Electric Mistress tidbit - Thanks Mike!

[Mark Hammer]

In a brief discussion with Mike Irwin this weekend, I learned something which surprised me and which seems to have eluded almost everyone except for MIke.  He said that the clocking system in the Deluxe Electric Mistress is hypertiangular.  The circuit built around the combination of the LM311 and 2N5087 alters the relationship between what the LFO puts out and what the clock delivers, and apparently produces what he described as a 1/x nonlinear clock sweep.

While it does not sweep as wide as some other flangers, this aspect of its sweep may well be one of the design aspects that so many players like about the DEM.  Interesting.

My own sense of this is that such a sweep is particularly useful at slower speeds, but if you're aiming for more bubbly or slow leslie chime-ey sorts of sounds, the sweep waveform/cycle may be of less importance.  Opinions?

[Ben N]

Hypertriangular?

1/x non-linear sweep?


Translate, please?

Thanks,
Ben

[Mark Hammer]

Standard 2-opamp LFOs produce a square and triangular output (the triangle output is used for sweep purposes).  When the triangle wave drives the clock generator (such as an MN3101/3102), the clock spends the same amount of time at each portion of the overall delay range.  

A "hypertriangular" (AKA parabolic, and likely a few other names) LFO is triangular at one end of the sweep and sinusoidal (round like a sine wave) at the other.  The consequence of this is that it slows down the sweep when it gets to the zone where you notice small transitions more, and then starts to pick up speed again as it moves out of that zone.  Think of it like a car that drives fast through the countryside and slows down when you reach a town/village where there's stuff to see.

This asymmetrical sweep has been used in several commercial products (e.g., Small Stone and Ross Phaser, PAiA Hyperflange) and produces a more satisfying sound at slower sweeps.  There are several scanned articles with examples of circuits that will do this at my site (http://hammer.ampage.org).

[Vsat]

Mark and list,
Think of the sound a linear (V/Hz)  VCO gives when swept by a triangle LFO, compared to an exponential (V/oct) VCO....starting from the low frequencies at the bottom of the sweep, a lot of octaves are covered very quickly, then things slow down a lot as the freq (pitch) rises... it becomes downright sluggish, and the last octave from 10 KHz to 20 KHz takes just as much time as the progression  from zero Hz to 10 KHz. This is because the "width in Hz" of a particular octave in the audio spectrum becomes larger the higher in freq you go. In contrast, an expo VCO spends an equal amount of time in each octave as it is swept up in freq... this is generally more satisfying sounding to the ear. A similar analysis can be applied to the movement of "center freq" or "cutoff freq" in a phaser or flanger. Linear sweep sounds sluggish, expo sweep sounds pretty good... but for "tape flange" "stratospheric" sweep an additional boost is required as you approach the top of sweep... using a 1/x or approx 1/x response in the sweep circuit can give this. The "hypertriangle" LFO uses a modified  triangle LFO waveform to sweep an otherwise normal BBD clock circuit to approximate the desired response. The triangle has a little "spike" at the top of each apex. The EH DEM uses a comparator-based 1/x clock in conjunction with an otherwise typical triangle LFO. The clock starts out slow, then speeds up (and the "rate of increase" speeds up at the same time...) as "x" becomes smaller. With this DEM circuit,  the clock is operating at it's fastest near the "bottom"  of the LFO triangle "small x" and slowest near the "top" of the LFO triangle "large x"... reverse to what one  would normally expect with LFO-driven circuits. This makes the delay time vary linearly with LFO voltage.... higher voltage, more delay... similar to how relative delay between tape machines will be varied in actual tape flanging, when the speed of one machine is controlled by hand. It is also a good response to use for pitch-shifting applications - a slow triangle will give a slow "square-wave" modulation of pitch... the exact same 1/x method is used in the Boss Dimension "C" to produce a "vibrato-less" chorus.
Regards, Mike

[Mark Hammer]

Then it appears I have it wrong.  :oops:

I was assuming that the more desirable arrangement was to have the change in delay time s t r e t c h e d at the long delay end, and compressed at the short-delay end of the sweep.  My assumption was that relocation of notches was the significant acoustical event being attended to.  If I've understood this last note correctly, you're saying that it works the other way; that such a sweep spend makes a quick turn-around at the long-delay end of the sweep and decellerates as it approaches zero delay, only to slowly speed up as it makes the turnaround.  From this perspective the acoustical event being attended to is the disappearance and reappearance of notches across the spectrum.

