Let's Design and Build a (mostly) Digital Theremin!

Mars Need Women

Spent an hour with PRAAT trying to get the formants of the singer / songwriter Poe, which as usual didn't work at all.  I love her voice.

Gave up and decided to edit one of my female presets to try to improve on things.  Female vocal is a real go-to, bread-and-butter type preset for the Theremin, you can't have enough of them.  Here's today's preset "fem_3":

And here is a short recording of it: [MP3].  (Much thanks to DanielMacKay for posting his repertoire!)

The formant levels and resonances seem to be almost as important as the formant frequencies.  The nasal formant (highest, here 3.42kHz) has a huge effect on the overall perception of the age, weight, etc. of the vocalist, I need to maybe put two up there and play around with them.

I've been working a little on the strings too: [MP3].  That's violin, viola, cello, then an octave up.  With some reverb.

[EDIT] Ha!  Moved the top formant up a bit to 3.62kHz, added a new one below it at 2.96kHz, a subtle but real increase in realism of the vocal sound.  Have a listen to my noodling: [MP3].  The room ambience is 100% from the resonator.   So the trick seems to be more formants...

[EDIT2] I'm in the mood to play this new voice! [MP3].  It can go really high and somewhat low, and the dynamics are fairly wide.  I wonder if we're just naturally attracted to complex resonances?  The ersatz room acoustics provided by the resonator in this patch make it a lot more interesting to me.

Thank you a lot for audio samples!
Nice preset.

I installed the new software and presets (3 strings and 3 female voices) this morning. The female voices are great. Vocals are incredibly sensitive to the most subtle changes in parameters, and I don't think it would take very long to fill 100 presets with uniquely identifiable voices.  We must have a heightened sensitivity to voices over other sounds - probably like our ability to read facial features and eyes even from a distance.

I'm finding myself torn every morning between playing the theremin and playing with the theremin.

"We must have a heightened sensitivity to voices over other sounds - probably like our ability to read facial features and eyes even from a distance."  - pitts8rh

I was saying exactly this to my wife yesterday evening (great minds).  We're hyper aware of small differences in voices, faces, gaits - anything that can identify someone.  I would imagine it's so that we can track cheaters in the social contract (good will is easier to bank than a year's worth of food and going it alone).

Glad you like the new voice!  It almost invites you to overuse dynamics: [MP3]

Black Is The New Black

I hesitate to post this picture because it looks somewhat worse than real life (and I've gone out of my way to make it more revealing by using a flash):

My printer has always done this weird ripply (believe it or not :-) thing on the first layer.  Little waves of plastic start forming, which show up as odd perpendicular seams.  These get ironed out by the next layer so I haven't been too concerned about them until now.  The z axis calibration height is always spot on, so I don't think it's that.  It's like it's feeding too much material, but not enough to fill in the gaps?  I don't understand what's going on here.  (Mostly the generic PETG profile on Cura v4.7.1, with 250C / 80C).

On top of this I've always struggled with my reel of black PETG.  It prints knobs and such OK, but the plastic waves get extra hairy, and the hairs make even larger imperfections as they get dragged around, consolidated, and ultimately incorporated into the final part.  On top of that, PETG isn't nearly as rigid as PLA, so these larger flat parts tend to "potato chip" without a lot of extra support.  If it weren't for the too low melting point of PLA I'd run screaming from PETG and never look back.  I suppose it's time to print more test parts, but where to start...

[EDIT] I should mention that the LCD window is way too big here because of a typo in my SCAD file, and it's since been fixed.  But I need to add some more stiffener ribs to the back to control the overall curling of the front plate.

The first layer with PETG is often more of a problem compared to PLA.  If you see waves forming in the first layer that seem to get worse with each adjacent line, try raising the height of the first layer.  If I print 0.2mm layer height with 0.16mm for the first layer for PLA, I'll increase that first layer to 0.18 or 0.2 for PETG.  As you may have read, PETG needs to lie on the surface as a nearly cylindrical extrusion where PLA wants to to be squished into a flattened bead.  I have seen the flaws in the above surface finish many times and they are almost certainly being caused by the nozzle dragging the extrusion loose from the surface, either on the current line or the adjacent one.  So try raising the first layer height by 0.2 to 0.4mm and playing with the first layer temperature too (I like hotter for a more fluid extrusion, but Prusa for example uses a cooler first layer).

