Ethernet (LAN) and 1-wire over the same cable?

[KLD]

After way too many days / nights spent over trying to get a serial port working on a router (so that I could connect a 1-wire sensor and use the router as a 1-wire-to-Ethernet bridge), I've now resorted to plan B. Let's hope it goes a little better. In my early days of researching 1-wire networks I read a note that in principle it should be possible to use the "spare" wires in an 8 conductor (4 pairs) UTP ethernet cable to transmit the 1-wire signal + power.
My application is that all the 1-wire network and hub is in one part of the building, whereas the electricity meter is in another. I like to add a 1-wire counter (Thanks, Eric!) to record the LED flashes on the leccy meter. Putting a dedicated cable in would be a major undertaking. There is, though, a LAN port nearby, which offers a direct cable link to the 1-wire hub.
In the 10BaseT and 100BaseT ethernet wiring only two pairs are used for data transmission. So, I've now taken the other two pairs (blue/blue-white and brown/brown-white), and connected them into the 1-wire network. On first tests it seems to work, but I obviously have to monitor it for a bit longer to be able to say anything about stability and interference on the ethernet side.

Klaus

[breezy]

Hi Klaus

I doubt that the ethernet interface will be bothered in the slightest. The "spare pairs" are frequently used for power-over-ethernet (POE). You need them for Gigabit ethernet though....

[Countrypaul]

If the length of the ethernet cable is much less than the maximum (100m?) then you should have few problems. If the cable quality is better (Cat5e or Cat6) again less likely to have problems, and if using 10BaseT rather than 100BaseT it should also be less likely for problems.

Paul

[KLD]

The cable run is about 20m in length, and connects to a Linksys NSLU2 (a.k.a. slug). For the time being, all my home LAN is 100BaseT, but the cables are relatively new and are CAT5e compatible (what it says on the box ;-)

Klaus

[WillemK]

Hi,
I´m new to this Forum so a small introduction :
My name is Willem and live in holland and have build a low CO² house with a geothermical heatpump. I am monitoring this with a (NSLU2) Slug running uNSLUng 6.10, OWFS and Temploggerd, a neat out of the box monitoring software that uses RRDTOOL databasing, to produce neat graphs and statistics, see the attached pic.
In my struggle to get the system going I read that because in CAT5 the blue an White blue pair is twisted more then the other pairs (amazing but true) therefor it´s best suited for the one-wire protocol.

It´s on page 59 of this designguide for 1-wire. http://www.1wire.org/Files/Articles/1-Wire-Design%20Guide%20v1.0.pdf
In here there is more you want to know about 1-wire.........

I also have a DS2423 counter connected, but when working I can get it going for about 30-45 minutes then the 1-wire network failes. Even at cablelenghts of less then a quarter meter. The frustration is very high. So high I´m thinking of switching to Arduino - It´s a shame because the slug is a very easy to set up system, that for not to bright persons like me is about the right level of competence.

Willem.

[Countrypaul]

Willem,

I think you are generalising the issue of the blue/white pair being more twisted more than you should.
The blue/white pair is twisted more than the other pairs in Belden cable according to the document you referenced. It does say check other types of cable as they may not be the same.

I have only read the document once, but my first impression (which may be completely wrong) is that length less than 0.5m may give issues. (That said on my breadboard the total length including 10 DS18S20 is less than 20cm at present and it works fine with an Arduino).

Paul

[KLD]

Hi Willem,

Welcome to the forum!
Glad to see there are more slug users out there, using it as a 1-wire host computer.
You don't say whether you operate your 1-wire devices in parasitic mode or whether you have a 5V supply?

You might have seen my other thread about the counter slave. This is a PIC microchip that another forum member (EricW) programmed to look like a DS2423 counter. I've got two on my network, and they are completely stable.

In your temperature plot, during periods when the HP is switched off, why is the "bron uit" temperature higher than the "bron in"? Are they mounted differently, or see different environments? How did you mount them onto (?) the pipe?

Klaus

[dhaslam]

Sounds a bit like faulty components.  I bought a batch of DS 18B20s from  Portugal recently that  either gave low readings or failed  completely.   Even the better ones from other sources fail if the location is too humid.   The other problem  can be wiring connections, the   CAT 5 cable has such tiny cores that  it is difficult to join them  properly.    The annoying thing is that when one component fails it  brings down the full network.   I was thinking of using separate cores in the cable to divide the network into separate sections but that would probably make the  overall cable length too long.       

