One Wire parasitic mode

[mellifera]

Hi MN,

Have you tried back tracking, ie. get your set up back to a state where you had it working (the original 10 sensors running in parasitic mode). Failing that, get a new DS18B20 and wire that up to your 1-wire adapter with a short length of cable and make sure that works. Slowly add more bits until you find where the tipping point is for your set up.

Networks are funny things really. You can ride a coach and horses through the specs and get away with all sorts of things; no termination, exceeding recommended lengths, connecting un-used pins to 0v, not bothering with pull-up resistors, but when it doesn't work you don't have a clue which one of your short cuts is causing you grief - if not all of them! 

It's difficult to recommend whether or not the Hobby Boards power injector would solve your problem as it's hard to say which particular problem needs solving and whether the power injector would solve it. You could "suck it and see" I suppose.

Have you now scrapped the DS1402D-DB8?
Are you plugging a RJ11 into the DS9490R yet?
How have you wired your additional DS18B20s into your existing 10 DS18B20s which have had their 5V legs cut off?
Are you sure you've wired all your new sensors correctly (one round the wrong way might stop all of them being able to communicate properly)?

I calculated the CRC value of ROM ID you posted, 17000001B9944C28, and it's fine so something is working. 

Hope this helps, Mike.

[MN]

Have you tried back tracking

Yes I can back track fine

Have you now scrapped the DS1402D-DB8?

Yes

Are you plugging a RJ11 into the DS9490R yet?

Yes

How have you wired your additional DS18B20s into your existing 10 DS18B20s which have had their 5V legs cut off?

I have fully wired them (no leg chopping) and wired the power to pin 1.
The whole 1-wire network is one long continuous loop

Are you sure you've wired all your new sensors correctly (one round the wrong way might stop all of them being able to communicate properly)?

I ’think’ so as it will read them but errors now and again – if I got one round the wrong way – I think it would stop reading completely
I could rig up to a portable and take out to the end of the 100m run and see if it works ok from there.

Any clues what

echo was incorrect

Means?  Can't find anything on the web giving any clues


Thinking more about chopping off the power leg.  Somewhere along the line I got confused between the DS18B20 (Temperature) and the DS201 (On/Off) chips.
The DS201’s don’t use the power line and can it can be chopped off.  What is confusing it that when I set up a test rig and connect a DS18B20 without the Power leg connected – it works fine.  So  I started to chop them all off!

The reason it probably works in my situation is that I read one sensor at a time – so it is quite slow.

Adding up the total network size it is probably 100m in total.
The new section is at the END of that 100m
So my guess is I am a the limits of what 1 wire can cope with
Also – for some unknown reason I didn’t use cat5 to the last run – I used alarm cable  – which might not be helping,

So

Sort term fix is to retry 10 times to get the temperature
This seems to work fine for about 8 hours then it hangs.
I will modify the software to reset the one wire adaptor on an error

Long term fix
I ordered a 6 channel hub from Hobby Boards.  This has the power injector built in.
My thoughts are to split the routing up after the 100m using this.  So one leg to the Solar Cylinder, one to the pump and another to the panels (via the Alarm Cable )
My understanding is that this should have the effect of shorting the wiring length

Long Long Term fix
Replace my chopped off DS18B20s!

I will let you know how I get on with it!!

Thanks for your help - much appreciated

MN

[kristen]

"This seems to work fine for about 8 hours then it hangs"

Sorry, haven't read the whole thread, so this may be useless / irelevant:

USB dongle-thingie isn't being "put to sleep" is it?

I would suspect software failure/issue if it runs for hours, rather than hardware / network problem.

Is PC's memory usage creeping up (leaking)?

[wookey]

Indeed. I had problems with the network hanging (no more readings, sometimes on all of it sometimes just on one branch) after some random number of hours (2-100). Turned out to be due to another process going potty and using up all memory until it was killed by kernel. Clearly this memory stress was bad for OWFS. Since I stopped that happening it's been rock solid. Not directly relevant to your case, but a data point.

I'ev had advice elsewhere that if your network stops working then being able to cycle the power to it is a good way to make it behave again. I can;t confirm this, having not yet tried.

[mellifera]

Hi MN,

MN: The whole 1-wire network is one long continuous loop

Sorry, being a bit thick here, but when you say "one long continuous loop", do you mean...
                        _________    _________    _________    _________ 
                       |         |  |         |  |         |  |         |
                       | DS18B20 |  | DS18B20 |  | DS18B20 |  | DS18B20 |
                       |_________|  |_________|  |_________|  |_________|
                          | | |        | | |        | | |        | | |   
5V O---+------------------+-|-|--------+-|-|--------+-|-|--------+-|-|-------.
DQ O---|-+------------------+-|----------+-|----------+-|----------+-|-----. |
0V O---|-|-+------------------+------------+------------+------------+---. | |
       | | |                                                             | | |
       | | |_____________________________________________________________| | |
       | |_________________________________________________________________| |
       |_____________________________________________________________________|

If you do, may I suggest removing the "bypass" and having your 1-wire network like a "daisy chain".
                        _________    _________    _________    _________ 
                       |         |  |         |  |         |  |         |
                       | DS18B20 |  | DS18B20 |  | DS18B20 |  | DS18B20 |
                       |_________|  |_________|  |_________|  |_________|
                          | | |        | | |        | | |        | | |   
5V O----------------------+-|-|--------+-|-|--------+-|-|--------' | |
DQ O------------------------+-|----------+-|----------+-|----------' |
0V O--------------------------+------------+------------+------------'

Also, can I recommend Maxim's application note AN148 "Guidelines for Reliable Long Line 1-Wire® Networks". 
http://www.maxim-ic.com/appnotes.cfm/an_pk/148
It talks about loading of the 1-wire network and typically what length and how many devices you can run with depending on what type of bus master you are using.

