dsi@unccvax.UUCP (Dataspan Inc) (09/07/85)
>Path: unccvax!mcnc!philabs!cmcl2!seismo!brl-tgr!almsa-1!control >From: "Leonard N. Foner" <FONER%MIT-OZ@MIT-MC.ARPA> >Subject: Schematics for stereo FM modulator needed Leonard Foner has asked for schematics to an stereo FM modulator. I thought I'd whip out my Optimod-FM 8000 service manual, but this is indeed a complicated home construction project. Instead, I have described in ten easy steps what one has to do in order to make a reasonably decent stereo generator that I've actually built and tested. It gives about 50 dB of separation when adjusted correctly, and doesn't sound half bad. If I'd have put it on just one breadboard and paid attention to detail, it might even do better. (Mail to ARPA always fails for me... sorry if this is boring!) DO NOT, repeat, DO NOT build this circuit and connect it to your 10 watt educational FM transmitter (or whatever). I know how tempting this is, but the Federal Cookie Company will have your hide. It does NOT meet broadcast specifications. Step 1: You need a 38 kHz (or harmonic thereof) master oscillator...sine wave, very high stability (4 hz) Step 2: Divide the 38 kHz signal in a JK flipflop or whatever. This will be your stereo pilot signal after you . . . Step 3: FILTER the 19 kHz square wave. (and I mean, filter) You want no more than 5% THD in this sine wave. Might as well make it critically coupled, double tuned and all that jazz, so you can then Step 4: ADJUST THE PHASE of the 19 kHz signal. You want to be able to set the 19 kHz signal in such a way that its zero crossing is coincident with one of the zero crocssings of the 38 kHz signal Step 5: Perform the same filtering trick with the 38 kHz signal. Step 6: To one MC 1496 doubly balanced modulator, add the necessary R's and C's to get it to be a DSB type device. In the "L" port you would apply the 38 kc stereo subcarrier. Provide an offset pot so that you can null out the subcarrier (remember, you are generating DSB!!!) The data sheet for this device in several manuals have test circuits which I copied verbatim in my prototype. In the "R" port you would apply the "L-R" audio (see below). The X (output) port would then be your stereophonic subcarrier. Step 7: Get some TL084 op amps (you want definition, right?) and build a 1:1 summer and 1:1 subtractor. This will make your L+R and L-R signal. Then, build a crude, but effective preemphasis filter, simply by putting a 16 k ohm resistor in the feedback loop of an op-amp, and to the (-) summing node a 16 k ohm in parallel with 4700 pf. Yes, I know that the frequency response goes up to lalaland, but you've fixed that somewhat by making sure that your input buffers are integrators with -3 dB points at 15 kHz. Not FCC grade, you see, but effective. You need preemphasis in both the L+R and L-R channels, and could do it to either the (L and R) by themselves, or the sum and difference (I selected the inputs). Step 8: The L+R signal modulates the main channel directly. The L-R signal goes to the MC1496 as above The 19 kHz signal modulates the main channel directly. Step 9: Sum all of this mess. Here is what you have to do: With L and R signals, equal phase and amplitude, you must not exceed 75 kHz deviation (including the pilot at 10 %. This is 100% modulation) With L and R signals of opposite phase and equal amplitude (just as before) the stereo subcarrier modulates the carrier 75 kHz just as before (the main channel should be quiet). The pilot subcarrier should be injected at 8 - 10 percent modulation. With an L or R only signal, and 10 % pilot modulation, the main channel should contribute 45% of the signal, and the stereophonic subcarrier should contribute the other 45% (at 100% modulation). Step 10: Now the fun part (calibrating this mess isn't the fun part) -- finding a modulator which will actually respond to all this mess. If you plan to use Mr. Microphone you'll have to get in there and fudge around with the direct FM input filter to the varactor diode. You really need flat response from DC to 75 kHz (at least) and good differential distortion characteristics, for proper "soundstage" and all that jazz. Step 10a: A spectrum analyzer will make your life infinitely easier, if you don't know what a baseband signal looks like in the time domain. If all you want to do is turn on people's stereo lights, a 19 kHz oscillator will do fine. BE CAREFUL with input filtering - if you plan to play CD's over an FM stereo transmitter, you'll find that some of them will cause the stereo decoder to "false" and attempt to track some of the transients rather than the pilot subcarrier. This sounds quite offensive on some radios. You may have to implement a higher order lowpass before sum and difference (and preemphasis), elliptic functions are preferred because of their improved group delay characteristics as you approach the stop band. THE SECRET TO GOOD STEREO SEPARATION is to make sure that the pilot subcarrier and the carrier part of the DSB modulated signal cross the axis at the exact same time. You adjust this not before the transmitter (which can do nasty things due to its group delay anomalies and high "Q" PA cavities) but with a high quality, calibrated receiver. Significant deviation from this will result in a very trashy stereo signal, regardless of signal strength. Happy stereo generation! If you do this right, you will get excellent performance for a total expenditure of $30-40, not counting some of the coils you'll need. Use TV horizontal oscillator coils (which have a very wide tuning range and high Q) available from good junked TV's everwhere. David Anthony CDE DataSpan, Inc