Howdy,
Just whipped up a SSWSMS alongside my Octave Up Sick Box variant (able to vary from half to full wave rectified), and while the OUSB does exactly what I want the SSWSMS is a bit underwhelming.
(https://nexp.pt/tpe123/swsh.gif)
It's built as per the schematic except for a dual 500k pot for the slew rate limiter section; didn't have space for two pots. I've done a bit of experimenting and got the transistor dialled in on a bias pot that will take from gated to oscillating with a full turn of the knob, and it's got sustain for days.
There are a few things that have me a little bit disappointed however - with the pot turned all the way down, it's just a fairly average fuzz with barely any treble whatsoever, and super low output. Turned all the way up it gets SIGNIFICANTLY louder, but the overall characteristic of the sound doesn't change that much. If I turn the bias to oscillate I can get some interesting stuff happening with tapped notes (lots of weird overtones and noise coming through the oscillation), but it's very much a one trick pony.
I am guessing I may have the shape pot wired backwards which isn't too much of a deal breaker - I don't mind which direction is square and which is triangle-ish.
I haven't fed a sine wave into it to see what it's doing with a simple waveform, but I had read plenty of reviews about how you could create massive amounts of tones depending on the interaction between all the pots. I figured I would be losing a bit of versatility with only a single shape pot, but still was expecting a little bit more than what it does.
Anybody done a bit of playing with it and have some suggestions about what could improve things?
Thanks!
Quote from: drdn0 on May 13, 2025, 03:05:42 AMgot the transistor dialled in on a bias pot
You probably refer on other than 25k trim-pot, yes..??
(e.g. inside NFB loop, in place of 500k resistor..)
Quote from: antonis on May 13, 2025, 04:47:56 AMQuote from: drdn0 on May 13, 2025, 03:05:42 AMgot the transistor dialled in on a bias pot
You probably refer on other than 25k trim-pot, yes..??
(e.g. inside NFB loop, in place of 500k resistor..)
I don't understand what you mean sorry - I am referring to the 25k pot, and bias probably isn't the right term for it.
I found that the sound would be fairly gated at 6k8 and have stacks of sustain at 7k4, so I've got a 6k8 resistor with a 660r pot (25k in parallel with 680r)
I made one for my modular system. And while it "works", yes it is underwhelming. But that's not a fault of the design. It is a fault of the input signal, whose amplitude and harmonic content changes too much in too brief a period to provide a consistently audible result of changes to rise and fall times. This stands in sharp contrast to the steady-state input from a synth oscillator.
I have not scoped the circuit, so I can't tell how consistently square the U1 output is, The Craig Anderton EPFM "Ultra-Fuzz" is a decent circuit for using a variable comparator to produce a nicely square output. Although the 47pf feedback cap in the first stage should probably be a larger value, like 100pf, to provide same modest filtering of the input bandwidth. And if you can compress the input before feeding it to the wave-shaper circuitry, that may help as well. The comparator provides a fixed output level, yes, but ideally one wants to be able to set the comparator voltage and then not have to think about it.
(https://img.photobucket.com/albums/v474/mhammer/ultra.gif)
Quote from: Mark Hammer on May 13, 2025, 12:00:50 PMI made one for my modular system. And while it "works", yes it is underwhelming. But that's not a fault of the design. It is a fault of the input signal, whose amplitude and harmonic content changes too much in too brief a period to provide a consistently audible result of changes to rise and fall times. This stands in sharp contrast to the steady-state input from a synth oscillator.
I have not scoped the circuit, so I can't tell how consistently square the U1 output is, The Craig Anderton EPFM "Ultra-Fuzz" is a decent circuit for using a variable comparator to produce a nicely square output. Although the 47pf feedback cap in the first stage should probably be a larger value, like 100pf, to provide same modest filtering of the input bandwidth. And if you can compress the input before feeding it to the wave-shaper circuitry, that may help as well. The comparator provides a fixed output level, yes, but ideally one wants to be able to set the comparator voltage and then not have to think about it.
(https://img.photobucket.com/albums/v474/mhammer/ultra.gif)
Ahh ok, so it's more just of a case that when it came out it was so novel that all the positive reviews are just because there weren't really any other options?
