SC/OFF: A linear scale-offset CV processing module, v1.0
Nov 26, 2019 2:55:30 GMT
admin, NightMachines, and 2 more like this
Post by young Protoboard on Nov 26, 2019 2:55:30 GMT
Buckle UP, 'cause this post is gonna fly by QUICK.
Why quick? Well, I've spent 4 hours total writing up beautiful, in-depth, discussion-rich posts about this module and this project, and the first draft got deleted because the website "lost connection to the cloud server" (
), and the second draft got erased when I accidentally refreshed my browser page (!!!), so this post will be 1. SHORT, and 2. CREATED IN AN OFFLINE WORD PROCESSOR INSTEAD OF IN BROWSER.
I digress, and through my tears of frustration, I present the SC/OFF.
-------
Hi, folks,
The SC/OFF's function is to linearly scale and offset a CV signal through AEM's CV range of 0-5V. Its function is described in the trick outlined here by user Gaëtan-- I have a Starter Rack 2, so I used my 2ATT/CV and the B mixer from MIXER 4-4 to accomplish what gaetan does with the 4ATTMIX. However, this patch requires 1.5 modules and two internal patch cables just to process 1 CV signal, using up all of my attenuators in the process. Not very efficient. Further, I use this patch so much that it warrants its own dedicated module in my rack.
To create the design, I reverse-engineered the modules within that patch; specifically, I examined closely the B mixer of MIXER 4-4, as attenuation is fairly simple (many props to NightMachines for their excellent intro DIY series!). The B mixer is depicted below. I found it to be too imprecise and unstable to use for this application; we have the opportunity to purpose-build our own module, so we might as well build it for decent performance, no?
The B mixer is the assuredly simple topmost group of 4k7 resistors on MIXER 4-4.
I'm not elaborating on design decisions in this post because every time I do, my draft explodes. So I'll just say that passive resistor networks like this do not "accurately" sum two input voltages. For audio signal combinations, it works well enough for its part count and simple design, however, the A side MIXER does so in a more precise, and buffered, manner, via op-amps. In fact, the v1.0 SC/OFF has more in common with the Audio only A side mixer than the B side, but I digress, as I can see my computer start to smoke.
So op-amps it is. Here's what I settled on as design v1.0:
Why this design? Because this project is cursed of course:
- I spent about two weeks of bench time revising and refining my op-amp-based designs due to unexpected, incorrect op-amp behavior, which I attributed to user error and lack of understanding
- The mystery errors were finally found to be caused by a curious probabilistic happening; I had bad batches both of ~50x LM358 ICs and ~30x MCP602 ICs. I had about a 20% chance of grabbing a correctly-functioning op-amp of either type, so I was spun in circles for days at a time, chasing phantoms through my breadboard.
- Three weeks of redesigning due do equipment failure is damn frustrating, and this was the first design I tried that worked properly on my breadboard. (I don't mind complicated and/or long-spanning projects, but effectively wasting 3 weeks? Yup, I'm gonna resume forward progress ASAP).
- The summing portion of the mixer is the most flexible summer I have found and used for this project so far. It should be fairly easy to modify this design to include desirable features.
- I'm more than ready to share my progress with this design as a kind of "checkpoint" and/or milestone, and free up some mental bandwidth for other co-developing projects.
This section left blank in remembrance of all nicely-formatted photo posts. 
What I like:
- It works!
- The "-" switch causes the attenuated input signal to be subtracted from the offset rather than added, which sort of "inverts" the analog signal, if you take a very liberal definition of "inverted." Very cool conceptually, and varies the CV signal in a way that I personally like.
- Knob spacing and front panel layout is acceptable.
- Fits on 1U (barely)
- Theoretical errors/distortions in voltage values are low due to proper op-amp selection.
What could be improved:
- 5 op-amps. 5? Seriously? I know I can make this way more parts-efficient in future versions.
- Two separate types of op-amp are used (MCP602 and LM358). This was necessary for proper input and output voltage ranges during signal processing, but can likely be avoided in future designs.
- Reduction in external parts may allow for a second SC/OFF to fit on 1U.
- A polarized fader knob for input attenuation, instead of a switch, would be great, i.e. the center point of the knob is zero signal passed, and turning left subtracts a growing input signal from the output, and turning right adds a growing input signal to the output. So, the "inverting" action is handled by the left side of the pot turn, rather than a switch.
- I have the wrong height of toggle switch; pretty sure I'll need a little tool to flip it through the front panel.
- Oh, man, I really should have planned the layout of my Protoboard better. That thing is an electronic spaghetti jungle.
- You can see that the top pot shaft has been melted slightly on its top. This was a soldering iron wound sustained at hour 2 of a 9-hour fabrication. I endeavor to do better, both in speed and precision.
