A large Western NY Company required the ability to apply controlled microwave and radiant heating profiles on
products made up of various chemical composition. There were several challenges that this application had to
- An easy to use interface allowing the user to create and run various temperature profiles, collect and display data.
- Allow the user to select certain characteristics about the sample under test and have the software automatically
determine the appropriate control parameters. That is, shield the operator from going through trial-and-error
selecting the appropriate PID values.
- Control both a microwave heating source and radiant heating source
- Autonomously control the temperature profile once the recipe is downloaded.
To keep cost down, most of the hardware needed to generate the microwave energy was reused, only the control
system was replaced to accommodate requested functionality.
The software developed by Manufacturing Automation Systems allows the user to create and run recipes
(i.e. temperature profiles) on samples placed inside an oven. The temperatures can vary anywhere from
25⁰C to 1000⁰C, and the recipe allows the user to create a custom temperature profile to be applied on the
sample. Data is then collected during the recipe duration, which can last for many hours.
The overall strategy in developing this application was to decouple the user interface, running on a
Windows PC, from the temperature control, running on a National Instruments Real-Time controller
(i.e. compactRIO). When the user is ready to run a sample, a recipe (previously created by the user)
is automatically downloaded from the host computer to the real-time target. The real-time controller
interprets the recipe, and begins to monitor the temperature, control the microwave and radiant source,
monitor various safety interlocks, collect and stream data. Status information, along with the data, are
bundled and transmitted back to the host computer, where the user is able to monitor test conditions.
Because the recipe is running on a real-time system, the host computer can be disconnected without
affecting the test or missing any safety interlocks. The screen shot below shows and example of the data
displayed to the operator during the test.
The software developed for the real-time controller was architected to use several state machines with several
independent loops running in parallel. For example, a safety loop monitored all interlocks during the test and
immediately shutdown the system if an error condition arose (i.e. overheating, oven door open, etc.). A
communication loop was responsible for receiving and transmitting data between the controller and the host computer.
The data transmitted back from the target to the host computer was directly streamed to a tab-delimited text file.
This file can then be easily opened and analyzed in Excel, as shown below:
The system is configured to capture data every 100 msec from all the thermocouples, microwave forward and reflected power,
and other sensors.
Another part of the application, exclusive to the host computer, provides the user with a recipe editor. The editor
allows the user to create a new recipe or modify existing recipes. The user can insert various steps to create the
temperature profile needed for a sample under test. The screen shot below show an example of several test step entries:
For each step in the profile, the user specifies various parameters, such as how the microwave is to apply power
(i.e. step to power level, ramp to temperature, soak at temperature, etc.), or which control algorithm to use
(open-loop, PID, PID with On-Off, etc.). One key objective that had to be met was the ability to shield the operator
from entering specific PID values when using PID control. This was accomplished building a database collection of
material compositions and their associated PID values. If an unknown material behaves similar to one of the known
entries in the database, then those PID gain values are selected. This method tries to remove the tedious trail-and-error
process of determining the optimal PID gain values for a material each time.
Some additional technical highlights:
- A manual diagnostic screen was developed to manually control the microwave power and monitor all temperature sensors
- A National Instruments Real-Time control system, with several I/O modules were used to collect data and control various outputs.
- All machine settings and recipe files were saved in a local SQL database.
- Both the host and target application were built using LabVIEW and LabVIEW Real-Time. The target application made extensive use of parallel execution of several state machines.