capacitance in long cables

Now one for the electronic gurus. The signals on our track are connected together by 3 core cable, mostly 1 sq.mm 3 core twin and earth. At one end what I'll call the message core is connected to ground via 1k resistance, at the other it is connected to 5v by 1k resistor, but of course there is no 5v before the 12v is established. There must be some capacitance between the cores, so what happens when I switch on? If the message cable wasn't connected to anything else I'd expect it to go to half of 12v, which would be embarrassing as it would exceed the 5v limit on the electronics. If I had a battery scope I'd just measure it, but I haven't, and we have no mains power. We do have a generator, but would that be good enough for my expensive scope?

I'm tempted to fit 4.7v zenners both ends, but with the lack of mains that isn't all that easy either. Would a capacitor to ground work? I imagine that if the combined capacitance core to core pu external cap is say double the core to core then I couldn't get more than 1/3 of the 12v. Anyone advise? Its way out of my understanding

Hello Duncan,

in your question you did not discribe what kind of electronic is on the other end.

From my experience with the point switches i made with such a low resistance on the end of the cable there are less problems. How long is the cable-

Duncan,

lets assume a 8mm OD 3core cable of 1mm sq Cu - this has typically around 97pf/meter tween cores.

take 20 meters - thats 1940pf, with a R of 1k to gnd and one to 5v.

A +12v pulse on an adjacent wire will spike the signal wire up to approx 10v, and down to approx -5v when the +12v turns off.

Add a 10n cap tween signal line and ground, and the spike reduces to +4.1v and +0.9v when the 12v is turned on then off.

Here's reasonable proof..

2x1k load, no cap to gnd

2x500ohm load, no cap

2x1k load, 10n cap

I don't quite understand why you think it would go to half 12v. The peak voltage on the signal cable would be dependant on the capacitance of the cable and how fast the voltages rise on the 5V and 12V. The examples I have seen on Google give capacitances of 100pF/m.

I assume the signalling you are talking about is manual on/off, not fast data signalling. I would connect a capacitor e.g 470nF on the signal wire. This would not only suppress any transients caused by the switch on, but also suppress any mains hum on the signal wire.

Adrian

*** I see Joseph beat me to it with a much better explanation

Edited By AdrianR on 11/09/2020 08:28:42

470 nF at 50 Hz is a reactance of 6.7k - not much hum reduction especially if inductively coupled.

I don't see why you are driving the line like this Duncan? Surely with a 1k pullup and a 1k termination the "high" level on the line is only 2.5V? Would be better to just terminate on a gate input, maybe with some additional protection diodes?

A bit more description of what we've got. The system is 3 aspect colour lights, controlled by Arduino Pro Mini in each signal. The 12v and ground are obvious, the third wire communicates between signals to tell a signal that the following one is on red and so to show yellow. At the driver end it is connected to ground via 1k, at the driven end it is connected to 5v via 1k, so most of the time it is at 2.5v but when it wants to send a message the driver end is grounded by an open collector transistor, actually a ULN2003 and it drops to ~ 0.6v, if the connection is lost, it rises to 5v and so it flashes an error code. I use 'message' to avoid confusion with 'signal', which is the complete box of tricks. Distance between signals is about 50 metres.

The system has been working fine for about 2 years, but we've suddenly had a spate of failures. The first 2 were after the long lock-down (no I'm not blaming coronavirus), and we've had 3 more in quick succession. These last 3 were in one batch of Arduinos. I'm coming to the conclusion that we've been sending too much voltage, because the nominal 12v is actually 13.8 when the battery is well charged, and so I'm fitting external voltage regulators to power the Arduino with 5v, but it would be fairly easy to add zenners/1M resistors/extra capacitors to the message wire.

Joe's post makes me think I should do something, adding zenners/1M resistors/extra capacitor would be possible, changing the circuitry around the Arduino not easy. Do I add zenners or caps, or both? I'm going to repeat Joe's experiment and will report back, in fact I've just remembered we have a 100m drum of cable waiting for the next phase of extension, so I can try it on that in the comfort of my workshop. Just one point on Joe's test, shouldn't there be a capacitor from the nominal 2.5v point to ground?

Thanks for all the replies.

Edited By duncan webster on 11/09/2020 11:36:30

Edited By duncan webster on 11/09/2020 11:42:49

Edited By duncan webster on 11/09/2020 11:59:26

...zener...one n...

Capacitors to swamp the coupling sound the best approach to me.

Many years ago I experienced capacitance in cables problem on a large fire alarm installation using MICCV cable (mineral insulated copper covered)
The system when installed was very unstable and produced spurious faults, Gent the manufacturer and system designer attended site and advised on value of capacitors to be fitted at various locations on the system, this action cured all problems when caps installed.

Emgee

Either zeners or caps - don't need both. If you had lightening in the area, zeners would be better, but then the ground is probably questionable with lightning around..

Caps are all that is needed - 10nf for 20m, 47nf for 100m good enough.. The simulation is a 'perfect' one - the voltages you get with a scope will most likely be a fair bit less, as I have not modeled the inductance of a 100meter length in the setup, nor taken any dielectric ( cable insulation) losses into the model - but that is becoming esoteric.. and offers little to the solution here.

Just one point on Joe's test, shouldn't there be a capacitor from the nominal 2.5v point to ground?

Uhmm, there is one....?? 10nf...3rd solution image..

Joe

Silly me, I meant a second 1940pf in fig one to replicate the message to ground capacitance of the cable in fig 1. Capacitors it is, I've got a big bag of 0.1uF, if these are too big I'll order some 47nF

Thanks again

Why not use chips specifically designed to drive and cope with long distance cables?