Thinking about it now, that makes more sense.  Especially when one thinks about the through zero point (where they exist).  Normally, one likes to savour the approach to that point.

Question to you, Steve, Ton, STM, and anyone else that understands flangers:  Is the "bounce" that Eventide and Jurgen Haible mimic electronically imposed at BOTH ends of the sweep cycle, or just on one side?  If applied to one sweep extreme, is it applied at the shortest delay or longest delay end?

I'm curious about how this interacts with the nonlinearities in the sweep.

Finally, possible to retrofit a 1/x conversion to the lowly BOSS BF-2?

[Vsat]

Mark, you got it right first time... the desired effect is to spend more time in the "bass-midrange" portion of sweep, and less time in the "midrange-treble" part of the sweep... as compared to the linear clock circuit. The linear clock spends way too much time in the "midrange-treble" part of the sweep, and not enough in the low end.

Interestingly, JFET phasers behave in the same "backwards" manner - with gate-source voltage at zero volts the notches are centered in the high end of the spectrum, when the gate-source is reverse-biased the notches are centered in the low end of the spectrum.
Regards, Mike
P.S. take a look at the DC-2 schematic and you may be able to replace the MN3101 in the BF-2 with  a similar circuit... the DC-2 clock  isn't particularly fast though. LM311 gets up to about 2 MHz if tweaked.

[StephenGiles]

Mark, the bounce occurs at both ends - both ends burning so to speak! However, when I built my TDA1022 Standard Electric Mistress (I must draw the circuit), I incorporated the Eventide bounce, slightly mucked about, and I found that a certain setting allowed the bounce to occur throughout the sweep - not an effect that could be used really (maybe inexperienced youth to impress a girl friend), but certainly different.

By the way as I type this I'm listening to Dick Gaughan's gig last night - strong stuff indeed! We probably understood 20% of his Edinburgh accent!
Still he has a magnificent site, well worth a look with translations of his songs!
http://www.dickalba.demon.co.uk/

Back to flangers, it's interesting what you say about the DEM, Mike, I did have that schematic explained to me in great detail back in the EH UK days, but long since forgotten. I have not seen any other flangers that use a 311 clock.
Stephen

[Arno van der Heijden]

Does this also apply to the original Electric Mistress?

http://www.fortunecity.com/tinpan/humperdinck/223/ehmist.gif

Mike,
How familiar are you with that version? My original Electric Mistress' LFO has a weird asymmetrical sweep. It can be heard on this sample:

http://www.student.tue.nl/T/A.C.v.d.Heijden/Mistress.wav (1.26mb)

I already asked about this a couple of times on this forum, but there's still no solution and I'm desparate.

The Mistress didn't work when I bought it and I had to replace the SAD1024 and some opamps because they were fried. I already changed out the other ic's, but that didn't help.

[Vsat]

Hi Arno,
Yes - the gif you posted uses a similar type of 1/x circuit... an LM339 comparator is used here instead of the LM311. The LM339 is a slower part... but I doubt EH was trying to get super-fast clock speed and extremely short delays.

This type of 1/x oscillator uses a PNP transistor current source to charge a timing cap - the voltage on the capacitor increases in a linear manner with time - one comparator input monitors the cap voltage, while the other comparator input is connected to the LFO - the cap gets discharged when it reaches the (instantaneous) threshold voltage provided by the LFO.

