Muff's Modules & More

roglok · Posted: Tue May 22, 2012 2:52 pm Post subject:

I just downloaded your python script from github, but I'm having a hard time getting it to run properly on a mac.

I'm hopeless as a programmer and never tried python, so there's not much I can do about it. The script runs and the interface seems to work, but I'm getting the following bugs after 16 steps have played:

ringroad1 · Posted: Tue May 22, 2012 3:04 pm Post subject:

hey roglok, looks like you're using Python 2.6 - think it needs Python 3.

Got it working happily here on Snow Leopard with Python 3 installed via Homebrew.

roglok · Posted: Tue May 22, 2012 3:37 pm Post subject:

hey, thanks for the hint. unfortunately it doesn't help. i installed python 3 and even cloned the git repository via command line, but i'm getting this mess after the pattern has run once:

ZoeB · Posted: Tue May 22, 2012 4:21 pm Post subject:

Thanks for trying this out, everyone!

I believe the filename bug is something I broke earlier on, and fixed earlier today. I wasn't very consistent with where it was defined, basically. Try "git pull origin master" (or probably simply "git pull") to get the latest version, and see if that works.

And the other news: I've implemented Sync24 output (changeable to Sync48 by changing a constant in the code). You have to slow it down to about 1BPM to actually see it in action. I wasn't sure if the run/stop signal could go up at any time, or if it had to politely wait for the next time the clock pulse starts to go high. I decided to play it safe and opt for the latter, to ensure (hopefully!) rock solid timing.

I'm not very up on hardware, but I'm hoping it should be as easy to implement as finding two more digital outs to assign the sync output to, and soldering them up to the appropriate pins of a DIN socket. That may be wishful thinking on my part, though. :)

Let me think about the other features later on. Nina (my partner) has pointed out to me how simple Sync24 input should be, so I'll try that next, probably later this week.

roglok · Posted: Tue May 22, 2012 4:55 pm Post subject:

brilliant! thanks for the fixes. works like a charm now thumbs up

ZoeB · Posted: Tue May 22, 2012 5:01 pm Post subject:

OK, a Sync24 input question... if you get the run/stop gate at any time other than when one of the clock pulse triggers is starting, are you supposed to start playing immediately or wait for the next clock pulse? For the input, I really need to follow the existing conventions if I want this sequencer to sync up properly with other devices...

roglok · Posted: Tue May 22, 2012 5:08 pm Post subject:

ZoeB · Posted: Wed May 23, 2012 2:23 am Post subject:

ZoeB · Posted: Thu May 24, 2012 6:29 am Post subject:

roglok · Posted: Thu May 24, 2012 8:11 am Post subject:

sounds great, zoe! i'm on vacation so i can't check it out atm, but will do so when i get back...

ZoeB · Posted: Thu May 24, 2012 12:06 pm Post subject:

OK then... does anyone have any more information on Tape Sync (frequency shift keying), or any examples?

It sounds like you take the Sync24 / DIN Sync clock pulse, and whenever it's high you output a high frequency tone, and whenever it's low you output a low frequency tone. (Hey, I think I finally understand what modems and early home computers' tape recorders do! Modulating and demodulating, that makes sense...) It's still just square waves, so still a boolean value being sent to a simple digital output. Fine, I think I've got that working. Some sources (An Introduction to Tape/MIDI Sync and the MQX-32 and Sync or Swim?) claim the frequencies are 2400Hz and 1200Hz respectively, while others (MC-4 service notes, page 4) say "approximately" 2100 and 1300. I'm guessing Roland are more authoritative on the matter, if a little vague. I'm not sure how accurate this needs to be...

