Hayden Allen and Andrew Yamada
Our group's focus was to design a new low pass filter for the vector network analyzer. Some of the rationale behind our specific project focus was group labor efficiency and project impact. In terms of work efficiency, our group did not want to duplicate any of the research that the other groups were doing. There was not a specific group that was tackling the low pass filter design. In terms of project impact, a better low pass filter design was imperative to sending a cleaner signal to the Analog to Digital Converter (ADC).
The original passive low pass filter from the 2017 vector network analyzer was sending an intermediate frequency of close to 500 Hz. Based on group 2's research, 500 Hz was a frequency that was heavily affected by radio frequency mixing. Group 2's research also pointed to 250 Hz as a good intermediate frequency target. Therefore our group wanted to design a new low pass filter that caused an intermediate frequency of around 250 Hz.
Our first couple of attempts at a redesign of the low pass filter were all passive. Originally we wanted to use the space on our current PCB. We did not want to redesign an entire PCB. However the PCB on hand only had contacts to support a passive filter of the fourth order. After trying many different configurations, the way the original passive RC low-pass filter was set up, we began to mess with some of the values of the resistors and capacitors. However, we felt that this type of filter was not ideal and could be made better using an active component. So looking into active low-pass filters, we found many different types of designs of filters such as Sallen-key, Cheby-Chev, Butterworth,... etc. We decided we wanted our filter to be 4th order so our gain can drop off at a steeper slope to make sure we cut off other unwated frequencies at a quicker rate. So we decided to use two of the same Sallen-Key 2nd Order Low-pass filter designs cascaded together using a non-inverting op-amps with unity gain setup. We decided to keep all of the resistors and capacitors used in the filters to be the same to be easier for calculations, costs and setup convenience.
Once we had the filter design finished we ran into the issue of meeting the range specifications of our ADC. Our ADC takes in data from the voltage range of 0V - 2.5V. With our reference as 0V a signal of amplitude of ±1V would clip off our data below 0V. So we needed to add a biasing circuit that will raise our signal to reference at halfway between our range for efficiency purposes. So we looked into different ways of biasing the circuit to get our desired results. We tried implementing different methods one being a split resistor method and another that included another opamp for buffer benefits. However, the split resistor method did not fit well with our design and is known to have a lot of noise and the use of another opamp would push up costs and PCB space and was not worth it. But by placing a capacitor in series with a voltage divider we were able to add the DC bias without affecting our filter. This circuit looks like a passive RC high pass filter as a small signal with the two resistors in parallel. So by choosing high values we make sure to not cutoff our lower frequencies.
