Nautilus Situation (Madbean project)

[Luqqas]

Hi there!

Just finished my Nautilus (Madbean project for Mu-Tron III). It works pretty much good. All modes and range and sweep. The effect works.

The only thing I found out of normal is that when I hit a string, as the notes sustain it makes like a "fizzy" noise... like a really low fuzz... it saturates. This happens in somes modes and depending if Range and Sweep mode is set.

Triple check everything. Accidentally discovered that if I disconnect the power and I turn ON the effect, the signal goes through. Is this normal? Does not supposed that if there is no power through the circuit non signal should pass?

Any opinion is welcome!

[Mark Hammer]

Yes, it is completely normal.

Keep in mind that the diodes (especially D2) will not pass any signal unless it exceeds the forward voltage of the diode.  While the signal is strong, it will always exceed the diode's forward voltage.  As the string starts to decay, the gain of that op-amp stage can help keep the signal high enough, but after a certain point, you end up with a lot of what is referred to as "envelope ripple".  The ripple consists of a few things.  One is the intermittent "micro-peaks" that exceed the diode Vf now and then.  Another is the undertones/beats that well-used strings can generate (pluck the string and watch for those blue and red illusions).  Those make the filter wiggle slightly at high speeds, and it sounds like distortion, because of the speed, but it isn't distortion.

There are generally two strategies for dealing with it.  C9 (4.7uf) is what is referred to as the "averaging" capacitor, which serves to both momentarily hold onto the envelope current produced, as well as smooth out the momentary variations in envelope signal.  If one makes the value larger, it will charge up and release more slowly, smoothing out any of the minute bumps/ripple in the envelope, just like smudging oil pastels with your thumb would.  The price you pay for that is that you can't get fast attack times, and the release can be much slower than you'd like.  If you make it smaller, it will get you the fast attack and release times, but the ripple may become more audible.

The compromise is to keep the fast attack, but play with the release time.  In this instance the 47k resistor, labelled R15 "bleeds off" what is stored in C9.  Consider replacing it with a 15k fixed resistor and 100k pot in series, to achieve a total resistance of 15k-115k, so that you can adjust how quickly C9 drains off.  When the release time is quick enough, the filter sweeps back down before the ripple sets in.

[Luqqas]

Yes, it is completely normal.

Keep in mind that the diodes (especially D2) will not pass any signal unless it exceeds the forward voltage of the diode.  While the signal is strong, it will always exceed the diode's forward voltage.  As the string starts to decay, the gain of that op-amp stage can help keep the signal high enough, but after a certain point, you end up with a lot of what is referred to as "envelope ripple".  The ripple consists of a few things.  One is the intermittent "micro-peaks" that exceed the diode Vf now and then.  Another is the undertones/beats that well-used strings can generate (pluck the string and watch for those blue and red illusions).  Those make the filter wiggle slightly at high speeds, and it sounds like distortion, because of the speed, but it isn't distortion.

There are generally two strategies for dealing with it.  C9 (4.7uf) is what is referred to as the "averaging" capacitor, which serves to both momentarily hold onto the envelope current produced, as well as smooth out the momentary variations in envelope signal.  If one makes the value larger, it will charge up and release more slowly, smoothing out any of the minute bumps/ripple in the envelope, just like smudging oil pastels with your thumb would.  The price you pay for that is that you can't get fast attack times, and the release can be much slower than you'd like.  If you make it smaller, it will get you the fast attack and release times, but the ripple may become more audible.

The compromise is to keep the fast attack, but play with the release time.  In this instance the 47k resistor, labelled R15 "bleeds off" what is stored in C9.  Consider replacing it with a 15k fixed resistor and 100k pot in series, to achieve a total resistance of 15k-115k, so that you can adjust how quickly C9 drains off.  When the release time is quick enough, the filter sweeps back down before the ripple sets in.

Incredible explanation. It's much more clear now. Release time and decay is set by default in this project.

I'm wondering if tantalum caps will provide better ripple rejection. Perhaps its a nonsense, but somewhere I read something about it.

Thanks. I'll try and report back what happens.

[Mark Hammer]

I'm wondering if tantalum caps will provide better ripple rejection. Perhaps its a nonsense, but somewhere I read something about it.

