Blackmagic DeckLink Audio out of Sync

[FranceBB]

Hi there,
I've been capturing a lot a tapes ever since I began the Remastering and Digitalization project (https://forum.doom9.org/showthread.php?t=181644) and I'm using a BlackMagic DeckLink card, however I'm facing an issue now: every now and then there are dropped fields which generate orphan fields and make the audio go out of sync. I'm capturing as lossless 720x576 25i BT601 directly from the SDI output of the VTR, as suggested by Derek (https://forum.doom9.org/showthread.php?t=181234) but sometimes when there are channel conditions or when the tape isn't playing smoothly, I get those horrible drops that make the audio go out of sync and I have to capture the very same tape again and again 'till it works. Do you have any idea why? It's particularly annoying 'cause I couldn't care less about drops, but at least I expect audio to be in sync... The Sony VTR is auto-referenced, which means that the reference SDI isn't connected to the Blackmagic card, could that be the problem? I thought it was because I was writing directly to one of our 35 TB RAID6 systems and that perhaps there were networking problems, but that wasn't the case 'cause I saved the file locally this time and out of 6 tapes, 1 still had out of sync audio due to dropped fields...

Any idea why?
Sample: https://we.tl/t-0EWmlA8Tgw
You can easily see at TC: 0:00:25.600, frame 640 marks the beginning of the drops. There's the speaker talking and you may not notice it 'cause you can't see his lips, but believe me, audio gets out of sync from there on.


Last but not least, I'm connecting the Reference to the BlackMagic card. I connected it from the BlackMagic Decklink card to the "Ref Video" of the VTR (Blue Cable). Did I do it correctly? And what about the 75 Ohm resistance inside? Is it needed or not? I mean, do I have to keep the switch on or set it to off? I'm sorry but VTR are really not something from my era.


VTR without SDI output only: https://i.imgur.com/oRuLDjT.jpg

VTR with the Reference SDI connected (top left, blue cable): https://i.imgur.com/H3HrlbY.jpg

Blackmagic Decklink card with SDI input (green cable, left) and Reference (blue cable, right): https://i.imgur.com/tOa6TcA.jpg

[Music Fan]

I'm not an expert and never used professional material but I believe a TBC could help (if you find one with SDI).
I noticed that when I put a standalone dvd-recorder between my camera and my capture card, the signal is much better, the video is more stable and there are much less dropped frames. I guess it has a built-in TBC which is quite good.

[FranceBB]

I believe a TBC could help (if you find one with SDI).
I noticed that when I put a standalone dvd-recorder between my camera and my capture card, the signal is much better, the video is more stable and there are much less dropped frames. I guess it has a built-in TBC which is quite good.

I see!
I just digged into old ancient documentations and apparently just like we manually use Bars-note to make sure our waveform monitors are working correctly and that our signal is in Limited TV Range during live events, that very same signal was used as "Reference" by old TBC to make sure levels were matching and that there weren't any dodgy things going on like sudden timecode changes and so on.
I didn't know VTR were so clever.
So now all I need to do is to bring the old SD bars-note reference to the office in which I'm doing the Remastering and connect it to the VTR so that it's gonna be referenced!

[Emulgator]

Just throwing this in:
https://forum.doom9.org/showthread.php?p=1873943#post1873943
https://forum.doom9.org/showthread.php?p=1873980#post1873980

boils down to: The Blackmagic Decklink Shuttle at least has no frame nor field TBC, just line TBC.
If the Decklink has, you may deduct from your specs/PCB.

Your video playback machine SONY DVW-A500P (a 35kg beast)
should have all it needs to get the best out of analog and digital Betacam tapes.

Maybe it is more than one TBCs involved and these interfere?
I had that once in CVBS S-Video land.

[FranceBB]

boils down to: The Blackmagic Decklink Shuttle at least has no frame nor field TBC, just line TBC.
If the Decklink has, you may deduct from your specs/PCB.


I see. Well, the Blackmagic card definitely doesn't have a Time Base Corrector turned on right now or at least, that's what I think. It does have an SDI input and a Reference Input, so until I'll bring the Bars-note SD signal there and connect it to the Reference, the poor card won't have anything to adjust levels...
Same goes for the VTR, it's currently set on "Ref Video" and in fact the stop sign blinks as it's complaining that it doesn't have a reference connected.
I already asked to bring the bars-note signal there, I hope they'll do it soon, but at the moment, I'm capturing without a reference. As to an external TBC, I don't have any. Should I purchase one or am I gonna be fine without it as long as the card and the VTR are referenced ('cause they have an internal one)?


Your video playback machine SONY DVW-A500P (a 35kg beast)
should have all it needs to get the best out of analog and digital Betacam tapes.


