Teaching aid for digital signal processing

Za učinkovito učenje digitalne obdelave signalov je potreben učni pripomoček, s katerim lahko enostavno izdelamo in preizkusimo digitalne sisteme. Izdelali smo razvojni sistem, ki zajema zvočni signal iz poljubnega predvajalnika in ga po digitalni obdelavi, ki jo sprogramira študent, rekonstruira in primerno ojači za predvajanje s slušalkami. Vhodna stopnja sistema je diferencialni ojačevalnik, ki odpravi morebitne sofazne motnje. Sledi mešalnik, ki lahko vhodnemu signalu doda signal enega razširitvenega modula. Razvili smo štiri razširitvene module: generator sinusnega signala, generator belega šuma, predojačevalnik za kitaro in ojačevalnik za mikrofon. Združen signal nato potuje skozi vhodni filter na analogno digitalni pretvornik (ang.: Analog to Digital Converter – ADC). Razvojni sistem ima namesto vgrajenega mikrokrmilnika, podnožje za razvojno ploščico NUCLEOG431KB, s čimer lahko bolj plastično prikažemo, da je digitalni filter le algoritem, ki teče na poljubnem računskem stroju. Mikrokrmilniški razvojni ploščici sledi digitalno analogni pretvornik (ang.: Digital to Analog Converter – DAC) in temu še ojačevalnik za slušalke. Ugotovili smo, da je šumni prag pred ADC in za DAC enak, in sicer znaša –120 dBV2/Hz. Več šuma je pri nižjih frekvencah zaradi vpliva omrežne frekvence. Razmerje med signalom in šumom sistema, za čisti sinusni signal z amplitudo 0,25 V, kar predstavlja zmerno glasnost, znaša 54 dB. Frekvenčni odziv vezja smo preizkušali z virom belega šuma, ki je pokazal, da sistem duši signale višje od 20 kHz. Dokazali smo, da ima generator sinusnega signala najboljši THD pri nižjih frekvencah, kjer znaša 0,68 %. Izmerili smo tudi frekvenčno karakteristiko mikrofona. Presluh med obema kanaloma je enak –54 dB. Prisotnost šuma in presluha sta dovolj majhna, da nista moteča pri poslušanju obdelanega signala.Effective learning of digital signal processing requires a learning tool that allows us to easily build and test digital systems. We designed a development system that captures the sound signal from any player. The sound signal is then digitally processed by program written by students. After that it is reconstructed and suitably amplified for playback with headphones. The input stage of the system is a differential amplifier that eliminates any common mode noise. Next comes the mixer, which can add the signal of one of the expansion modules to the input signal. We have developed four expansion modules: a sine wave generator, a white noise generator, a guitar preamplifier, and a microphone amplifier. The combined signal then travels through an input filter to an analog to digital converter (ADC). Instead of an embeded microcontroller, the development system has a socket for the NUCLEO-G431KB development board. That allows us to show transparently that a digital filter is just an algorithm that runs on any computing machine. The microcontroller development board is followed by a digital-to-analog converter (DAC) and a headphone amplifier. We found that the noise floor before the ADC and after the DAC is the same, i.e. –120 dBV2/Hz . There is more noise at lower frequencies due to the influence of the mains frequency. The signal-to-noise ratio of the system, for a pure sinusoidal signal with an amplitude of 0,25 V which corresponds to moderate volume is 54 dB. The frequency response of the circuit was tested by a white noise generator, which showed that the system attenuates signals higher than 20 kHz. We proved that the sine signal generator has the best THD at lower frequencies, where it is equal to 0,68 %. We also measured the frequency response of the microphone. The crosstalk between the two channels is equal to –54 dB. The presence of noise and crosstalk is small enough to not affect the perceived quality of the processed signal