Yes, having the pairs twisted minimises interfere with other pairs. At any instant the currents in the two wires of the pair are going in opposite directions, so any radiation is self-cancelling. You could argue that's also the case with regular parallel wires, but it's not quite the same. The twisted construction keeps each pair bound together, and separate from other pairs. It works!
What it doesn't do is prevent the possible problem of capacitance in a long cable from making the unit think a button is pressed when it isn't. But it should minimise potential interference between the circuits for each button.
—
Rick Jones (Treasurer), former 26X & 19 owner, Isle of Wight
Thanks Rick, I think, in my situation, there is no current flowing (not planned anyway, until a button is pressed) but a small voltage present, generated by a chip on the PCB which is waiting to see if the button IS pressed.
Ive tried to understand it by googling but have to admit that the complexity has put me off.
I'm trying to understand but in the meantime I think I'll just cut the end of one of my Cat5e cables and wire it up and see what happens.
At the end of the day I can always rip a multicore apart and make my own flat multicore with enough space between the cores to avoid this issue; it's not as if it needs to look pretty:) Methinks a gaffer tape sandwich would work.
—
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
Ok, so I attached my cat5e 2m cable to one side (3 buttons - 6 contacts) of the autopilot.
Result is non-destructive havoc:)
Even one of the non-linked buttons was "disabled". Semi-random effects all over.
I tried to logically snip one of each pair at a time (ie those connected to my cat5e) and observe the effect on all the buttons.
It becomes clear that there is crosstalk not only between the button pairs but also between non paired wires (ie creating new unplanned circuits which confuses the PCB).
Any other suggestions would be welcome
Just thinking back over the timeline using my 6 core unshielded, untwisted (7 strands of 0.1mm per core)
1. Using my 4 metre length I connected one button, ie two wires with a view to touching together the open ends to see if it would "operate" the button remotely.
The system acted as though the button was pressed even without a physical circuit (crosstalk of some description)
2. I halved the cable length to 2 metres and the adverse effect disappeared.
3. With joy I soldered the remaining 4 wires of the left hand set of 3 buttons BUT was disappointed as the adverse effects returned. Great care taken re soldered joints.
4. Shortening the cable to 9" again causes the problems to disappear, even with all 6 wires connected.
This pattern seems to confirm that crosstalk is occurring between most, if not all cables BUT that it is "cable length" related. And presumably cable proximity related.
5. Similar results from substituting a cat5e cable (UTP).
Im wondering if a multicore with each core being shielded would help but I'm not sure whether it would.
I guess since it's not too expensive I could always just try it.
Any thoughts?
—
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
I think from your experience that any kind of long cable is going to give problems. How about an alternative approach?
Install miniature relays close to the AP, so that when closed each one activates its own button, i.e. contacts connected to your existing short soldered wires. Your remote panel will then have six buttons each operating the relevant relay.
This also means you only need a 7 core cable (+12V power to all buttons, 1 connection back to each relay coil, other side of coils commoned to ground). If you've already got 12V where the panel will be fitted you only need 6 core.
You can get miniature automotive relays pretty cheaply.
—
Rick Jones (Treasurer), former 26X & 19 owner, Isle of Wight
Ok this whole thing of Crosstalk between cables has fascinated me and I've been googling it quite a bit.
The business of cable separation looked interesting and when I cut up my spare cat6 cable I observed that it had a plastic device inserted longitudinally, in the form of a "+" which had the effect of separating the 4 twisted pairs by about 1mm.
That got me thinking. I had noticed that the cable length was having an effect on my Crosstalk - when i shortened the same cable from 4 metres to 2 metres the effectve Crosstalk disappeared.
Now it became clear that a tiny separation between wires could have a similar effect.
So, I stripped the cover off my 3m cat5e (UTP - 4 twisted pairs with no shielding), soldered each pair together, wired these up to two of my autopilot's buttons.
At this stage I kept each of the 4 pairs well apart so no surprises that it worked ok - of course my issue is that I need to feed these wires up the pipe into my binnacle.
Next I got my wife to hold the wiring so it was as closely compacted as she could get it and it still worked.
Anyway, finally, as you can see from the attached pic I then twisted the whole lot together, with both ends fixed to ensure the whole length was twisted together as tightly as I could manage and it still worked - clearly a tiny separation is all that's needed.
At the same time I've been googling relays and, whilst I loved your idea in theory Rick, there were so many to choose from and, since an excellent wireless remote is available for £150 I thought I'd press on thinking through a total wiring solution.
It then occurred to me that I could get away with just 4 out of the 6 buttons on the remote - the extra two buttons only needed for convenience or calibration, which I can do off-line.
So I think what I'll do is make up 4 pairs of twisted pairs from the redundant cat5 & 6 cables I have levy from my house install, shove them up into my binnacle and see what happens. It probably won't too pretty but if it works...
The issue of EMC (Electro Magnetic Compatability) is VERY complex. The good news is it can be broken down into easy(ish) bits and then fairly easy to avoid/cure problems.