I've built a stack of other Escobedo designs and they've all done exactly what was promised, so I was really surprised that this one doesn't quite cut it
No. I suspect people who knew how to use it liked it and said so. It does what it says it does. The trick is to feed it what it needs.
I say this having spent much of the day in the garden, transplanting things, and finally spreading composted manure along with the potting soil, so that the things I want to grow will...this time. It's not the seeds, or the circuit design, but the environment around them.
As well, do consider playing around with the value of that .02uf cap to ground after the diodes. Remember that the "slope" created by varying the charge-up of that cap will be more audible for some frequencies than others. It's quite possible that you may be producing only modest variations in rise and fall rates, given the notes you feed it.
Ok, so did some scoping. Things are...strange. I knew it wouldn't be perfect manipulation of even a clean sine wave, but even so this isn't what I expected.
Signal in was 1v at 100/300/1k/3k.
I can get something approximating a square wave in with the slope knob all the way down - but only at 70mv out. As soon as I increase the frequency it turns into a sharkfin then triangle, but the output drops even further.
With the slope knob all the way up I get huge outputs, but a low duty cycle wonky triangle with increased duty cycle until 1khz, then it turns into something else.
First run of images is with the slope turned all the way down, second run is turned all the way up.
(https://i.ibb.co/Z1BbQpNr/image.png)
(https://i.ibb.co/35KNKMnN/image.png)
(https://i.ibb.co/VYtNwHZs/image.png)
(https://i.ibb.co/chG4mhbk/image.png)
(https://i.ibb.co/fVdDVPPC/image.png)
(https://i.ibb.co/KpszfJw5/image.png)
(https://i.ibb.co/TxXFx1vS/image.png)
(https://i.ibb.co/XxRBVFRb/image.png)
Funnily enough, with the OUSB side I can go from a half to full wave rectified square which looks great!
This looks like slew rate limiting so maybe a better op amp or lighter filtering would change it.
Ok, so did a little more poking around today.
On the output of first op-amp, I have a beautiful square wave. After the pot everything gets a little odd so I did some measuring - one of the pot tracks is 300k. One of the pot tracks is 500k. I am going to make a guess that's not helping things whatsoever.
Continuing further, after the AR gen everything starts sucking.
Did a little bit of wafer swapping to get both the pot tapers around the same which got the wave closer to a square, then played around with the cap until I got something that worked well with a sine wave from 100hz to 3khz - which happened to be 3n3. Even with the AR gen cap disconnected the leading edge is about half a volt above the trailing edge above 300hz or so, so I suspect I've got a weird diode thing happening - just waiting for my meter to charge and I'll do some measuring.
I then noticed the output buffer wasn't REALLY buffering, and found out that that op-amp section is seemingly scuffed as I was losing 50% of output voltage over the pot. Replaced with a different op-amp and no voltage drop now.
Getting there!
Ok, had some odd schottky's - from the same batch of BAT41s, half have a forward voltage (at 1ma) of 300mv +/-20mv, half half a forward voltage of 400mv +/- 20mv. All marked BAT41, all tested under the same conditions. 400mv matches the BAT41 datasheet, but ~300 doesn't so I've got no idea what they are.
Swapped to a matching pair and things improved, but not fully.
Finally, while poking around I increased the output capacitor to 470n, and suddenly I have a perfectly square wave at all frequencies, and a perfect triangle from 100-1k, and a little bit wonky above 1k. I'm still bleeding a lot of signal (at full triangle and 1khz I'm getting about a volt out now), but it's a HUGE improvement over what I originally had.
Quote from: drdn0 on May 16, 2025, 12:43:46 AMFinally, while poking around I increased the output capacitor to 470n, and suddenly I have a perfectly square wave at all frequencies, and a perfect triangle from 100-1k, and a little bit wonky above 1k. I'm still bleeding a lot of signal (at full triangle and 1khz I'm getting about a volt out now), but it's a HUGE improvement over what I originally had.
The time constant created by the 500K pot sections and the cap aren't related to the input frequency. It's an integrator with a constant slope, effectively. That means that once you're at the point you have a perfect triangle, if you go up an octave, you'll get half as much output. That's just a side-effect / limitation of the design.
HTH
I know it's only using a sine wave in, but all functions work properly. Sounds pretty wild with guitar (and even gnarlier on bass), with lots of textures of fuzz + ring mod/octave thrown in.