TVS Diode really required?

I don't know anything about model trains but this sure seems like one or two pieces of the puzzle are not fitting together.

The "short" or "fault" you talk about seems to be one rail shorting to another rail. Assuming you only have one power supply for your setup this short would mean low (near zero) voltage and high current (probably infinity reduced by the short circuit impedance of the transformer). Thus, it would seem you want to "fault to safe" much, much faster than the 4-5 seconds the circuit currently opens up in.

The 1.5KE36CA is a very fast, rather high power pair of silicon avalanche diodes (basically zeners) arranged in series. The typical nondestructive breakdown is 30.8 VDC BIPOLAR. For suppression of ESD or other surges this family of devices is often an excellent choice. But for what you are describing I just can't make the leg bone attach to the hip bone. I wonder if the 1.5KE36CA TVS is just being recommended to suppress the transient spike (spark) from the short? It will do that quite well at about a 31V level (the 18 VAC nominal is probably peaking around _/- 25VDC). I wonder if the assumption/assertion that the TVS will decrease your "fail to safe" time might be a misunderstanding?

Shorts or faults are usually derails. An ‘all stop’ when a derail occurs is desirable for several reasons, primary reason here is remove power to the engine, this intended to eliminate power surge tuyphat may damage the electronics in the engine. Many of these are in the neighborhood of $500 bucks each.

A usual operating voltage is about 14 or 15 volts, amps drawn vary but is usually around 2A for a freight train, 3A for a passenger train (lights in cars).

The 4-5 seconds to trip is the breaker that is part of the transformer. This transformer was common until the late ‘80s when UL imposed some new rules for train transformers, these rules brought about the currently produced type of ‘chopped sine wave’ transformers. I have some engines that won’t run on the new transformers, thus the pure sine wave equipment is used here.

I wonder if the assumption/assertion that the TVS will decrease your "fail to safe" time might be a misunderstanding? Very likely a misunderstanding of mine. It seems to lessen the time to safe, this may just be wishful thinking on my part. Is the diode being recommended because it shunts the power (this is installed at the output terminals of the transformer), limiting current to the track? I get full voltage on track (measured with VOM) when shorted. In a few seconds the breaker opens up in response to the short (excessive amp draw).

What about a fast acting automotive fuse?

I considered this, and may yet use the fuse.

Only resistance to this is frequency - as in frequency of event. Would need a carton of fuses on hand. Sometimes goes several weeks without a short, sometimes have several in an hour.

Lemme see... fuses about a buck, one locomotive $680 bucks...

Cheap insurance...no matter what setup you go with, I would still have a fuse...

I agree.

Transformer output is max is 18 VAC, not DC. These 3-rail trains are AC.

A search tells me an auto fuse will act quite differently AC vs DC. Can someone provide a chart, formula, ??? that might give at least a starting point to arrive at correct value of blade fuse?

Bonus question: I have seek auto blade fuses that are resettable mini circuit breaker. Is this a gimmicky trinkety solution? I am guessing when I say a melt fuse is simply going to be more reliable, anyone with experience to the contrary?

..."Generally, fuses have a DC voltage rating that is half of the maximum AC voltage rating."...

...."DC Amps and AC amps are the exact same thing, they are the measurement of electrons past a given point, the difference is that the electrons of AC go back and forth (alternating) and DC go only in one direction (direct)."Sep 1, 2008

Good find. Thanks for this.

I’m sort of liking this idea.

This interrupts the circuit on detection of short, a fast acting CB with a variable value?

I’ve not been real comfortable with the entire existing system (circuit breaker trip), breakers can and do wear out. Will they ALWAYS fail to open circuit?

GA vote from me.

"Follow that with a comparator"

Agree with everything you have said (in both posts): just thought I'd add that the negative side (WLOG) of the comparator could be driven at a slightly higher level than that relating to the current current (STET). That way the power could be gradually increased without tripping the circuit, but, any spike in current even within the normal operating parameters would cause a trip.

This warrants a little bit of discussion.

It is very common to watch the ammeter bounce around as a train goes around the track with the throttle at a steady state. Hard to quantify with analog needle meter, but jumps from about 2A to 5A then right back to 2A are not uncommon. This is due to joints in track, dust or dirt in the collector (the rolling power contact at center rail), debris on track, track crossover...

Will the scheme be flexible enough to tolerate these spikes?

"Will the scheme be flexible enough to tolerate these spikes?"

BruceFlorida already covered this with a low pass filter.

On the whole I think that his idea of using an arduino or raspberry pie may ultimately be more flexible and easier to debug than an increasingly complex analogue solution.

