Few sensor readouts, then nothing

jlaiho

New Member
Hi everybody.
Ok, here's the situation:

I recently bought the following equipment:
- 1 x RJ-45 Coupler
- 1 x Adaptor Cable RJ45 to RJ11 3ft (1m)
- 2 x Humidity/Temp Case
- 2 x Humidity/Temp
- 1 x 1-Wire USB Adaptor

They've been sitting on the shelf for a week before I installed them on site.
The configuration is as follows:

[Computer] --> [1-wire USB Adaptor] --> [RJ11 to RJ45 cable] --> [RJ45 coupler] --> Cat5e RJ45 cable, 5 m (~16 ft) --> [Humidity/Temp Sensor] --> Cat5e RJ45 cable, 10 m (~ 33 ft) --> [Humidity/Temp Sensor]

As you now might expect, both sensor boxes are jumpered so they are using parasitic power.

Everything was going according to my plan: I installed the drivers (version 4.03 beta) with the USB Adaptor disconnected. Once successfully installed, I plugged the Adaptor in and ran the "Default 1-wire Net" program and pressed auto-detect. The Adaptor was detected successfully in USB1.

Obviously the next step was to check whether I get any readings out from the sensors, so I ran the "OneWireViewer" program. All the sensors were detected as they should.
I selected the first DS2438 device on the list and clicked on the "Real-Time Temperature" tab. It successfully polled the temperature once and then threw the following exception:

ERROR: TemperatureViewer (D9000000F1D81C26) Error reading device! com.dalsemi.onewire.OneWireException: 1-Wire Adapter communication exception
ERROR: OneWireViewer (null) 1-Wire exception: com.dalsemi.onewire.adapter.OneWireIOException: native TMEX error -12


After throwing those exceptions, it suddenly read the temperature once again, this time successfully three times in a row and then it started throwing the same exceptions again and never recovered from it afterwards. Restarting the program revealed 0 devices. Only way to recover from this situation is either disconnecting the USB Adaptor, disabling and then enabling it from the Device Manager or by machine reboot. Doing so brings the sensors back but only until they're polled for readings again.

I'm using a program called "LogTemp", but it suffers the same symptoms as seen in the OWV.

The computer specs are as follows:
- ASUS CUSL2 rev. 1.02
- Pentium III 866MHz
- 512MB RAM
- 40GB HDD

Believe it or not, this ancient machine is running Windows Server 2003 R2, with Service Pack 2. :rolleyes:
I'm not sure if this is of any importance, but USB ports in this machine are all 1.1's. According to the Device Manager, the USB Adaptor is taking about 100mA current so it's not a case of port overload.
At present, it's running Java 1.5.0 Update 22, the last of the 1.5.0 series.

I'm running to a brick wall with this problem.
Any thoughts, comments or ideas are very welcome.

Thanks in advance and Best Regards,
Juha
 
I would suggest some basic troubleshooting to determine if its a cable/device failure and where.

Take your network down to the minimum and test each device.

[Computer] --> [1-wire USB Adaptor] --> [RJ11 to RJ45 cable] --> [Humidity/Temp Sensor]

If that's stable, check the other sensor.

Then start building it back up piece by piece.
 
Take your network down to the minimum and test each device.

[Computer] --> [1-wire USB Adaptor] --> [RJ11 to RJ45 cable] --> [Humidity/Temp Sensor]

If that's stable, check the other sensor.

Then start building it back up piece by piece.

Thanks for the quick reply and thoughts, sda!

I just finished reducing the network down to a minimum, bringing it down to a "starting level" you suggested. The result was not all that surprising: I got a total of six consecutive readings from the sensor box but the seventh attempt failed and never recovered until I quit the program and disabled and re-enabled the Adaptor. Tests on the second box gave the same end result: it failed but after four consecutive readings.

Based on the result, I suspect either the USB Adaptor or the ancient computer's ancient USB 1.1 ports themselves may be at fault.
I'll continue narrowing down the list of suspects by trying the USB Adaptor on a different computer (a lot newer one having a handful of USB 2.0 ports) and check if there's any difference. I'll return with the results soon.

Are there any more ideas or suggestions?
 
Try using a serial adapter type RS-232 9097. I have two connected via a Digi box coming into a 2003 SP2 server. I have three HB combo sensors on one 1-wire serially connected 9097. I had "issues" running it on parasitic power so I am currently supplying additional power; hence a separate 1-wire network for just these devices. (BTW using two 9097's, two Temp-08's and one Temp-05 for my 1-wire "stuff").
 