The first thing is there are two issues. The vulnerability of your equipment to incorrect operations caused by other EM signals. And the amount of stray signals you generate. Ideally you need to treat both but preventing the stray signal is better than curing it later.
Unwanted signals can find their way into other circuits by a number of means. RFI (Radio Frequency Interference), Magnetically coupled from one circuit to another, Capacitively coupled, Inductively coupled and Conducted from one to another.
In general High voltages, High currents and Fast changing signals produce the worst sources of interference.
Very small signals, both low voltages, and low currents are most susceptible to being overwhelmed by the interference (and misbehaving as a result).
Large Area loops produce really good antennas to both send and receive interference, so always think where the return current for any signal will go.
The suggestion above to use another 12v supply to save one core in a cable is a bad idea. If the current flows from a 12v supply and back down a signal wire and they are not in the same cable you have made a large area loop that may well cause problems.
Signal cables use twisted pairs (in part) so that the circuit is in very close separation to avoid loops. It would be best to use one pair for each push button.
Alternatively they can have a screen around all the cores to help avoid EMC problems. Normally (but not in all cases) best to connect one end of the screen to 0v or Earth at the source of the signals and leave the other end unconnected.
Relays (and most other wound components) produce big voltage spikes when they turn off. A 12v relay can easily give -1000v spikes. These can cause havoc. DC relays should always have a diode fitted across the coil to absorb the spike. AC contactors can similarly have suppressor capacitors connected.
Hope this helps more than it confuses.
Dave.
P.S. On an unrelated note CAT5 cable normally has fine solid cores that are not very good in a moving corrosion prone environment. I would recommend multi core screened 0.25mm2 stranded tinned cables. Maplin and other sell them by the metre.
P.P.S the dielectric cross in CAT5 /6 cables is there to ensure a consistent performance at very high frequencies for data transmission lines, it won't have an effect on slow pushbutton signals.
I think the issue here is that although the application is superficially "slow push-button", the unit is digital, using a micro-processor (like most modern stuff). This means the button state is being sampled by the processor using high frequency pulses, so you end up with the same problems as any RF transmission line.
This is why I thought maybe relays could be a solution.
—
Rick Jones (Treasurer), former 26X & 19 owner, Isle of Wight
Thanks Rick, I think I've misunderstood something here.
I thought you had to send each part of the circuit down the two wires of a twisted pair to get the benefits of unwanted crosstalk.
I don't see how "one wire per contact" in an Ethernet cable will be any better than my existing, non-twisted, multicore.
Is it a feature of the twisting that provides the protection?
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
Yes, having the pairs twisted minimises interfere with other pairs. At any instant the currents in the two wires of the pair are going in opposite directions, so any radiation is self-cancelling. You could argue that's also the case with regular parallel wires, but it's not quite the same. The twisted construction keeps each pair bound together, and separate from other pairs. It works!
What it doesn't do is prevent the possible problem of capacitance in a long cable from making the unit think a button is pressed when it isn't. But it should minimise potential interference between the circuits for each button.
Rick Jones (Treasurer), former 26X & 19 owner, Isle of Wight
Thanks Rick, I think, in my situation, there is no current flowing (not planned anyway, until a button is pressed) but a small voltage present, generated by a chip on the PCB which is waiting to see if the button IS pressed.
Ive tried to understand it by googling but have to admit that the complexity has put me off.
I'm trying to understand but in the meantime I think I'll just cut the end of one of my Cat5e cables and wire it up and see what happens.
At the end of the day I can always rip a multicore apart and make my own flat multicore with enough space between the cores to avoid this issue; it's not as if it needs to look pretty:) Methinks a gaffer tape sandwich would work.
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
Ok, so I attached my cat5e 2m cable to one side (3 buttons - 6 contacts) of the autopilot.
Result is non-destructive havoc:)
Even one of the non-linked buttons was "disabled". Semi-random effects all over.
I tried to logically snip one of each pair at a time (ie those connected to my cat5e) and observe the effect on all the buttons.
It becomes clear that there is crosstalk not only between the button pairs but also between non paired wires (ie creating new unplanned circuits which confuses the PCB).
Any other suggestions would be welcome
Just thinking back over the timeline using my 6 core unshielded, untwisted (7 strands of 0.1mm per core)
1. Using my 4 metre length I connected one button, ie two wires with a view to touching together the open ends to see if it would "operate" the button remotely.
The system acted as though the button was pressed even without a physical circuit (crosstalk of some description)
2. I halved the cable length to 2 metres and the adverse effect disappeared.
3. With joy I soldered the remaining 4 wires of the left hand set of 3 buttons BUT was disappointed as the adverse effects returned. Great care taken re soldered joints.
4. Shortening the cable to 9" again causes the problems to disappear, even with all 6 wires connected.
This pattern seems to confirm that crosstalk is occurring between most, if not all cables BUT that it is "cable length" related. And presumably cable proximity related.
5. Similar results from substituting a cat5e cable (UTP).
Im wondering if a multicore with each core being shielded would help but I'm not sure whether it would.
