Search the Community
Showing results for tags 'test bench'.
Dear colleagues, I'd like to share my first project using Unistream. It consists in a test bench for an important manufacturer of industrial HVAC systems and compressors in a new factory in Spain. When a new compressor is finished in the production line, it has to pass the test bench in order to be suitable for selling. These compressors can reach up to 250 KVA of power in 50 or 60 Hz and there are three independent lines: - The first line is for machines from 200 to 460 volts in 50 and 60 Hz and up to 250 KVA. - The second line is for machines from 12 to 48 volts in 50 and 60 Hz - The third line is for machines from 12 to 48 volts in DC. I used Unistream mainly for three reasons: - I needed to communicate with 5 Modbus RTU slaves (Power meters). Comms are quite easy with Unilogic. - I needed to use function blocks to control the great amount of contactors during the commutations. I can also manage them individually in manual mode with a unique faceplate indexing the data of the instance based on the object pressed in the screen. I wanted to integrate also interlock management and the fail-to-start alarms so the function block was almost mandatory. - I had a big datalogging requirement with the test results with the powers, currents, voltages, power factor from all the power meters to allow the operator to analyze the data from the test. There was requirements for trends and for exporting the data to CSV files. So the control consists in a sequence of steps which also requires intervention of the operator to select the frequency, and voltage. Once this is selected, the PLC will close the contactor with the 50 Hz or will start the speed drive with 60 Hz. It will also close the correspondent output of the transformer with the selected voltage and then it will start the fine tuning by regulating a variac up or down acting as a voltage stabilizer. I tried to do this with a PID but it’s quiet complicate since the variac is not analogic. You have to increase or decrease the voltage on it through two digital outputs (up and down). Trying to tune this PID resulted into impossible or not reaching a minimum error between the setpoint and the output or continuously oscillating. So I created my own PID where I sent a continuous pulse or small pulses depending on the distance between SP and MV. With this and a configurable minimum error I could set the MV with less than 0.5 volt error. The part with the communications was quite tricky as I couldn’t communicate with the 5 Modbus RTU slaves at the speed I needed and I had continuous congestions making the RS485 network to crash. Since the values from the power meters were used as the MV for my PID, I needed more speed so we added the TCP module to three of them. Then I had Modbus TCP and RTU but the change in the network performance was very important. The datalogging part was also challenging, because of the amount of data. I added a “record” mode on each line to start the datalogging, create a new CSV and close it once the test was completed. I’d like to thank to my colleagues in Grupo Galiana (Spain) for developing and building this amazing machine and the Unitronics forum people for the help. It was really challenging but the effort was well worthwhile. Thanks to everyone. Best Regards