If you choose to print on the higher end of the PETG range, or beyond it at 260C like I often do, make sure that you use hairspray on your flex plate.  PETG can bond a little too well at high temps, and it is possible to tear out chunks of the PEI surface unless it is "seasoned" with a film of something.

Hairs and fine stringing lead to blobbing (as can an improperly tightened nozzle).  Almost invisible wispy strings get intercepted by the nozzle and create a growing blob.  The hairs come from inadequate retraction, bad travel paths (try turning "combing" on), travel speeds too low, and unfortunately just the inherently stringy characteristic of melted PETG.  I've already suggested trying higher temps to reduce stringing, but you have to combine that with fast travel moves to minimize oozing.  Make sure that your extruder calibration is valid for the measured diameter of you filament, and if not, you can temporarily modify it.  Your Ender 3 may have a flow percentage adjustment, but if not use M503 to look at your current stepper calibrations and scale the "E" factor as needed. As you might guess minor errors in the nozzle diameter can account for huge discrepancies between what you think the extruder calibration should be and what it actually needs to be. 

As a rule my PETG profiles in Cura all seem to want to have the "Flow" parameter (under Materials) set to 94-96%, for reasons I have never understood.  This also minimizes the nozzle dragging and accumulation of excess material on the nozzle.  "Initial Layer Flow" just below it is usually left at 100% or even a little higher if I want more fill, but again beware of aggravating the waviness problem described above.  Trying to get a glass smooth fill for the first layer is probably not a good idea, especially with PETG.

I hated PETG initially as well but now it's nearly all I print and I actually don't like going back to PLA.  You do have to tweeze the nozzle between prints and at the beginning, but other than that it works really well.  If you need a visual reference, go to the Prusa website and take a look at some of their assembly manuals that have pictures of their own PETG printed parts.  They just shoot these things out and they are some of the cleanest examples that I've ever seen.

PS: Higher bed temps (like 90C) will help prevent potato chipping, but in some cases it also helps to put a clear plastic bag over the whole printer (minus the display) to keep the environment warm.  I've saved a couple clear bags from electronics for this purpose.

Mexican Hairless

"As a rule my PETG profiles in Cura all seem to want have the "Flow" parameter (under Materials) set to 94-96%, for reasons I have never understood.  This also minimizes the nozzle dragging and accumulation of excess material on the nozzle."  - pitts8rh

Nailed it!  Roger, you are a veritable plastic whisperer!  I can only guess at how many reels of filament you've gone though to get such deep 3D printing knowledge!  Thank you!  Lowering the flow to 92 made all the difference in the world: virtually all the nozzle snot, hairs, and dragged around detritus have disappeared.  It's like night and day.  I wonder how many 3D printers there are out there sitting idle because the users just couldn't get past various issues?  You should consult (for profit)...

I made a credit card shape in SCAD with a thickness of 2 coarse 0.28mm layers, and played with the Ender | Tune | Flow % (never did this before) until it looked like not too much and not too little, then stuck this value in Cura.  It's hard to get around to these sorts of experiments, you just want to print the next pretty thing and not waste too much plastic, but in doing so you end up wasting plastic...

This morning I modified the LCD & encoders panel, mainly by adding more stiffening ribs to combat the potato chipping, as well as support posts for the main PCB:

Then printed it out 1/2 scale to see the lay of the land:

Above is the worst picture I could take of it, with direct flash bounce.  This is hot off the grill with no knife, drill, countersink, reamer, or sand paper clean-up.  It looks much nicer in person and is almost ruler flat.  So I've got a full-scale one printing now, the first layer took an hour, the whole thing will take over 8.  For that I'm going with 2 bottom layers (it prints upside down so the face is 2 layers), 3 side and top layers, and 20% infill.

[EDIT] Ah, here we go:

It all came out fine except the LCD window still needs work.  I'm going to make the opening angle less severe, remove the corner radii, and bring the display closer to the front.  But this is good enough for P3, and most of my worries about PETG are laid to rest (though I wish it were as rigid as PLA).

I'm not having a very productive morning.  I'm stuck on a couple of things:

1) I'm trying to back up some presets that I've been working on, and I think I'm doing something wrong with the commands.  I created a directory "presets_favorites" then switched to that directory, which is empty.  I do a "14 cello_muted.dlp stof" in an attempt to store the contents of slot 14 as the new file "cello_muted", but nothing happens with either return or space after the command.  I wasn't sure if the .dlp extension was needed so I left it off, with the same results.  Next I moved back into the system directory (in case something was wrong with my newly added directory name), did a "ktoe" to load the slot 14 contents into the editor, then tried "etof" both with and without an optional file name. I still just get a space when I hit return, and no file is generated.

