External sync drifting when syncing to tape

Cubase
1

Hi there

a while back I got myself a tascam msr 16 tape machine and have been working on a fully analogue project - I’m only using Cubase for controlling my external analogue midi synths. I’ve striped the tape and the smpte timecode goes to my jl cooper pps-100 which is then connected via midi to my Rme fireface 400 box. When I hit play on the tape machine Cubase starts and is perfectly in time, but then gradually drifts so that by the end of a 4min song it can be 20+ milliseconds. I’m using 30fps on the jl cooper and in cubase. The drift directly relates to the tape machine warming up.

in an attempt to workaround, which works reasonably well, i record the synth drums from a drum machine that is itself the master (I cue it with a count in from cubase), then I record the drum tracks back to the pc and line the cubase project up to it via warp. This works well if the machine has settled and there isn’t too great a pause between setting up warp and hitting record of the next item to be laid down (ill often have to retransfer the drums into cubase just to see how far it’s drifting at that moment).

Cubase will be start and be perfectly in time when timecode begins and will stop when the timecode stops, but it’s almost as if it it’s not paying attention in between.

I had the exact same problem when I had cubase 6 which was on a windows 7 pc (my current PC is windows 11). the timecode source is set to midi timecode, lock frames 2 drop frames 3, though I’ve had various different settings. I can rule out the Jlcooper box too, as I previously had an EES one and the RME fireface has a built in MTC convertor also - same problem with all devices. I’ve swapped midi cables and also have a midex 3 which has given me the same results. Years ago I used ataris and even a very early windows PC (probably 97) and never had this problem.

So i can only see two things left, either Cubase or me doing something stupid and not getting my head around it properly. i didn’t find the manual much help for my particular recording situation.

I would greatly appreciate if anyone has any ideas.

Mark

2

I guess one needs a tape machine in order to test if this is a reproducable issue in Cubase.
Did you stripe the tape with the SMPTE code generated by Cubase?
Please also drop the info which Cubase version you are using exactly, e.g. Cubase Pro 14.0.41.

3

Hi Johnny, thanks for getting back.

I have previously striped the tape with Cubase’s own generator, in fact that was what I’d done originally when I got the tape machine. However the problem was there with that too, these days i use the JL cooper device to stripe it as it will be that device that’s reading it.

I anticipated the tape machine question last night, so I got my JL cooper box and put it under my PC screen, activated it’s free run button which sends out Midi time code via the midi cable to the fireface/pc box. Unlike my other devices the Jl cooper has a time display on the front so I could use that (via slow-mo iphone video)to compare with what Cubase is showing (btw no audio files plug ins or instruments in the cubase project, buffer size 96samples/3.19ms in/4.354 out. At the start there’s possibly a 1 frame lag (33ms at 30fps), it’s difficult to be too precise because of display refresh rates and no doubt a host of other stuff, but by 4 mins its about 2 frames and at 10.00.03 cubase is reading 10.00.08 possibly 5 frames (166ms). I will run the two for a longer stretch just to make sure that this isn’t just a display discrepancy, but the fact that the initial discrepancy increases through time suggests not to me.

So not just a tape machine, possibly when Cubase (or the Cubase on my pc only) is being a slave to any device via MTC at least.

I suppose, when I think about it, the other culprits could be windows, firewire (which isn’t technically supported in 11, but i have a working card and drivers), though my fireface midi is really tight - 1 to 2 milliseconds. There is always the option of purchasing an additional usb midi interface I suppose.

The point is, regardless of how many milliseconds , I can hear the discrepancy and it makes it very difficult layering things on afterwards. I also have a Nagra T audio machine which has the facility to follow/sync to timecode far better than this - which kind of blows my mind.

One final important point I think, there are no error messages anywhere, Cubase’s timeline runs, the JLcooper box locks in when it receives code - everything looks like it’s working except it seems that other than for start and end - each is doing their own thing.

My version of Cubase is 14.0.32 and I was previously up to date with v6.5? on the other computer.

btw 20mins running now jl coop says 20.00.03 cubase says 20.00.15 so approx 12 frames 396ms.

best

Mark

4

So we could focus on Cubase’s behaviour as a MTC slave and put the SMPTE part aside for a moment.

