4049 in Ultraflanger, Why ?

[Yuan Han]

Looking at John Hollis's UltraFlanger:

http://www.hollis.co.uk/john/ultraflanger3.jpg

Why is there a need for 4049 ? Seems like it just inverts the signal from the 2 outputs from 4046 and then send them into the BBD ?

Is it because MN3007 is clocked fast (for short delay) and thus a 4049 is required to "buffer" the signal ? (buffer in what sense ??)

thanks

[puretube]

the 4049 makes really steep sloped squarewaves
from the not perfect 4046 outputs,
and offers low enough output impedance
(or call it: delivers enough current)
for driving the capacitive BBD`s clock-inputs.

[Yuan Han]

mmm, but in zombie chorus, the 4049 is not used. and the 4046 is connected straight to the BBD

Or does it just sound better when the slope is steep ?  So that the LFO spends more time at the plateau, hence sounding "wider"

[puretube]

chorus prob. got slower clock;
slopes of 4046 degrade at higher freq;
BBDs demand exact 50% duty-cycle steep clocks -
otherwise more noise or malfunction.

[Mark Hammer]

To expand on Ton's note, remember that a BBD is really just a bunch of FETs and caps.  Switch one set of FETs, and the charge stored up in the caps can "bleed off" to another set of caps.  The caps themselves don't really "care" where they bleed to.  Could be forward, could be backward.  What allows it to go in a specific direction and ACT like a shift register is the sharpness of the switching action of the FETs.  If the FETs were controlled like those in a phase-shift stage, slowly increasing then decreasing in resistance, the BBD would not function as intended.  Consequently the clock pulse acting on the FETs has to be truly square and not siinusoidal, trapezoidal, or triangular.

If you look at the specs/datasheet for the MN3007, you'll see that they don't indicate/recommend clocking it above 100khz or so.   That doesn't really yield a short enough practical delay time for great flanging.  It IS sufficient delay time for chorus, but not for achieving "the big jet" like you'll find on an A/DA or some others.  Now, you CAN achieve shorter delays at the same clock rate by using fewer delay stages (e.g., a BBD with 256 or 512 stages), but that also tends to constrain the width of sweeps because what it takes to squeeze 10msec delay out of 256 stages just ends of reducing signal quality.

As Ton points out, the clock inputs on the BBD have a certain input capacitance.  As long as the clock signal remains below a certain frequency, that capacitance will not alter the shape of the clock signal.  It will be as nice and square when it hits the input pin as when it left the MN3101.  Once the clock signal exceeds a certain frequency, that input capacitance starts to "un-square" the clock signal.

What permits the clock signal to exceed that apparent limit is the amount of current packed into it.  The MN3101 and 3102 are limited in terms of the amount of current they can deliver to the BBD in the clock signal.  The specs you see listed are based in the assumption that you'll have an MN3101 directly connected to an MN3007 (or 3102 to a 3207), since they obviously can't make any other specific assumptions about what you'll drive it with.  

Anything that can INCREASE the current of what's coming out of the MN3101, however, permits it to deliver a clock signal the BBD *sees* as nice and crisp.  Consequently, buffer stages between clock generator and BBD are recommended for flangers or any other application where you need ultrashort delay times or fast clock rates produced (e.g., squeezing out chorus-range delay times from an MN3005, or worse, two in series).  One way of delivering more current is to parallel several invertor sections, such as those on a 4049 or 4069.  I'm sure there are other ways, but these are cheap and convenient.

Using this method, Mike Irwin has assured me (and demonstrated) that he can clock an MN3007 well out past 1mhz and darn close to 2mhz, yielding considerably shorter delays than a standard 3101/3007 combo.  You will note the presence of such buffering on the A/DA, which is also reknowned for how wide a sweep and short a delay it can generate.  The SAD-1024 was usually adopted for applications where ultra-short delays were desired because it is spec'd to clock up to 1mhz in the absence of such buffering.  So, while you can make an MN3007 behave much like an SAD-1024, *if* you are willing to put in the extra parts, the MN3101/3007 combo became a simple and "close enough for rock and roll" approach to design.

[puretube]

tnx Mark, for taking more time than me... (you said it!)

Have a happy easter!

[Yuan Han]

Wow !

Thanks Mark and Puretube !

Very useful information

Happy easter!