I have found the one wire system  a bit disappointing, too complicated to  program and not reliable enough to use for control systems.   The  seasonal store uses mechanical thermostats and  the one wire system just monitors temperatures.         
 

[WillemK]

Hi Klaus, Hi DHaslam,

Thanks for the reply.

@Klaus, yes I saw the thread and I´m more then intrested in the chips but still have to figure out what halfbee exactly is.
About the HP, first let me translate the dutch, bron=source= earthcollector and woning=house.
So when you look at 16:00 hours the HP start, the upper red, house in gets warm and with it the returnhouse warms up. At the same time the source out and in gradually cooling down, depicting the Delta T of the source.
When the WP shuts down the source in heats up, I guess by the heat op the compressor, and because the outside pump is shut too, the water is sationairy and heats up.
What you can see is when making hot water, (15:30) the delta T of the source is collapsing, hence the bad efficiënty at high temps.

@dhaslam :

I used 18B20 from poland and the readings were spot on. I have now, as a test 18S20 from singapore, which give low readings. But I can adjust that within temploggerd. I use RJ11 6p connectors on the CAT5 and can move the cables any possible amount, without getting a failure.
Another thing is when a sensor is bad, you just remove it, start OWFS and the system is running again - so that is very easy. When the couter is just within the network everything is OK, but as soon as I start programming the cron, and executing things get bad. So I think it has to do with the amount of traffic on the network. but there i´ve reached my summit of knowlidge. The source of my system is a group of people in Holland that have fish in outside pools who were intrested in logging the temps. But the lost intrest, so I can´t aske them anymore.

Any help would be much apriciated.

This morning, my temperture logger (I have 2 sets of everything, just to rule out any part of the system) which was running 3 weeks without a breakdown went down........

Willem.

[WillemK]

Paul,

Sorry, i missed out your reply. Thanks for the input. Concerning the blue and white pair, I checked 2 different brands of Cat5e cable, both not Belkin, and the both show the blue pair being more twisted then the others. ISo I guesses it being general. As I understand the manual, (and not being nativ to the language it´s written in) the most important thing to look at is the daisy-chain or star topology. Since Iwent for the daisychain things went more stable, but not perfect.
The people who designed my set-up all worked with hobbyboards 6 channel hub, and none of them had the same problem. on the other hand, I hear of people running 100 meters of telephonecable without problems. I run 5 sensors on 15 meter in daisychain the The hub is $46,--. That would by far be the most expensive part in my set-up.

I just got a tip by a friend : http://forum.brewtroller.com/showthread.php?t=958

Some folks making beer and using the one-wire too. Their suggestion is to add a 10K resistor at the end of the daisychain. Mmm - I got my senser solderd very beautiful at the end. Well let´s see what I can do.


Willem.

[skyewright]

Even the better ones from other sources fail if the location is too humid.

For my 1-wire soil temp sensors I took a belt & braces approach. After soldering them on to the wires I treated the sensor and connections and a little way back up the insulation with a couple of coats of "Liquid Electrical Tape", then once that was dry I coated the whole lot with epoxy resin! Tests indoors suggested that these coats didn't slow down response times much even for air temps; where they are now, measuring soil temps at 100mm, 300mm and 1000mm below ground level, temp changes are pretty slow anyway!

Those sensors have been buried for around 14 months. At times they will have been below the water table. So far so good...

[Countrypaul]

Willem,

Have you read this?

http://www.maxim-ic.com/app-notes/index.mvp/id/148/CMP/ELK5

Alot of overlap with your document, but may help as it may coer some other details.

What part of the Netherlands are you located in?

Paul

[WillemK]

Paul,

I know this document, I´ve put in the RC filter in as suggested there.  I´m totally lost when it comes to the timing of the signals.

I originate from Amsterdam, but moved away 5 years ago to Venlo, on the border of Germany.

BTW I make my own sensors by moulding the TO92 with epoxy in a Stainless steel cap. Is that how it´s done among members of this forum?

Willem.