Question: Does more random modulation improve modulated delays?

[ElectricDruid]

I'm interested to see where this goes. Random modulation is one of my pet topics. I've often tried to include S&H waveforms and random slopes waveforms on LFOs I've done.

Seems to me with the "tape warble" application, you'd need some idea of what you're trying to copy exactly. The general feeling seems to be that there might be several periodic variations in frequency going on from various bits of the mechanism gumming up a bit - so maybe several LFOs mixed at very low levels would be ideal. And I wouldn't worry it being too subtle. Subtle is good.

Tom

[Mark Hammer]

Agree on all counts.

The first step is always to provide a good detailed description of what the objective is.  Once you have that, the technology required to achieve it becomes much clearer.

[pinkjimiphoton]

well, i think we'd have to consider drift. echoplexes get hot. the hotter they get, the worse they get. if ya use a tape loop, you may be off almost a half a step sometimes from it. thats a significant modulation.
but yeah, slippage, bindage, variations in line voltage all seem to affect it somewhat. i would think a long shallow sine for the wow part would work, and at least one or two random modulations that are very subtle but controlled by envelope i think would be a good way to do it. then ya get the subtle tape echo kind of modulations, and perhaps you could control how bad the "@#$%-up" is from how hard ya hit it.

way above my paygrade still. i'm still learning to make a fuzzbox that don't suck

[Scruffie]

I tried injecting just a noise source today, it led to some very interesting ring mod type sounds.

I would say you have to consider the actual tape going round the loop and what might be affecting it. I'm only guessing on its action here but you have your sine wave modulation matched to the delay time but from the tape stretching etc. that would vary slightly as it goes round the loop so you need a long random modulation on top of the sine, then you have the passing of the heads and reels where each pass probably leads to some unintended extra wobble which would show up as spikes that also match to the delay time but not consistently.

Or, i'm talking out of my posterior.

[highwater]

Theoretically, any tape mechanism will have a significant number of parts which can be out-of-round/bent (sinusoidal variation), and/or that the tape can stick/slip on (sawtooth-oidal variation). Most of those parts can (and will) do both.

Presumably, if you have an appropriate sinusoidal LFO for every contact-point that rotates as the tape goes-by, and a sawtooth-oidal LFO for every point (or, at-least, most points) that the tape touches, it should be possible to emulate any amount of wear/damage/manufacturing tolerance/etc. of an actual tape-based delay. (the characteristics of the write/read/erase heads, and the tape itself, are also theoretically emulate-able, but that's an entirely separate discussion)

As the saying goes, "in theory, practice and theory are the same. In practice..."

I suspect that emulating all of these things with analog hardware would end-up being more complicated than just building or repairing a real tape-based delay... at-least as a DIY project. A cheap microcontroller running and mixing several digital LFOs may-or-may-not be practical.

In any case, I doubt that random modulation (whether directly or by randomly changing an LFO's rate) will give anything that sounds "more" like a real tape-delay... although it could certainly give an interesting sound in it's own right.

[Ice-9]

Tape echo machines where  the tape is on a loop and driven around the record and playback heads will often have  a motor drive wheel a small pulley and a larger pulley, to mimic the W&F of this type of unit first work out the individual speed / frequency of these moving parts, new v's old will also be a factor in the amount of variable modulation that should be reproduced.

[Mark Hammer]

I guess the experiment that needs to be conducted is to feed a tape-based delay with a steady-state signal, and then plot the observed deviations in speed, based on the deviations in pitch.

Should I assume that some of the "big" pedal-makers have undertaken such experiments and implemented their findings?  Have any of you come across a commercial delay that you feel emulates tape (near) ideally?

[robthequiet]

One of the things, and this might be more or less trivial, is that the tape traversing the PB head gap will all be physically synchronous. A simulation using analog parts may introduce phase shift at different frequencies. Not saying it's a bad thing, but I wonder if quality of the effect is/is not noticeable. Digital might come closer if this were the case. Howard over at Catalinbread did a comparison between his pedal and an Echoplex, might be worth a listen.

If the idea is to create a tape-like circuit, the tape head preamp might be something to model, or not. At least the Echoplex provides an example to ponder.

[Mark Hammer]

My sense is that ultimately, this will lead to microcontrollers being used to do the modulation, using look-up tables.

[StephenGiles]

My sense is that ultimately, this will lead to microcontrollers being used to do the modulation, using look-up tables.

Heeeeeeeeeeeeeelp!!!

[Digital Larry]

Stuff like this brings out my cynical side, so just keep that in mind...

