Ah, the boards that @woodslanding linked to say that you should provide your own 5V say that on-board 5V doesn’t work over 12V motor supply.
Probably because it’s a 7805 so at higher voltages it’s dropping so much voltage it’ll overheat if there’s much load on the 5V. Good catch.
Bingo, looks like a 7805. Looks like you can change a jumper to supply it yourself. I don’t know how clean the power needs to be, whether you can use a buck converter or need a linear regulator.
okay, yikes step away for a few days and I forgot half of what I was doing. Hope you all had a nice holiday!
Well, and I just learned there is a non-musical use for pulse-width modulation! Who knew?
So if I understand correctly (and that’s hardly likely) I will need 4 connections to the teensy for each motor:
Motor1 - Digital
Motor2 - Digital
MotorEnable - PWM, to set the speed.
EncoderIN - Digital, I assume, as it looks like a square wave.
This means, for an 8-string instrument (which I think this one will be… The second one might be 10, if I get that far…) I’ll need 32 connections?
The teensy has 34, great!
Except that doesn’t leave much for the 6-10 analog ins I’ll need for pedals and levers.
So I guess I will need some kind of multiplexing of those digital signals?
Encoders are usually two pins, often referred to as A and B. “Quadrature encoders” have four states, and by tracking the changes you know not only that they changed, but in what direction.
You generally need to use interrupt handling for decoding quadrature encoders. You put an interrupt service routine on both the input pins, and every time it is called, you read both pins and compare to the previous state, and then you know direction and can add or subtract one to/from the previous absolute position. Here’s one easy approach:
This might seem silly, but… One of the easiest ways to control more inputs is yet another microcontroller. An ESP32, ESP8266, or Raspberry Pi Pico are all inexpensive and you can connect them with a serial connection (typically SPI) and have them talk to each other. (Or just use another teensy, but you don’t need the 4.1’s 600Mhz processing power there) That is a form of multiplexing, but it’s more flexible than a hard-wired multiplexer.
Quadrature decoding is a great use for a secondary microcontroller. Every time the absolute position changes, it just reports the axis and the new absolute position to your main controller.
Also, using secondary microcontrollers like this makes it easy to expand to more strings; if you run out of pins, you can just add another secondary microcontroller for the additional channels you need.
1 Like
I doubt you could do all this with one controller. As @mcdanlj suggests I would put a small controller on each string channel connected on a buss like SPI or I2C. The string controllers can all run the same code leaving the main code for computing more sophisticated elements of the problem and the UI.
2 Likes
From a comment in a recent HaD post about building a stepper motor replacement from a gear motor, take a look at this library:
While it includes the ability to read STEP/DIR signals, it also has a serial protocol. It includes sensorless homing. It might be useful to read, if not directly to use.
I dont remember who pointed me to this gear motor encoder but I thought it was a good setup. I may use it for a curtain controller.
@donkjr I think it might have been @dougl quite a few comments back within this mega-thread… 
Good idea… Would one board per 2 strings/motors make sense? It fits well with the motor control board layout.
so all the wiring from a daughter board to the main board could go over a single serial connection? Easier from a wiring perspective for sure…
I will… my understanding was that for my application a PID motor would work better than a stepper, but regardless, it will be helpful to look at how the serial protocol works.
Edit: looks like he is using it as a servo after all… and that is the exact setup of my motor, except I didn’t have to add my own encoder. I like how he fit the control board in the same case with the motor. very compact… Reading about ‘backlash’ in the comments, I’m not sure I understand what that is exactly… I’ll have to read up on that. Does PID account for that?
reading this: XMoto DC as Servomotor | Details | Hackaday.io I’m wondering if this is not the same motor I have after all. It looks the same, but this is a ‘high power’ version, probably not available with a built-in encoder… this one will take 1.6A, and IIRC the one I ordered takes 500ma. Presumably, if the one I got doesn’t have enough power, I can go this route. This post is only 4 days old, so I’ll be following with interest!
If this solution doesn’t have the torque needed, I’ll probably be looking at a worm gear motor like that. 8 of those would take up a lot of space, though, so I’m hoping the smaller motor works. And the rpms are pretty low for my application, although some of them may get close enough.
Totally depends on the number of pins on the controller you use.
You’ll need one SPI channel per daughterboard. I²C you can have more on one bus but it’s slower. SPI is simpler and it’s what I’d use.
I’m pretty sure that the motor control daughterboards, even with slow CPUs, will not run out of compute power even if they are controlling lots of strings. I’d use as few daughterboards as I could get away with.
The HaD post was about something like what you are doing (though not as small) but I was linking to a library mentioned in a comment. That library implements both PID control and trapezoidal motion planning. I haven’t read the source code but if I understand correctly, it would probably be relatively easy to modify it so that instead of using STEP/DIR lines, you just send it motion commands over an SPI connection. It would be a starting point, not something you would use as-is, was at least the idea I had…
First read up from Wikipedia on what backlash is in engineering:
You don’t really have to worry about backlash at all because your mechanism is always under tension in one direction. This means that your gear train (if you have one) is always being tensioned in one direction, and the play between the gears never… comes into play!
PID is for movement management based on a sensor input for where you are and knowing where you want to go… Watch the video I linked to way back in comment 16 a few weeks ago to understand PID — it’s the best thing I’ve found yet for understanding what PID does for you. 
1 Like
okay, that makes sense.
