Ménage à Quatre
It's fun to make a gaggle of voices using the triple oscillator with pitch big offsets and the pitch preview pressed into service (with no pitch correction) as the fourth:
https://d-lev.com/audio/2021-08-13_were_in_the_mood.mp3
Playing intentionally rather lame here (I meant to do that! /Pee-wee Herman).
[EDIT] It's weird how multiple detuned anything ends up sounding like an accordion.
Light Gray Is The New White
Resolved all the 3D printing issues except for the slight discolorations here and there. Uncle! So I'm switching to light gray PETG, which is less revealing:
Tons of trials on the left, new light gray PETG on the right. For some reason when printing full size in Prusa the hole diameters and such work out without compensation, which is weird but good I guess. I think too that the translucency of the white PETG leaked too much light between the LEDs, which blurred the focus of the pitch indication. So light gray is the order of the day. Wish it were a bit lighter, but quite happy to have a good alternative to pick here.
Trouble In The Fields (RIP Nancy G.)
Until now, all D-Lev development has taken place downstairs in our house, the floor is a concrete slab half submerged in a hillside, and the hill itself obscures almost all RF emanating from NYC - a fairly ideal Theremin environment that could possibly produce designs which are hot house flowers.
When I move the units upstairs for show-and-tell I've noticed some instability in the pitch field. It doesn't seem to be broadcast AM RF nor ground related, but these things can be quite ephemeral, and I'm trying to investigate it. It seems to come and go with the time of day, but the upstairs is never quite as stable as downstairs, and sometimes it's radically worse, jumping all around.
I took Roger's P3 around to all the outlets upstairs but none seemed to make any difference. I also watched it with a frequency counter, and watched my 8 transistor oscillator running upstairs too. This all needs much more rigorous and controlled testing, but the 8 transistor oscillator seemed to be more stable.
Yesterday I played with some oscillators upstairs, lost much of the day figuring out I was looking at a spurious ground loop issue (scope was double grounding things). Today I watched two oscillators on the scope and noticed the P3 would really mess with them, but they didn't mess with each other too much. Perhaps the pitch axis operating frequency should be below that of the volume axis, to rule out harmonics interference. Lower frequencies have lower parasitic C & L in the oscillator, so the behavior is more ideal and less "ragged" too.
My somewhat unfounded fear is that a major redesign of the front end may be coming down the pike, which on the physical side might only involve the AFE and perhaps the coil. On the software / firmware side the FPGA axis code may require a very different approach. One theory is the C divider on the AFE is letting noise into the system. And I've noticed since the very beginning of my Theremin experimentation that lower capacitance loading of the coil tends to correlate directly with higher instability, which is another reason I push plates as they have higher intrinsic capacitance.
I've been winding the coils with 30 and 32 AWG, but today I went ahead and ordered some 34 and 36 AWG for experimenting. 30 AWG is falling off a log to wind, 32 AWG is harder to start and keep the turns tight together, and everything above that is increasingly difficult - 38 AWG is a total bear. I'm thinking 4mH for the pitch and 2mH for the volume. If you're in the US, I can recommend Remington for magnet wire, good quality and they get it to you fast:
https://www.ebay.com/usr/remington_industries
This reminds me an EMC test session, the guy at the EMC lab told us (in French) something like "OK let's go to the green field". I thouht that the so called "green filed" was a specific anechoic chamber but we took a car an drove for 30 km in the countryside. We arrived to a place in the middle of nowhere with a mast on which there was an antenna. We put our device on a table, made some test and decided plenty of redesign to do on our almost finished product.
Oh dear! The fields...
If you're in the US, I can recommend Remington for magnet wire, good quality and they get it to you fast:https://www.ebay.com/usr/remington_industries
Ooof! I've just had a rather bad experience with them. Price was right, free shipping was nice, but 2 of my rolls were unusable. My customer service experience hasn't been awesome, so I may go back to using Consolidated.
I find that rewinding the thinner gauge wires onto a dedicated spool helps a lot. That way you can control the tension going onto the spool, and it's much easier to use some sort of power tool for the winding.
For really delicate stuff (40 awg or smaller) I have a modified spinning reel from a cheap fishing setup for winding. I made a little spindle using an old electric drill chuck, and the feed is two pinch rollers from old cassette recorders. By increasing the "pinch" between the rollers, I can adjust the tension to make a smooth wire feed. For winding gauge heavier than 32, I have a cheapo harbor freight electric driver that has really low torque, so it is very controllable...I can usually wind off the spool that comes with the wire. The problem I find I have is that the OG spool has enough twist in the wire that as you wind back, it can exacerbate that, and the wire breaks in a way that seems rather inexplicable.
