UBS with temp sensor (DS18B20)- wiring question

[keiran 0]

Hi All. Been pulling my hair out with these temperature sensors! so id be grateful for your guys expert electrical opinion (im a novice). 

 

yesterday, i had 1 (out of the 4) sensors working well. Obviously with three other not working, this was not ideal. This was up until yesterday. I think the problem at that point was bad connections on the other three wires. So yesterday re-wired, without fancy splicing connectors, old-school stripping the wires and twisting them together. 

 

Now i have a new problem - 0 sensors are detected!!   (haha winning).  

So, im out of options, and i need help - in the focused form of two questions : 

 

QUESTION 1/   "Do we, or dont we, need a resistor?" 

Forums here seem to have anecdotes of both ways. I followed the fibaro manual so i dont. The DS18B20 product spec (

Please login or register to see this link.

) seems to indicate i *DO* need a resistor (per sensor?)... at least in connecting upto a arduno, or a pi (which im not - im going to this fibaro FGBS001 unit). There is zero mention of this resistor wiring in the fibaro unit manual though! (

Please login or register to see this link.

). Further googling and i hit this (

Please login or register to see this link.

) - a sentence mentioned there got my attention, after explaining why the resistor was needed there was a trailing sentence ".... as an alternative, the DS18B20 may also be powered by an external supply on VDD." might be the get out of jail for this fibaro unit? Are we indeed powering via the Vdd? (Vdd is my red -> orange -> brown wiring in my below diagram) ....  

 

QUESTION 2/  "is this wiring pattern ok, or do i need a bus design"

Amidst other things im suspecting the wiring design. Heres how ive wired it (note the distances in orange metres). Ive used cat6 and picked my pairs carefully (the solid brow, orange, green to retain cross-talk protection). what im wondering, is this "hub and spoke" / "point to point" wiring design ok? or do i need a single cat6 bus, upon which i tap at various drop-off points for the 4x sensors? Ive no idea if these two wiring patterns are equivalent. 

Please login or register to see this image.

/monthly_2020_05/image.png.e45df59f549482940bb46d43c41252be.png" />

 

 

 

 

 

Today, i might unwire everything at the FGBS001 unit end and just try putting on cat6's .... one at a time and see if they pop up... .until they, well, all disappear. 

can i test/debug this with a multimeter i wonder?  (go to the sensors and see if i get a reading on the orange cat6 copper - should i expect to see 150mA on a working one?)

keen for any thoughts you may have. 

cheers

Keiran. 

[RH_Dreambox 179]

@keiran

There are two things that are important for a good function.
1) All sensors must be connected before including the UBS to the Zwave network. If you connect additional sensors, you must remove UBS from the network and then include it again.
2) Sensor connection cables should be slightly different in length. If all the cables are the same length, it can become conflicts when transmitting data.

Edit:

And you can't have too long cables, max 30m.

Edited May 3, 2020 by RH_Dreambox

[keiran 0]

1 hour ago, RH_Dreambox said:

@keiran

There are two things that are important for a good function.
1) All sensors must be connected before including the UBS to the Zwave network. If you connect additional sensors, you must remove UBS from the network and then include it again.
2) Sensor connection cables should be slightly different in length. If all the cables are the same length, it can become conflicts when transmitting data.

Edit:

And you can't have too long cables, max 30m.

Thanks!
 

RE

1) yup no probs there. Ive been Removing and re-adding like crazy over the past two weeks trying diff things, so yup understand that. 
2) hmm, just how diff in terms of length are we talking? They will be a little (50cm) diff likely. 
 

re max 30m - in my case with a star witing, i assume each leg / ‘spoke’ in my cable Star can be that long. Have heard people running upto 100m on cat5/6 choosing the wires like i did (1 core of the twisted pair that buffers against EM interference / crosstalk) but im not sure if that was a single run/bus. 
 

so, do you use resistors? 

Edited May 3, 2020 by keiran

[RH_Dreambox 179]

No I use no resistor. The problem may be that you have several long cables. I once tested how long a cable could be and ended up at approx. 32m. But for my UBS I have 4 sensors where the cables are 18, 4, 2 and 0.5 m.

And 50 cm diff in length is OK, but the total length = ???

Edit:

Is it possible that you can use a single long cable (max. 30m) with several sensors mounted alongside it?

Edited May 3, 2020 by RH_Dreambox

[keiran 0]

8 hours ago, RH_Dreambox said:

No I use no resistor. The problem may be that you have several long cables. I once tested how long a cable could be and ended up at approx. 32m. But for my UBS I have 4 sensors where the cables are 18, 4, 2 and 0.5 m.

And 50 cm diff in length is OK, but the total length = ???

Edit:

Is it possible that you can use a single long cable (max. 30m) with several sensors mounted alongside it?

in my wiring ‘star’, i have approx 2x15m and another 2x28m. That other wiring pattern you suggest (more like a ‘bus’ i guess) is possible. This way, i would end up with 1x15m (one sensor) + 1x28m (and this 28m one has 3 drops for the other three sensors), which would take about 50m off the total distance of cat6 cabling. Worth a shot you reckon? 
 

thanks so much for your help!

[RH_Dreambox 179]

In addition to my UBS with 4 sensors, I have another 1-wire network that is logged into a Raspberry (RRD database).
Both my 1-wire networks are "bus" type and the total length of the Raspberry network is probably a bit over 40 m distributed on 4 sensors and one text display.
Please tell us how you solved your 1-wire network.

[BurnerMan 3]

Maybe until now you have found your way around, but for future use i'll try to answer

 

No need for a resistor. The 1-Wire network is designed to work as a bus, rather than a star. This means you should design your network so that the sensor devices are connected in one long line starting at the master and not so that there are several cables from the master heading out to sensors in all directions.The maximum network length depends on many factors so it is not easy to give a firm guide as to what will work and what will not. As a general rule is not recommend you build each bus on your network with more than a total of 100m of cable, but Fibaro says only 30m in total. However cable quality and routing play a big part in determining how well a network will work. Cat5e or Cat6 cable, and that you route it as far away as possible from mains cables and appliances (especially electrically noisy ones).

If you can't change that then you need 4 UBS to solve your problem.

[Gabor1 3]

Hi Keiran,

I assume, you are Hungarian according to the webshop you have mentioned...

 

I kind of runed to the same issue like you. My experiance is...

I have tried with resistor, didnt worked on long distance wires, over 10m it probably eated up the juice.

Then I have a test without resistor, it worked on test bench fine with 20m wiring.

Then, i installed on site. the wires was above the false ceiling, on the cable tray running together with 230Vwires around them. fail over 10m, probably elecrical interference... I had no chance to try with shielded wires. The connections was the same like yours.

Did you found a solution for long distance?