Reading up here:
Looks like his motor is more similar to mine than I thought. Looks to be better built, but at 12v the amperage is cut in half. And the figures for torque on higher rpm gearings look similar to mine. But building the daughter-board right on to the motor does make a lot of sense from a physical layout perspective.
From what I can tell, it’s very much what I’m doing, at least from a hardware standpoint, right? and that motor is only 2mm wider than mine… but I will look at the library as well.
2 Likes
I didn’t realize it was so close to what you were doing in size. Cool!
Boy, a ton of info and links here: dcservo/README.md at master · misan/dcservo · GitHub
never mind the actual code.
Thanks! I’ll be studying in the library for a while now…
3 Likes
Wow, my google-fu is failing me as I looked and looked before finding something close and then mashing into my own version: This is what I’d posted about in Nov '21 and now realize I had not shown the motors I’ve been using or else @donkjr would have been made aware of them long ago. Use DC motor/encoders as stepper motors with step/dir signals - work in progress
2 Likes
Looking into small boards to run the PID code, and wondering if this one would work:
Very compact, which is good, and seems to be flexible and fast. Doesn’t require an adapter to program… I see I2C on the pinout diagram…
Now I notice it is backordered. Guess I should get something that’s actually available, at least for now 
I’m most familiar with this: Teensy LC (Low Cost)
which is only a couple of dollars more than the Arduino Pro Mini, and has built-in usb. However, since the usb adapter only costs a couple of dollars, that’s maybe no big deal.
I’d personally be looking at the teensy, ESP32 and maybe Raspberry Pi Pico, if I were doing this myself now. The ESP32 has the ability to do fail-safe firmware updates, even over the air with wifi. Some folks use it for robot arms to run only power on slip rings and data over wifi or Bluetooth for more free maneuverability. I’m sure there are lots of reasonable options that I’m not thinking of…
Wow, there are a lot of options. The pico looks great, and low cost. (not that any of these are prohibitive… although costs do add up!) I can get it locally for $4, but if I’m only buying one to start with, it’s still cheaper to pay $8 online.
It supports programming in Python, which piqued my curiosity. I’ve managed to go all this time without learning Python (I’m familiar with Java, Javascript, Lisp, C, C++, Delphi, Lua and VB) and so I started reading up on it. I see why people like it. Seems like Lua borrowed a lot from it. That said, the libraries you’ve linked to seem to be in Arduno, and that’s what the Teensies use. I’ve never had issue with using Arduino, which seems similar to Java. I’ll just hope the semicolon key on my laptop holds up.
The Esp32 with wireless is interesting… but it seems like this board is really new, and still being developed at a lower level. Might be a bit deep for me to get into? I’m assuming too, that I could prototype with one board, and move to another with minimal changes to code, correct? Once I am scaling up, the wireless might look more attractive… It looks like I will need 6 connections for each encoder motor, (not including power and ground) and the pinout on the pico Pi Pico Pinout and Power Pins - Raspberry Pi Spy looks like it could support 4 motors and still have pins left over for the I2C connection… Am I missing anything?
Okay, reading more on the Pololu motor mentioned above, it looks to be quite a lot more powerful than the one I ordered from ali. So I think I will order one. They also make a hall encoder to fit it. Reading more on the site, I came upon this:
Looks like a more compact option than the ones I see on Amazon, and it can handle 12 volts just fine… reading on the specs, there are two different wiring diagrams for it. One uses TWO pwm sources, rather than 1 pwm and one switch. Wondering what the difference is between the two… EDIT: okay, it’s down there in the readme for the hackaday link, with links to explanations of both modes.
My motor has arrived in the US, but not cleared customs yet.
1 Like
It’s actually C++, it just magically imports some header files before compiling.
And yes, Java does descend from C++.
The Raspberry Pi Pico is new, less than a year old, but the ESP32 is well established. They do continue to develop it actively.
The 16-bit ESP8266 was introduced to be a wifi modem for other processors and was adapted to be a microcontroller of its own, and then they came out with a 32-bit version as a more general-purpose system from the start, a bit over five years ago. (The odd thing is that the CPU isn’t the most important part of the platform, and they have two CPU architectures in the ESP32 family. The Tensilca CPUs are in the original ESP32 CPU platform, the RISC-V CPUs are in some of the newer parts. I’d stick with the Tensilica-based versions if you want an established platform.)
My only Raspberry Pi Pico experience so far wasn’t great, but I was trying to have it talk over USB to a Raspberry Pi 4, and the problem appeared to be on the 4; I was able to talk to the Pico fine from my Linux laptop. So I am not inclined to blame the Pico there. However, the Pico has a lot less power than the ESP32 and is a lot newer of a platfform.
As far as I know, Pololu established the “step stick” format for stepper drivers used on many 3D printers today, and has been long established — over twenty years. It’s well respected.
If you are using the Arduino platform, that’s going to be generally correct, especially if you keep your pin mappings symbolic, so that when you attach to different pins on a different CPU, it’s a matter of changing one definition, not search-and-replace. Probably this is something you already know, but just in case I thought I’d mention it.
Well, I ended up buying a teensy LC, because I’m familiar with it. I may move over to the esp32 when I build the actual machine, but I got a bit bewildered by the options, having never heard of it before. Also ordered a motor driver board, and a teensy 4.1.
Yes, I’m not a great programmer (although I was a Java Swing developer for a couple of years) but my programming is light years ahead of my EE skills-- so yeah, you can assume I know the basics of programming, but please don’t assume I know anything about electronics 
1 Like