt supercaps in series with values between 6 and 10 Farads work well.

Well…, until they run out of juice. Then the sound unit may “sputter” struggling for enough voltage to keep running. The “sputter” can vary from a couple times to many times drawn out in intervals, and while this doesn’t seem to do any harm, it’s annoying.

Then when power is returned to the engine, the cap needs a moment(s) to recharge before you realize sound, and this too can cause some sputtering.

Of course neither of these issues raise their heads so long as you repower the engine before the caps run out of juice. But I, and many others have found these inconveniences well worth the price of not dealing with the gel battery.

With 6 Farads in series (2 Farads total), I get about 75 seconds using one manfacturer’s caps. I came across some 6.8 Farad units from another maker and figured these should give about 85 seconds. But I’m only getting about 55 seconds (in a different installation). It could be the impedience of the speaker, volume of the unit, etc. or the caps themselves as I purposely selected units with a higher internal resistance.

Be that as it may…

In actuality, one only needs ~30-45 seconds to let the engine spool down and go into idle and have the crew “do their thing” and then we can let the system turn off. So, we just need to shut off the cap before it runs out of juice.

My multimeter shows that the sputtering will start when the voltage gets below ~4.5 volts at the cap. So if we use a 6 volt relay, that will typically operate down to about 4.5 volts and let the cap power the relay while running one leg of the cap though the relay armature to the board, the system will shut itself off at about 4.5 volts until it sees power from the wheels which puts ~7 volts to the board. Instead of now stating from 0 volts, the cap still has ~4.5 volts on it. (Recognzine that caps don’t charge in a linear fashion except between about 1/3 and 2/3 of capacity and it takes more time to get from 0 to 1/3 as opposed to 1/3 - 2/3. Think sine wave.) This is also probably easier on the sound system as it now doesn’t see a “short” each time it has to start charging the cap from nothingness, except maybe the first cycle of the day.

If we want to raise this voltage a bit, to avoid sputtering and have power left in the cap when the engine restarts, we add a resistor in series with the relay windings to reduce its sensitivity.

But the relay consumes power and will shorten the useful life of the supercap. So, we need to find really low current relays that will trip at about 4.5 volts. I’m thinking that these will be the ticket and put in an order. They are rated at 6 volts and only use ~13 millamps, or a bit over half that of an LED.

I’ll let people know how it goes after they arrive and their trial.

http://www.ebay.com/itm/371499680215

Todd please do. My Sierras all now have NiMh batteries, but they need a good charge at the beginning of the season, and they need topped off before each run. I suspect that the ones in my F3 are not holding a charge like they should anymore. A plug and play solution (removing the need for a good charge) would be very helpful.

The relays arrived yesterday and I wired up the circuit. These relays are rated at 6 volts with a 450 ohm coil (13 milliamps). They have beaucoup hysteresis and once tripped, don’t shut off until about 1.4-1.5 volts.

This is lower than desired as the card will still sputter. But it won’t sputter as long, and it “fires-up” much faster because it is not starting at 0 volts. It also reduces the current pull though the board created by the capacitors (approximately 1.3-1.4 amps on start-up with a drained capacitor).

I can "desensitize the relay coil so it drops out sooner by adding a resistor in series to drop the current. At a value 985 - 990 ohms, the relay will cut the sound just before the board sputters, and still provide enough time to go through the “shut-down routine.” Lesser values provide a longer shut down idle time but will sputter a bit.

But, at a value of over ~600 ohms added to the coil, the relay will not power up from the ~7 volts put out by the board.

So we can leave it here and live with some sputter…, or the next step to to try using a diode(s) to route more current to the relay when powered from the board, but run through the resistors when powered from the caps.

Or, I’m open to suggestion.

Uh oh, chronic hysteresis.

Problem is solved! Only required a slight revision to the wiring with no additional parts.

The system now starts as normal and will run off the supercap until the voltage ramps down to 4.6 volts (adjustable) at which time it will shut itself off. Under power, the Sierra board runs at 6.9 - 7.0 volts to the cap.

When it re-energizes, it then starts at 4.6 volts, rather than 0 volts, so is ready to go and the board doesn’t see the large amp draw that is does with a discharged cap.

With 2.3 Farads (three 6.8 Farad caps), the board goes through spool down and air release before cutting off. I’m satisfied and will draw up a schematic.

OK…, the schematic is included. This will shut down the sound system before the voltage runs too low and would also be an ideal way to implement the Sierra with battery power. All parts combined cost <$6 from ebay sources.

Todd, I think I understand. But the caps will leak a little in the off state, through the pot. If I am reading it right. So the instant on is only for a certain amount of time.

I like that. I will have to see about getting those parts, when I place my next electronic parts order.

Edit, due to more studying of, and a better understanding of the schematic

The leakage is very slow. And, there are two poles on the relay. I ganged them because of the initial current surge, but one set could also break the circuit between the cap and pot or ground, totally isolating it. This is a good idea that I will probably implement. Or, I’ll take another look and see if there is a way around this through wiring. Thanks.

The coil first receives full board voltage until it activates. This gets around the hysteresis issue. As soon as it fires, it latches itself through the other terminal and pot.

If you find the relay “chatters” (mine didn’t), it is not making it to the other terminal before it goes back to rest state. If this is the case, put a capacitor across its coil. Use a 16 volt capacitor and start with a small value (e.g., 16 mfd) increasing until the relay latches and the chatter goes away. Watch the polarity on the cap and the relay is polarity sensitive.

Put a voltmeter on the cap and adjust the pot so that the relay opens at ~4.6 Volts. If the pot is adjusted incorrectly either the system will motorboat because you’re letting too much voltage to the relay holding it open so the voltage to the board gets too low and motorboats, or the sound cuts off earlier than it could because you are opening the relay before you needed to to avoid motorboating. The actual value of the pot is ~1K, but this is very sensitive.