It definitely is a beast (this picture is from one that got replaced few days ago due to a faulty electronic card):

https://i.imgur.com/8xcJiJ8.png

[Blue_MiSfit]

I don't believe a TBC is of any use when you're working with an SDI output. I think it's solely useful when capturing analog video, but I could be wrong about that :)

Even if you're capturing from an analog Beta(SP) your VTR will do a great job of digitizing it. The DVW-A500 is a monster :D. I.. A frame sync on the other hand is often useful, but generally not for cases like this, especially with short cable runs. If you have one laying around like an AJA FS1 it might not hurt to throw it inline...

You shouldn't need to do anything more than connect SDI out from the VTR to SDI in on the capture card, plus an RS422 serial port for deck control if you care to use it. That should literally be it.

I'd makes sure you're on the latest stable driver and firmware for your card to start. Also maybe use a different SDI cable just in case. Also, are you sure your local storage can keep up quite reliably with the video format you're capturing? Uncompressed 576i 4:2:2 10 bit is still like 220 Mbps, which isn't much, but may be a challenge if your disks are nearly full or there's other stuff happening on your system :) Try capturing into H.264 or something as a validation.

Do you have a different VTR you can try with as well? As a last resort you _might_ get better results using a second VTR to dub to a fresh Digibeta tape, and then capturing that.

[FranceBB]

Also, are you sure your local storage can keep up quite reliably with the video format you're capturing? Uncompressed 576i 4:2:2 10 bit is still like 220 Mbps, which isn't much, but may be a challenge if your disks are nearly full or there's other stuff happening on your system :) Try capturing into H.264 or something as a validation.

That was the very first thing I took care of when I setup the machine: a proper dedicated 1TB Sata III SSD which is empty and not used by Windows (which has its own SSD connected on another SataIII port). Files are then automatically moved to the MAM by an automation I made with FFASTrans (https://forum.doom9.org/showthread.php?t=176655).

The DVW-A500 is a monster :D.


Yeah, it really is xD


A frame sync on the other hand is often useful, but generally not for cases like this, especially with short cable runs. If you have one laying around like an AJA FS1 it might not hurt to throw it inline...


I gotta ask if we have one somewhere...


You shouldn't need to do anything more than connect SDI out from the VTR to SDI in on the capture card


Yeah, that's what I thought before getting those sync issues...


I'd makes sure you're on the latest stable driver and firmware for your card to start. Also maybe use a different SDI cable just in case.


Yep, the card is updated as it was one of the very first things I did. I can try to see whether a different SDI cable makes any difference, but I don't think it's the one causing the issue...




Try capturing into H.264 or something as a validation.


Well it only happens once in a while, so it's really hard to spot. For instance, it happened several times during the hours-long Agassi vs Ivanisevic tennis game, but only 1 time on 24 episodes of "Juke Box" which last 30 minutes each. I could leave it to record something for several hours and see if it loses the sync, but still...
As to the CPU is generally 12% when capturing in lossless uncompressed yv16 8bit 'cause it isn't something intensive... (it's a very old 4c/4th Xeon with 32 GB of RAM, but it seems to deal with it just fine).
Speaking of which, I'm capturing in 8bit which is 160 Mbit/s; do you think 10bit would really be better given that the content doesn't really have much detail or critical gradients? I mean, banding is literally the last of my problems on those tapes full of spots, scratches, raimbowning, dot crawl, grain and many other things...
I mean, should I change and capture in 10bit instead?



Do you have a different VTR you can try with as well? As a last resort you _might_ get better results using a second VTR to dub to a fresh Digibeta tape, and then capturing that.

Plenty of them (where plenty is a number ranging from 35 to 100 'cause some of them may or may not work anymore), but I would need to see if they're working and carry them all the way up to the third floor where I'm currently located... and it's summer... I don't wanna make my life miserable again by sweating the hell out of my shirt, carrying a 77 pound beast from -2 Building 1 to the 3rd floor of Building 2 xD

[Emulgator]

If I had such a wealth of playback machines:
I would have like 4 set up, swap machines as I find dropouts and have at least 2..3 good captures per source left.
And have fun with the exhaust of those 2kW X-}.
A bit safer than having someone finding faults later in the process and blaming the capture guy.
Are those orphaned fields at the same place everytime ? One machine may find its way better than another.
...And well I may be in the queue for a working machine for the few beta tapes I will have to capture myself.
Around 250 to 450.-€ for a DVW-A500 now.

[Blue_MiSfit]

Yes, you should do everything in 10 bit 4:2:2 these days unless you really don't care :) Minimize those format conversions, keep everything SDI compatible.

Any chance you could try a different computer in case it's some weird bus latency issue? I've seen problems relating to unstable capture on certain motherboards and had to disable bits of integrated crap or just use a different board alltogether. These types of issues can also cause latency / other problems in audio production apps. Believe it or not, this is one area where Apple generally does a very good job.

