Gentlemen,

I want to sync SMPTE striped analog tape machines to a computer recording
application. I currently have an Roland SBX-80 SMPTE to MIDI converter.
The recording application will only accept MTC or midi clock for sync. Will
this unit give the same sync accuracy as the newer SMPTE to MTC converters?
It's difficult to find information on the SBX-80 since it is no longer
supported by Roland. This particular unit is not your typical midi clock
generator and I don't believe this unit lets the midi clock "free run" as
many units do. Of course I don't want to purchase something I don't need as
well! Anyone have knowledge or experience using this unit? I will do
some tests for sync accuracy but thought if someone has already had done
this, why not ask!

Thanks,
Jeff

When chasing an analog machine, it is important for the analog deck
and the DAW to resolve to the same ext video sync source. aka
blackburst. This requires a syncronizer on the analog machine so the
capstan motor will resolve, and a SMPTE to MTC converter with external
video sync input. Then simply patch the SMPTE off the analog deck
into the converter box, and set your software app to chase.

Some people have had success without the ext. video option by using a
SMPTE to MTC converter alone, but the devices will surely "drift" away
from each other after initial trigger.

Mark






.. On Fri, 22 Aug 2003 20:57:31 GMT, "Jeffrey Landgraf"
wrote:

Gentlemen,

I want to sync SMPTE striped analog tape machines to a computer recording
application. I currently have an Roland SBX-80 SMPTE to MIDI converter.
The recording application will only accept MTC or midi clock for sync. Will
this unit give the same sync accuracy as the newer SMPTE to MTC converters?
It's difficult to find information on the SBX-80 since it is no longer
supported by Roland. This particular unit is not your typical midi clock
generator and I don't believe this unit lets the midi clock "free run" as
many units do. Of course I don't want to purchase something I don't need as
well! Anyone have knowledge or experience using this unit? I will do
some tests for sync accuracy but thought if someone has already had done
this, why not ask!

Thanks,
Jeff

"Jeffrey Landgraf" wrote in message
m...
Gentlemen,

I want to sync SMPTE striped analog tape machines to a computer recording
application. I currently have an Roland SBX-80 SMPTE to MIDI converter.
The recording application will only accept MTC or midi clock for sync.
Will
this unit give the same sync accuracy as the newer SMPTE to MTC
converters?
It's difficult to find information on the SBX-80 since it is no longer
supported by Roland. This particular unit is not your typical midi clock
generator and I don't believe this unit lets the midi clock "free run" as
many units do. Of course I don't want to purchase something I don't need
as
well! Anyone have knowledge or experience using this unit? I will do
some tests for sync accuracy but thought if someone has already had done
this, why not ask!

Thanks,
Jeff


Hi,

Why would you want the converter to 'free run'?
I think your SMPTE to MTC converter would be ok,
but there may be problems in how your software can handle
MTC vs word clock discrepansies.

I think there is units (motu,aardvark?) that convert SMPTE to word clock and
MTC simultaneously, word clock output varies with the wow&flutter of the
tape machine
and software happily follows, but I have no personal experience.

My personal solution was (PC with Sonar & Fostex G24S tape machine)
to record SMPTE to PC as an audio track and feed that track to the
SMPTE input of the tape machine. ie. PC was SMPTE master and
the tape machine a slave. That worked out fine.

If you have enough analog inputs in your audio card you could
transfer all the tape tracks to computer in one pass and forget
all about SMPTE and suffer a little less.

-CS

In article ,
"Jeffrey Landgraf" wrote:

I want to sync SMPTE striped analog tape machines to a computer recording
application. I currently have an Roland SBX-80 SMPTE to MIDI converter.
The recording application will only accept MTC or midi clock for sync. Will
this unit give the same sync accuracy as the newer SMPTE to MTC converters?

As good as. The sync accuracy is more likely to be limited by your
computer's response to MIDI.


It's difficult to find information on the SBX-80 since it is no longer
supported by Roland. This particular unit is not your typical midi clock
generator and I don't believe this unit lets the midi clock "free run" as
many units do. Of course I don't want to purchase something I don't need as
well! Anyone have knowledge or experience using this unit?

The SBX-80 predates MTC. It will play a MIDI Clock between a set start and
stop time. More complex tempo maps are possible, but not required for your
purpose.

These days it is easier to transfer the audio to hard disk and do
everything further there.

Graham,

When you say "as good as" what are you basing this on? Have you used the
SBX-80 unit or have specific knowledge of it? Yes, I agree, if I had 16
channels of AD I'd simply do a transfer. Unfortunately, I have 2 SMPTE
sync'd 8 track machines and would like to move 7 channels at a time of
pre-existing audio to the computer for overdubs and mixing. I had hoped
someone (hopefully you!) had used this box for this purpose already. More
information would be great. By the way I'm using Samplitude (RME AD/DA) and
will investigate their midi sync feature as well.

