The MINT Backend | PrincetonCMB |
![]() Main · Science · Instrument · Backend · Channelizer · Design » Software · Transmission line · Power Splitter · Bandpass Filter · Mixer · Amplifier · Jumper · SMA Connector · Enclosure · Fabrication · Performance · Future Revisions · Correlator · Receiver · Optics · Data Acquisition · Local Oscillator · Base · Analysis · Results · Publications · Team |
Software
Microwave Simulation The design of MINT channelizer revolves tightly around the HP (Agilent) EEsof Design Software website . There are many different versions of EEsof, all of them performing slightly different tasks. Lately, however, Aglinet has been trying to unify all of the programs into one non-redundant package known as ADS (Advanced Design System). ADS is very large (1 GIG of Hard Drive Space!) with many different parts. The package still "feels" like it is many different programs stuck together with some parts not communicating well with other parts. Agilent is working on better integration. Different parts of ADS are disabled depending on the cost of the license, so close consultation with the sales people is critical. The current version that we are using is ADS 1.3 . The software offers two levels of simulation, one in which components are entered in a circuit diagram, and the circuit is simulated using analytical equations. This is sometimes called "linear" simulation. The other type of simulation requires the physical layout of the device. The simulator then performs a 2-D full E&M computation of fields and currents. The user selects the gridsize which, ultimately affects the simulaiton time and accuracy. In general, we used a cellsize of roughly 1/30th of the wavelength.
Minor Calculations There were some calculations that needed to be performed many times to determine the starting point for the physical dimensions of the components. We used MathCad as an easy way to enter large equations in graphical form. If I had to start over, I'd probably use Mathematica since it seems to be more widely accepted, but I used MathCad because of legacy. Excel was also used to solve non-ivertable equations. CAD Although EEsof provides a layout program, it is not quite good enough for a final design to ship to a manufacturer. This is where a traditional CAD package comes handy. We use both Vellum and AutoCad. Vellum is more intuitive but AutoCad is ultimatly more usefull. ADS will produce a DXF file in a special third-party-looking plugin, which can be read by AutoCad. We used AutoCad to place some of the final holes and to arrange the layers in a form readable by the manufaturer. Vellum was used to draw the enclosure. Files Final version of large 2-layer board:
P2layer.dxf
P2layer.dwg
|