[sdiy] Techniques for Multiplying MIDI Clock frequency?

Gordonjcp gordonjcp at gjcp.net
Sun Dec 19 14:55:22 CET 2021

On Sun, Dec 19, 2021 at 09:50:42PM +1000, Adam Inglis (synthDIY) wrote:
> > On 19 Dec 2021, at 20:47, Gordonjcp <gordonjcp at gjcp.net> wrote:
> > 
> > So I guess the MPC60 must use some sort of software PLL to evenly double up the incoming clock pulses to get 48ppcn clocks.  The Roland W30 has something like 192ppcn and also responds to external MIDI clock, so it must do some very clever measurement - or, perhaps, when you get that fine you don't really notice a bit of wibbly-wobbly clocking.
> I’ve not used a W30, but I understand it’s sequencer runs the standard Roland “Super MRC” software that runs on the MC-300, MC-500 and MC-50 midi sequencers, and that runs at 96ppqn, not 192.

It could be.  Both of mine are stored in a mate's workshop just now until I find somewhere with enough space for anything bigger than an ES1 and a pair of headphones.

> Also swing or shuffle aren’t actually even mentioned in the manual, and there is no specific setting for this, you have to do a little trick with the quantize function.

They're not, but you can nudge the timing of individual notes.  It's kind of a shame that you can't do this on-the-fly, but there you go.  The Alesis HR16 can quantise with swing but can't apply swing on playback.

The maths around applying swing to notes that might not have started out bang on the grid probably get a bit too interesting for 8-bit MCUs!

> Gordon you said 67% sounds “lumpy”… but that’s just triplets! It shouldn’t sound lumpy… unless not everyone is in sync!

Bump-dadump-dadump-da-dump, rather than "timing all over the place".  It's hard to convey in an email.

> I take a look at some of this stuff here
> https://mezzoauto.blogspot.com/2017/08/swing-conversion.html <https://mezzoauto.blogspot.com/2017/08/swing-conversion.html>
> WRT to Spiros original enquiry, I don’t understand why you need to drastically change your clock PPQN numbers in order to experiment with polymeter rhythms… but then, I’m not using modular gear. 
> What you definitely do need is rock solid clocking at whatever your chosen resolution, so that any subtleties like the push and pull of swing aren’t lost in the jitter.

I guess it's so that you can divide "strange" note lengths.  I've always assumed that 24ppcn was chosen because it divides nicely for 2/4, 3/4, 4/4, 6/8, and 12/8, and in both simple and compound time you can have triplet notes, and you don't have any leftovers at the end of the bar.  With 5/4 you'd have 19.2 clocks per crotchet, with a clock left over at the end making the last note in the bar a bawhair longer.  Now you could position that spare clock anywhere but I reckon you'd always hear it!  That might give a bit of a noticeable "pulse" to the bar, although if you're dicking about with polymeter are you really going to notice or care?

Dividing into 7/4 and 9/4 give increasing amounts of "leftovers" to cope with.  I guess this is where you'd use a large Euclidean sequence generator to spread the error out across the beats in the bar.  Either way the greater the number of clocks, the better the chance of dividing the bar up evenly so that four crotchets fit as well as your polymetric part.  You could also just pick meters where you have a number of clock pulses that divides evenly into both, but my maths is nowhere good enough for that.


More information about the Synth-diy mailing list