I haven't carefully looked at that particular LFO circuit you show, but asymmetry in triangle LFO outputs is usually caused by having the "fake ground" reference on each op amp improperly centered - the baasic triangle LFO uses an op amp connected as an integrator and another op amp connected as a Schmitt trigger-comparator -  the "Schmitt trigger" op amp will have a square wave output at the (+) and (-) clipping levels for the particular op amp and supply voltage used. To get a nice 50:50 triangle the "fake ground" has to be set to a voltage exactly half-way between the (+) and (-) clipping levels of the square wave generated by the Schmitt op amp. Some level-shifting of the triangle output may need to be done to fit the input voltage range expected by the LM311/LM339 oscillator.
Regards, Mike

[Vsat]

Arno,
The LFO circuit in the gif is certainly unusual... it doesn't use  the normal op amp circuit, rather it uses the two left-over comparators in the LM339, along with the CD4013 dual D-F/F. The thing to remember about the LM339 is that has open-collector outputs, so presents a very high output impedance unless it is in the low state. Your  mp3 is interesting... it certainly shows the 1/x response is working... but the triangle is not 50:50 as you mentioned! I would look at the operation of the LFO with an oscilloscope.
Regards, Mike

[Arno van der Heijden]

Mike,
Thanks for the help.  (Sorry to hijack you thread Mark  :oops: )
I will try to understand the circuit with your explanation. However, I'm not an advanced enough diyer to have an oscilloscope. I may try one of the pc based ones.
In the meantime, is it possible to check the operation with my DMM?

[puretube]

[OT]

I prefer the 2-OTA SmallStone hyperbolic (to add another name  ...)
LFO for "normal" flangers...

one current job that keeps me busy: quadrature hyper...
(guess I`ll have to use the diode approach, that Mark got in his vaults  :wink: ).

[DiyFreaque]

Hmmm.....quadrature hyper.  I wonder if a combination of the docs on Mark's ampage site would work?  

There's the Thomas Henry design for the quadrature function generator, which puts out triangle waves in quadrature (I made a module out of that for a friend - works very well, very stable), then there's the article from Polyphony using a sort of half wave sine distortion circuit to change a triangle into a hypertriangle.  The output of each phase of TH's design is 10Vp-p, which is what's required by the FET based hypertriangular convertor.  You could put the hypertriangle converter at the output of each triangle wave, and have a hypertriangular quadrature function generator (say *that* 10 times fast).

Cheers,
Scott

[Mark Hammer]

hypertriangular quadrature function generator (say *that* 10 times fast).t

If I'm not mistaken, that was a cut on Isaac Hayes album "Hot Buttered Soul" :wink:  (That's for Ton)

[StephenGiles]

I lived in a road called Lullington Garth for 14 years - that's a bit of a tongue twister after a few drinks!
Stephen

[puretube]

Hyperbolicsyllabicsesquedalymistic

...

yes, he was way ahead of his time, then...

my favorite I.H. tune is: "Good Love 69-9-69" from his "Black Moses" album, though.

or maybe: "Do Your Thing" (the best tremolo demonstration song since: "Crimson and Clover").

POLL:
which best wah demonstration song was first: "Papa was a Rolling Stone", or: "Theme from Shaft" ?

(being aware, that "C & C" was more than just mere tremolo, and that "White Room" was before the 2 others...).

[puretube]

BTW.: any of you guys (sorry,  rather: gentlemen)
ever analyzed [or closely scoped] the 2-OTA SmallStone LFO?


Gez?

:wink:

[Eric H]

Since the best was "Rainy Day/Still Rainin," The question is irrelevant ;-)

POLL:
which best wah demonstration song was first: "Papa was a Rolling Stone", or: "Theme from Shaft" ?

(being aware, that "C & C" was more than just mere tremolo, and that "White Room" was before the 2 others...).

[Mark Hammer]

First time *I* heard one was "Tales of Brave Ulysses".  In retrospect, it was a pretty lame use of the thing, with very little comprehension on Clapton's part about what it could do.  He simply moved it back and forth and played, with no real coordination.  Still, grabbed my attention and made me want one.

If you listen to Manu Dubango's "Soul Makossa" you'll want a Schaller for sure.