I think I've been through this before and I fear that there is some obvious key point that I am missing this (early) morning.

2) I had to do a double/triple take on this one:  I was trying to change the octave register for one of your default wind instrument presets (slot -9 on my system) and it would not change up or down without a complete change in the sound.  The sound becomes thin or distorted going up or down, as if the oscillator is shifting but the filtering or some other process isn't.  I haven't run into any others that do this, but then I don't often try to switch registers. 

I've done the usual steps of restarting both the theremin and the librarian app with no luck.  What's going on?

{EDIT]

BTW, your sample print above looks almost textbook.  If you ever move to a textured plate you will need to increase the first layer flow for proper fill.  In lieu of using a textured plate another very simple solution for turning that part into a presentable panel is to use some stick-on (or glue on) hardwood veneer or a plastic graphic overlay.

Regarding reduced flow rates, I wonder if nozzle diameters are not as accurate as we think - that's the only reason I can think of why properly calibrated extruders (taking into account the measured filament diameters as well) would need to use flow rates other than 100%. But still, PLA prints correctly with 100% flow, so it must be a PETG thing.  I've never really found any information on this or a suggestion that others have to reduce flow for PETG.

"I do a "14 cello_muted.dlp stof" in an attempt to store the contents of slot 14 as the new file "cello_muted", but nothing happens with either return or space after the command."  - pitts8rh

You would do a "cello_muted.dlp 14 stop ".  The commands take the form "[from] [to] [cmd] " and the 'f' has since changed to 'p' ("preset file") since "file" is rather ambiguous at this point (there are bank files too).

"I wasn't sure if the .dlp extension was needed so I left it off, with the same results."

IIRC it doesn't matter, the librarian will add the extension if it isn't there.

"Next I moved back into the system directory (in case something was wrong with my newly added directory name), did a "ktoe" to load the slot 14 contents into the editor, then tried "etof" both with and without an optional file name."

Try "[preset file] etop ". 

You can also do this by naming the editor preset: "[preset file] rene ".  Then do a raw "etop ".

The "help " works everywhere now, and all the commands are in there.  Sorry to have changed them on you, it won't happen again!

"I was trying to change the octave register for one of your default wind instrument presets (slot -9 on my system) and it would not change up or down without a complete change in the sound.  The sound becomes thin or distorted going up or down, as if the oscillator is shifting but the filtering or some other process isn't.  I haven't run into any others that do this, but then I don't often try to switch registers."

The octave control changed to a custom type recently to give it more low-end range [-7:3].  In the librarian if you do a "ver " do you get "9c9d6d1"?  That's the latest SW version.

Also, "oct" only changes the oscillator register, and not the frequencies of any of the filters.  The clarinet presets in particular rely on filter frequencies lining up with the oscillator harmonics, hence the need for multiple presets to change the register.  I think the general case is you don't want filter frequencies shifting with "oct" (doesn't happen in nature too often).

"Regarding reduced flow rates, I wonder if nozzle diameters are not as accurate as we think - that's the only reason I can think of why properly calibrated extruders (taking into account the measured filament diameters as well) would need to use flow rates other than 100%. But still, PLA prints correctly with 100% flow, so it must be a PETG thing.  I've never really found any information on this or a suggestion that others have to reduce flow for PETG."

I wonder if it has to do with the extruder grip?  When switching from PLA to PETG I noticed the end of the PETG, which I'd used previously, had fairly deep notches in it from the extruder drive digging in.  PETG is more malleable than PLA, so maybe when it goes through the extruder it gets elongated and somehow more material ends up at the nozzle?  I dunno...

[EDIT] "Any thoughts on the register shift weirdness on preset -9?"  See above.

My screw up!  I keep a printed help sheet next to the theremin and I was using the old one (now in the trash).  Next week when I forget and ask you the same question please just direct me to this post.  I've learned to document and organize important information over the years, but the real fun starts when you start forgetting that any of this documentation even exists or where it might be. Time for Uncle Billy's string tied around the finger, I guess.

Any thoughts on the register shift weirdness on preset -9?