How to trace a short?

[gordon s]

Those that have read about my layout, Eastwood Town will know this has been an ongoing saga, with several attempts to build a large continuous layout. That is a subject on its own and can be found in the ET thread, but having spent most of the day on my back underneath the boards wiring and religiously checking every track and Tortoise connection, I started to think about how you trace a short and if there are some shortcuts to help. I recall only too well spending days trying to find a short a few years back and that taught me a lesson, although I have yet to come up with a methodology, that is tried and tested, so would be grateful for some advice from our electronics experts. I guess it is simpler on a small layout, but when faced with 100's of feet of track on a large layout, where do you start?

 

My logic is to split everything up into sections and keep switching bits off until I get a clear, but this is far more complex with DCC where even if you separate the track, the bus is still connected. Desoldering every dropper in turn seems a crazy approach and there must be easier methods. My last build split the bus into sections controlled by 10A switches and had isolation gaps to keep the track in matching section. That at least allowed me to reduce the fault tracing to 10' sections, but it was still a bit of trial and error.

 

This may sound daft, but can you locate a fault using a meter? Do the resistance/voltage/current change the closer you get to a fault? Apologies if these are off the wall questions, but I'd really like to understand and simplify the process of fault tracing.

 

As it is now, I have the bus split into sections and check every joint with a continuity meter after I have soldered it so I am fairly confident I can build the layout without a wiring fault, but what to do if a fault develops later is all a bit of trial and error for me.

 

There must be a sensible method to trace a fault, so over to you guys for ideas. If nothing else it will preserve my sanity and prevent the loss of days from my life in future...

[smokebox]

Split the circuits, track and DCC bus in two. find which half has the fault then split this half in two repeat until you have located the fault.

[gordon s]

Thanks Smokebox and in essence that's what I do. I tend to treat it as a mechanical problem and end up separating tracks and cutting wires to divide the layout down, but that involves a fair bit of rework to make good all the disconnected wires and I wondered if there was another method, perhaps using a meter or tester of some description.

 

On the latest version of ET the track is isolated into sections and the bus is switchable to each section. I just wondered if there was another method other than the divide/divide/divide process.

[Theakerr]

Whilst I run a DC system, my methodology is much the same as you have described. I use cab control with individual blocks, which is very similar to your description of your DCC system. I did have an interesting situation that took me a while to find and that was a Deltic sitting over a block junction with wheels in two blocks. This was a section of the layout built many moons ago and i thought the block change was a bit further down.

[Pannier Tank]

When wiring your layout if may be useful to have a Multimeter across the Feed / Track being wired, set to Ohms Range with a Buzzer. Then if you wire in a short you'll know straight away!

[gordon s]

I find it interesting that the divide/divide/divide approach still appears to be the logical method, but can't recall reading about this when I first ventured down the DCC route. Unless I missed it, everyone spoke of running a couple of bus wires around the layout and then attaching droppers, preferably to every rail, but no mention was made of fault tracing and the need to separate the whole layout into sections. Mention was made of power districts, but I got the impression this was related to current demands rather than fault finding.

 

At present I have droppers from each rail to a sub bus for each board or main section. I then run an additional bus wire around the layout and then feed each sub bus via a 10A switch. This allows me to switch each section off and eliminates the need to systematically remove droppers until the short can be identified.

 

I was sure in these days of electrical testing, there may be another way of doing things, but haven't come across anything as yet. Maybe it doesn't exist after all....

[gordon s]

Thanks David, I do something similar when actually wiring the layout in the first place, but I'm curious about what happens afterwards.

 

We've all had a layout running perfectly one day, then something happens and you are faced with tracing a fault. My worst scenario was pinning down copper clad sleepers. I suddenly had a short and after hours of desoldering droppers eventually traced the fault to one particular area and lifted some point work. Of course one I lifted the turnout, the fault was gone, yet the turnout itself tested perfectly. Put everything back and the fault returned. It took ages before I had a doh! moment and realised the track pins on each end of the turnout sleepers were shorting through to the back of the double sided PCB strip and causing a dead short. I guess movement or temperature change had caused the sleeper to move and make contact with the pin. A real beginners mistake, but at the time nearly drove me insane.