Back with the results.
The list of suspects has now been narrowed down to just one culprit: The ancient computer and its ancient USB ports.

I installed the 1-wire drivers to my Windows XP-driven laptop that's about 2 years old, it sits on a docking station and has plenty of USB 2.0 ports available on both the laptop and the station. Installation went without problems, as usual. After reboot, I ran the the "Default 1-wire Net" program. Again, the Adaptor was auto-detected successfully in USB1.

Crunch time. I ran the "OneWireViewer" program. As with the server before, I selected the first DS2438 device on the list and clicked on the "Real-Time Temperature" tab... This time, no exceptions were thrown, no errors were invoked, only successful readings one after another from the sensor box. Also the second box reported temperature and humidity values without problems. There were no problems even after 15 minutes, so I'd call that success. :)

What I conclude from this test is that USB 1.1 and the DS9490R-A are incompatible, at least on this machine (a system with Intel 82815 Northbridge and Intel 82801BA Southbridge). I have a reason to believe it's a bandwidth issue. USB 1.1 has a maximum transfer rate of 12 Mbps. According to the Device Manager, the system reserved bandwidth on the bus is 23% (which is roughly 2.76Mbps), thus leaving about 9.24Mbps bandwidth for the rest of the connected equipment. Since I'm not an expert of 1-wire and its specs, I can only guess where the problem is. Luckily all modern day (and a few older) systems are equipped with 2.0's now, so there's no reason to be concerned about the USB Adaptor. It will work.

Since I'm not ready to scrap the server just yet, I ordered a 1-Wire Serial Adaptor (DS9097U-A) in hopes of getting the sensor network up and running through an available COM port. Let's hope the network goes up without any problems.
Should problems arise, you'll read about it here.
 
The Digi box that I utilize with the serial 9097 is connected to PC via USB 1.1. (two of these set up).

I would tend to believe that the USB 9097 device should work both with USB 1.1 and 2.0 - odd that it doesn't work.
 
1 wire operates at about 16k bits/sec in regular mode, 160k bits/sec in overdrive mode.
Easily within USB 1.1 limits.

Did you install the latest drivers for your mobo when you installed 2003?

You could always get a usb 2 card and try that.

Sounds like its "one of those things" and it'd be easier to throw a few bucks against it rather than spend a lot of time trying to figure out exactly what's wrong.
 
1 wire operates at about 16k bits/sec in regular mode, 160k bits/sec in overdrive mode.
Easily within USB 1.1 limits.

Did you install the latest drivers for your mobo when you installed 2003?

You could always get a usb 2 card and try that.

Sounds like its "one of those things" and it'd be easier to throw a few bucks against it rather than spend a lot of time trying to figure out exactly what's wrong.

Thank you all for your responses, they've been helpful.

The mobo is extremely old. In fact it's so old that Intel chipset driver installer decided not to install anything. Instead it told me that out-of-the-box drivers of the OS are actually newer than the ones the installer had to offer. Since there was no possibility to force install the old drivers, I'm forced to stick with what I have.

USB 2 card crossed my mind, too, but I had to ditch that idea because the server case is a 1U-sized and so strangely made, that it accomodates only one PCI slot - and that slot is already being occupied by a 3Com 3C905C-TX NIC card. So that's a dead end as well.

I have come to a conclusion that this is just as you said, "one of those things" or just plain bad luck with old hardware with surprising incompatibility issues. Since this computer has an available COM port, I've shelled out a few bucks for the DS9097U-A Serial Adaptor and It should arrive soon. If that doesn't work either, I'll scrap the server and start from square one.
 
Here's the latest update:

The DS9097U arrived. I installed it and it worked right out of the package - Well... sort of.
Everything worked without problems in OWV but Logtemp ran into some strange problems, where it started producing error message similar to "1-Wire network not found" after three-four readouts, then came back, took a few reads more and failed again.

Here's the fun part: In order to make those errors disappear completely I had to set the 1-wire network run at Overdrive speed and reduce FIFO buffers on the COM port to the lowest setting possible. I don't know why I needed to go this far, but I think it has something to do with the age of the computer.

The network's been running solid for 24 hours now. Let's hope things keep running smoothly from now on.
 
Back
Top