Thanks for your help on this,
Regards,
Jeff


"Graham Hinton" wrote in message
...
In article ,
"Jeffrey Landgraf" wrote:

I want to sync SMPTE striped analog tape machines to a computer recording
application. I currently have an Roland SBX-80 SMPTE to MIDI converter.
The recording application will only accept MTC or midi clock for sync.
Will
this unit give the same sync accuracy as the newer SMPTE to MTC
converters?

As good as. The sync accuracy is more likely to be limited by your
computer's response to MIDI.


It's difficult to find information on the SBX-80 since it is no longer
supported by Roland. This particular unit is not your typical midi clock
generator and I don't believe this unit lets the midi clock "free run" as
many units do. Of course I don't want to purchase something I don't need
as
well! Anyone have knowledge or experience using this unit?

The SBX-80 predates MTC. It will play a MIDI Clock between a set start and
stop time. More complex tempo maps are possible, but not required for your
purpose.

These days it is easier to transfer the audio to hard disk and do
everything further there.

In article ,
"Jeffrey Landgraf" wrote:

When you say "as good as" what are you basing this on?

I chose my words carefully there
MTC is not really good enough for for the accuracy you would want when
synching two ATRs at bit lock. It should be, it could be, but
implementations are poor.
The SBX-80 is from a different era, before everything with MIDI sockets had
to be made cheap regardless of whether it still worked or not.

Have you used the
SBX-80 unit or have specific knowledge of it?

Yes, it is not really what you want though. It was intended to slave drum
machines and early sequencers to audio. In the days of single tasking eight
bit CPUs the word "latency" was never heard. In the days of multi tasking
32 bit CPUs asking A/D converters to slave their clock to incoming MIDI
Tempo Clock is asking a lot.

You were told almost the right answer earlier, what you really need is both
ATRs slaved to video/LTC and a master clock for A/D also derived from this
source.


Yes, I agree, if I had 16
channels of AD I'd simply do a transfer. Unfortunately, I have 2 SMPTE
sync'd 8 track machines and would like to move 7 channels at a time of
pre-existing audio to the computer for overdubs and mixing.

Two passes of two tapes never lines up exactly. Similarly two passes of the
SMPTE track will not either so it is no use deriving the clock from that.

I'd get it down by somebody who can and move the data files to your
computer. It will save a long of hassle in the long run and be cheaper than
buying the gear for one job. OTOH if you have a lot of tapes to transfer,
buy the gear.

In article ,
Bob Olhsson wrote:

I've never had the slightest problem getting a consistent,
sample-accurate frame rollover using MTC except with Pro Tools where
there is some kind of a strange + or - 5 sample slop that doesn't exist
when using other applications with the very same hardware.

1/48000th of a second is pretty fine resolution if you ask me!

Resolution is not the same as accuracy.
The resolution of LTC is 1/2400th of a second and that of MTC 1/600th of a
sec. Given free running tape and commercial LTC-MTC converters and layers
of software heirarchy in the receiving computer you don't get 1/48000th
second accuracy.

I made an audio test CD a couple of years ago comprising one channel of
timecode and one channel of a tone exactly in sync, ie it had 1/44100th
second resolution AND accuracy. Anybody can try this using Cubase to
generate a timecode file and an accurate tone generator like Make A Test
Tone (don't use the Pro Tools one - it is not accurate). This was used to
test a lot of current MIDI interfaces and software using the tone as a
'scope trigger. Not one MTC converter got the quarter frames correctly on
quarter frame boundaries and there was a fair amount of jitter on the
message positions. Then audio and MIDI output events in the computer were
compared to the sub-frame positions they were positioned at. They were all
over the place. The purpose of the CD was to provide a stable reference,
rather than have the variations of tape speed complicate the measurements.

Any hardware/software system that locks to a lower frequency reference is
some form of damped PLL. There is always a tradeoff between short term
accuracy and long term accuracy.
The true test of "consistent, sample-accurate frame rollover using MTC"
would be to digitise the same track twice from ATR locking to its LTC and
then to null out completely the two passes. If you can do that I want to
know what equipment you are using!

In article ,
Bob Olhsson wrote:

Accuracy of the frame rollover is the only thing that counts. More bits
only potentially buys you a faster lock up. In practice, most gear that
only supports LTC has even less resolution under the hood than MTC.

Are we talking about the same thing?
I'm not sure what you mean by frame rollover, to me it means the second
boundary and that is not the only thing that counts.
Yes, there is a lot of dubiously designed gear out there though.

Remember that timecode ONLY identifies position and not speed which
must always be calculated from the reference signal.

