Why is there such disparity in analog tuning routines?

Discussions about anything analog, digital, MIDI, synth technology, techniques, theories and more.
User avatar
Sir Ruff
Synth Explorer
Synth Explorer
Posts: 3519
Joined: Tue May 09, 2006 11:55 pm
Gear: Two persimmon modulators and a frequency adjudicator.
Band: Ruff in the jungle
Location: Philadelphia

Why is there such disparity in analog tuning routines?

Post by Sir Ruff » Tue Dec 14, 2010 3:28 am

I've been wondering for a while why there is so much difference in the length of time it takes different analog synth to get "in tune" with their respective auto-tune routines. The oberheim xpander for example takes forever to tune JUST the oscillators (let alone the filters, PW, VCA, etc), but it then rarely (for me) needs to be done-tuning is extremely consistent from session to session, regardless time span or temp changes in between - I haven't done it in maybe 3 months. The mks80 on the other hand auto-tunes almost instantly, but is always out of tune when powering up, and needs to be retuned after any given period of time, especially after warming up.

Obviously there's different oscillators involved and different tuning algorithms involved for each synth, but there doesn't seem to be a pattern of tune times getting necessarily shorter or tuning being more stable. The OBMX, which uses the same oscillators as the xpander, but is younger, still seems to go out of tune regularly. The most recent analog synth I can think of - the andromeda - needs to be tuned fairly often, and also takes a long time per voice (I realize there are twice as many voices to tune). Regardless of warm-up times and VCO chips, it seems like there would be only so many ways to tune, and that by now, tuning should've been nailed to a T. This is not even mentioning other older synths (such as the yamaha CS series), that hold their tuning surprisingly well, despite the lack of any auto-tune at all.

I've compiled a list of synths with autotune that I've used and the "rough" tuning times.

prophet 5 rev. 3.3 - <5 secs?
OBXA (8 voices) - <3 secs
Xpander - 20 secs? (oscs only - whole routine takes maybe a minute)
Prophet 600 - <22 secs
Roland mks80 rev. 5 - <1 sec.
OBMX (haven't used, but am guessing relatively quick)
Andromeda - 60 secs?

So I guess my two questions are: why is there so much difference in the timings, and how can some synths (like the xpander) have almost perfect tuning without the need for any autotune, and others need it constantly?
Last edited by Sir Ruff on Wed Dec 15, 2010 5:36 pm, edited 3 times in total.
Do you even post on vse bro?

maindeglorie
Active Member
Active Member
Posts: 613
Joined: Mon Nov 27, 2006 3:17 am
Gear: Voyager Select,Little Phatty,ARP Solus,Omni II,Yamaha CS-40M,Prophet 5,Prophet 08,OB-Xa,MaxiKorg,Nord Lead 3,Nord Wave,Eurorack modular, etc.
Location: Pennsylvania (NEPA)

Re: Why is there such disparity in analog tuning routines?

Post by maindeglorie » Tue Dec 14, 2010 3:59 am

My Prophet 5 does indeed tune in 5 seconds and does a great job at staying in tune.
It only needs a 5 minute warmup time too, which is unbelievable to me considering that my F***ING Andromeda takes 20-25 minutes of warmup then takes about 60 seconds to autotune. Not too great for live use.

maindeglorie
Active Member
Active Member
Posts: 613
Joined: Mon Nov 27, 2006 3:17 am
Gear: Voyager Select,Little Phatty,ARP Solus,Omni II,Yamaha CS-40M,Prophet 5,Prophet 08,OB-Xa,MaxiKorg,Nord Lead 3,Nord Wave,Eurorack modular, etc.
Location: Pennsylvania (NEPA)

Re: Why is there such disparity in analog tuning routines?

Post by maindeglorie » Tue Dec 14, 2010 4:02 am

Also, I'd like to state that the ARP Solus is the most rock solid vintage synth in regard to staying in tune I have ever owned. It takes just minutes to warm up and then stays like a boulder.