[FranceBB]

Yes, you should do everything in 10 bit 4:2:2 these days unless you really don't care :) Minimize those format conversions, keep everything SDI compatible.

Alright. I've started capturing in v210. So far so good, all 10 tapes are in sync. For the records, next week they're gonna bring an old TBC that is only collecting dust right now. I hope it will help the situation improve even more.


A bit safer than having someone finding faults later in the process and blaming the capture guy.


Yeah, true, although I'm QC 'em all myself while capturing so I guess they're gonna be fine. On the 25th of August one of my colleagues is gonna join me so that we can speed-up the process, I'm gonna work in the morning, he's gonna work in the evening and we're gonna be active 16h a day and leave the other hours for the VTR to rest xD



Are those orphaned fields at the same place everytime ? One machine may find its way better than another.


Not really, no, different captures produce different results most of the time. "Most of the time" 'cause among the 140 tapes I've already captured only 1 tape produced the same result every single time (three times in a row), so I had to manually edit the final file, but I was able to fix all the others that had issues by recording them twice, so I think that your "best of three" solution would work.


And well I may be in the queue for a working machine for the few beta tapes I will have to capture myself.

Well, all I can say is go for Sony and DO NOT go for Thompson or other brands. This is from a fairly demagnetized Tape from the 80s (Thompson on the left, Sony on the right): https://we.tl/t-ypJVoxYrXH


As you can see, the difference is ENORMOUS. I never expected Sony DVW-A500 to be so good, but apparently it really is, so hats off and a big fat "thank you" to Sony Engineers from the 70s/80s for their incredible electronic work.


Any chance you could try a different computer in case it's some weird bus latency issue? I've seen problems relating to unstable capture on certain motherboards and had to disable bits of integrated crap or just use a different board alltogether. These types of issues can also cause latency / other problems in audio production apps.

I see... sure, I can do that, but right now I'm gonna wait 'till it happens with v210 as well. If it does, then I'm gonna try another workstation. Looking inside the BIOS, I didn't really see anything dodgy activated, though.


EDIT: Oh, look at that, my 1'000 post on Doom9! I'm moved! :')

[Music Fan]

This is from a fairly demagnetized Tape from the 80s (Thompson on the left, Sony on the right): https://we.tl/t-ypJVoxYrXH
Could you post a capture ? Because 658 MB is quite big.
:thanks:

[FranceBB]

Could you post a capture ? Because 658 MB is quite big.
:thanks:

A capture...?
Oh you mean a screenshot! Sure thing!

Thompson VTR on the left, Sony VTR on the right:

https://i.imgur.com/DoT9DVf.png
https://i.imgur.com/qKRIj29.png
https://i.imgur.com/MvB6OW2.png
https://i.imgur.com/yFqnIot.png
https://i.imgur.com/Ela8aEk.png
https://i.imgur.com/KocX943.png
https://i.imgur.com/klAfkaw.png
https://i.imgur.com/Mu8crkC.png
https://i.imgur.com/quLTox7.png
https://i.imgur.com/V7mZsQ8.png



As you can see, the Thompson VTR makes no effort at all on trying to compensate oxide fading occurred on the tape and it just outputs what it reads. Besides, in the parts in which the tape is completely ruined, it doesn't do any kind of smart correction using adjacent fields or anything, it just outputs crap. This includes, stuttering, shacking, channel conditions and some scenes in which there's chroma only but almost no luma at all for whatever reason, while the Sony VTR makes a HUGE effort in trying to compensate the deteriorated tape in the best possible way and if it's really completely spoiled and it can't read anything, it just outputs a duplicated field from the former one.

[Emulgator]

Just downloaded your .mov sample.
Remarkable difference ! Many thanks for that comparison.
That particular Thompson machine seems to be just defective,
beside the tape-introduced artifacts I see AGC/PLL pumping heavily already in sawtooth manner within the colorbar intro.
BTW, I see progressive frames, where has the interlacing gone ?
Ah, I see, 50fps. You uploaded your results post QTGMC, I guess.

[FranceBB]

Ah, I see, 50fps.

Yes, it would have been a mess to make a side-by-side comparison keeping the original fields.


That particular Thompson machine seems to be just defective,
beside the tape-introduced artifacts I see AGC/PLL pumping heavily already in sawtooth manner within the colorbar intro.

Not really. Those colorbars were recorded in the tape, they were not generated by the Thompson itself, so you can imagine that the problem was kinda persistent across the whole tape. As to the VTR being defective, it could be if it wasn't for the fact that I tried a Digital BetaCAM from 2007 on it and it played just fine. I think Thompson engineers only cared about Digital and didn't really pay much attention on the Analog counterpart which was older, so they limited their VTR to just play what was on the analog tape without any kind of adjustments. They probably thought that Digital BetaCAM were "the future" and focused on them only.

