Multi-head "drum/tape" delay
Last updated
Last updated
One of the things I love about the FV-1 is how many different things you can do in the realm of "ambience". It can make some pretty good sounding reverbs. Delays are like falling off a log.
One DIY FV-1 project I found early on in my FV-1 quest was "Piet's Echotapper" which was dedicated to recreating all the wonderful multi-head drum/tape delays as used by the Shadows (and others) in the 60's. An SKRM module was available from Spin while they were still selling those. I bought one!
The Echotapper was a well documented DIY project, but it struck me as odd to be so dedicated to a specific type of sound, never mind limiting yourself to mimicing actual products from the real world!
Much of this work wound up in the StanleyFX "Blue Nebula".
I'm fairly sure that some of these programs use a "pot skip" structure, which allows you to turn individual playback "heads" (aka delay line taps) on and off depending on the pot position. I am considering making a "multi-head delay" block which would differ from the existing delay blocks in SpinCAD Designer by the addition of the ability to bring heads in and out with a control pot. Nothing in SpinCAD Designer currently uses a pot skip structure.
So that's a great idea, and doesn't seem too hard. I'll use it as a "how-to-add-a-new-block" tutorial. But in the meantime, how would I do that? So here's what I did. Here's a link to the SpinCAD file.
The heart of this, as so many things I do, is the Three-Tap Delay.
Longer delay times didn't sound that great. Note that here, I have the taps set to a 1:2:3 ratio. I think most drum/tape delays don't do that, but you'd have to do some research to figure all that out. The great thing is that you are not constrained by either physics or history here!
The lower end of the delay range is set by the circled parameter in the "Scale/Offset" block. Multiply this by the delay Time in the Three-Tap block to get the total delay when the pot is at zero. The scaled and smoothed pot value is sent to all three delay tap time inputs so that they track together, allowing you to change the "size" of the delay. You may find that a more constrained range of values suits your taste. Make it happen!
The smoother is set by ear to give some pitch bend when you change the delay time, which is totally optional. I like it so I put it in all the time.
I put all-pass blocks in the input to the delay and in the feedback loop to "smear" the sound. All-passes turn clicks into "puffs" for lack of a better word.
I'm also working on a video where I show you what happens in a reverb algorithm if you set all all-pass coefficients to 0. It's pretty interesting! The point here is that you should also try this without the all-pass blocks to hear the difference. I've linked a "click" simulator wave file you can use.
Next, I show how I manage to use a pot to bring in the heads "one at a time" or close to it.
I use the "Power" block (under "Controls") to warp control signals one way or the other. The top Power block which goes to input 1 gain on both the mixers is set to "5", "Invert", and "Flip". The actual function is y = (1 - (1 - x) ^5). This comes in faster than the pot signal, so the taps attached to input 1 on the 2 mixers will come in first.
The pot signal goes directly to input 2 of the delay mixers, so that is a straight line from 0 to 1 and therefore no need for a power block.
Finally, the lower Power block is set to "5" only so implements y = x^5. This goes to input gain 3 of both mixer blocks. Notice that I did not attach the delay taps to the mixers in the same order. So this means that the taps will come in on left and right with different blends.
The balance between the tap levels on both sides are further controlled by the mixer Input Gain settings.
There's no explicit "mix" control here. With the "heads" pot (#1) all the way at 0, the mixer gains are all set at 0. At 25%, you clearly hear one head. At 50%, two, at 75% all three. When it is at 1.0 (well OK, it's a little less than 1.0), then all the heads are turned up as far as they will go. You can also have more elaborate control schemes where certain heads fade out as others come in, but I wanted to just keep it straightforward here.
I used Shelving filters on the input and in the feedback loop just for the sake of trying those out. My high frequency hearing is pretty bad so I can't always tell what those things sound like.
The feedback is tapped off the longest delay tap, but that is also not a necessary constraint. Try taking the feedback from the shorter delay taps and see what happens.
Another interesting thing to try is to drop in a "Cube" distortion (from the "Wave Shaper" menu) between FB Out and the Shelving Lowpass.
The Cube distortion is a "mild soft clipper". Now with high levels of feedback (pot 2) you start getting an overloaded sound through the delay. Note that this positioning means your first echo is always clean! To use a single distortion block that affected both the input and the feedback, you'd have to add another 2:1 mixer before the Three-Tap delay and take the feedback into that rather than using the built-in path of the delay block. Here's the SpinCAD patch.
Fine tune the maximum gain in the loop using the control panel of "FB In 1".
Another thing I like doing with delays is more aggressively cutting out low end from the feedback loop. Combined with some distortion, this begins to have a surreal effect as the sonic integrity deteriorates. I compare it to the visual of a guy looking at a bunch of reflections in a funhouse mirror, but he's getting older in each successive one. I didn't really aggressively cut out low end in the supplied patch, but adjusting the shelf (deeper) and frequency (higher) in the shelving highpass will get you there.
You can also experiment with putting a limiter in the delay feedback loop which can keep oscillations going for a really long time. I won't say it's impossible to distort but it's worth further experimentation. Also try the other limiters under the "Dynamics" section.
Here's the SpinCAD patch.