User avatar
tallowwaters
Moderator
Moderator
Posts: 4998
Joined: Thu Aug 25, 2005 4:11 am
Gear: LC-MS/MS
Location: snake's belly in a wagon rut

Re: Why is there such disparity in analog tuning routines?

Post by tallowwaters » Tue Dec 14, 2010 4:32 am

maindeglorie wrote:Also, I'd like to state that the ARP Solus is the most rock solid vintage synth in regard to staying in tune I have ever owned. It takes just minutes to warm up and then stays like a boulder.
Sounds like Asian food.
Brains can be used like a "stress ball," but only once.

maindeglorie
Active Member
Active Member
Posts: 613
Joined: Mon Nov 27, 2006 3:17 am
Gear: Voyager Select,Little Phatty,ARP Solus,Omni II,Yamaha CS-40M,Prophet 5,Prophet 08,OB-Xa,MaxiKorg,Nord Lead 3,Nord Wave,Eurorack modular, etc.
Location: Pennsylvania (NEPA)

Re: Why is there such disparity in analog tuning routines?

Post by maindeglorie » Tue Dec 14, 2010 5:00 am

ARP & MSG

That's a great evening right there Tallow! :D

User avatar
b3groover
Senior Member
Senior Member
Posts: 965
Joined: Thu Apr 10, 2008 7:41 am
Real name: Jim
Band: THEO
Location: Lansing, MI
Contact:

Re: Why is there such disparity in analog tuning routines?

Post by b3groover » Tue Dec 14, 2010 5:53 am

The Andromeda needs to warm up for at least 15 to 20 minutes before you even bother to tune it. I've noticed that mine stays in tune pretty well in my home studio as long as I give it time to settle.

The Moog Voyager needs about 15 to 20 minutes to warm up as well.

The ATC-x is always in tune; not sure why.

User avatar
garranimal
Synth Explorer
Synth Explorer
Posts: 1728
Joined: Mon Sep 27, 2004 8:57 pm
Location: USA
Contact:

Re: Why is there such disparity in analog tuning routines?

Post by garranimal » Tue Dec 14, 2010 6:32 am

Wow, I'm taking a break at this moment while my A6 tunes and what topic do I find?

In defense of Xpander, Matrix-12, A6: these have background routines running. So after the warm up and initial tuning they're usually good...well a few straggler oscillator aside once in a while. For the A6 we know what that Motorola is capable. The Oberheims have processors that were state-of-the-art at the time they came out anyway, and I can faintly hear the oscillators, filters etc during the tuning routine - sounds like a a beep-and-a-squeek....love that funny little sound.

The Memorymoog took 8-10 seconds to autotune. But I don't think it has any background tuning routines and therefore frequently necessary even after warm-up. Perhaps background routines would have been much more than that poor Z-80 processor could muster, it already had so much it was trying to do.

My Jupe-6 takes 1-2 seconds. That's in line with the MKS-80 that was mentioned.

Otherwise, as for the reason for the wide disparity I couldn't speculate.

User avatar
StepLogik
Expert Member
Expert Member
Posts: 1492
Joined: Fri Dec 29, 2006 4:16 am
Gear: Gauntlets of Ogre Power, Vorpal Sword, Chain Mail +3, Ring of Invisibility, Staff of the Magi, Boots of Speed
Location: Boston, MA

Re: Why is there such disparity in analog tuning routines?

Post by StepLogik » Tue Dec 14, 2010 2:56 pm

i'm sure a lot of it has to do with the algorithm implemented to perform the tuning.

does anyone know exactly what the tuning routines are actually doing? i'm guessing the CPU is sampling the output of the osc using the A/D converter and somehow comparing that to some internal reference?

FYI both my Jup-8 and Jup-6 tune up quickly, usually a few seconds.

User avatar
b3groover
Senior Member
Senior Member
Posts: 965
Joined: Thu Apr 10, 2008 7:41 am
Real name: Jim
Band: THEO
Location: Lansing, MI
Contact:

Re: Why is there such disparity in analog tuning routines?

