I did a song the other day using the Shruthi-1 and I noticed something: the tuning was off.
I have checked recently and it was fine, but now its off.
After stumbling into the tuning menu (hard to get into that menu), i found the amount i had to set it to get it back in tune was -31 (it was at 0)
Is there a trim pot inside somewhere where i can mechanically tune it?
No, there is no trimpot to adjust, for the simple reason that the oscillators are digital. They don’t go out of tune depending on temperature like analog oscillators. So the only thing that the tuning depends on is the quartz oscillator clocking the MCU ; and the quartz oscillators have a a ridiculously low drift / error.
So I suspect it might be something else that has changed in between:
- Did the patch use detuning or any modulation routed to pitch? I got fooled once with a brass patch that has a small amount of velocity mapped to pitch. My master keyboard plays on average at a velocity of 100, by default logic creates notes at a velocity of 80 ; and it was enough to cause a difference.
- Have you sent any CC / NPRN that could have affected tuning?
thanks for the response. both of these are great ideas, but i checked both of them and they werent the answer.
i also thought that perhaps i was hearing the “tune” of the filter (which is analog) but that isnt it either.
ideally, you should adjust “tune” from the init patch, correct?
i have done this, and it works out to be -10 for mine. i suppose everyones is different?
My Shruthi’s main tuning is set to -10 too.
Is this a Europe/US thing? I read somewhere that ‘concert pitch’ is different in the US and Europe…
Interesting… Which procedure do you use to check the master tune? Guitar tuner? Software?
The tables are computed to yield a 440 Hz A4. In version 0.92 and above, the oscillators code has a + / - 3 cents imprecision. The earlier versions of the firmware were less accurate (+ / - 24 cents in the lowest octaves, + / - 6 cents in the highest ones).
Will do measurements tonight!
Interestingly, the difference between A4 = 440 Hz and A4 = 442 Hz is 8 cents.
Actually, -8 is better. I checked with fmit ( http://home.gna.org/fmit/ )
Also, don’t even try to tune your Shruthi with the “pad” wavetable
it’s 8% at middle A, but will drift further out as pitch increases, presumably (since pitch is exponential).
Ah, no, hang on… the difference would get smaller, as a proportion of the distance between notes, as pitch gets higher…
Yup, still at -10 today
Im checking it with the Korg DTR-2000 digital tuner. It seems to be fairly accurate, as its what i tune my other synths to.
I guess it doesnt really matter, just good to know (now that i know!)
Its interesting that hi/low notes seem to drift further out.
For me, these tunings are what i found:
C0 = -19
C1 = -18
C2 = -17
C3 = -14
C4 = -12
C5 = -10
C6 = -14
C7 = -18
C8 = -18
(i tested using INIT patch)
I’ve spent a great part of my evening investigating the cause of this issue.
I have found two things.
The first one is that there’s some jitter on the AVR timers. Not a problem, but it contributes to the “character” of the oscillators.
The second one is that there is indeed a timing bug causing a 0.4% error in pitch (8.6 pts on the tuning page). For the technical minded people among you, this is due to the timer in phase correct PWM module counting 0, 1, 2, 3, 4, 5 … 255, 254, … , 1, 0, 1, 2 (picture a triangle wave), and thus counting 510 steps per cycle instead of 512, and the overflow interrupt is thus called 512 / 510 more often than it would be if you were just halving the frequency of the fast PWM mode. I’ve just found that in the ATMega datasheet.
So the actual sample rate of the Shruthi waveform generation is not 39.06 kHz but 39.21kHz. One small step towards hi-fi!
The firmware files of v0.93 have been updated.