Is there any way to add a second conductor ahead of or behind the primary power conductor?This would prevent spikes and increase engine life,and give smoother operation.There would be no spike unless both of the contacts break at the exact same time;a redundant contact.

I used this method in the 1970's to decrease traveling track fed 6 way hoist brush problems,and it is now a requirement in the NEC for all track fed devices.

(I should have patented it,but I was young and dumb and didn't know how.)

Looking at this again I'll add a little:

1. Everything I described will work as fast as the components will work. I don't know if your equipment has a smooth draw of current or is "spikey". As described above you might be at risk of nuisance trips. Since you currently trip in 5 seconds without smoke you probably don't want a goal of microsecond tripping. Thus, a simple low pass filter is a good idea between the differential amplifier and the comparator.
2. The simplistic circuit I described is only good for one half of the AC supply. Based upon your current 5 second response time this simplification should be acceptable.
3. The AC on the rails might be floating and might be ground based. My guess is that it is floating. This isn't a problem, especially since anything you might build would probably be powered by a floating "wall wart". This is probably a BYOG (Bring Your Own Ground) party.
4. Digikey and Mouser have been low on parts for months now. If you design something be sure to check availability before going too far.
5. If you design something be sure to add a power LED and an operate/fault LED. Nothing is worse than to put a lot of effort into something that should be fun and then spend time standing there wondering why the darn thing won't work.
6. With a small "ask" you can probably get several designs from CR4 members. I'd do it but I'm so swamped right now I shouldn't even be writing this.
7. I haven't looked at the Arduino offerings or other hobby offerings in a long time. Except for the resistor and wire you might be able to do all this with an Arduino board and a "shield" board, or some other "off the shelf hobby board". I don't know if you have ever played wth Arduino's but it is hard to beat the low amount of build effort of buying a $10 - $20 board from Amazon, Digikey, Mouser or others and only needing to run a few wires for the build to be done. It could also (programming required) allow easy adjustment of the voltage and time values for tripping. (Since you would have a microcontroller you could also make red LEDs blink back and forth where the highway has a railroad crossing.)
8. An "analog only" solution might be easy using an op-amp shield board (Amazon, others) and skipping the Arduino board. Once again, the ease of sending $19 to Amazon is an addictive way to "build".

No wall warts here.

To perhaps better illustrate what is going on here:

Dammit! Images/photos don’t want to load.

I’ll work on this.

I have images of train sound cards, one from 1984 production, another of (more or less) current production. 1984 card has 22 surface components on a single layer board, new version has 2 prom chips, seemingly countless surface components, and a number of (to me) unidentifiable bits on a multilayer board.

An aside, I run three of these transformers, each capable of independent control of two separate tracks. I have seen layouts with dozens of tracks and yards.

Tracks have connecting points where a train can cross from one to another. At these crossovers there is an isolation of the center rails. This allows for ‘power districts’ to operate independently from one another.

That is the way a modern feedback controller works.

GA

You might find some ideas here.

https://www.modeltrainforum.com/threads/auxiliary-fast-acting-circuit-breaker.20049/

https://www.mouser.com/datasheet/2/657/ourproducts_snapak_series_datasheet-1528227.pdf

<...when there is a short...>

It is possible to design these out. The principles applied in UK 2-rail finescale modelling practice are as follows:

• Points [switch rails, closure rails and crossing rail assemblies] worked as pairs wherever possible, as in the prototype.
• Where needed on the prototype, trap points to be provided on the model, and operable.
• Points arranged so that the necessary rail joint for the closure rails occurs away from the heels of the switches.
• Crossing power polarity dependent of the position of the switches.
• Signals as prototype, interlocked with points' position so as to prevent conflicting moves.
• A signal in the clear position switches track power onto the rails in advance of the signal.
• Signal only clears if the track in rear and the track in advance of it are connected to the same controller.
• Separate functions for Signaller and Driver.
• Driver drives to the signals set up by the Signaller.

With these principles applied the occurrence of <...shorts...> drops through the floor, with attendant joy for the modeller and the viewer.

_{Been there. Done it. Secondhand T-shirt now available on Vinted (other clothing trading sites are available).}

Good to hear from another train guy.

Differences between the OO gauge 2-rail and the O ga 3-rail are many, but your comment shows there are also many similarities.

Building a layout presents the largest number of opportunities for shorts. After track adjustments, as you say, derails are quite few indeed. But, there are occasional issues, particularly when running two or three trains on the same track at the same time.

I like this comment a lot.

I have been on CR4 sabbatical, and I see today quite a bit less activity than years past.

Shame, this.