[Blue_MiSfit]

Go Sony! That's awesome :)

[Music Fan]

A capture...?
Oh you mean a screenshot!
Yes, I used the same word than in french :o

Sure thing!
Thanks, amazing results !

[FranceBB]

Derek, out of curiosity, we're ready to move from the one with a single SDI (which was effectively a test) to the one with 4 SDI + a Reference.
The only thing I'm wondering, though, is: is there a way to remote all 4 VTRs connected to it? 'cause it looks like we can use 4 SDI but only remote 1 VTR, which isn't really ideal...

[Blue_MiSfit]

Is your goal to capture from 4 VTRs in parallel each with their own deck control?

[FranceBB]

Is your goal to capture from 4 VTRs in parallel each with their own deck control?

Well yeah, 'cause so far I've been using only one as a test.
I was looking at the card you suggested but it seems that I can't have a deck control for them all.
I would really like to avoid to monitor 4 VTRs at the same time as I would rather pick the starting and ending timecodes and let them run automatically...

[Emulgator]

The Sony DVW-A500P user manual mentions the control panel BDKW-515 which can control 2 machines
and the Parallel Interface Kit BDKW-509 (50-pol (Centronics ?) cable).
If equipped, maybe you can use that to hook up 4, at least 2 DVW-A500P machines in parallel...

[StainlessS]

Parallel Interface Kit BDKW-509 (50-pol (Centronics ?) cable
Centronics, was a company name, used as a generic term for parallel printer [EDIT: port or] cable (as is Hoover to mean vacuum cleaner).

There are many types of parallel ports, but the term has become most closely associated with the printer port or Centronics port found on most personal computers from the 1970s through the 2000s.
It was an industry de facto standard for many years, and was finally standardized as IEEE 1284 in the late 1990s, which defined the Enhanced Parallel Port (EEP) and Extended Capability Port (ECP) bi-directional versions.
Today, the parallel port interface is virtually non-existent because of the rise of Universal Serial Bus (USB) devices, along with network printing using Ethernet and Wi-Fi connected printers.
https://en.wikipedia.org/wiki/Parallel_port

https://i.postimg.cc/LY2yVX1z/225px-Parallel-computer-printer-port.jpg (https://postimg.cc/LY2yVX1z)

EDIT: Google finds nothing for "BDKW-509". [ Not even in this thread ??? (well it does find a web page full of letters and numbers "bdkw ;509", but thats all) ]

Google:- https://www.google.co.uk/search?source=hp&ei=419KX9vlBYmyUoCyqPAN&q=%22BDKW-509%22&oq=%22BDKW-509%22&gs_lcp=CgZwc3ktYWIQA1DUEVjUEWCLFmgBcAB4AIABZYgBZZIBAzAuMZgBAKABAqABAaoBB2d3cy13aXqwAQA&sclient=psy-ab&ved=0ahUKEwib4_vsycDrAhUJmRQKHQAZCt4Q4dUDCAg&uact=5
Bing:- https://www.bing.com/search?q=BDKW-509&qs=n&form=QBRE&sp=-1&pq=bdkw-509&sc=0-8&sk=&cvid=2ECBC4D2B1284005B9E5D0A10997D7F0
Yandex:- https://yandex.com/search/?text=%22BDKW-509%22&lr=10393
Swisscows:- https://swisscows.com/web?query=%22BDKW-509%22&region=en-GB
Duckduckgo:- https://duckduckgo.com/?q=%22BDKW-509%22&t=hk&ia=web

[Emulgator]

http://www.classicbroadcast.de/Sony-DVW-A500-Manual.pdf
https://docs.sony.com/release/bkdw515_380054104_om.pdf
https://pro.sony/s3/cms-static-content/operation-manual/3800542041.pdf
Poor Google ;-)
And I assumed wrongly, the 50-pin wasn't a Centronics-50, it was a D-Sub 50 female
(OM 10000+ en, 1995.05.13, page 44/156, paragraph 19 and page 146/156)

[Blue_MiSfit]

You can also install multiple RS422 serial controllers in your PC. Then your capture software will need to be able to run 4 capture sessions in parallel complete with deck control for each, of course!

At this point you're making a video ingest server :D

[FranceBB]

I finally inserted the 9 pin cable for the remote and it's working just fine! :D
I was so happy, but then, today, as I was working on a 1989 BetaCAM, something bad happened... :(

https://i.imgur.com/8u0Ua4H.png
https://i.imgur.com/ga0wBQ3.png
https://i.imgur.com/deNYWWO.png

[Emulgator]

Send it to me and have it cleaned.