1. metrology in a fast paced society
2013 NCSLI International Workshop and Symposium
metrology in a fast paced society
July 14-18, Nashville, Tennessee
INVESTIGATION ON DIGITAL CONTROL CONCEPTS FOR DYNAMIC APPLICATIONS OF ELECTROMAGNETIC FORCE COMPENSATED BALANCES
Hanna Weis ¹
Irina Gushchina ¹
Dr. Arvid Amthor ¹
Dr. Falko Hilbrunner ²
Christian Diethold ¹
Prof. Thomas Fröhlich ¹
¹ Technische Universität Ilmenau, Ilmenau, Germany
² Sartorius Weighing Technology GmbH, Göttingen, Germany

2. Outline Introduction State of the art: analog control loop
Improve performance: digital control loops
Controller based
FPGA based
Design of the controller
Results
Conclusions

3. Introduction – How does an EMC load cell work?
1 – Weighing pan
2 – Load suspension
3 – Transmission lever
4 – Coupling member with flexure hinges
5 – Parallel lever system with flexure hinges
6 – Bearing of transmission lever
7 – Position indicator
8 – Electrodynamic actuator

4. State of the art: analog control loop
Lever moves: Different areas of photo diodes are shaded → two signals proportional to lighted areas
Differential amplifier: transformation to one position proportional signal
Fed into controller → controller computes current to compensate lever deflection
Current generates Lorentz force acting upon lever
Measure current, digitize, convert to mass and display
Lever moves: Different areas of photo diodes are shaded → two signals proportional to lighted areas

5. Analog control Three goals for the controller design:
Derive mass as fast as possible
Smooth and stable mass signal required for high resolution
Lever should never touch mechanical stops (leads to nonlinear behavior)
Analog controllers:
Realization as P, PI, PID or PIDT1 controllers
Just few parameters to realize contrary demands
Good tradeoff can be found for systems with PT1 or PT2 behavior
Optimization potential for systems with high frequency resonances restricted due to limited design degrees of freedom → limited measurement speed

6. Increase speed: digital control
Application of EMC load cells for high speed purposes like filling plants
→ short measurement time and high resolution needed
→ Improve controller performance
Advantage of digital controllers:
Not restricted to 3 to 4 parameters
Any concept realizable e.g.:
State space controllers
Observer based approaches
Online adjustment of parameters
Self learning strategies
Implementation and usage of a priori knowledge
→ Suitable hardware for implementation needed

7. Digital control loop Additional hardware needed:
A/D- and D/A-converters
Voltage driven current source (provide necessary current)
Control system (real-time system):
Controller-based
FPGA-based

8. Controller based real-time system
PXI-Chassis:
RT controller: Controller algorithm
ADC/DAC module
Host PC: Host program
PXI-Bus
TCP/IP

9. Controller based real-time system
Host PC:
Hardware:
Windows PC
Software:
Windows
User interface realized with GUI
Connect and disconnect RT-target
Start/stop measurement
Set control algorithm and filters
Specify sampling frequency, measurement range and channels
Display data
Real-time Chassis:
Hardware:
PXI-controller NI PXIe-8102 RT
ADC-module NI PXI 6289 (18/16bit)
Software:
Real-time OS
Program deployed as DLL on controller
Runs control algorithm
Saves data
Realizes communication between:
Controller and ADC-module
Controller and host

10. Controller program on RT-target
Distributing thread:
Establish connection with host
Receive configuration from host
Wait for commands from host to:
Start / stop
→enable / disable other threads
Main thread:
Runs in a hardware timed loop:
Read / write data from / to ADC / DAC
Perform controller algorithm
Put data to DAC queue
Trigger saving and sending thread
Wait for next hardware trigger
Sending thread:
Wait for sending event
Copy fraction of data from last iteration to local array
Send via TCP
Saving thread:
Wait for saving event
Copy data from last iteration to local array
Write data to file

11. Realized specifications
Control loop runs with 10 kHz
Speed limited by bus latencies transferring single values to / from DAC / ADC
Input resolution 18 bit
Output resolution 16 bit
Problem: Delay of one sample between channels (not solved by manufacturer)
→ leads to phase distortion
Improvement: Realize controller on FPGA (Field Programmable Gate Array)

12. FPGA based system PXI-Chassis FPGA: Controller algorithm
ADC/DAC module onboard
RT controller: runs host program
Display
PXI-Bus

13. FPGA based real-time system
Advantages of FPGA:
Direct interface to ADC and DAC modules → no bus latencies
Easy to reconfigure
Applied modular system:
PXI chassis with PC-based real-time controller (NI PXIe-8102 RT)
FPGA-module (NI PXI-7854R) with Virtex-5-LX110 FPGA
Several ADC and DAC modules onboard (16 bit each)
Very good synchrony of channels, low jitter, high sampling rates (>500 kHz)
Host program running on RT controller
Implementation on FPGA with soft core LiSARD: realizes floating point processor
→ realize control algorithm with double precision

