[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Date Index][Thread Index][Author Index]

Re: Syncing the EDP to the Repeater



sorry for copying the mail below whole, but it was from a while ago so I 
figured the context was needed again. I was concerned about this problem 
quite a bit, and one of my reasons to get a repeater was just to figure 
this one out. (how's that for dedication? :)

Anyway, it is true that when you try to sync an Echoplex to the Repeater's 
midi clock out there will be problems. The Echoplex loop time ends up 
slightly longer than the Repeater, and then shifts against the Repeater's 
loop as they play together. I thought this was really weird, since I had 
never seen anything like it with the EDP.

I did some analysis of the Repeater's midi clock output, and sure enough 
there is a subtle bug in the Repeater clock out.

If you aren't familiar with how midi clock works, basically it uses the 
ancient 24 pulses per quarter note method that dates back to the days of 
analog synths. in midi though, the "pulses" are actually a particular 
System Realtime byte. The clock generator sends 24 of these per beat at 
the 
given tempo. The clock pulses are normally spaced evenly in time so the 
receiver can use whichever pulse it needs to time most any sort of rhythm. 
so the time interval between these clock pulses should be equal to 1/24th 
of a beat.

I made a little max patch that times the interval between midi clock 
pulses 
in milliseconds, and saves the data it collects into a text file. With the 
Repeater clock, I see the 23rd pulse of every beat is delayed by an entire 
interval. The 24th pulse then comes immediately after that, with no 
interval at all. so the 23rd pulse is practically right on top of it. So 
if 
the tempo is 125 BPM, the usual interval between pulses is 20ms. Between 
the 22nd and 23rd pulses the interval is 40ms, and between the 23rd and 
24th pulse is less than 1ms. (Max says the delay is 0ms between these, 
since it resolves to 1ms.) That means these last two pulses in the beat 
are 
as close together as two midi commands can be.

The echoplex apparently decides this situation of two clock pulses so 
close 
together and completely different from all the other pulses is nonsense, 
and doesn't count one of them. It counts the first pulse in the next beat 
as the 24th pulse. So it ends up thinking the beat is 1/24th of a beat 
longer than the Repeater thinks it is. This is why a 2 bar loop at 120BPM 
shows as 4.1 seconds long on the EDP instead of 4.0 seconds.

I tried the midi clock outputs from some other devices into my max patch 
to 
get some comparison data and they did not have any problem like this. All 
the pulses were evenly spaced as they should be.

My guess is, this delayed pulse also why drum machines and sequencers 
synchronized to Repeater's midi clock out sound a little wobbly, even 
though they don't always get off so much like the echoplex does. Some 
notes 
probably get shifted around a little while it tries to sort the weird 
clock 
pulses out.

hopefully electrix fixes this one quick, since it is really a drag. The 
only work around I can see here is either use something else for a midi 
clock to both Repeater and EDP, or slave the Repeater to the EDP midi 
clock 
out.

hope this helps,
kim



At 07:27 AM 8/31/2001, muman wrote:
>Hello all,
>Sorry about repeating (repeating) my problem. I have a number of
>excuses, but I won't bore you with any of them.
>
> >
> > does Repeater actually generate MIDI clock out? I looked through the
> > repeater feature lists and manual and saw mention of sync to midi
> > clock in, but couldn't find anything about clock out. It's hard for me
> > to imagine it doesn't do that, so I'm guessing I missed it?
>
>Yes, I'm getting stable clock out. 3 other Electrix components and
>an emu sound module sync accurately.
>
> > Assuming it does, you should see the sync LED flashing on the EDP to
> > indicate it is receiving clock. This is the decimal point on the lower
> > right of the display. It should flash where the Echoplex thinks the
> > beginning of the loop is. Are you seeing that?
>
>Yes, I'm seeing the flash.
>
> > once the echoplex has started receiving sync and the LED has flashed,
> > pressing Record will put you into a waiting period. The display will
> > show "ooo".  It waits for the next startpoint to come along, at which
> > point it automatically starts Recording. When you press Record again,
> > it again shows the "ooo" and waits for the start point again, at which
> > point it stops recording. Did it do that much?
>
>Yes, it also appears to correctly start the recording.
>
> > Then, assuming the EDP continues to receive a good midi clock it will
> > continue to sync to it by aligning it's start point with the midi
> > clock. You can tell when it actually adjusts itself, because two
> > decimal points will flash.
>
>Here is where I run into a problem.
>I just created an empty loop on the EDP of 4.1 (Repeater =
>120BPM, 2 measures). The loop actually started on the 2nd
>measure 2nd beat, than drifts one half beat every cycle.
>
> > If the incoming midi clock tempo stays within a reasonable window, the
> > echoplex will continue to sync to it. This is to handle drift between
> > machines. (I forget the size of the window exactly, somewhere around
> > +/- 5BPM). If the clock changes by more than that, the Echoplex
> > assumes you intend for things to go out of sync and stops trying to
> > sync itself to the source. With Repeater you can change loop tempo a
> > lot, which I assume would change the midi clock out tempo too? If you
> > are doing that and Repeater is the clock master, you will get mixed
> > results with other devices. Some things will follow it and some won't.
>
>During the sync process, I am not changing the Repeater loop
>tempo.
>
> > also, are you using the current software for the Echoplex,
> > LoopIIIv5.0? That made a lot of improvements for midi sync. Most
> > importantly, it allowed the echoplex to continue keeping itself in
> > sync after the initial loop was recorded.  The older software just
> > sync'd the initial loop and then ran free, which tended to have
> > problems with drift.
>
>I am using Loop 3, v5.0
>
>Thank you for your help.
>Louis Hesselt-van-Dinter

______________________________________________________________________
Kim Flint                     | Looper's Delight
kflint@loopers-delight.com    | http://www.loopers-delight.com