Post by b3groover » Tue Dec 14, 2010 3:05 pm

garranimal wrote:Wow, I'm taking a break at this moment while my A6 tunes and what topic do I find?

In defense of Xpander, Matrix-12, A6: these have background routines running.
Yes, but with the A6, as you know, you can turn those routines off. I have them turned off in my A6 and once it's warmed up, it stays in tune just fine. As long as the temperature in my home stays relatively stable (a feat for this 100 year old house!) the tuning stays stable for a few weeks at least.

User avatar
Sir Ruff
Synth Explorer
Synth Explorer
Posts: 3519
Joined: Tue May 09, 2006 11:55 pm
Gear: Two persimmon modulators and a frequency adjudicator.
Band: Ruff in the jungle
Location: Philadelphia

Re: Why is there such disparity in analog tuning routines?

Post by Sir Ruff » Tue Dec 14, 2010 6:13 pm

garranimal wrote:Wow, I'm taking a break at this moment while my A6 tunes and what topic do I find?

In defense of Xpander, Matrix-12, ... these have background routines running.
Are you sure? I've never read that anywhere, though that certainly would explain why it remains so stable-I mean, you turn it on and tada, it's ready to go, zero warm-up time needed. I actually don't auto-tune it on purpose just to give the voices a little subtle "breadth". But I'm still not sure there's background tuning...

Roland seems to have nailed the tuning the best-regardless of discrete oscs vs. chips... which would remove that factor as a variable.
Do you even post on vse bro?

themilford
Senior Member
Senior Member
Posts: 819
Joined: Sun Jan 10, 2010 1:19 am
Real name: David
Gear: Oberheim SYSTEM: OB8, DSX, DMX, DXa, Moog Taurus 3, Fender Bass IV LH (CS), Rickenbacker 620/6 (LH), Crumar Trilogy needing major repair.
Band: Season Finale
Location: Brooklyn, NY
Contact:

Re: Why is there such disparity in analog tuning routines?

Post by themilford » Wed Dec 15, 2010 1:43 pm

Sir Ruff wrote: Prophet 600 - <2 secs
Weird, the one I just bought takes about 22 seconds

The Akai AX-60 is about 12 secs.

User avatar
Sir Ruff
Synth Explorer
Synth Explorer
Posts: 3519
Joined: Tue May 09, 2006 11:55 pm
Gear: Two persimmon modulators and a frequency adjudicator.
Band: Ruff in the jungle
Location: Philadelphia

Re: Why is there such disparity in analog tuning routines?

Post by Sir Ruff » Wed Dec 15, 2010 5:36 pm

themilford wrote:
Sir Ruff wrote: Prophet 600 - <2 secs
Weird, the one I just bought takes about 22 seconds

The Akai AX-60 is about 12 secs.
hmm.. I was just quoting based on memory, so yours is probably closer to reality. But that still sure seems long! Ditto for the ax-60.
Do you even post on vse bro?

User avatar
calyx93
Active Member
Active Member
Posts: 448
Joined: Tue Aug 16, 2005 5:48 am
Gear: PPG Wave 2.2, Waldorf μWave x3, Roland JP6+V-Synth, SCI P5+600, Ensoniq SQ-80, Kawai K-3
Location: Winston-Salem, NC

Re: Why is there such disparity in analog tuning routines?

Post by calyx93 » Wed Dec 15, 2010 5:57 pm

My P600 is anywhere from 10 sec to over 20 sec. and the P5 is nearly 15 sec. Both take longer when warmed up from cold and are much shorter once warmed up - maybe three or four seconds each.

The JP6 and JP8 are MUCH shorter - just a second or two. Once warmed up it's as fast as pressing the button - done.

Strange how the discrete JP8 is so much faster at this than the chip-based Prophets.
“I have stretched ropes from steeple to steeple; garlands from window to window; golden chains from star to star, and I dance.” - Rimbaud

User avatar
garranimal
Synth Explorer
Synth Explorer
Posts: 1728
Joined: Mon Sep 27, 2004 8:57 pm
Location: USA
Contact:

Re: Why is there such disparity in analog tuning routines?

