We are trying to read - with a CR1000 - the output of load cells that are monitoring a weighing lysimeter. The cells are rated at 0.5 kg resolution; output is around 8 mV at the actual 5V excitation, corresponding to a load of more or less 40000 kg.
We are not interested in the absolute weight of the lysimeter, only in relative variations (one hour to another), so I was wondering which is the best way to offset the voltage measurement (for example, applying -8mV) to keep the measurement inside the 2.5mV range and get the better possible resolution. I suppose it is a bridge measurement instruction that I should use, but I'm not sure which one.
On the other hand, even in the 2.5 m range, the resolution of 0.33 uV of the cr1000 would equal circa 2 kg, seems to me.
The system has an independent display that outputs the measurement to peripherals via a RS232, so I suppose we could read the digital output with a control port; nevertheless I would like to try a direct analog reading before, for simplicity, because I never worked with serial output sensors.
Thanks in advance to anyone who could help.
LT
Do you have more details of the sensor, e.g. a link to a datasheet. You need to know if the sensor is an unconditioned bridge or a conditioned voltage output. The latter seems likely if the unit is capable of directly measuring the sensor and outputting data as RS232.
What I can tell you though is the maximum resolution of the CR1000 is about 1 in 7500 for as a single differntial measurement or 1 in 15000 for a full bridge measurement at best. This will mean you will never get near 1 kg resolution for the analogue measurement of the sensor unless you take repeated measurements and average.
If the RS232 output gives higher resolution you may be better to use that. However, you need to be careful in interpreting that resolution as the normal errors of linearity, repeatability and in particular temperature drift would normally be much larger than resolving 0.5 in 40000, i.e. 0.001% of full scale, with most load cells.
Thank you Andrew, I will follow your advice and I'll start looking into digital data acquisition through RS232 output. I'm not finding too much information on digital data aquisition in the website or in the CR1000 manual, though. Do you know of any tutorial or example program for reading and parsing the RS232 inputs (maybe in the manual of another sensor) that I'm missing? Before starting another thread...:)
This forum is really helpful, and I deeply appreciate the effort you CS people put in answering our questions.
LT
The latest versions of the CR1000 manual have a much expanded section on serial sensors (see section 11.8). Unfortunately there is no standard for any sensor output so each one requires a different solution.
We are releasing a number of tutorials on-line and a serial sensor one is planned but not yet complete. Monitor this page to watch out for new ones:
http://www.campbellsci.com/19_1_773
A general tip, if you can, use the serialinrecord command to read data into a string from sensors which are automatically spitting out data and then if you need to (some just store the string) use splitstr to split the data into variables.
The latest logger operating systems also have a new terminal "W" command that lets you monitor ("sniff") the communications going on through another serial port. This is very useful in seeing the data being sent and received by from the logger/sensor. This lets you check you have the baud rate set right and that the sensor is responding as it is supposed to.
Thanks again.
LT
In addition to Andrew's advice, use the SerialOpen instruction between BeginProg and Scan. SerialOpen configures the communications port with settings required by your sensor. You will need to know how your sensor is setup to configure SerialOpen. Put that one instruction in a program then use the Terminal Mode "W" command to see if communications between the sensor and logger are working. Then go on to the SerialInRecord and SplitString instructions.
Regards,
Janet