LTC can be the reference itself and that may be measured on a bit, byte or
frame basis. MTC has to be measured too, but the reliability of that
measurement is highly suspect if it is made after trickling through OMS and
MacOS before it is done in the high level application software. If the MTC
were stable to start with and the audio bit clock were generated by a
dedicated "gear box" device it can be done, but not with general purpose
MIDI ports on a general purpose computer.


There has never been such a thing as "plug and play" timecode!

On that we concur.

In article , Graham Hinton
wrote:

LTC can be the reference itself and that may be measured on a bit, byte or
frame basis.

LTC should NEVER be used as a reference. It has never been specced for
that and is frequently a sure way to introduce excessive wow, flutter
and jitter into any analog or digital audio recording.

Again, the only thing that counts is the rollover being right on the
money. This is the instant the frame actually changes. If the timecode,
the audio and the picture are all consistently in the proper location
at the same time to the sample, you are in sync. If any one or all
three aren't, you are not in sync. It's really just that simple. You
can't be more or less in sync, you are either locked up in sync or
you're running wild. Operating perfectly in sync using MIDI time code
is done every day at numerous post facilities around the world.

--
Bob Olhsson Audio Mastery, Nashville TN 615.385.8051
Mastering, Audio for Picture, Mix Evaluation and Quality Control
http://www.hyperback.com/olhsson.html
Over 40 years making people sound better than they ever imagined!

Graham Hinton wrote:

In article ,
Bob Olhsson wrote:

Accuracy of the frame rollover is the only thing that counts. More bits
only potentially buys you a faster lock up. In practice, most gear that
only supports LTC has even less resolution under the hood than MTC.

Are we talking about the same thing?
I'm not sure what you mean by frame rollover, to me it means the second
boundary and that is not the only thing that counts.
Yes, there is a lot of dubiously designed gear out there though.

Remember that timecode ONLY identifies position and not speed which
must always be calculated from the reference signal.

LTC can be the reference itself and that may be measured on a bit, byte or
frame basis. MTC has to be measured too, but the reliability of that
measurement is highly suspect if it is made after trickling through OMS and
MacOS before it is done in the high level application software. If the MTC
were stable to start with and the audio bit clock were generated by a
dedicated "gear box" device it can be done, but not with general purpose
MIDI ports on a general purpose computer.



The Windows driver kit supplies a high-resolution ( 1 ms ) timestamp
for MIDI events, including MTC events. To my knowlege, this timestamp
is derived from the hardware. There'll be jitter on actual time of
receipt of events, but it can be resolved to millisecond
resolution against the timestamp.

Nothing like this exists on a Mac?

There has never been such a thing as "plug and play" timecode!

On that we concur.


--
Les Cargill

In article ,
Bob Olhsson wrote:

LTC should NEVER be used as a reference. It has never been specced for
that and is frequently a sure way to introduce excessive wow, flutter
and jitter into any analog or digital audio recording.

You are talking about syncing to video. LTC has been used as a reference
for audio multitrack synchronisation and automation for some 25+ years and
I don't know about excessive, it is not much worse than the master machine.

For two audio machines a constant sub-frame offset is no use, better than
bit lock is required.

In article ,
Les Cargill wrote:

The Windows driver kit supplies a high-resolution ( 1 ms ) timestamp
for MIDI events, including MTC events. To my knowlege, this timestamp
is derived from the hardware. There'll be jitter on actual time of
receipt of events, but it can be resolved to millisecond
resolution against the timestamp.

Nothing like this exists on a Mac?

There are manufacturers' claims and then there is a reality gap.
Timestamping may be derived from a hardware counter, but the time that is
done is when the incoming data is serviced and that can be tens or even
hundreds of milliseconds (worse case) after the event occurs.

OMS and OS X Core MIDI claim 1ms accuracy, but that is not realised with
USB or multiport serial interfaces. Some of the proprietary timestamping
schemes which preload the interfaces have some very bizarre algorithms and
are data dependent. 0.5ms may be obtained using FreeMIDI and individual
serial ports, but very few people have that sort of setup.

The times quoted are usually referring to playback under optimum
conditions, not the delay of playing a keyboard through the combined delays
of interface, system driver and application. A quick comparison of playing
a module from a keyboard directly and then routing via a computer, Mac or
PC, will demonstrate that 1ms timing is wishful thinking.

This is what makes locking to incoming MTC laughable. Errors of the order
of milliseconds on a quarter frame period are a very large percentage.

In article ,
Les Cargill wrote:

I guess the Soundblaster I use is probably pretty close to a single
serial port, although I wonder about drivers that exhibit that
kind of delay variation. It's got it's own friggin' interrupt,
fercryinoutloud.

That would be optimum, depending on the interrupt latency of coarse. On a
Mac the only serial port left is the vestigial one for the modem. Then
serial support for MIDI got removed between 10.2 and 10.3 although there
were not actually any drivers for 10.2.


