Sony PCM-7030/50 Synchronization Issues Part II of II

Sony PCM-7030/50 Synchronization Issues
(Part II of II)

by
Evan T. Chen
July 21, 1995

Nowadays, saying that a DAT is striped with 29.97 timecode or 30 timecode is just not enough information. The big issue is what type of external signal the DAT was referenced to and the settings on the DAT machine at the time of recording. This article describes an experiment carried out by the Creative Director at New York's Soundtrack and myself in an attempt to demystify the Sony PCM-7030/50.

Nomenclature issue: "timecode" consists of timecode "type" and timecode "rate." There are only four timecode types: 30 non-drop frame (ndf), 30 drop-frame (df), 25, and 24. There exists only one drop-frame timecode type, 30 df; hence, when someone speaks of "drop-frame" timecode, 30 df timecode type is implied. Similarly, does "non-drop frame" timecode imply 30 ndf? Technically no, since "25" and "24" are also "non-drop frame" timecodes, but who's that picky? There are only four timecode rates (or speeds): 30 frames/second (fps), NTSC color video which is approximately 29.97 fps, 25 fps, and 24 fps. This is the syntax adopted by Timeline, and for a more introductory explanation that doesn't quite live up to its name see "SMPTE Made Simple" in the appendix of Timeline's MicroLynx manual.

One method of completely specifying timecode is to specify first the timecode type and then the timecode rate (or vise versa and that's exactly the problem with this method). For example 30/29.97 means 30 frame timecode type running at NTSC color video. Some engineers may claim that 29.97 is the fifth timecode type, and that is partially true. Before timecode DAT players 29.97 timecode meant 30 timecode type running at 29.97 fps. However, with these newer machines and relentless software upgrades, one may now specify 29.97 timecode "type" running at 30fps. Hence, 29.97 timecode type is identical to 30 timecode type in all respects except for one -- its name. For instance, 30/30 timecode is functionally the same as 29.97/30 timecode except for the label. In this article "30 ndf timecode" refers to 30 ndf timecode type running at real-world 30 fps, whereas "29.97 ndf timecode" refers to 30 ndf timecode type running at real-world NTSC color video, which is approximately 29.97 fps. Any reference to the rate "29.97" means the rate of NTSC color video.

In Part I of this article (April 17th) I summarized my findings on successfully pulling up and down audio by 0.1% by using 59.94Hz and 60.00Hz pilot tones. (If one looks in the 7030/7050 Quick Reference handbook, this particular case is NOT covered.) The same results can be achieved by using an NTSC monochrome video signal running at 30Hz, but since Soundtrack has no means of accurately synthesizing such a signal this is out of the question. (A Video Sync Generator option card for the MicroLynx or an NTSC monochrome card for the Grass Valley 8500 would possibly remedy this.) Recently an old question concerning 0.1% pull-up/down re-surfaced: is the timecode rate pulled up/down as well as the audio? Remember, the April 17th experiment in Part I only tested audio. A second test was devised.

An analog sine wave was sent from the built-in oscillator on the Sony Mixing Console in Studio D into the left "Analog In" on the Sony PCM-7030 and measured at 10006Hz. Approximately 3 minutes of this test tone along with 29.97 ndf timecode was recorded onto the DAT referenced to housesync (NTSC color blackburst). This same tone was then recorded at -3VU, 7.5ips onto the left track of a 29.97 ndf pre-striped 1/4" tape resolved to housesync running off of a CTTC Studer A810 controlled by a Lynx. By measuring the change in frequency of the test tone playing back from the 1/4" as it chased to a timecode-only master, in this instance the 7030 playing our test DAT, and as the various 7030 parameters were altered, e.g. rEF-tcF, Sync nrr, etc., the effect on the timecode rate could be determined. Hence, a second Lynx was configured to run as the code-only master off of the 7030's timecode, the A810's Lynx was placed into chase mode, and the experiment proceeded. Furthermore, by measuring the change in frequency of the test tone playing back from the 7030 (as opposed to the Studer), again as its various parameters were changed, the effect on the rate of the audio could also be determined. This was also carried out. The results are as follows:

