janzen@pldvax.DEC (Tom J. LMO2-1/E5 279-5421) (03/27/86)
It was asked by Sam, why do RAM-based delays change pitch when the delay length is change. I also have a DOD Digitech 1.9s digital delay. I am using, incidentally, a spring reverb, a CCD reverb,and the digital delay in my performance on Saturday 29 Mar at Mobius. I agree they all have different sounds. The spring blurs and filters the sound; the CCD is noisy, and the Digital one is just an echo w/o filtering. Anyway, the reason that Digital delays change pitch when you change the delay depends on how you do it. the digitech delays have two controls (the cheap ones) for delay length. One of them, the button, which all have the same function despite having different labels, (flange, chorus, double, delay), merely select different lengths of RAM to use before the address counter overflows. There is only one binary counter in my Digitek. On Delay, it uses the whole RAM. On Double, (I'd have to look at the front panel, at home to see the ms reading by the button) it uses some 2**n fraction, like maybe 1/4. Chorus might use 1/16, Flange maybe 1/32. Read the front panel. It's all there in front of you. Now. The other way to change the delay is to turn the delay vernier knob, while the buttons select a certain range of delay. This is common in instrumentation; first you select the range, then you use the vernier to get exactly the value you want. Anyway, the vernier does change the amount of RAM you use; it changes the RATE, in and out, than I mean THAT data go in and out of the RAM. on the Digitech, data must go into the RAM at the same rate it goes out, because there is only one counter which points at the same time to the Read address (it must read first) and the Write address. There is only one master clock. As you turn the knob, the Read/Write master clock changes frequency. However, some data still in the RAM was put in at another sampling rate, before you turned the knob. So, until the RAM is full of new data at one sampling rate, reached after you stop fiddling with the knob, the output is going out at a different rate than it went it, thus , a pitch change. Turn the delay vernier up, and the pitch goes down, because you're slowing the master clock to less than it was when data which is still in the RAM was put in there. Turn the delay to a smaller delay, and the pitch rises because you're making the sample clock be higher than it was when the data in the RAM went in. Soon, the data in the RAM catches up though. A harmonizer must have separate clocks and counters or a microprocessor to constantly change the pitch. The digitech cannot be economically changed intoa harmonizer, in my opinion. It is like an old TV; everything is melded together in an integration (I don't mean chips) of function that is the opposite of modular. I use this effect in my performance at Mobius in Boston on 29 MAR, incidentally. Chorusing, flanging, and swooping are all the same technical effect. The Digitech Frequency modulates the frequency of the master/sample clock. On flanging, only a tiny RAM is used, like maybe 2K words, so there is not enough time for the read data to have been put in at a much different rate. With feedback, it contructively/destructively interferes at different frequencies at different times with itself, oi la, flanging. Chorusing use more RAM, so the read data was put in at a significantly different rate than it goes out, so its pitch changes as the modulated clock, (same modulating circuit and waveform as flanging) and you get vibrato. With feedback it is out of tune with itself, changing the pitch different with the clock modulator, oi la, chorusing. Swooping is the same modulation circuit, but uses more RAM, so that the difference between the rate a particular string of data went in is much different from how it went in, so it swoops in pitch. So, flanging, chorusing and swooping are all the same electronic circuit. Tom Janzen DEC 111 Locke Marlboro MA 01752