My first auto-wah prototype build, feedback appreciated

[jatalahd]

Hi,

I wanted to make my own design of an auto-wah and see what comes about. There is basically no new inventions in the design, it takes bits and pieces from here and there, but one thing I wanted to accomplish was to create a wah effect (for bass mainly) that combines the LFO and envelope control into one. The key idea is this: when the string is plucked, the envelope rises fast and causes a Schmitt trigger to reset the LFO to start a new cycle from the "initial state". The following simulation clarifies the idea, the blue trace is the envelope, red trace the LFO and the green trace is the Schmitt trigger which goes high at 6 V and low at 3V.


This kind of gives a control for the attack and release times of the wah-sweep.

The current schematic looks like this:


Explanation: The input signal from the guitar is split at node 2 to two paths. The upper path goes to a Tow-Thomas biquad filter, which is the actuator for the wah-sweep and the low-pass filtered output to the guitar amplifier is taken from node F8 at XOP3. The Q-factor of the filter is controlled with R1 and gain with R4. The sweep is implemented using two NSL-32SR3 optocouplers connected in series with R2 and R3, which affect the frequency tuning of the filter. The lower control path does the envelope detection with Q1 and Q2 and the envelope signal is available at node 7. Q3 and Q4 implement the Schmitt trigger which controls Q5 as a switch that sets the initial state (empties the cap Cosc) to the relaxation oscillator XOP4. J1 and J2 are JFETs with pinch-off voltage around -1.8V (BF245A for example). The J2 is sourcing current to the diodes of two NSL-32SR3. The idea is to have the J2 in complete pinch-off when the control signal from the LFO is at minimum, thereby hoping to get zero current to the optocouplers. I chose the NSL-32SR3 because of its low current demand and relatively fast response time even in the "darkness" direction.

Difficulties: major problem was to find the sweet spot of the optocouplers to give maximum sweep on the filter. Hopefully I can still find a more sweeter spot, since the sound is still not quite what I was hoping for. Matched optocouplers would have helped i presume, but this could be fixed by taking an additional J3 to control the other NSL-part with different bias current. There are so many parameters to tweak in the filter so more experiments are needed. Currently it seems that the range for good sounds is very narrow in the filter settings, which is nasty. I was hoping to get a wide range of tunability.

Bugs: In the prototype build there is a nasty "click" problem which appears to follow the frequency of the oscillator. I have not yet found a way to fix this issue. Maybe one solution could be to not use the same quad op-amp to share the LFO and the filter...

Anyway, if someone is interested to hear what this sounds like, I recorded a youtube video from my sound check: http://youtu.be/JKmkryHx9KU

All comments, suggestions for improvement and other feedback are warmly welcome!

[PRR]

Dark blue on black with two shades of grey gridlines??

Curious what was in the murk, I transcribed it to ink on paper:

http://i.imgur.com/GxPvPfb.gif

I see an input but no output.

The input is strangely biased-up mid-way on the battery. I assume for simulation not reality.

> sweet spot of the optocouplers

I am sure the two parallel LEDs will not share current 50:50 exact. Use two JFETs and two trim resistors. Or put the LEDs in series. Use much higher LED current to get the LDRs away from their dim/dark region down into the 5K-50K zone (and adjust all R and C in the filter(?) to the same scale).

If that is really a LM324 carrying audio, hang 2K from Out to ground on /1 /2 /3 to stay out of the '324's nasty class B action; or use TL072/074.

[jatalahd]

Thank you, Paul, for commenting. Yes, the output was not drawn in the schematics, one can take the output from nodes F8, F6 (both low-pass outputs) or F4 (band-pass outputs). In the prototype build I am taking the output directly from F8 without any coupling capacitor nor additional output resistor. I guess this is not a good approach... And actually I intended to bias the input mid-way, not only for simulation. The 4.5 V acts as the "real" ground point for the audio path from the input to the output, since the effective audio path goes through the filter section only. I am using the whole 9 V range only in the envelope and oscillator parts of the circuit.

I will try your suggestion on the LED connections on the coming weekend when I have time to concentrate on this again. Basically I did do elementary measurements on the NSL to get it somewhere near the range you mentioned, and I was very surprised to notice that I actually needed to use a very low LED current to get there. According to the NSL-32SR3 specs, the 10k - 100k zone is obtained using 0.1 - 0.01 mA current on the LED --> 10 V/100000 ohm = 0.1 mA,  so I should be in the correct range, but I will try to tweak it a bit more. I also wanted to make use of the very high dark resistance of NSL-32SR3 (25Meg) to maximise the sweep. In this I did not succeed that well, but it was my first time building a circuit using optocouplers so I am just happy to see them working almost as intended.

The tips on the op-amp are very much appreciated! Thanks again!

[PRR]

> to make use of the very high dark resistance

That's like, my garden hose faucet "can" be turned to a very small trickle, but it is MUCH more controllable in the 100%-10% range. I can wash the horse, I can wash the Corgi, but it isn't much good for washing an ant.

Also that 25 Meg point may take many seconds, even minutes, to reach. Tedious for a static effect and counterproductive for any dynamic effect.

Sock with 10mA, find the "light" resistance. Perhaps near 1K-2K. You have a very good 10X range up from there (20K), and maybe 100X (200K). There are few audio sweepers which really need more than 10X or 100X range.