Thanks for all of the research into this Todd.

I use 5 volt, and 12 volt relays to control my railroad, and I have never had them chatter because they didn’t throw fast enough. But that is a good tip you gave to stop the chattering, if it occurs.

I changed it up a bit using the other set of terminals on the relay to disconnect the cap.

It won’t work like this because each time the relay tries to trigger, the draw on the cap reduces the voltage to the relay to where it “un-triggers” and the relay chatters. So the charge on the cap takes too long before the board goes into normal mode, if it can even get there. If the cap is “pre-charged” (could be a momentary switch) to where it fires the relay, it is not a problem and when it switches off, it disconnects so does hold the charge.

This doesn’t happen with the drawn schematic because the cap charges though the pot (@ ~1,000 ohms) when the relay is off, and directly from the board when the relay is on. I did notice that from a dead cap, there is a bit of chatter as the cap nears ~1.5 volts, but it quickly gets beyond this and if the cap doesn’t drop to this level, it doesn’t happen.

Perhaps a cap on the relay coil would solve it, but I think there there is a better way. I think that a diode could feed the cap directly from the board so it gets full power, -0.7 volts, until the relay fires at which point the cap can top off the last 0.7 volts through the relay armature. This also has the advantage in that now we are not drawing the current of the discharged cap through the relay armature which shouldn’t close until the cap has a few volts on it.

I need to play with it a bit more, but work calls. (http://largescalecentral.com/externals/tinymce/plugins/emoticons/img/smiley-undecided.gif)

OK it’s fixed.

As I previously noted, quite often placing a cap across the relay coil can prevent chattering. I revised the schematic slightly using the second armature on the relay and a 10 mfd cap on the relay coil.

The 10 mfd cap charges right up and prevents the relay from chattering. Now when the relay drops out, the supercap is isolated and will hold its charge indefinitely.

I found that the value of the pot is fairly critical and my pots weren’t giving good consistency. Also, some of the relays show better consistency in their drop-out than others.

We want the relay to consistently drop out when the cap is between 4.45 and 4.6 volts, with lower being better in that it lets the sound run longer before disconnecting. Below 4.40-4.45 volts, the Sierra SoundTraxx units will sputter.

Most of the referenced Aromat relays (I checked eight of them a dozen times each after breaking them in >100 times each) dropout fairly consistently within 0.1-0.2 volt, but some will sometimes spuriously go lower on occasion. If that happens, there would be a bit of sputter from the sound unit, but it still disconnects and never drains down and starts right up as compared to when relay isn’t in there. Buy a couple and check them for the most consistent. The others are still fine and well within spec and you will eventually find use for them.

The pots I have didn’t work well in that they were inconsistent in that they could also result in sputtering or not connecting the relay on occasion, especially when/after handling. So I figured out the value of the pot and replaced it with fixed value resistors. Now it consistently breaks at 4.47 volts, just before the sputter. You need to figure this resistance out for each relay as even a few ohms can make a difference.

Finally, I changed the value of C2 to 30 mfd. It works fine, maybe better than the 10 mfd, and I had it.

If it’s of interest to some, maybe I could make a video showing an engine equipped with supercaps and no relay, and the engine with the relay.

You replaced the pot with fixed value resistors? What are the values?

They will vary with the relay but seem to be in the range of about 850 to 1K ohms. I’ve buttoned up the engine so can’t see the colors of the two resistors. Actually, the colors only provide an indication of the actual resistance and for best results, these parts must be hand matched because 10% tolerance is more than enough to make or break this assembly. A quality, precision pot would probably work OK.

Luckily, the “break voltage” (i.e., 4.47-4.48 volts) on the first engine I’ve done is on the lower end of what is tolerable and if it gets any lower, I can add a couple ohms in series to raise it. More resistance added translates to more voltage necessary to keep the relay open, and a higher voltage available when the board “un-couples.” But at this point, I’ve “cycled” the engine many times and it has yet to sputter.

I just figured about what I needed using the pot, then went through my bags of resistors to try to put near that resistance together, and tried a couple combinations to get where I wanted. The final resistance when the parts are soldered together seems a bit lower than when alligator clipped together, so shoot for ~4.6 volts on the cap on the bench before soldering it up.

I opened the engine to replace one of the body screws, so I read the resistors. I used an 800 and a 100 ohm resistor that totalled 897 ohms for this relay.

I’ve now done a second one and in this case the resistance came in at ~1,940 ohms to get the proper break point.

This works very well, and even after days the caps remain charged. But I found that even a few ohms one way or the other can make a difference. And, when you’re dealing with this many ohms, even squeezing the alligator clips on the “test resistor” can make a difference of a few ohms.

You need to clean the leads on the test resistors and alligator clips to get a proper assessment of what the break voltage will be and dedicate some time to hand matching these components.

It is also possible that the heat from soldering the components directly to the relay may have ramifications on the release voltage and I am now using IC sockets on which to mount the relays, that take 10 of the 14 pins on the socket.

Also, if you wait long enough (~ a minute), the Sierra will “shut itself down” with a “Neeeyooooo” but continue to draw current, rather than just “go silent.” If you use 6 Farad caps, at a loud volume it will just go silent because the system doesn’t stay active long enough to reach the “Neeeyooooo” point. At low volume the system consumes less power and will reach this point. If you want to reach this point at a loud volume, use the 10 Farad caps.

While the 10 Farad caps take longer to charge, once they’ve cycled and have a charge on them, they won’t drop below the set voltage and should also be instantaneous to recover.