ce1zzes@prism.gatech.EDU (Eric Sheppard) (08/25/89)
Where can I find a suitable circuit for the FM *Stereo* transmitter? Normal FM modulation is easily accomplished, yet I'm at a loss for duplicating the L+R encoding scheme used in stereo modulation. Eric Sheppard Georgia Institute of Technology, Atlanta Georgia, 30332 uucp: ...!{allegra,amd,hplabs,seismo,ut-ngp}!gatech!prism!ce1zzes ARPA: ce1zzes@prism.gatech.edu -- Eric Sheppard Georgia Institute of Technology, Atlanta Georgia, 30332 uucp: ...!{allegra,amd,hplabs,seismo,ut-ngp}!gatech!prism!ce1zzes ARPA: ce1zzes@prism.gatech.edu
ISW@cup.portal.com (Isaac S Wingfield) (08/26/89)
Eric Sheppard writes: >Where can I find a suitable circuit for the FM *Stereo* transmitter? >Normal FM modulation is easily accomplished, yet I'm at a loss for duplicating >the L+R encoding scheme used in stereo modulation. You might try your local radio station; when I was in that business, we always provided schematics for service purposes. In any event... The classical way to generate the composite stereo signal is to: 1) Sum the L and R channels, in phase (L+R) 2) Invert R and sum to obtain the "difference" (L-R) 3) Provide a double sideband suppressed carrier modulator using a carrier frequency of 38kHz 4) Modulate the 38 kHz with the (L-R) 5) Provide a 19 kHz "pilot" phase locked to the (not transmitted) 38 kHz carrier, at about the 10% amplitude level, and phased so that positive zero crossings coincide. (This to allow the receiver to accurately regenerate the 38kHz carrier for demodulation) 6) Sum all this stuff together, and jam it into the modulation hole on the transmitter (that is, (L+R),(L-R), and pilot) If that sounds too hard, another technique which provides exactly the same result and is much easier, is: 1) Using an analog multiplexer, alternately select the L and R channels at a 38kHz rate 2) Use a phase linear low-pass filter to remove all sampling artifacts above 53kHz (53 is 38 + 15, the highest audio frequency transmitted in composite stereo) 3) Sum this signal with the "pilot" mentioned above Because of the pilot at 19kHz, it is essential that no audio frequencies from the source be allowed to interfere (causes the decoder to lose lock), so *matched* 15kHz low pass filters are required in the audio input lines. ALso, if the output 53 kHz filter mentioned above is not phase linear, channel separation will suffer (maybe badly). Further, depending on how the transmitter's modulator works, it may not be phase linear to 53kHz, again causing poor separation. Incidentally, recovery of the L and R information can be the inverse of either above technique: A) Bandpass and demodulate the 38kHz, and sum as (L+R)+(L-R)=2L and difference as (L+R)-(L-R)=2R B) Use an analog mux to switch the incoming composite signal alternately to the L and R outputs at 38kHz, phase locked to the pilot Although these descriptions are straight forward, as usual in real-world engineering, it's the second-order stuff that gets you, and in the case of composite stereo generators it's amplitude and phase matching and linearity in the filters, and imperfect commutation if using the switching method. It's no easy task to do this well enough to meet the FCC's (admittedly ridiculous) set of specs, which includes (if memory serves) a separation of 29.7 dB at any audio frequency from 50Hz to 15kHz. Think about it.... The L and R channels and their associated filters must be matched such that if the phase of one channel is inverted, the baseband (L+R) output is 30dB down from either channel alone (or, if in phase, the 38kHz DSB modulation is similarly down). Good luck, Isaac isw@cup.portal.com