The main thing I'm not sure about, though, is how Tape Sync knows when to start and stop. This is a separate signal entirely in DIN Sync, whereas Tape Sync only uses one tape channel. The MC-8 owner's manual, pages 107 and 108, make mention of starting and stopping the track, with a bit of breathing room around it. Maybe it just says this to ensure you actually leave the clock pulse running long enough, but I'm hoping the equivalent of the run/stop signal gets passed along somehow. That would certainly seem to be the point of synchronisation, to not just match the speed but also the start time. Plus the manual says this modifies the carrier, insinuating either that it's further modified once you start playback, or that it wasn't modulated for an audible medium beforehand maybe (as in, maybe it sends out the raw Sync24 until the sequence is started)? Perhaps the sync signal is only output to the Sync24 or Tape Sync output in a mutually exclusive manner, so the manual's saying it's neither modulated nor output to tape at all until the song's started? Wouldn't that mean the slave device would only know the correct tempo after the song's already started? I have no idea...

So does anyone know how Tape Sync works, or better yet, have a .wav or .aif file example of it in action that I could dissect?

Thanks!

ringroad1 · Posted: Thu May 24, 2012 2:09 pm Post subject:

Couldn't work out where to put them so hopefully this'll do:

http://stevewoodward.posterous.com/roland-mc-4-sync-tones

It should be fairly obvious when I'm starting the sequencer.

darenager · Posted: Thu May 24, 2012 2:19 pm Post subject:

IIRC tape sync uses an initialising tone before anything starts to sync, then once the tone starts to modulate the slave will start. Just curious as to why you are thinking of implementing tape sync?

ZoeB · Posted: Thu May 24, 2012 3:08 pm Post subject:

bananeurysm · Posted: Thu May 24, 2012 3:21 pm Post subject:

Sweet! Looking forward to seeing this in action!

I sincerely hope you're not emulating the mc202 sequencer. I have a LOT of experience using it - and it only gets more and more frustrating the more you use it. I nominate it for worst internal sequencer ever. x0x on the otherhand is pretty nice!

ZoeB · Posted: Fri May 25, 2012 3:18 am Post subject:

ZoeB · Posted: Fri May 25, 2012 5:20 am Post subject:

Oh, I think I just properly worked out how the syncing works. I think it might be much simpler than I realised. What I've got Stepper currently doing is timing the distance between the last two pulses, and extrapolating the tempo from that, but that's no good if the song has to start before it's heard some pulses (so fine for DIN sync, but no good for tape sync). But now I'm thinking that's really overly complicating things, that the pulses aren't used to synchronise clocks among different machines after all; maybe the pulses are the clock, and the slave's internal clock gets disabled... maybe the slave machine just has to count pulses from the master, ignoring entirely how long they take to arrive, and if it's 24 pulses per quarter note, and the device only plays sixteenth notes, then it merely has to only increment a step once every six pulses. Hence, it doesn't need to be led in to get adjusted to the tempo, and the timing's bang on no matter how much you change the tempo half way through a song. It's a simple and elegant solution, great.

The only point at which it breaks down is if you need to update something more than 24 times per quarter note, such as portamento. Which further explains why the TB-303 implements sliding in hardware, not software (TB-303 service notes, page 5). So does that mean the TB-303's slide length, the time it takes to glide from one note to the next, isn't just independent of the pitch distance, but also independent of the tempo? And if that's the case, does anyone know how long it takes? I've currently got Stepper sliding only for the last half of a sixteenth note, so it takes a 32nd of a bar exactly. This seemed like an elegant solution until it had to be hardware and interface with other machines...

I guess I can refactor most of Stepper to work the simple way, but keep the portamento working the complex way. That way, the timing of everything except the portamento will be absolutely spot on always, and it will still be able to have tempo dictated portamento lengths, that may be slightly out if it's slaving to tape sync and there's portamento on the very first note. That's probably an acceptable trade off...

ringroad1 · Posted: Fri May 25, 2012 12:59 pm Post subject:

There's some discussion of the portamento/slide of the 303 and 202 over at ye olde hyperreal - seems like the 303 is constant time, as you suspect. Dunno how long it takes tho.

Robin Whittle wrote some good stuff on the triggering of the 303 slide which might come in handy:
http://www.firstpr.com.au/rwi/dfish/303-slide.html

Like I say, personally I'd take a look at the 202 slide circuit, but if you are trying to make several voices I can see why you'd be tempted to do it in software.

ZoeB · Posted: Fri May 25, 2012 1:27 pm Post subject: