ELEKTOR FORUMS

Answering my own post. I found a second problem - I re-read the DS18B20 datasheet, and saw that the Vdd connection had to be grounded for parasitic power operation - I was only connecting to 2 of the pins - ground and DQ. Now the sensors produce the correct readings continually. Note that it still needs the lower value pull-up resistor else you get only 127.9.

Tony

I did some temperature tests on a sensor, and have discovered that the NET_THERMO code displays the wrong temperature below 0 deg C. The data sheet for the DS18B20 shows that the ADC is a signed two's complement number, so below 0 deg C 4096 needs to be subtracted from the reading to display the correct temperature. The software actually displays it as -(reading) whereas it should be (reading-4096). I note that the log file shows that temperatures below 0degC have two '-' signs before the number - I wonder if that confuses things - Excel shows the values as +ve.
As a matter of interest, I tried putting the sensor in cold nitrogen just above a layer of liquid nitrogen, and it still works down to -127C - way below the spec of -55C. It still seems to be doing something below -127 but the value changes to a small positive number which reduces from 126.6 down to 106 when the N.R. (no reply) value is returned. Warming up gives the same sequence, so the sensor was not damaged by this brutal treatment.

Thanks Jan - I had overlooked the Corrections tab.

Tony

[Taurus2 is Carlo Tauraso, the designer of the project]

I've developed a new release (v1.2) of NET-THERMO that fixes the negative temperature value problem and replaces the log file
separator character. During temperature monitoring you can find
the data packet in the StatusBar. So, have a look at the attached
image (form-new.tiff) in order to understand the temperature
conversion process. The transmitted data packet is 254 136.
Convert it to bin: 11111110 10001000. To calculate the real value,
invert it by replacing all the 1s by 0 and 0s by 1, 00000001
01110111 and then add 1. The result is:101111000 bin = 376 dec. Multiply it by 0,0625 and you have 23,5 C the correct negative temperature value.
I confirm that the firmware is for Microchip C18 Compiler.
Some DS18B20s need more current using parasite power, so you can reduce the R2 resistor value till the minimum limit
of 1,5K. For long lines and many slaves it should be necessary to use an external FET as explained in the article.
The thermo-snake needs only two wires because you connect together Vdd and Gnd directly on the sensor.

Thanks to Tony18M and John-H for their contributions.

I have just downloaded the 070122-11.zip, and it contains just a single *.msi file dated as of 10.10.2008 with no source code or *.hex file. After instalation, only the NET-THERMO.EXE is installed.

Please advise.

For simple projects, they may provide the source code or hex file
Complex and full fledged projects expect the DIYers like me and you to start learning how to write software or perhaps pay for a pre-programmed (locked though) chip.

Distinguished editorial, why is not possible to obtain the source code of firmware and software for PC?
This undermines the credibility of such a distinguished magazine ... : (

I am using the thermo snake project to monitor temperatures over a day and I'm finding the amount of data being logged makes producing graphs difficult - can anyone advise me on how to increase the timing between logged temperatures (to 10 seconds) in the software? or alternatively has anyone developed a software program for the project that they would be prepared to share? Kind regards,
Brett.

sorry does anyone know of a way to use teh DS18B20's direct with a datalogger ?
again thank you

How long is your wire?

I found a problem that the maximum length of the twin wire was only 30 m. That is too short for my purposes.

Does anyone know wha tis maximum length of wire ? What tricks there are to lenghten the wire.


regards
Kimmo