The weird thing is that unlike a clock signal MTC should not drift as the position is transmitted in every (full) message.
Which raises the question: Does Cubase receive the messages on time or is any component between Cubase and the JL “loosing it” over time?
Do you have a midi monitor software that could replace Cubase in the setup? Then you can run your test again.

5

Thanks for the suggestion. I downloaded midiview and ran the JL cooper device for a short time, here is a screen shot below. Don’t really know what I’m looking for, but all the messages are the two hexidecimal numbers F1 followed by 41,50,60,76 and the odd ascending variation, there’s nothing that stands out immediately

best

mark

jlcooper to rme pc
jlcooper to rme pc1864×1440 193 KB

6

It sounds like Cubase is communicating with “address” signals, but the actual word clock (the higher frequency 44k, etc, “teeth on the gear”) is not tracking.

Do you have the “word clock out” of your JLC synchronizer connected to the “word clock in” of your sound card, with the driver set to “external sync”?

1 Like
7

Hi there

thanks for your response. It shouldn’t require word clock as I’m only synchronizing midi, not audio. I was only using the audio from the tape machine as a reference as to how far out the midi sync was. the JLcooper device doesn’t have word clock as it was only built for syncing midi devices, tape machines and sequencers/ataris/macs/pcs.

best

Mark

8

These are my thoughts as well.

@SwarfFire Your screenshot shows MTC quarter frame messages that are sent by your clock source when it is running. They consist of 8 individual messages sent individually each quarter frame.
You can decode them like this:

F1 0x  // x=least significant nibble of frames
F1 1x  // x=most significant nibble of frames
F1 2x  // x=least significant nibble of seconds
F1 3x  // x=most significant nibble of seconds
F1 4x  // x=least significant nibble of minutes
F1 5x  // x=most significant nibble of minutes
F1 6x  // x=least significant nibble of hours
F1 7x  // x=most significant nibble of hours + frame rate

It takes 2 frames to transmit and update the full time.

1 Like
9

Thanks for the info. I’m wondering if it might be something to do with the drivers. i just discovered that my old midex 3.1 no longer passes MTC, though I can still record MIDI with it. I had previously used it to test the timecode issue on my old 32bit windows 7 machine. The fact is both the midex and the RME fireface are both legacy devices and although I still had this problem when I was using windows 7 (the midex was legacy even then), I think I need to be able to rule this out - so I have a modern usb midi interface coming tomorrow hopefully.

However it is strange, as the tape timecode will make Cubase start and end so something must be getting through.

best

Mark

10

The RME manual suggests to have a word clock connection from the JL to the Fireface additionally to the midi connection. I was also surprised reading it but maybe it helps.

11

The bit I read relates to LTC and would be a separate sync input within Cubase’s Timecode source dialogue (ASIO audio device vs MTC). It’s not clear that it is for the purpose I’m using it, there might be some video machines with word clock and use LTC for positional information, but I’m not aware of Tape recorders having it. Certainly my JlC box does not have a word clock out. I saw a new device today for converting LTC to MTC called DoreMIDI, but there is no word clock on that either, so I’m pretty certain that wordclock is not involved. Bear in mind also, as I said before - several broadcast tape machines can sync very well to SMPTE/LTC no word clock required (obviously I wouldn’t split a true stereo track between them, but the timing is good enough for drums and dialogue).

I received a new midi interface today which is USB, but no joy - the problem still remains.

So to recap i have tried 3 different LTC to MTC convertors, 3 different midi interfaces, two different PCs and two different (and quite far apart) versions of Cubase.

Unfortunately, I can’t help feeling that it’s something stupid I am doing or misunderstanding. My next port of call is to see if I can get a demo of another midi sequencer to try this out on.

12

I can’t see anything that you’d be doing wrongly.

That seems llike a good idea.

I really wish I could test your case myself but I do not have the necessary equipment.

13

Bottom line first….my Cubendo systems sync flawlessly to my tape machines as slaves. Every time. All the time.

My tape machines can also slave to Cubendos (although that’s because I also own synchronizers for that).

I’m extremely familiar with the half-inch msr16 (used to sell those as a Tascam dealer in the 80s/90s).

I still own a poor-people-smpte around here somewhere (the jl cooper “pps”)….always thought Cooper had a good sense of humor with that one :slight_smile: I bought it new in the 80s or whatever to sync an unruly piece of equipment to the larger tape machine sync system.

Not familiar with the rme interface, but I did take a look at its back connections for sync.

