1-Wire Topology Assistance

[Tico]

I'm planning to install a number of 240V electric window openers around the house and request some advice.

There will be a power cable that will run to each window and back to the hub (240V 4-core/0.75mm²). Each cable has earth, neutral and two actives for the respective direction. I'm happy with the wiring for this and the switching methodology in the Miniserver.

I'll run Cat5 to the window to provide for a physical switch at the location (neutral/open/close). This will use 3-cores of the Cat5 (+24VDC and two returns for the respective digital input).

It's the remaining 5-cores of the Cat5 where I'm a bit unsure of the best way to provide for a 1-wire reed-switch at each window. I'm planning 1-wire because of limitations with the number of digital inputs I have and lack of space for extra extensions.

Option 1: The Cat5 will be star-wired for each window, however I recognise the limitations of 1-wire for this topology. I therefore plan to daisy-chain the 1-wire bus at the hub to maintain a linear topology overall. At the extremities of each window, the 1-wire bus would likewise be daisy-chained. I believe the suggested wiring is GND/DATA on 1 pair (OUT), GND/DATA on 1 pair (IN) and the last core as +5V.

Option 2: Just use a star-wire topology for the 1-wire and address the result if I run into issues.

Option 3: I refer to the attached image from the German Loxwiki - https://www.loxwiki.eu/pages/viewpage.action?pageId=917700

What I'm unsure of is whether the switches indicated above are electrically separate from the 1-wire bus? If they are electrically separate, I could use these grouped together in the hub.

If they're not electrically separate, I anticipate they would have the same limitations as a star-wired topology?

In the Loxwiki article, some DS variants (eg. DS2401) seem to use the switch as a break in the Data wire.

[Duncan]

the DS2401 is likely to be quite unreliable using the topology you are planning for.

the DS2438 doesnt physically connect or disconnect anything from the bus, unlike other 1-wire solutions. It contains a couple of A/D converters which can be configured as switches - it requires a little logic in loxone config to compare the received voltage and therefore detect switching.

if you are going to use a daisychain topology using star wired cat cable, your total length will be twice the distance from your 1-wire extension to each sensor, and could rapidly build up to the point where it may not be very reliable. If your cat5 is cat7 and therefore individual pairs are screened, its likely to work using a twisted pair for +5v and data and the screen for 0v.

your basic problem is a lack of digial inputs - although it might be a slight bit more expensive, i would consider using either a DI extension or even a few of the knx inputs/button sensors such as this one:

https://www.eibmarkt.com/cgi-bin/eibmarkt.storefront/5b3f1be9001526742747ac1e040205a1/Product/View/NS6518301

or

http://www.datec.ch/en/binary-input-8-channels,-flush-mounted.html which also has a din-rail mount available

[Tico]

Thanks.

It's a fairly easy option to upgrade to Cat7 for the project. The reed switch for each window is not essential (limit switches are built into the electric winders). It just gives me a bit more utility for programming blinds. ie. Rather than just drive the window closed if closing the blinds, I can check if they need to be.

My additional motivation for using 1-wire is I've already got the extension there and it's connected to only two temperature sensors. I'm also of the mindset to eke out everything I can from existing resources (using spare cores of Cat cable etc.).

I may need up to 10 reed switches. Each run will be an average of 15-20 metres.

You mentioned "If your cat5 is cat7 and therefore individual pairs are screened, its likely to work using a twisted pair for +5v and data and the screen for 0v.". 

Which option is 'its' referring to (all with Cat7) -

Option 1: All star-wired topology,

Option 2: Daisychain topology as described previously (I need to bring the Cat cable back to the hub for the switch input),

Option 3: Split the daisy-chain into smaller runs. As a possible example, a daisy-chain of 5 sensors (existing on a physical star-wire), total length about 150-200 metres. Then duplicated as two stubs.

I've also read that DATA and GND should be twisted against each other (but I don't fully understand the specifics of Cat7 and how that impacts shielding in this application). Is the following a roughly correct guess -

1. Cat 5: The GND twisted against the DATA provides a pseudo shield by virtue of the twist coverage?

2. Cat 7: The shield around the DATA provides a true shield for the entire length?

Cheers.

[Duncan]

'its' refers to option 2 if using cat7

a 1-wire bus has an effective maximum of about 500m with a good driver (we dont know anything about the loxone driver inside the extension), each connector or stub adds additional virtual length called weight, which reduces the amount of real cable you can use. 10 switches at 2 x 20m (signal and return) is already 400m plus connections and stubs so you are close to the max.

twisted pair only really works for differential systems, such as network and dmx, which have a + and - signal relative to 0v

1wire doesnt, so to shield the data signal as much as possible, the data should be one of the strands within the shield, and the shield/screen should be the 0v so that any interference signal is shorted/referenced to the 0v.