14. Realized structure Host: Start / Stop measurement Start FPGA
Send controller configuration with FIFO DMA (Direct Memory Access)
Receive sampled data from
FIFO DMA
Compute weight
Display Data
Save Data
Receive user commands
FPGA:
Read coefficients from FIFO DMA
Receive commands from host
Execute control loop:
Read Data from ADC 1 (position signal), write to FIFO DMA
Perform control algorithm on soft core
Write data to DAC, write to FIFO DMA
Read data from ADC 2 (current), write to FIFO DMA
Execute loop until host sends „stop“
Communication
PXI-Bus
Data exchange
via FIFO DMA

15. Realized specifications
Control loop runs with up to 350 kHz
Speed limited only by size of program running on soft core
Input resolution 16 bit
Output resolution 16 bit
Very stable execution
PXI bus only used for communication between host program and FPGA
→ Now we have a suitable system for the implementation of digital controllers

16. Controller design
Existing: Several papers about digital controllers for EMC load cells
Reduction of measurement time by ~ 30 % (Pfeiffer et. al.)
Very complex control algorithm
Controller concept needed for EMC load cell with max. load 220 g
Mechanical resonance at ~ 475 Hz (transmission lever)
→ unstable for common PID controllers
→ analog PIDT1 controller realized
→ stable but slow
For comparison purposes: design of a digital PIDT1-controller

17. Digital PIDT1-controller
Blue – balance
Magenta – controller
Green – open loop
Red – closed loop
Structural limitations:
controller acts in low frequency range
Increase bandwidth: Increase gain
Resonance should not cross 0dB

18. Digital polynomial controller
Controller design: System transfer function known well
Compensate resonances
→ reduction to (ideal) PT2 behavior
→ only PIDT1-controller needed for fast control loop
→ polynomial controller of order 7
→ Bandwidth: increase from ~ 90 Hz to ~370 Hz
Blue – balance
Magenta – controller
Green – open loop
Red – closed loop

19. Conclusion
Possible to implement digital control concept to EMC load cell
With digital controllers measurement speed can be increased significantly
Therefore: Suitable hardware needed
Comparison of controller based and FPGA based hardware
Problems of controller based hardware omitted with FPGA
Realization of control loops up to 350 kHz
Simple controller design realized
Outlook: with more advanced controller design further improvement in speed and performance

20. References
F. Hilbrunner, H. Weis, R. Petzold, T. Fröhlich, G. Jäger, Investigation in the Impedance-Frequency-Response for a Dynamic Behaviour Description of Electromagnetic Force Compensated Load-Cells, XX IMEKO World Congress, Busan, Republic of Korea, 2012
I. Gushchina, B. Däne, A. Moskalev, W. Fengler, Untersuchung zur FPGA-Implementierung von Mess- und Regelungsalgorithmen, EKA2012 Beschreibungsmittel, Methoden, Werkzeuge und Anwendungen, Magdeburg, pp , ISBN: , 2012
A. Pacholik, J. Klöckner, M. Müller, I. Gushchina, W. Fengler, LiSARD: LabVIEW Integrated Softcore Architecture for Reconfigurable Devices, ReConFig 2011.International Conference on ReConFigurable Computing and FPGAs, Cancun (Mexico), pp , ISBN: , 2011
R. Maier, G. Schmidt, Advanced Digital Control and Filtering for High-Precision Weighing Cells, Proceedings of the International Conference on Advanced Mechatronics, p , 1989
R. Maier, G. Schmidt, Integrated Digital Control and filtering for an Electrodynamically Compensated Weighing Cell, IEEE Transactions on Instrumentation and Measurement, Vol. 38, No. 5, p , 1989
W. Balachandran, M. Halimic, M. Hodzic, M. Tariq, Y. Enab, F. Cecelja, Optimal digital Control and Filtering for Dynamic Weighing Systems, Proceedings of Instrumentation and Measurement Technology Conference IMTC, IEEE, 1995
W. Becker, P. Siebert, Elektromechanische Kompensationswaage mit digitaler Regelung, wägen + dosieren, Nr. 6, p. 2-7, 1991
W. Becker, P. Siebert, Elektromechanische Kompensationswaage mit modellgestützter Messung des Wägegutes, tm Technisches Messen 58, Oldenbourg Verlag, P , 1991
A. Pfeiffer, G. Schmidt, Integrated H∞ Design for Filtering and Control Operations of a High-Precision Weighing Cell, Proceeding of the Asian Control Conference, Vol. 2. p , 1994
A. Pfeiffer, Intergierter κ∞-Regler/Filterentwurf für elektromechanische Präzisionswaagen, Dissertationsschrift, Düsseldorf, VDI-Verlag, 1996

21. Aknowledgement
This project is funded by the Federal Ministry of Education and Research (BMBF):
InnoProfile Transfer:
“Neuartige Anwendungsfelder innovativer Kraftmess- und Wägetechnik”
in cooperation with:
Sartorius Weighing Technology GmbH
SIOS Meßtechnik GmbH
PAARI Waagen- und Anlagenbau GmbH
driveXpert.

22. Thank you for your attention