The obvious choice are the RS232 1488 and 1489 devices. Feed power along two wires and signal on the third.

If I've understood correctly the system has been working fine for two years, but has suddenly had a number of failures? This is put down to the applied supply voltage. Has the system been run from a different supply, or flat batteries, over the last two years? If not, then why would the same supply voltage suddenly cause failures? Has anything else changed? Sticking Zeners/capacitors in without understanding the issue is a job creation scheme for monkeys.

Andrew

It worked for 2 years, then we had the lockdown and all signals were taken down and put away. When we put them out again, 2 died. No change to power supply. I mended those 2 with a new batch of Arduinos, and they failed, mended one of those yet again and that failed. The last 2 have blown on board regulators, which led me to look into the supply voltage requirements. Nano and Uno quote up to 20v, mini quotes 12v. We supply 13.8 via a diode, so 13.2. Theory is that we've been lucky for 2 years, but the latest batch are not as tolerant of over-voltage. Whilst I'm at it I'm trying to pin down any further potential issues, hence this query

If I started again I'd probably put a transistor on the input, but I've got 15 of these things on dedicated pcb, mods not easy, so an external bodge is called for. A capacitor seems the easy way to me.

Not really fair, or very nice, that answer, Andrew...

Duncan's question was :

Now one for the electronic gurus............................... There must be some capacitance between the cores, so what happens when I switch on? If the message cable wasn't connected to anything else I'd expect it to go to half of 12v, which would be embarrassing as it would exceed the 5v limit on the electronics. If I had a battery scope I'd just measure it, but I haven't, and we have no mains power. We do have a generator, but would that be good enough for my expensive scope?

Duncan did not ask why it is not working now, after being OK for two years...

No problem suggesting alternatives to Duncan's signalling solution and implementation, but that's not what was asked, no?

A reasonable answer to his question does not monkeys of us make..

Joe

Ok, so the Arduino has changed - makes sense. Simplest solution is to stick two or three more diodes in the supply line. If interference on the 5V line hasn't been a problem I wouldn't worry about trying to fix it unless it has been proved to be such.

Andrew

Edited By Andrew Johnston on 11/09/2020 14:45:55

Nothing personal intended and I apologise. If I've understood Duncan's follow on post correctly there never was a signal problem, so nothing to solve.

I'm happy to hold my hand up and say it's all my fault. Some years ago, after a disagreement on this forum about the Fast Fourier Transform, I resolved to never get involved with electronics threads. Stupidly I broke that rule. I must stick to my own rules.

Andrew

Connecting any modern electronics directly to long lengths of cable is asking for failures. There are lots of possible causes but in this case a change in the earthing arrangments, intentional or otherwise, caused by removing and re-insalling is a possibility.
Adding capacitors to the ends of the message line has some advantages but will slow down the maximum communication rate. Adding zeners (6.2 V 1.3 W on message and 16V 5W on 12V would be a good starting point) will give protection with minimum effect on communication. Really the system need to be desigined for the environment incluging ground loops and common mode voltages. The ideal solution would be isolted DC-DC coverters powering the Arduinos and optical isolation on the message line. Care would also have to taken with the lamp drivers.

Using 1k at each end is a little odd. Ideally the pull-up/down resistors should match the impedance of the cable probably 100-300 ohms. There should be just two, one at each end of the cable. is it a one transmitter multiple receiver setup or does each signal transmit and receive?

Robert G8RPI.

Hello Duncan,

for me is a privat rule i do not use the built in 5 V regulator . I use an outside 7805 with a 1 N4007 Diode at the input.,so reverse polarity is no problem. And it is important to block the 7805 with the 10 µf Capacitors.

To the input of the arduino you have to decide to us a pullup or pull down resistor, this depends on the logic of your signal. In your aplication you use a nearly static signal so you can have a much larger capacitor i would use 1 µf on the logic line. The best way is to transmit with an optocoupler at the input of the arduino.

this is what i would do. Several years ago we had at our club some problems with a broadcast transmitter at medium wave. The cable to the point act as antenna and give some strange behaviour. I kept the input low by a resistor and the problem was gone.

I'm with Andrew on this one. Duncans fuller posts at 11.27 and 14.21 suggest an issue with swapping to different Arduino's as the primary issue. Extra diode(s) is one way of going, personally I would tend towards DC-DC converters local to each Arduino.

regards Martin

So here's where I'm up to, cut through the RAW pin on the Aduino with a grinding disc on a Dremel, un-solder the remains of the pin from the board and fit a 78L05 regulator using some convenient spare holes which were for a feature we dropped, then solder capacitors from both inputs to ground on the back of the board

One down and working, 14 to go! It works on my bench in the workshop. Will give me something to do during the latest lockdown (one of or members has tested positive for the Coronavirus, so no playing trains at the track for a bit, and our local infection rate is ~100/100,000). Some more evidence just emerged, the solar panel man has disclosed that the system at the end of the track where we've been having all the trouble runs 0.6V higher than the other end, which has been running for 2 years at least.

To add some final clarification, the message line isn't sending high speed data, when a train enters section it falls from 2.5v to 0v, 10 seconds later (depending on how fast the train is going) it returns to 2.5v. If we lose connection (track circuit jumper getting cut usually) it goes up to 5v, which the Arduino detects and sets the relevant light flashing. This makes it not that easy to use opto isolators etc, I suppose an op-amp buffer would be more robust than the Arduino, but not at all easy to back fit to the board.

Some pictures:

Edited By duncan webster on 14/09/2020 21:35:16