The fastest way I’d personally force “your exact” system to behave….and I mean immediately with no more drift….

is to (1) buy a used motu dtp (digital timepiece…NOT midi timepiece). Mine cost $1000 new. Now, I have several that I bought used over the years on ebay for $25 each or so.

Then…..

Stripe new smpte to an entire ms16 reel (chnl 16), using the dtp stripe capability (30, non drop). Then…

Msr16 rca out of trk 16 to smpte-in on dtp.

Rme 400 is your main interface on the Cubase Windows machine…right? Ok..then Wordclock in and outs of rme to/from dtp (despite your comments….suspend your disbelief that this is useful/necessary).

Connect EITHER a spdif…OR 5pin midi out from the rme400 to the spdif OR 5pin midi in of the dtp….that is ONLY so that dtp will constantly SEE Cubase’s positional information for keeping EVERYTHING tightly locked.

Set dtp switches to be master-master…rme will slave to wc..ie, dtp will do the heavy locking…smpte as address master…wc to keep the rme at sample accurate to dtp..and the spdif/5pin …dtp will manage all.

Setp Cubase to ext mtc sync (as you’re likely already doing)….verifying in Cubase setup that you’re indeed routing mtc to the rme (which the dtp then sees).

There’s the general gist of how I’d do the main setup.

Of course, then bringing in any number of other peripheral “stuff” if desired…..all funneled through the dtp…..piece of cake.

14

Hi there,

thankyou for your response. I shall keep the digital timepiece in mind, anything that solves this would be great. I should point out though that I should be able to make cubase run from MTC - it’s in the manual, my Nagra T-Audio tape machine can do it without any box so i’m scratching my head as to how my cubase system can’t.

The word clock you are using is from the DTP which I would be slaving my RME to, is not coming from the tape , it’s coming from the LTC to MTC convertor (DTP) in the system. After your post I am now very tempted to get a DTP, but the one thing that’s putting me off is the RME is legacy firewire and firewire operation in windows seemed grudging at best back in the day and I only just got it working in the new windows 11 setup, so I’d be a bit worried I’d be building my house on sand. However , as I say it is very tempting and I shall look at some prices.

So I’m torn between just working around the problem (pushing the midi sync to another device, which may solve my problem completely or show that my firewire isn’t up to it), or actually fixing the issue that Cubase on my system wont sync to the incoming MTC which it says it should.

I tried an experiment this morning which has given me mixed signals tbh. I ran the SMPTE output of the JLC into input 4 of my fireface , which accepts LTC. the specific LTC tab on the fireface pc dialog showed the MTC counter and the input state which said Valid, 30FPS and Full frame. I ran the JLC device for 10mins again compared its front panel time with the fireface LTC dialog time - there was no drift. However the fly in the ointment was when I went to make the LTC the master clock source on the front page of the fireface dialogue, the monitor beneath just kept saying no lock and Cubase did not budge (i’d set ASIO device as the sync master). So why would the SMPTE output from the JLC register correctly in one part of the Fireface dialogue, but still not lock?

So still scratching my head and a little worried that buying something else might still not solve the problem as there is a bigger issue in my system that would still be there.

best

Mark

15

Your MIDI monitor screenshot shows you’re using 30fps drop-frame. Have you tried with other frame rates?

16

Are you sure it’s drop frame, cos that’s a very interesting statement mlib. I have tried different frame rates to no avail in the past. The JLC box says it’s 30 - the other options are 29.97/24/25/30D, cubase is set to 30fps, I’ve never knowingly used drop frame and Cubase usually tells you if there is a mismatch. I will run some more tests just now.

17

My apologies. No it shows 30fps non-drop.
Perhaps it is still worth trying other frame rates.

18

No apologies necessary, i would have loved if you’d been correct - it at least would have been another clue.

19

Lazy Man’s Chrome AI:

Yes, Cubase can sync to MIDI Timecode (MTC) without a dedicated Word Clock connection, allowing it to act as a slave to an external master device (e.g., hardware sequencer, tape machine) for start, stop, and location. However, without a shared digital clock (Word Clock/SPDIF/ADAT) for reference, the audio playback may eventually drift out of sync due to clock discrepancies between devices.

20

Yes, thankyou that’s my understanding too and for this particular application i don’t need the audio to be aligned

best

Mark