The two sections "1-Wire Network Limitations" and "What Makes a Reliable 1-Wire Network?" are very relavent to this community. It makes the point that due to the nature of how the 1-wire bus works and devices are searched for and discovered, the transition between working and not working is abrupt and un-graceful.

AN148: Simply put, a network that appears to be successful because all the devices are reliably found in the search algorithm, can actually be near failure. Minor degradation can suddenly produce seemingly catastrophic failures—all it takes is one faulty bit to make a search stop, and parts disappear.

The application note also talks about "Switched Networks" which is what yours (MN) will become if you use Hobby Boards' Hub. It may well help if your problem is an overloaded network (overloaded means either too much cable OR too many devices).

MN: Any clues what "echo was incorrect" Means?

Not really, but I would think the message comes from the software using the 1-wire network and it, perhaps, writes some data to one/each/all of the devices and reads it back (as a so-called echo) and checks the data is un-changed.

Good luck.
Cheers, Mike.

[MN]

Mike

Sorry I mean it is like a ‘daisy chain’
What I meant was it goes to each device and back – so there are no branches
Each device has 6 wires 3 going in and 3 going back until the last one where it does not come back!
In effect the ‘Linear topology from that document.  Once the hub is added I am guessing it will be switched

That document has many things in it I do not understand, especially in the appendix!

One thing I did that was different

I wired up the CAT5 slightly differently
On the old wiring  I had the data and ground in one twisted pair (and the return in another)
For small sections (<1 Metre) I have some wired up with the data and return data on one twisted pair and the ground and ground return on another
Would that make ANY difference?

Should there be a reading of 5v constantly on the power pin or does the controller only put out power when it is needed?

MN

[martin W]

Save starting a new thread here as this is related to the op topic (I think)

I also have the DS9490R (USB 1-wire thingy) currently with 2x temperture sensors in parasitic mode (basicall wired up for powered, but I havent got a 6 wire RJ11 cable yet).

I want to add around 10-12 more temperture sensors and I am now confused if I need to add resistors in the line?

I am using these RJ45 couplers (http://www.ebuyer.com/product/113528) and just run a small trailing wire from the internal wiring to the temperture sensors. This has worked fine for my 2x temperture sensors, but I need ot know the correct? way to do this for 10 plus sensors.

I was thinking of aorund a 4 inch trailing wire 'T'd from coupler to the temperture sensors. I think I read somewhere that 2" for max recommended wire length, but would 4" be OK? Also do I need a resistor on each sensor or what?


Next... where is the cheapest to buy 10-15 temperture sensors - I bought my 2 from Homechip - DS18B20, anywhere work out cheaper for the UK?

Also where can I buy a 6 wire RJ11 lead, so I can use the 5volt line on pin 1 of the DS9490R?

[Pat_]

ebay?<- click

[MN]

I want to add around 10-12 more temperture sensors and I am now confused if I need to add resistors in the line?

I have 15 temperature sensors and have no resistors in the line

but I need ot know the correct? way

Not sure there is a correct way – I just use cat 5 cable soldered onto the DS18B20s and dipped in liquid electrical tape.
http://www.navitron.org.uk/forum/index.php/topic,6704.0.html

Next... where is the cheapest to buy 10-15 temperture sensors - I bought my 2 from Homechip - DS18B20, anywhere work out cheaper for the UK?

I got 20 from china from ebay – can not remember who exactly but I remember they turned up really quickly

Also where can I buy a 6 wire RJ11 lead, so I can use the 5volt line on pin 1 of the DS9490R?

I got mine from that box under the workbench that says ‘old phone stuff’ that Mrs MN tells me to put out for the bin men....

It was actually off an old weather station like:
http://www.ukweathershop.co.uk/acatalog/weathershop_1001.html

But there must be a cheaper way – if you find a source let us know

MN

[martin W]

Pat, thanks for that... didn't think of the fleabay.....

bought 20... so have loads of fun to come.

[Pat_]

Let us know when they come. I'm planning to order some soon.

[StBarnabas]

MN
have you got a sketch of your system? Maybe there is one somewhere that I have missed?

[wyleu]

There's an approved power connection at the head end involving shotttky diodes and other such specialist stuff. I've had problems with parasitic mode with the temp sensors only working in batches. 1,2,3,4 work  and 5,6 don't and then ten minutes later 3,4,5,6 work and 1,2 don't. You do need to strap the +V pin to 0V cos it get's far more twitchy if you don't.

For the slight extra cost, get a power injector. Hobbyboards sell them and you can string any number ( 30 +) on without a worry.
As has been said you get faster and more accurate reads with powered.

Given a lot of these sensors end up in some fairly in accessible places the last thing you want to worry about is is it the chip or the wire or the power supply or the interface that's causing the problem.

Don't forget to put some form of return wire on remote sensors. Otherwise you have no way of extending a run and you might be temped to branch it and that is another cause of problems.