Most of my projects use spindles that are <25mm in diameter (lots around 5mm) so excessive twist plays a big part in those. I have found that very gentle annealing (100 degrees for 3 hours) relieves the stress enough to make the wire easier to handle and it doesn't break from twist, but it is easier to break from too much winding tension after annealing.
Do you use enameled or polyurethane magnet wire? I have way better luck with the polyurethane. I've tried the polyamidimide and polyimide types, but they were a bit spendy for my wacky experiments!
"We put our device on a table, made some test and decided plenty of redesign to do on our almost finished product." - Mr_Dham
Ooh, my condolences. Sounds like your device got sent to the cornfield ala The Twilight Zone: https://youtu.be/QxTMbIxEj-E
"Ooof! I've just had a rather bad experience with them. Price was right, free shipping was nice, but 2 of my rolls were unusable. My customer service experience hasn't been awesome, so I may go back to using Consolidated" - Flounderguts
Yow, thanks for relating your experience! Roger recommended Remington to me so I can't imagine he's had any issues. They ship real fast and even refunded my sales tax, but I guess we'll see. I only bought one other spool of wire from them, a 1/2 lb of 30 AWG, which seemed OK on the two volume coils I wound: https://www.ebay.com/itm/370810507659. Previous to that I bought several smaller spools of various gauges from Tech Fixx which were also fine: https://www.ebay.com/str/techfixxmagnetwire. I always buy the single coat 155C stuff to avoid stripping the fine insulation, but beyond that I haven't paid much attention to the formulation - from now on I will!
The Consolidated site (https://www.conwire.com/) is super obnoxious with all the popups, forced me to bail before I could give any of their wares a good look-see, a shame.
Flounderguts, have you tried Tech Fixx wire? I like that they have various color options, as well as single and double coat.
I admit that I order from Consolidated from the paper catalog...I just looked at the website, and it IS obnoxious.
Hum / Tracking Filter Overhaul
While looking into the upstairs noise issue, I became fixated on the hum and pitch tracking filter, and think I've got a better arrangement of it. Previously it had six notch filters, one at each harmonic of mains hum (60Hz, 120Hz, 180Hz, 240Hz, 300Hz, 360Hz - or the sequence starting at 50Hz for Euro power). The damping of the first two was 0.25, and 0.125 for the remaining, so as to keep the passband fairly flat. Preceding this was a 4th order low pass filter with critical damping (2 and 1) which was generally kept 4 octaves below the pitch being played, but limited to a minimum of 10Hz and a maximum of 200Hz.
After playing around with it for many days with no ideal arrangement jumping out at me, I finally decided yesterday to convert two of the notch filters to a second 4th order low pass filter, giving an 8th order roll-off. The remaining notch filters were given equal damping of 0.25, which imparts a natural roll-off to the passband (why not?):
Above is just the notch filter response. Along with this, I was able to make the 8th order LPF track a mere one octave below the pitch while fairly effectively suppressing the "traveling hump" of internal interference (associated with drooping +5V supply on the D-Lev). I also lowered the maximum cutoff to 100Hz with no obvious sluggishness. Haven't tested it upstairs yet, but it probably can't make things worse (famous last words).
Job #1 is not letting noise in in the first place, job #2 is filtering the noise that gets in to within an inch of harming the responsiveness.
Journey To The Center Of The Tuner
Had a minor brainwave yesterday to use the center pitch circle LED, which is normally always lit, as a mute indicator, with logic opposite that of the dedicated mute LED, where lit=mute. So the center LED is dark when muted, and lit otherwise, which doesn't get in the way of using it as an aid to playing. Like the dedicated mute LED, it also indicates ACAL in progress via a 2Hz flashing (so you can count them to know the progress), and preset write pending via a more frenetic flashing sequence (to get your attention). My plan was to use the 7-segment decimal place as a mute LED, but this is much more direct and less fussy, plus it dims with the rest of the display. And it's aided by Roger's notion of making the center LED a different color. Don't know why this solution didn't dawn on me years ago, it's so simple.
Making 5 LED tuners for the D-Lev kits today and it feels like I'm finally getting back into the groove.
You must be logged in to post a reply. Please log in or register for a new account.
