NEMA 17 Toolhead -- how to control

So this is a 2D "printer"- I'm guessing, laying down ceramics or icing or something more useful than plastic in a single layer- and you just want to convert "laser power" to "extrusion rate". That seems... Extremely easy?

There may be off-the-shelf PWM-frequency converters that allow you to set the min, max and resolution, but it's probably not a difficult electronics problem either.

You could use a low pass filter to convert the PWM to a voltage, or a microcontroller board that allows PWM "analog" input and true analog voltage out. There are digital frequency generator ICs like the AD9833 that would allow you to use an A/D converter to tell it what frequency to output, or you could probably write a simple sketch that reads a PWM input and outputs it in an SPI/I2C serial output to such a chip.

For really hacky and simple with on-hand parts, you could use a voltage-controlled device (ie. a BJT transistor) to control the current into the capacitor governing a 555 timer's frequency, with a schmitt trigger on the output to crisp up the waveform.

There are probably any number of similar possibilities, and I'm sure many conversations out there on Google for PWM-frequency converter options. Totally doable! Probably a fun project too.

 

Driver should be ok, but hard to tell without a decent range of reviews. Usually there's a black-case one for about $13 with reliable-looking good reviews. Claims to be a TB6600, probably isn't, but works fine for NEMA 17s.

Duplicate link though, what's the converter? I'll check it out.

If you're accustomed to thinking in programming, I think there'd be a way of doing it all in a single sketch, from a simple RC low pass filter, on an Adafruit Cortex M4 board. They have analog in and frequency control over their PWM output, so you could program it in like, 30 minutes!

 

Oh, yeah, an LM331 breakout board is a solid option, as long as you can get the 0-10V VFD-type signal out of something. 10kHz, maxed out, at 200 steps/rev, would give 50 rpm. Depending on your extruder type/diameter, that could be useful, or possibly not, I don't know.

A Metro M4 would give you a traditional Arduino Uno form factor, but you could get more reliable connections soldering to the Express. Up to you. I've used the M4 Express, it's nice. I don't know what it's frequency range is, though, and this doc page- PWMOut – Output a Pulse Width Modulated signal — Adafruit CircuitPython 0.0.0 documentation - isn't much help. With a 120MHz CPU, I'd be surprised if it were lower than 10kHz though, but with embedded type hardware, writing in lower-level C might be a better option.

 

Yeah, Mysweety is the one, couldn't remember the name. Used a couple of those.

The voltage-frequency converter, the LM331 board, is good for the task, except you need at least a 20kHz output and you really need a 0-5V/TTL input. Might be worth looking at LM331 example circuits to see if you can DIY it, or just hack say one of the resistors on the board or something, to get you what you want.

The actual chip will use 0-5V and go up to 100kHz, apparently: http://www.ti.com/document-viewer/LM331/datasheet

If I thought the CAM/"slicing" on the 3D printer/Marlin board mentioned earlier was gonna be very workable, I'd probably just recommend that, but I think that'd turn into a nightmare on the software side, depending on how much variability you need on the extruder speed over a run. This is just a little bit of hardware tinkering and you're set.

 

Scratch that: the chip does need 0-10V, will take up to 15V for power. That makes the project a little more complex using that IC. You need a fairly precise voltage doubler, which would be an LC+op-amp circuit, then the LM331 circuit, and then a frequency doubler, which I think is some kind of comparator/flip-flop circuit or something, IIRC.

I can't see if it's possible to get higher stock frequency out of the ATSAMD51 driving the M4 boards either. It seems to be in the ~kHz range.

Doesn't seem like one of those issues that should be super tricky, but perhaps not.

Best solution I can see right now is either use Stepper library on Arduino at a high refresh rate (poll AnalogIn, set speed, fire a few steps, repeat), or find a 3rd party variable frequency generator library- there are a few, for different purposes- to add to your IDE and include that. Arduino seems to be more the way to go, though; 8/16MHz is plenty for this purpose. Plus side; off-brand Unos are $11 on Amazon.

Edit because I can't leave well enough alone: Arduino is the way to go. This library: Google Code Archive - Long-term storage for Google Code Project Hosting. - is perfect for this application. Non-blocking, tens of kHz range.

 

Yeah LM331 is an old school analog IC like the 555, should be easy to make work. The 8-pin DIP packages are $4 or so on Digikey if you want to try your own.

AnalogRead(), with a two-stage RC low-pass filter to smooth the PWM. That Tone library page is well-written, looks super easy to implement. Then just have the thing whizzing around and around at maybe 1-5kHz (depending on how long it takes to set the Tone() pin) and you have a 256-stepped output of 0-65kHz (depending on how you parameterize it in code, of course, that's just the 16-bit int limitation). Long as you're not trying to accelerate the stepper faster than your reader sketch is cycling, should be good!

Yeah, if you filter the PWM signal so it's just analog 0-5V, then double that fairly precisely with a 2x gain (or maybe more, 5-10X even, depending on how much the RC filter attenuates the signal) op-amp circuit (a cheap LM358 should be plenty good enough for that, I think, depending on the voltage you have available to drive the op-amp, it's not rail-to-rail), then you also get your 0-10V input. If you can modify the LM331 circuit to be able to adjust your max output frequency, then this is also a pretty easy analog-only option.

In any case, the first order of business is to make a nice, reasonably accurate low-pass filter. This is discussed in a lot of places online, largely for the same reasons we are, though there's also a lot of talk in the DIY subwoofer crossover community that I've seen in the past. This'll all be on Google results in 6 months too! If you have access to an oscilloscope to see what it's doing, it'll be much easier, but I think enough people have talked about it elsewhere in enough detail that you can just plug a couple of resistors and capacitors into a breadboard and see what happens.

 

Depending on the viscosity/texture and rate of deposition of what you're dispensing, I don't know that you'll need to go as far as a NEMA 23. NEMA 17s are pretty underrated as steppers go. But maybe, you'll find out soon enough, I guess!

Another option on this type of project is a low ratio 12V gearmotor with a PID control loop and an H-bridge output. Essentially a servo spindle.

The voltage on the Arduino output should be 5V (or 3.3V if it's a 3.3V board) though, shouldn't need an op-amp amplifying/buffering the output. Not sure what's going on there, unless you're just measuring with a multimeter, in which case some combination of the frequency and pulse width of the variable frequency output may be the culprit. An oscilloscope should show 5V peaks, in theory.

The op-amp was to ensure full resolution on the input, but may not be strictly necessary, depending how critical the speed is. You can probably do in-software scaling in the meantime since the RC filter is likely attenuating the signal; you may only be getting somewhere in the region of 0-2V analog input or something. The Arduino, if you're scaling 0-5V to 0-20kHz output, is naturally going to limit its output frequency since you're not inputting that full scale. Setting your AnalogRead to 0-300 on the A/D converter (or whatever a multimeter suggests is appropriate) and scaling that to your full 0-20kHz would do the job, though at the cost of reduced resolution to make up for the limited input bandwidth.