You forgot "HOMEWORK HELP"

A crowbar circuit or what they normally call as "fuse blower" may possibly solve your issue, since it handles over voltage or surge in power supplies.. similar to below.

Will this act any faster/better/stronger than the 1.5KE36CA diode already installed?

The speed sensitivity will always be dependent upon the fastest components used, as it can be used in conjunction with your zener diode..

The beauty about the crowbar ckt is, it stops and prevent the power which actually as designed is a load protection..

The crowbar will "snap to a short circuit". The 1.5KE36CA will never do this.

Think of the 1.5KE36CA only shorting circuit excess voltages down to approximately 31 volts. It is two zener like devices in series and it essentially provides as much conduction as necessary to regulate excessive voltages down to approximately 31V.

The fault condition for the crowbar is quickly going to near zero volts. This will help protect things from excess voltage but will expose your power source to a "more perfect" short circuit. This would probably help protect "other stuff" but will probably be a little rougher on your power supply. If you have a series fuse as shown in on of the posts you will blow the fuse much faster but you still will have the time delay of the fusing material warming up and dripping.

The model train forum that I frequent is populated with enthusiasts of widely disparate tech savvy. One fellow in particular is an authorized repair agent for two of the big manufacturers, and freely offers all sorts of advice and solutions to problems. He confirmed the TVS diode captioned in OP as correct for application.

So, he offers advice to members for whom plug and play solutions are a challenge to grasp and accomplish, guys who believe $4 bucks is a budget blowing expense, some ESL members who may also fall into one (or both) of the first categories, to the other end of the hobby spectrum (people like us). So I believe the solution I started with here is probably a ‘Works pretty well, easy to understand, easy to do, and is fairly inexpensive’ solution.

All this said, if an in-line auto fuse would do the trick I believe he would offer that up. I tend to agree with SE, even if only moderately effective a fuse is a cheap addition to a belt and suspenders method of protection, a method I generally like.

A crowbar circuit is starting to sound like a better solution than the diode. I figured there had to be something better.

Jeez, I like this place! Many Thanks to everyone! Pony rides and lollipops for everybody, compliments of the appreciative Doorman.

Can you elaborate on the “rougher on power supply” part of your comment?

Rougher on the power supply since a "dead short" across the output of the power supply means the maximum value of current, and all the power dissipated in the circuit will be dissipated inside the power supply. Assumes negligible voltage drop across the "short" or the wiring.

Iout = Vsupply / Rpowersupply

Rpowersupply is the internal resistance of the power supply (a,k.a. source impedance), usually a fairly low resistance value.

Having a 31V TVS (Zener) across the output means that the circuit current is much lower because the 31V is subtracted from the power supply output voltage. The math would look like this:

Iout = (Vsupply - Vzener) / (Rpowersupply)

For the sake of simplicity, I am ignoring the true VI curve of the Zener, but there is some dynamic resistance of the Zener that is added to the source impedance of the power supply which further reduces the current and hence internal heating in the power supply.

In all this rough order analysis, I am also ignoring the impedance of the transformer. This has an impact on the fault currents if you short the secondary. Since most model train power supplies are a transformer, you might also enjoy reading the following for more insight:

https://americas.hammondpowersolutions.com/en/resources/faq/general/transformer-short-circuit-considerations

Many years ago I built something that had several short circuits per day as part of normal operation. I used magnetic hydraulic circuit breakers and I was very happy with how well they worked and lasted. Everything about that project was wrong for your application and the parts did more than give cancer to California mice. But, the experience might be useful.

With only a couple of minutes of looking I was impressed by the Sensata Airpax "SnaPak" series of breakers.

A data sheet is available at https://www.sensata.com/sites/default/files/media/documents/2018-04-25/ourproducts_snapak_series_datasheet.pdf. They are offered in "instant", "fast" and "slow" versions. None will be as fast as previously described circuits but you can easily be 10X or more faster than your current setup. Pretty simple, pretty cheap, big step up in performance.

I found the above item in stock at Digikey but I suspect they are also available from one or more of Mouser, Newark, Allied and others. I only spent about 5 minutes looking. There are many other options for you to consider. The parametric search engines at Digikey and Mouser provide some help in narrowing down the candidates.