Tantalums may well be better for hanging on to a stored voltage, but in this particular instance, you don't want to hang on to anything.  Rather, you want it to leak out at a rate that works best for you.  In that respect any garden-variety electrolytic will be good enough, and the onus shifts from the cap to the suitability of the "bleed-off resistance".

As with so many tidbits of info one gleans in this hobby, context is important.  Something that is true and important in one context can be relatively moot in another.  The hard part seems not to be amassing factoids, but keeping abreast of the nature of the context, and the relationship between the factoid and that context.

On a related note, when this noise gate project first appeared in Modern Recording magazine ( http://hammer.ampage.org/files/GainesNoiseGate.PDF ), it was followed by a couple of letters to the editor in the month or two that followed, concerning what the letter-writers considered to be annoying "distortion".  The designer's suggestion was to increase the value of the averaging capacitor, as I've done here, and for precisely the same reasons. 

Whenever one is dealing with an effect that uses a sidechain and envelope control (autowah, limiter, compressor, noise-gate) - especially control over fairly quick events - ripple can result in something that sounds like distortion.  Use of LDRs tends to improve immunity to ripple, compared to say, envelope control of an FET or OTA, or bipolar transistor, or switch, because the LDR has a certain built-in sluggishness that "smudges" ripple even further.  But even LDRs have their limits, and nobody is going to sacrifice the responsiveness of the effect in rabid pursuit of eliminating all possible ripple.

[Luqqas]

But even LDRs have their limits, and nobody is going to sacrifice the responsiveness of the effect in rabid pursuit of eliminating all possible ripple.

That's the tricky point or debate.

Well, this is what I get:

I've tried with 15K fixed resistor in series with 100K pot.
Also with 10 uF in C9 instead 4.7 uF.
Also with bass (earlier just with guitar).

With bass and guitar ripple disappear just a little. It's more noticeable when Range switch its in [funny... I don't know when is "Up" and when is "Down"] but... referring to schematic, when Pin 10 of IC2 is connected to R14, R15, R16 an C9.

When is connected in the other Range position -Pin 10 to R18 and R19 and the other half of the switch to R11- ripple is practically off.

Here is the totally weird thing (just realized now, not before):

In one of the position for Range switch if I only left my hand over the string, not hitting a string, just slowly touching it, I hear a ripple before hit a note o effect itself making the envelope.

Imagine that with my left hand a play a B on 3rd string... I hear the ripple.

So I hear before and after.

[Mark Hammer]

When the switch connects R18/R19 to pin 10, that is the "down" position.  R18/R19 provide a constant DC voltage to IC2_C (by dividing down +9V), and the envelope voltage is subtracted from that.  Keep in mind that the higher the DC voltage coming out of IC2_C, the brighter the LED in the optoisolator shines, the lower the LDR resistance, and the higher the filter is set.  So, the "up"-sweep position starts from little or no DC voltage coming out of IC2_C, leaving those LEDs dark, the LDRs at max resistance, and the filter set low.  Strumming makes the LEDs light up and the filter sweep upwards.  In the "down" position, the LEDs are lit up by that DC voltage applied to pin 10, and the envelope takes away from the DC offset, dimming the LEDs, and making the filter go down, before you "run out of" envelope voltage and the LEDs drift back to max brightness.

Bass will usually provide more audible ripple, if only because the beats it produces are of greater amplitude.  But because the ripple is a lower frequency in general, it won't sound like distortion to most people.

Difficult to say in an authoritative way why the ripple is more noticeable in the one position than the other, but my best guess is that if the filter returns to a corner/resonant frequency that is generally higher than the upper end of treble on the bass, the filter may well BE moving quickly back and forth in response to ripple, but it will be doing so above the treble end of the bass signal.  One of those if-a-tree-falls-in-the-forest-and-there's-nobody-to-hear-it-did-it-make-a-sound things. 

Since the "starting point" of the downward sweep is set by the amount that R18/R19 has divided down the supply voltage, a person could easily make that starting point adjustable.  R18+R19 = 300k.  Making each of them 100k, and sticking a 100k (lin) pot in the middle (whose wiper goes to the switch, and on to pin 10), would allow a person to select higher and lower "sweep-starts", in addition to the stock starting point.  My guess is that, if playing bass, a lower start-point would make ripple become more audible (although it may be a more usable sound for bass).