Re: Repeater: Persistent Problems with Sync

[Kim Flint <kflint@loopers-delight.com>]

At 11:53 PM 9/9/2001, Mark Sottilaro wrote:
>on 9/9/01 10:48 PM, Charles Cohen at ccohen@voicenet.com wrote:
>
> > I bought the repeater with the intention of syncing it to other MIDI 
> rhythmic
> > sound making devices, and it just isn't working for me.
> >
> > Try this experiment to make it obvious:
> > Slave a drum machine to Repeater MIDI out.

etc.

>Are you sure this is happening?  Try it again and make sure the Repeater
>hasn't been put to sleep prior to your test.  Unplug it and start it 
>fresh.
>I'm using the Repeater slaved to a Roland MC-307 and I've found NO midi
>synch problems at all.

Charles was talking about clock out, and you are talking about clock in. 
Pretty big difference to me. Brett was talking about sync to midi clock in 
problems. I'm having accuracy problems both ways. All directly out of 
power 
up on the Repeater. I also saw some bugs where it did some weird stuff and 
I had to power cycle it to fix them, although I haven't been putting mine 
to sleep so it wasn't that.

I do really like the idea of time stretching loops to match a changing 
input tempo from midi sync. That actually does seem to work, at least as 
well as sync works on repeater, and it's pretty cool. but I do have a 
"gotta walk before you run" feeling here. Sync just has to work better at 
the basic level.

Before the "it don't seem to work right" problems, I have one issue with 
just the basic way midi sync in works that is driving me nuts. I really 
want Repeater to automate this a little better for me so I can be busy 
with 
something else while repeater captures the loop. I basically want to be 
able to tell it that the loop is to be 2 bars or 4 bars or 9 bars or 
whatever at a given time signature. Then I want to push Record and have 
the 
Repeater capture that much time based on the incoming midi clock, and 
start 
looping it. I don't want to have to press it myself exactly at the end, 
like seems to be required now. (or am I missing something?) If I push too 
early by an eighth note or more, it just goes ahead and makes that the 
loop, and tells me I'm in 3/4 time or 5/4 or whatever, and the resulting 
loop is a mess.  I want to be lazier about that end point so I can focus 
on 
other things, and know that the machine will do what machines are good at, 
which is grab precise amounts of loop time that I tell it to do up front 
and then lock it in sync... This is something I really prefer about sync 
on 
the Echoplex, you can be really imprecise about it as the user and barely 
pay any attention to the echoplex, and the machine takes care of locking 
it 
in precisely. On the Repeater though, I'm constantly having trouble with 
this.

here are the actual problems I found:

For midi clock out:

- Repeater always sends clock! is there any way to turn it off? Right when 
you boot up it is sending clock. In every sync mode it sends clock! I want 
to choose!

- There is something weird about Repeater's midi clock out. I'm going to 
analyze it later with Max, but I haven't had a chance for that yet. Some 
devices are fine and some devices get screwed up. One notable example is 
the EDP, which syncs like a champ to other stuff I have, including my 
"gold 
standard" clock devices. but not to the Repeater. It gets the loop length 
slightly long, as if it missed a few clock pulses. Then it shows it's sync 
LED like it is syncing fine to that slightly slower tempo, so whatever 
Repeater is sending is consistent. Real weird, I haven't seen that one 
before. I'll be investigating this more, so more later. I do notice when I 
use slow tempo and set the echoplex to only grab one beat or one eighth 
note using Repeater's clock, the resulting loop times will vary by as much 
as 40 ms! The same test with the alesis drum machine to the edp is dead on 
every time, at least to the 5ms of resolution I have in the EDP display 
for 
this case. This could be a problem specific between the echoplex and 
repeater, or it could be a problem everywhere that the echoplex just 
happens to reveal. Most devices only care about the tempo from one beat to 
another, so they don't have to keep track of many clock pulses and they 
will be forgiving of a missing pulse or two. With Loop sync though, you 
have to count up a lot more clock pulses to get the length right, and the 
smallest errors really add up as the loop cycles. So it is hard to say 
what 
the problem really is here until I look into it more.

- Really more of an echoplex problem, but repeater happens to do the thing 
that causes it: There is the annoying problem of the current echoplex 
software where it is a little dumb about StartSong messages, and uses the 
first midi clock it gets for it's reference of the startpoint of the loop. 
This is fine in a lot of cases, since many devices don't send clock until 
they start playing anyway. But Repeater always sends clock, and uses the 
StartSong message to note the beginning. (which is arguably the better way 
to do it, but obviously not compatible with everything, and it would be 
nice if the clock out wasn't always on or could be selectable.) So the 
startpoint of the echoplex doesn't match the loop startpoint of repeater, 
but that can be worked around by changing the startpoint reference on the 
echoplex. Or you could just not care, because it is a loop and as long as 
the lengths are equal it is ok. (although they currently are not equal 
between EDP and repeater). The next version of the EDP software is far, 
far 
better about this sort of startsong thing....

- If I have midi sync out going for a while to a drum sequencer, I also 
notice weird things where some drum hits on the receiver are a little off 
one way or another for a beat, then it comes back, same as Charles 
noticed. 
It seems a little unstable. Maybe these are the same missing pulses that 
are screwing up the echoplex.

- My old alesis drum machine gets the tempo correctly from Repeater. I 
know 
it has very solid time, so it is good for testing these things. (I use it 
all the time to test sync on other devices, because it is such a good 
reference.)  I used Repeater to send it a clock at 120BPM, and recorded 
the 
sync'd drum pattern back into repeater. Then I let the two play together 
and see how well sync'd they are. Repeater output is consistently off 
significantly from the drum machine, so that I am hearing a lot of 
flamming 
between the two drum tracks. That means the sync is not very tight. It 
wanders around quite a bit too, but they don't come together, so it isn't 
very stable somehow. Same test on the EDP is very solid. For me, this sort 
of loose sync is a serious problem that Electrix really needs to fix.

[This drum machine sync method is one of my standard sync tests. I beat 
the 
EDP to death with this type of test. Passing means I will only hear minor 
phase problems, because midi clock is not accurate enough to sync to the 
sample level. The most perfect midi clock sync will still give phase 
problems in this test. How much phase problem and how much it moves around 
is interesting. But I should never hear the drums flam. That means the 
sync 
is not very tight. And if I do hear flams, I don't want to hear it 
changing 
a lot because that means the clock has serious jitter problems. ]


Midi Clock in:

- similar test with the drum machine, other direction. drum machine sends 
clock to Repeater, plays drum sequence to Repeater. Record drum sequence 
in 
repeater and play back with the original sequence, in sync. Same problem, 
Repeater flams pretty badly and wanders around the beat. Same test on the 
EDP is dead on and you barely hear any phase change.

- As noted above, it is hard to capture a loop in sync at the right time 
signature without being pretty precise about your tapping. using midi sync 
should eliminate the need for that precision. I frequently get it wrong.

- if you are off by an eighth note when capturing the loop, Repeater 
suddenly doubles the tempo indication. So a 125BPM input clock was read as 
250BPM because I happened to close the loop a little closer to the 15th 
8th 
note instead of the 16th. I couldn't figure out how to convince it to go 
back to the real tempo without doing a power cycle, so I guess this is a 
bug. Really hard on fast tempos to get it right. With a 234BPM clock I 
wound up with repeater at 468BPM! I don't think anybody can play that 
fast. :-)

kim



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