I guess since it's not too expensive I could always just try it.
Any thoughts?
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
I think from your experience that any kind of long cable is going to give problems. How about an alternative approach?
Install miniature relays close to the AP, so that when closed each one activates its own button, i.e. contacts connected to your existing short soldered wires. Your remote panel will then have six buttons each operating the relevant relay.
This also means you only need a 7 core cable (+12V power to all buttons, 1 connection back to each relay coil, other side of coils commoned to ground). If you've already got 12V where the panel will be fitted you only need 6 core.
You can get miniature automotive relays pretty cheaply.
Rick Jones (Treasurer), former 26X & 19 owner, Isle of Wight
Rick, you are a genius, thank you!
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
Ok this whole thing of Crosstalk between cables has fascinated me and I've been googling it quite a bit.
The business of cable separation looked interesting and when I cut up my spare cat6 cable I observed that it had a plastic device inserted longitudinally, in the form of a "+" which had the effect of separating the 4 twisted pairs by about 1mm.
That got me thinking. I had noticed that the cable length was having an effect on my Crosstalk - when i shortened the same cable from 4 metres to 2 metres the effectve Crosstalk disappeared.
Now it became clear that a tiny separation between wires could have a similar effect.
So, I stripped the cover off my 3m cat5e (UTP - 4 twisted pairs with no shielding), soldered each pair together, wired these up to two of my autopilot's buttons.
At this stage I kept each of the 4 pairs well apart so no surprises that it worked ok - of course my issue is that I need to feed these wires up the pipe into my binnacle.
Next I got my wife to hold the wiring so it was as closely compacted as she could get it and it still worked.
Anyway, finally, as you can see from the attached pic I then twisted the whole lot together, with both ends fixed to ensure the whole length was twisted together as tightly as I could manage and it still worked - clearly a tiny separation is all that's needed.
At the same time I've been googling relays and, whilst I loved your idea in theory Rick, there were so many to choose from and, since an excellent wireless remote is available for £150 I thought I'd press on thinking through a total wiring solution.
It then occurred to me that I could get away with just 4 out of the 6 buttons on the remote - the extra two buttons only needed for convenience or calibration, which I can do off-line.
So I think what I'll do is make up 4 pairs of twisted pairs from the redundant cat5 & 6 cables I have levy from my house install, shove them up into my binnacle and see what happens. It probably won't too pretty but if it works...
image_0
MacGregor 26M 2009 - Sky's the Limit - Suzuki DF50
The issue of EMC (Electro Magnetic Compatability) is VERY complex. The good news is it can be broken down into easy(ish) bits and then fairly easy to avoid/cure problems.
The first thing is there are two issues. The vulnerability of your equipment to incorrect operations caused by other EM signals. And the amount of stray signals you generate. Ideally you need to treat both but preventing the stray signal is better than curing it later.
Unwanted signals can find their way into other circuits by a number of means. RFI (Radio Frequency Interference), Magnetically coupled from one circuit to another, Capacitively coupled, Inductively coupled and Conducted from one to another.
In general High voltages, High currents and Fast changing signals produce the worst sources of interference.
Very small signals, both low voltages, and low currents are most susceptible to being overwhelmed by the interference (and misbehaving as a result).
Large Area loops produce really good antennas to both send and receive interference, so always think where the return current for any signal will go.
The suggestion above to use another 12v supply to save one core in a cable is a bad idea. If the current flows from a 12v supply and back down a signal wire and they are not in the same cable you have made a large area loop that may well cause problems.
Signal cables use twisted pairs (in part) so that the circuit is in very close separation to avoid loops. It would be best to use one pair for each push button.
Alternatively they can have a screen around all the cores to help avoid EMC problems. Normally (but not in all cases) best to connect one end of the screen to 0v or Earth at the source of the signals and leave the other end unconnected.
Relays (and most other wound components) produce big voltage spikes when they turn off. A 12v relay can easily give -1000v spikes. These can cause havoc. DC relays should always have a diode fitted across the coil to absorb the spike. AC contactors can similarly have suppressor capacitors connected.
Hope this helps more than it confuses.
Dave.
P.S. On an unrelated note CAT5 cable normally has fine solid cores that are not very good in a moving corrosion prone environment. I would recommend multi core screened 0.25mm2 stranded tinned cables. Maplin and other sell them by the metre.
Dave Newton Sailbadthesinner
P.P.S the dielectric cross in CAT5 /6 cables is there to ensure a consistent performance at very high frequencies for data transmission lines, it won't have an effect on slow pushbutton signals.
Dave.
Dave Newton Sailbadthesinner
I think the issue here is that although the application is superficially "slow push-button", the unit is digital, using a micro-processor (like most modern stuff). This means the button state is being sampled by the processor using high frequency pulses, so you end up with the same problems as any RF transmission line.
This is why I thought maybe relays could be a solution.
Rick Jones (Treasurer), former 26X & 19 owner, Isle of Wight
Pages