Post by garranimal » Wed Dec 15, 2010 8:56 pm

b3groover wrote:Yes, but with the A6, as you know, you can turn those routines off. I have them turned off in my A6 and once it's warmed up, it stays in tune just fine. As long as the temperature in my home stays relatively stable (a feat for this 100 year old house!) the tuning stays stable for a few weeks at least.
Did you turn off both the background routines and the osc temperature tuning routines? I just turned off both of them and will see later today how it does.

User avatar
Synthaholic
Expert Member
Expert Member
Posts: 1206
Joined: Tue Aug 21, 2007 5:43 pm
Gear: Motif XS6, TX802, D-550, A6
Location: NH

Re: Why is there such disparity in analog tuning routines?

Post by Synthaholic » Wed Dec 15, 2010 9:46 pm

A lot of it has to do with how the tuning algorithms work, and how many bands/frequencies are tuned.

Basically how auto-tuning works: the CPU sets the voice being tuned to a certain frequency, then it "listens" to the voice to see how far out of tune it is. It then adds/subtracts an offset to that voice's tuning table, applies that offset, (on some synths) re-samples the voice, and re-corrects a number of times (getting closer each time) until it's in tune. Repeat for each oscillator. The tuning adjustment offsets are stored in the synth's memory and are used to adjust the oscillator's CVs so they play in tune.

Some synths tune each oscillator at several frequencies, to adjust for tracking discrepancies. For example, it might tune each octave over the range of the oscillator. If you watch the A6 display during the auto-tune, you'll see it tune "frequency 1" of voices 1, 2, 3... up through 16, then "frequency 2", then "frequency 3", each frequency tuning faster and faster until it's finished. In this case, there's a tuning offset for each octave (or whatever interval the tuning system uses) stored in the synth's memory. I suppose an extreme tuning algorithm could tune each note individually for each oscillator, but that would probably take a while.

How long the tuning takes depends on (a) how fast the CPU can sample the oscillator's frequency, (b) how many "tries" it makes to tune each oscillator (it may just apply a calculated offset once per oscillator and be done, or it may use a successive approximation, tweaking the tuning closer and closer until it's in tune), (c) how many frequencies it tunes, and of course, (d) how many oscillators there are.

Filter tuning (such as on the A6) is done in a similar fashion: the filter is set to self oscillate and then that signal is sampled by the CPU to tune the filter. I don't know how the A6 tunes filter 1 which doesn't self oscillate, but IIRC it takes a lot longer for the A6 to tune filter 1 than filter 2. It probably sets the filter to bandpass, max resonance, sets the oscillator to the same frequency, then tunes the filter for maximum output.

Some synths, like the Prophet and Jupiters, may do a more thorough tuning on power-up (to get the tuning tables initialized), and then re-tuning is just a quick pass when you hit the autotune button. The A6 stores its tuning tables in non-volatile memory, so there's no need to re-tune on every power-up.

On the A6, the background tuning works like the auto-tuning, but it's performed continuously in the background (unused voices are re-tuned in succession). Temperature tuning simply reads the temperature of the chip using an on-chip sensor and then applies a pre-programmed tuning offset to the oscillators based on the typical temperature drift of the chips used in the A6. It's not as precise as auto-tuning or background tuning, but it does help make the tuning more stable while allowing some drift to still occur to make the synth more organic sounding. I run my A6 with background tuning off, temp tuning on, and I only have to hit autotune maybe once a month. It's a bit out of tune when first powered on, but after about 20 minutes, it usually settles down nicely.
Two VCO: thanks to the push rods, one can choose several forms of waves at the same time!
(from a Babelfish translation of a Jupiter-6 site)

Yamaha: Motif XS6, TX802 Roland: D-550 Alesis: A6 Andromeda

Post Reply