[an error occurred while processing this directive]
  Crableg Resonator Tutorial (Part 1)
  From Design to Testing
     
    Purpose of this document

This tutorial provides an overview of the series of steps from the design of a crableg resonator through the testing of the same bonded die. The second part of this tutorial (Part 2) will focus on testing the bonded die using the MIT Microvision system. As such, it may serve as a reference for new users in learning the system. Data collected on a crableg resonator die at Carnegie Mellon is detailed in Part 3.

Background

Designing a properly functioning MEMS device has become increasingly more complicated as the layouts have become more complex. Few people have the time or money required to wait for a design to be fabricated to then realize a mistake was made in the layout causing thier device to not work as they intended, or to not work at all. Time and effort invested in design can help save time, and money,in the end.

Design includes hand analysis, simulation, and layout. Hand analysis invloves first order calculations and average values. Simulation utilizes software to factor in second and higher order effects. Carnegie Mellon's NODAS is one such simulation tool. Layout may be done by hand, utilizing p-cells, or even auto-generated from a simulation schematic.

After the design is included on a fabrication run, the die will be returned a few months later from the foundry. After the necessary post-processing steps, the die is ready to be tested. The MIT Microvision system is one of many instruments that can be used for testing. This system optically measures the motion of a MEMS device in the x, y, and z planes. Bode plots of the response can identify the resonant frequency and other properties of the device.

Comparision of the simulation and test results can indicate the accuracy of the simulation. The simulation most likely will need to be redone with the actual device dimensions which may be slightly different than the layout dimensions.

This tutorial will follow through these steps for a simple crableg resonator test structure.

Hand Analysis

Hand analysis, sometimes referred as 'back of the envolope' calculations, utilizes first order calculations and average values to characterize the performance of a device. A few hand calculations for a crableg resonator may include the spring constant, the force generated by the comb drive, AC displacement amplitude, DC displacement amplitude, resonant frequency, and quality factor. A few of these calculations are shown here in PDF format.

Simulation

Carnegie Mellon's NODAS was used to simulate and automatically generate the layout for this crableg resonator. NODAS divides a MEMS device into a few simple components - beams, plates, comb drives, etc. The first step is to build a schematic of the device as shown here.

NODAS Crableg Schematic
NODAS Crableg Schematic (click on image for larger view)

The crableg resonator schematic is then drawn as a symbol view that is incorporated into a testbed setup schematic (shown below). The crableg resonator testbed includes a DC source connected to the plate and an AC source to one of the comb drives. The simulation can include an eletrical output, however we are more interested in the mechanical displacement. That is what the MIT Microvision system will measure.

NODAS Testbed Schematic
NODAS Testbed Schematic (click on image for larger view)

Using this testbed schematic, several different analysis types can be performed. An 'AC Analysis' is important for identifying the resonant frequency. The results are shown below for an AC voltage (2.5V and 5V) applied to the comb drive with a 25V DC voltage applied to the plate. The output plotted is the displacement of the node identified in the schematic. For a 5V AC drive, the output is the red line with a resonant frequency at 52.4 kHz with a peak amplitude of 4.79 micron. For a 2.5V AC drive, the output is the yellow line with a resonant frequency at 52.4 kHz with a peak amplitude of 2.39 micron.
Note: Exact values from the AMS run may not have been used for this simulation. The values for metal thickness, Young's Modulus, etc. ought to be verified.

NODAS Simulation Results
NODAS Simulation Results (click on image for larger view)

Layout

The last item needed is a layout of the design. NODAS has the ability to automatically generate the layout from the schematic that we built. The automatically generated layout is shown below. This layout would be instantiated on a chip and connected to bond pads.

Crableg Layout
Crableg Layout (click on image for larger view)

The actual die that was manufactured is shown below. This die is commonly referred to as Salil Desai's MIT Resonator chip because Salil, now at MIT, designed the larger six degree-of-freedom resonators on the left side of the chip. Five instantiations of the crableg resonator are included on the right hand side. There are three bond pad connections for the resonator: the plate, the right comb drive, and the left comb drive. Two other important bond pad connections are the substrate and the metal 3 field. This die was included on a tape out to AMS.

Salil Desai's MIT Resonator Die
Salil Desai's MIT Resonator Die (click on image for larger view)

Now that the device is designed, it is ready to be tested after it is manufactured, released, and packaged. The second part of this tutorial steps through the testing of this device using the MIT Microvision system.


Continue to Part 2

Continue to Part 3




[an error occurred while processing this directive]