 

It's this scenario I'm interested in and how to trace this type of fault that just suddenly appears..

[Tony_S]

I think awareness of the fault finding problem and the advantage of being able to isolate sections may not have been adequately discussed when DCC wiring started. A lot of layouts just evolve. The specification for the wiring on our club modules was written by someone who is an electronic engineer and so the ability to isolate a section even if a fault develops after the layout is erected was designed in. However I think Gordon is hoping for a gadget that will tell him the fault is x cm along the track.

[skipepsi]

I think what you want is a meter similar to the one used to check the charging system on a car that measures current or volts without being connected to the wire, as I don't understand electronics I am not sure that one exists to measure such small amounts of current/volts. I am sure it could be done by an expert maybe someone else could comment.

[34theletterbetweenB&D]

It's easy enough to design the gadget in theory, question is would the the gear to build it be affordable. What you need is a fast clock and pulse emitter and detector. The time from pulse emission to detection, can then give you the length of the path from the signal input point to signal output point. (It will still help if the layout has some structured approach to wiring to enable it to be tested section by section by single paths.)

 

Since the electric shoots along at near light speed, 3.10⁸ meters per second, a clock capable of measuring nanoseconds will deliver decent resolution, should enable the short site to be located within half a metre. Is there affordable kit? 'Physicsman' who posts here might well know.

[Dan6470]

Hi Gordon,

 

I’ve thought about the problem of fault finding in the past and I think the best possible way is to design the layout in sections that can be easily isolated otherwise it becomes quite difficult to fault find with a basic mulitmeter. Sure you can see that there is a short but where is it?

 

The rails themselves have a resistive element, same as a piece of wire and if you know the cross sectional area then the resistance per linear metre can be calculated which would give an indication of how far away from your test point the short is (*). The problem is that if you have a lot of rails wired in parallel it can become confusing and may require further calculations. Other methods exist for finding shorts in long lengths of wire which in essence is what the rails are but 20 or 30 metres isn’t consider long. I am sure that a telecoms engineer will come along soon and explain how they fault find long lengths of wire using complexed waveforms and capacitance readings but as far as I can see any of these methods are only going to advise how far from your test point the fault is.

 

So in answer to your question, it would seem that unless you build fault finding circuitry into the layout, the easiest way to find the fault is as suggested by Smokebox, basically divide and conquer.

 

Best regards

Dan

 

(*) A quick check with my multimeter on a 1m length of peco code 100 gives a reading of 0.1ohms/rail (and I’m none too sure about the accuracy of the meter) so you would need 5m of track before the meter would read 1ohm between the two rails with a short at one end.

[Gordon H]

It's easy enough to design the gadget in theory, question is would the the gear to build it be affordable. What you need is a fast clock and pulse emitter and detector. The time from pulse emission to detection, can then give you the length of the path from the signal input point to signal output point. (It will still help if the layout has some structured approach to wiring to enable it to be tested section by section by single paths.)

What you are describing is known as a 'Time Domain Reflectometer' - and is unlikely to be affordable.

Another method for tracing the current path along conductors is to use a thing called a 'Toneohm'. These apply current pulses to the wiring and have a current sensing probe which you move along the current path until you lose (or find) them. A quick Google search implies that they are mostly used for detecting PCB shorts these days, though the Model 580 appears to be of the type I am talking about. My experience was with the Model 700 over 20 years ago.

[gordon s]

Thanks for all your input guys. Thankfully I'm doing the right thing, but as I said I'm not an electronics or electrical expert, so simply apply mechanical logic to the problem. I am surprised this is still the preferred method and just felt there was an easier method using technology. At least I know now to keep building my layout in sections, even though it is DCC.