I love any questions, suggestions, or comments! And I look forward to improving this design further; increasing functionality and features, reducing part count. It will be updated as I get back to it in between other module projects.
Thanks for reading.
-yPb
Why quick? Well, I've spent 4 hours total writing up beautiful, in-depth, discussion-rich posts about this module and this project, and the first draft got deleted because the website "lost connection to the cloud server" (
), and the second draft got erased when I accidentally refreshed my browser page (!!!), so this post will be 1. SHORT, and 2. CREATED IN AN OFFLINE WORD PROCESSOR INSTEAD OF IN BROWSER. I digress, and through my tears of frustration, I present the SC/OFF.
-------
Hi, folks,
The SC/OFF's function is to linearly scale and offset a CV signal through AEM's CV range of 0-5V. Its function is described in the trick outlined here by user Gaëtan-- I have a Starter Rack 2, so I used my 2ATT/CV and the B mixer from MIXER 4-4 to accomplish what gaetan does with the 4ATTMIX. However, this patch requires 1.5 modules and two internal patch cables just to process 1 CV signal, using up all of my attenuators in the process. Not very efficient. Further, I use this patch so much that it warrants its own dedicated module in my rack.
To create the design, I reverse-engineered the modules within that patch; specifically, I examined closely the B mixer of MIXER 4-4, as attenuation is fairly simple (many props to NightMachines for their excellent intro DIY series!). The B mixer is depicted below. I found it to be too imprecise and unstable to use for this application; we have the opportunity to purpose-build our own module, so we might as well build it for decent performance, no?
The B mixer is the assuredly simple topmost group of 4k7 resistors on MIXER 4-4.
I'm not elaborating on design decisions in this post because every time I do, my draft explodes. So I'll just say that passive resistor networks like this do not "accurately" sum two input voltages. For audio signal combinations, it works well enough for its part count and simple design, however, the A side MIXER does so in a more precise, and buffered, manner, via op-amps. In fact, the v1.0 SC/OFF has more in common with the Audio only A side mixer than the B side, but I digress, as I can see my computer start to smoke.
So op-amps it is. Here's what I settled on as design v1.0:
Why this design? Because this project is cursed of course:
- I spent about two weeks of bench time revising and refining my op-amp-based designs due to unexpected, incorrect op-amp behavior, which I attributed to user error and lack of understanding
- The mystery errors were finally found to be caused by a curious probabilistic happening; I had bad batches both of ~50x LM358 ICs and ~30x MCP602 ICs. I had about a 20% chance of grabbing a correctly-functioning op-amp of either type, so I was spun in circles for days at a time, chasing phantoms through my breadboard.
- Three weeks of redesigning due do equipment failure is damn frustrating, and this was the first design I tried that worked properly on my breadboard. (I don't mind complicated and/or long-spanning projects, but effectively wasting 3 weeks? Yup, I'm gonna resume forward progress ASAP).
- The summing portion of the mixer is the most flexible summer I have found and used for this project so far. It should be fairly easy to modify this design to include desirable features.
- I'm more than ready to share my progress with this design as a kind of "checkpoint" and/or milestone, and free up some mental bandwidth for other co-developing projects.
This section left blank in remembrance of all nicely-formatted photo posts. What I like:
- It works!
- The "-" switch causes the attenuated input signal to be subtracted from the offset rather than added, which sort of "inverts" the analog signal, if you take a very liberal definition of "inverted." Very cool conceptually, and varies the CV signal in a way that I personally like.
- Knob spacing and front panel layout is acceptable.
- Fits on 1U (barely)
- Theoretical errors/distortions in voltage values are low due to proper op-amp selection.
What could be improved:
- 5 op-amps. 5? Seriously? I know I can make this way more parts-efficient in future versions.
- Two separate types of op-amp are used (MCP602 and LM358). This was necessary for proper input and output voltage ranges during signal processing, but can likely be avoided in future designs.
- Reduction in external parts may allow for a second SC/OFF to fit on 1U.
- A polarized fader knob for input attenuation, instead of a switch, would be great, i.e. the center point of the knob is zero signal passed, and turning left subtracts a growing input signal from the output, and turning right adds a growing input signal to the output. So, the "inverting" action is handled by the left side of the pot turn, rather than a switch.
- I have the wrong height of toggle switch; pretty sure I'll need a little tool to flip it through the front panel.
- Oh, man, I really should have planned the layout of my Protoboard better. That thing is an electronic spaghetti jungle.
- You can see that the top pot shaft has been melted slightly on its top. This was a soldering iron wound sustained at hour 2 of a 9-hour fabrication. I endeavor to do better, both in speed and precision.
I love any questions, suggestions, or comments! And I look forward to improving this design further; increasing functionality and features, reducing part count. It will be updated as I get back to it in between other module projects.
Thanks for reading.
-yPb