I would still use the 1.5KE36CA TVS diode(s). The goals of the breaker and the TVS are totally different and both are important. In fact, from a technical point of view, it would be beneficial to use several diodes scattered at different physical locations. Human electrostatic discharge (ESD) and shorted train rail sparks can have very high voltage and current spikes. The total energy is very low but down in the range of nanoseconds and microseconds things can be rather nasty. An iron core transformer won't be damaged but some of your modern electronics items might. That stuff you call train rail I call inductor. Given a choice between a TVS and an inductor, the high frequency energy events will prefer to travel harmlessly through the TVS. Thus, by placing several TVS diodes at different locations in your distributed network of 3 rail O gauge inductors you can harmlessly dissipate the high energy event before it travels to equipment down the line. The train has been running in your basement for years so you don't actually "need" to do this, but everything I said is true and at a buck each a couple of extra TVS diodes might eliminate that once in a lifetime damage to a $400 controller. Just a little insurance policy suggestion from one who likes to go overboard with protection.

As I have said before, don't get too deep into something until you check stock. This isn't like the good old days where you could actually buy what people are offering for sale.

This sounds like something to investigate. Thanks very much.

Thanks to everyone, this is very helpful. A personal issue has arisen, and I will be absent for a while. I shall return, so don’t hesitate to continue without me.

See you in a little while.

I am liking this idea as well.

A common practice is to connect the track rails with a wire pair every 15 feet (track length) or so. Track itself is a material that is sort of resistive and needs these feeders for a more steady performance. A subset of that practice is to provide switched control of each of these power lines, especially with analog control of the trains. From the data sheet provided it appears I can select an appropriate breaker to function as the on-off for individual power blocks within power districts.

As the thread title says, is a transient voltage suppressor diode really the answer to this little problem? From comments it appears a TVS is indeed a belt, and benefits of suspenders are available from other varied circuits or components.

This may be useful in your case.

Link below:

https://www.amazon.com/dp/B07P87TG8F/ref=redir_mobile_desktop?_encoding=UTF8&aaxitk=77f5d6e50434754b1acb2a3c99d11e76&hsa_cr_id=8710065600201&pd_rd_plhdr=t&pd_rd_r=2f557072-60df-4f59-86bf-136989fb2ab9&pd_rd_w=CJKaE&pd_rd_wg=LzzI5&ref_=sbx_be_s_sparkle_mcd_asin_0_title

Does the train derail at the same place every time?If so,you could try embedding a permanent magnet at the trouble spot,or spots,to hold the train tighter as it passes.

just thinking....

Derailments are sporadic (time and locations) and unpredictable. During a layout assembly any identifiable trouble spot is reworked.

Hello everyone, I am returned.

Derailments are but one source of track fault /short circuit.

Stuff drops from scenery bits that may prove to be a little conductive. Metal pieces can be rattled off of cars, shorting the track. Tubular track pieces (used here) use a paper isolator strip to physically separate the center rail from the remainder of the conductive elements, and these can wear through or slip a little bit.

There are anecdotes of visitors to large model train displays tossing a handful of paper clips, coins, nails... onto the layout.

A small soft bristle brush ahead of the engine to clean the track of debris may be helpful.Every time a track arcs,it leaves a tiny pit mark on the track when it crosses this spot again,it repeats the arc,enlarging the pitting.These small pits can cause a variable resistance and hence,voltage spikes to occur.Initial polishing of the track could be done with a fine flint paper gently rubbing the track as the train moves.The flint paper could polish,and the brush could remove the dust.Perhaps a special car could be fitted with this "Track Cleaner"?This,along with redundant double brushes should keep the track pristine.You would only have to run the Maintenance car periodically.

Track cleaner car should be on the roster of every model railroad. This (cleaning the track) is well established as good practice.

I use a retired boxcar frame dragging around a ScotchBrite pad. As you say, only an occasional tour around the track is required. If I get a snag, this indicates a closer inspection and remedy are required.

I did not know they made one.

I am speaking from my past experiences with track fed hoists,cranes,vacuum cleaners etc.I have no experience with model trains,but it seemed like a similar problem;Well,almost no experience.I had a model train when I was a kid,over 65 years ago,and I did solder around the sections using copper wire.This cured a lot of problems with unscheduled stops. All of The track sections were simply socket-ed into each other and were very problematic.

To my knowledge there are no track cleaner cars in production today.

Lionel made one for a while, two rotary motion pads with a solvent drip sort of a thing. Discontinued, lots of problems is what I hear. These were a dedicated self-propelled unit, could be had in several paint schemes to go along with whatever road you are modeling.

A specialty car was pretty popular amongst train people but is also no longer produced. Trackman 2000 is the model number. Guy made them in cottage industry fashion, retired, nobody kept the operation going.

You could go a whole nuther way with battery and wireless control...?

There are many images and videos of some VERY elaborate train layouts, and you have posted a couple of excellent examples of what these hobbyists are doing.

Thanks for sharing.

Yeah, that’s a whole ‘nuther kettle of fish.