 

Two wires? Who's kidding who?......but I still prefer it.

[34theletterbetweenB&D]

What you are describing is known as a 'Time Domain Reflectometer' - and is unlikely to be affordable...

This guy reckons a peripheral - presumably to a PC, he doesn't specify! - can be built for relatively little.

 

http://www.ecnmag.com/Articles/2009/07/Low-Cost,-High-Resolution-Time-Measurement-Application/

[APOLLO]

Remove every piece of rolling stock - locos first. I usually find a duff loco to be a major source of short circuits.

 

Good luck

 

Brit15

[gordon s]

Not to mention the needle files, steel rules and cable reels that have been left behind when I'm working....

[Kal]

HI

 

A TDR is very useful for finding shorts or open circuit or even frequency "suck out" on long cables, and they are used extensively in Coaxial cable installations underground. Every Cable TV company will have one somewhere and its optical equivalent the OTDR.

 

Problem is, they are really used for serial cables and when you have multiple branches, it becomes difficult to read without disconnecting the branches, and even more so on small sub 10 meter installations.

 

Technically it is possible and you would even be able to find dry solder joints, but is it practicable on a railway layout? I would guess not, but happy to be proved wrong.

 

Regards

 

Kal

 

PS, Just for completeness there are also SSTDR (spread spectrum) which are much more expensive tools, and they can do parallel evaluation and are accurate down to about 500mm, Having just looked, you can get a handheld SSTDR for as little as $400, a lot cheaper than in my day.

[noiseboy72]

A low ohms meter will read faults down to 0.05 ohms. You might pick one up from an army surplus store as they are not that common these days. A pat tester has one built in, but it is difficult to isolate the function.

 

I find even with my £9.99 meter I can get down to the nearest 20CM or so. I move the probes together until I get the lowest reading and work from there.

 

I have had points with sitcking frog polarity switches, intermittant open circuit droppers and a soldered rail joint that was acting as a low pass filter, giving 1 Ohm when tested, but showing a 10V volt drop at DCC frequencies! That one took some finding...

[StuartM]

I'm afraid the only way to really find faults is the divide it method.

When building a layout you should think about this and build in electrical sections ie: outer loop, inner loop, depot, sidings etc

each section should be electrically and mechanically isolated from each other and then wired to circuit breakers or links/switch's which if your working in dcc can then all be connected to a central connection point. This way you can switch each section in or out, which will narrow the fault down to a particular section.

Once you've proved the fault to a particular section you can then start looking for the problem.

This method can be taken further eg by each siding having it's own isolation switch (dpdt). so the process of elimination can continue to a single section of track.

It makes for makes for more wiring but it also makes for easier fault diagnostics in the years to come.

As for using a meter or a scope, I see no advantage over such short distances, instead just use 2 AA battery's a 3v bulb and two flying leads, when the leads are placed across the rails, if there is a short, the bulb will light up, if there is no short the bulb will remain off. This also makes for a cheap continuity tester

[davefrk]

Hi, and I know of a club that had an 'untraceable' short.... Connecting a car battery for a moment certainly found the out of place wire feed.....

Not to be recommended so don't try this at home. I'll not mention which club but they'll know who they are.

 

Dave Franks.

[Edwin_m]

One unexpected consequence of wiring my layout for current block detection is that any short is traceable down to an individual track circuit. The diodes in the block detectors introduce enough voltage drop from one section to the next that my multimeter on ohms setting (which only puts a volt or so into the track) only registers the short if it is put across the rails on the section where the short is.

 

If you plan to do this make sure you do the coin test though, as the drop through the diodes may also defeat the short circuit cut-out!

[Mattmeister]

I am an apprentice in the electronics industry, so not an 'expert'. But, whilst I was building my layout I came across a few electrical faults. What I did was I bought a cheap multimeter with a continuity buzzer function. I then put one lead on a wire and the other on the other end . If it beeps, you have a circuit. If it does not beep then you don't have a circuit. Hope this helps.

 

Matt