You mean on the complex interfaces? Maybe so. With one port and one MIDI
device, I'm seeing pretty much that.

Yes, the old serial multiports and the new USB ones are pretty bad. The
bigger they are, the worse it gets.


That's probably close enough for a lot of applications, though.

Yes, if you are locking to video you know what the rate is so it can be
averaged over a long period.

But I suspect
the multiport interfaces that aren't hardware-switched do incur some
jitter, and the drivers probably add some, too.

Yes. The only hope is that someone will come out with a serious FireWire
interface. There are FireWire devices with MIDI appearing now, but the
FireWire chips split MIDI off to a microprocessor and so replicate all the
problems of the old multiports.

In article , Graham Hinton
wrote:

You are talking about syncing to video.

NO, I'm talking about how timecode works. Automation and MIDI
instruments are triggered events that do not involve speed however
continuous audio always does. Tape machines can use a number of
references besides video sync.

--
Bob Olhsson Audio Mastery, Nashville TN 615.385.8051
Mastering, Audio for Picture, Mix Evaluation and Quality Control
http://www.hyperback.com/olhsson.html
Over 40 years making people sound better than they ever imagined!

In article ,
Bob Olhsson wrote:

NO, I'm talking about how timecode works. Automation and MIDI
instruments are triggered events that do not involve speed however
continuous audio always does.

Only in sound for picture. If the ATR is master, automation and MIDI events
have to be scaled to match the relative time speed. In the extreme case you
get people like Peter Gabriel changing the final tempo of a track from the
multitrack varispeed and expecting everything else to follow.

The automation devices I know about all generate their timing from a higher
multiple of the incoming reference and then divide back down again so it
doesn't matter if the timecode is a different flavour or being varispeeded,
it will still follow in proportion. Are you telling me that there are
plenty that don't work like that?

In article , Graham Hinton
wrote:

Only in sound for picture. If the ATR is master, automation and MIDI events
have to be scaled to match the relative time speed.

No, they are triggered by tape location. Speed can be inferred from
timecode chages but you NEVER want timecode controlling speed because
none of the gear using time code was designed to do that other than as
an emergency fix as an alternative to jamming sync.

--
Bob Olhsson Audio Mastery, Nashville TN 615.385.8051
Mastering, Audio for Picture, Mix Evaluation and Quality Control
http://www.hyperback.com/olhsson.html
Over 40 years making people sound better than they ever imagined!

In article ,
Bob Olhsson wrote:

In article , Graham Hinton
wrote:

Only in sound for picture. If the ATR is master, automation and MIDI events
have to be scaled to match the relative time speed.

No, they are triggered by tape location. Speed can be inferred from
timecode chages but you NEVER want timecode controlling speed because
none of the gear using time code was designed to do that other than as
an emergency fix as an alternative to jamming sync.

What you are saying is that YOU don't want that in sound for picture, which
is fair enough because you don't varispeed it, apart from pull up/downs. In
using timecode from a master ATR through a device like an SBX-80 (which is
where this thread started) to control a MIDI tempo based device, like a
drum machine, I most certainly DO want it to follow the speed, otherwise it
will not be in time with the music on the tape.

Maybe you call that following the location? I prefer to call that speed
because I'm more using to dealing with the clock edges and PLLs. They are
locked together on the tape so it doesn't make any difference how you view
it.
For devices that are only frame accurate, like I think you are talking
about,they will still follow and were not accurate enough for musical
timing anyway.

I don't know where you have got the idea from that no timecode gear was
designed to do this. I personally have designed sub-frame accurate
automation devices for SSL, Amek and motionworks (to name a few) and have
designed timecode reader chips specifically for these sorts of
applications.

In article ,
Bob Olhsson wrote:

When the reference is right on the money and the frame rollovers are
correct to the 48kHz. sample frame after frame for hours on end, that's
accurate by any measure! The ONLY difference "only frame accurate"
makes is how quickly the system can achieve precise syncronization at
whatever speed you choose.

True as far as a system locked to a MSG goes, but that is not what we were
talking about.
How do you think it is possible to generate a sample clock on a sound card
when the only reference is MTC or a MIDI Clock trickling through software
on a general purpose computer OS? You can't, not without errors, which is
why I said don't solve the problem that way.
I answered the OP's question, you are talking about something else out of
context.

In article , Graham Hinton
wrote:

How do you think it is possible to generate a sample clock on a sound card
when the only reference is MTC or a MIDI Clock trickling through software
on a general purpose computer OS? You can't, not without errors, which is
why I said don't solve the problem that way.

And my point was that you can't using just LTC any more than you can
MTC.

--
Bob Olhsson Audio Mastery, Nashville TN 615.385.8051
Mastering, Audio for Picture, Mix Evaluation and Quality Control
http://www.hyperback.com/olhsson.html
Over 40 years making people sound better than they ever imagined!