As mentioned earlier in the thread, I think smaller speed variations (flutter) are of a stick-slip mechanism, which is probably like a small sawtooth with "random" amplitude, and the wow comes from things being either not round or off center (as in an oil-can delay).  Somebody making a tape machine (my first job after uni was at Ampex, so I got to play with a fair number of them) would do as much as possible within budget to eliminate these artifacts, while someone making a digital product purporting to emulate such a thing will HAVE to make them audible or you will feel ripped off.  I listened to cheap cassettes for years and years and generally I didn't notice these things to the point where they were distracting.  If anything I noticed the high noise floor. 

One of the cooler things I cooked up in my DSP lab is a flanger with a noise + S/H to deliver the random steps, but the rate at which the steps change is very slow.  Also, they don't jump around quickly, they run through a low pass so they can take a second or two to get where they are going.  The end result is that 60% of the time it is NOT moving at all, and when it is moving, it is going at a slow rate for a little while then it stops.  I know I shouldn't give away all these secrets!  But I doubt most people would like this as a guitar effect because it sort of sounds like you have an idiot periodically adjusting your tone controls to random places.  The rate can also be randomized but IME it's good to constrain this to a relatively narrow range.  It really sounds cool on drum tracks or in a break where you want something weird to happen for a little while.

I think it's cool that people "emulate" old things but it also drives me up a tree that some people think this is the holy grail of new technology, to emulate old stuff.  I also think that a lot of people take shortcuts and throw filtered noise onto a modulation signal and decide "there now I have wow and flutter" without really analyzing some old crappy piece of gear.

One of the more interesting bits of old crap I'd like to emulate (see, I'm not immune, I have feelings too) is a Fender oil-can delay I heard on YouTube.  There is the pitch wobble caused by the off center delay wheel and that is easy to emulate.  But when you slow it way down, the sound totally goes to hell and I think this is the result of the oil on the wheel starting to drip off and munging the spatial arrangement of the electrons which form your signal.  But is it truly random or is it subject to Wesson's third law of vegetable oil?  Every now and then these things come up for sale and WOW (pun intended) they are expensive.  If I found one for about $200 I might grab it though.

[Mark Hammer]

Which is why I titled the thread with a question.  I'm agnostic as to whether it is "better".  But if it's a sonic objective for someone, and represents an "improvement" for them in their context, how best to accomplish it?

So far, I think the consensus is that obvious periodic is less than optimal, but so is truly random.  Subtle quasi-periodic seems to be the optimal.  But as you illustrate, how one defines that aperiodic/quasi-periodic can vary substantially, and is probably a matter of taste in the end anyway.

Like yourself, I don't think that emulation of old technology is any sort of holy grail.  On the other hand, sometimes you hear a tone/sound you like and you think "Hmmm, how can *I* replicate that?"  I always liked the solo guitar sound on the Joe-Meek-produced song by the Honeycombs "Have I the Right".  A very distinctive kind of vibrato trill.  I'd love to know how to replicate it....though I'm pretty sure it won't be via modulated echo. 

"Wesson's third law of vegetable oil".  You is funny, Larry. 

[ElectricDruid]

Theoretically, any tape mechanism will have a significant number of parts which can be out-of-round/bent (sinusoidal variation), and/or that the tape can stick/slip on (sawtooth-oidal variation). Most of those parts can (and will) do both.

Presumably, if you have an appropriate sinusoidal LFO for every contact-point that rotates as the tape goes-by, and a sawtooth-oidal LFO for every point (or, at-least, most points) that the tape touches, it should be possible to emulate any amount of wear/damage/manufacturing tolerance/etc. of an actual tape-based delay. (the characteristics of the write/read/erase heads, and the tape itself, are also theoretically emulate-able, but that's an entirely separate discussion)

As the saying goes, "in theory, practice and theory are the same. In practice..."

I suspect that emulating all of these things with analog hardware would end-up being more complicated than just building or repairing a real tape-based delay... at-least as a DIY project. A cheap microcontroller running and mixing several digital LFOs may-or-may-not be practical.

In any case, I doubt that random modulation (whether directly or by randomly changing an LFO's rate) will give anything that sounds "more" like a real tape-delay... although it could certainly give an interesting sound in it's own right.

I'd say that if you've got a decent idea of what exactly you were trying to emulate, you could certainly do a good clone with a cheap microcontroller. You can definitely run several LFOs on one, mix the outputs and spit it out. The toughest bit it seems to me is knowing what you're trying to do. That said, how many people *really* know what an out-of-spec genuine tape delay sounds like? Aren't we really just trying to get something that gets us in the ballpark and fools most of the people most of the time?