caseSync SourcerEF-tcFVari-SpeedSync nrrCommentsTimecode
"frequency"
Audio
"frequency"
1Video29.97 ndf0on-10006 Hz10006 Hz
2Video29.97 ndf0offwide10006 Hz10006 Hz
3Video29.97 ndf+0.2%on-10026 Hz10026 Hz
4Video29.97 ndf+0.2%offwide10026 Hz10026 Hz
5Video30 ndf0onVideo (blinks)10016 Hz10006 Hz
6Video30 ndf0offwide, -0.1%10006 Hz9996 Hz
7Video30 ndf+0.2%onVideo (blinks)ErrorError
8Video30 ndf+0.2%offwide10026 Hz10016 Hz
9Int29.97 ndf0on-10006 Hz10006 Hz
10Int29.97 ndf0offwide10006 Hz10006 Hz
11Int29.97 ndf+0.1%on-10016 Hz10016 Hz
12Int29.97 ndf+0.1%offwide10016 Hz10016 Hz
13Int30 ndf0on-10016 Hz10006 Hz
14Int30 ndf0offwide10016 Hz10006 Hz
15Int30 ndf+0.1%on-10026 Hz10016 Hz
16Int30 ndf+0.1%offwide10026 Hz10016 Hz

In terms of percentage change this is what we have:

caseSync SourcerEF-tcFVari-SpeedSync nrrCommentsTimecode
%change
Audio
%change
1Video29.97 ndf0on-0.00.0
2Video29.97 ndf0offwide0.00.0
3Video29.97 ndf+0.2%on-+0.2+0.2
4Video29.97 ndf+0.2%offwide+0.2+0.2
5Video30 ndf0onVideo (blinks)+0.10.0
6Video30 ndf0offwide, -0.1%0.0-0.1
7Video30 ndf+0.2%onVideo (blinks)ErrorError
8Video30 ndf+0.2%offwide+0.2+0.1
9Int29.97 ndf0on-0.00.0
10Int29.97 ndf0offwide0.00.0
11Int29.97 ndf+0.1%on-+0.1+0.1
12Int29.97 ndf+0.1%offwide+0.1+0.1
13Int30 ndf0on-+0.10.0
14Int30 ndf0offwide+0.10.0
15Int30 ndf+0.1%on-+0.2+0.1
16Int30 ndf+0.1%offwide+0.2+0.1

Remember that our test tone was a 10006 Hz sine wave recorded with 29.97 ndf timecode onto a DAT referenced to NTSC blackburst. We also assumed that the 7030 had no real-time pitch-shifting DSP capabilities that could alter the audio frequency without changing the duration.

With the DAT's sync source set to internal oscillator, the Sync nrr parameter has no effect on our outcome (see cases 9 - 16). For these cases Vari-Speed effects both audio playback rate and timecode rate, as in case 11. rEF-tcF only effects timecode rate, independent of the audio playback speed. For example compare cases 9 and 13. Changing rEF-tcF from 29.97 ndf to 30 ndf raises the timecode rate by 0.1%; however, the audio playback rate remains the same. In case 15 setting the rEF-tcF to 30 ndf first raises the timecode rate 0.1%, and then Vari-Speeding the DAT +0.1% increases the timecode rate an additional 0.1%.

It is my opinion that had we recorded the 10k test tone with 29.97 ndf timecode referenced to its internal oscillator, our results would have been similar. But also remember that we never tested a DAT recorded with 30 ndf timecode.

With the DAT's sync source set to Video and referencing the 7030 to NTSC color blackburst, setting the rEF-tcF to 29.97 ndf (our nominal setting) produces expected results, i.e. 29.97 ndf timecode and the same-pitched audio. Notice while referenced to Video, Vari-Speed only works in divisions of 0.2%. Setting the rEF-tcF to 30 ndf produces more interesting results. With Sync nrr "on," the 7030 will only lock to an external video signal if its speed is within a certain tolerance specified by the rEF-tcF setting (I believe it's 100ppm but don't quote me). In cases 5 and 7 the 7030 sees a 29.97 external video signal which it compares to its highly accurate internal oscillator and confirms that it is beyond its rEF-tcF setting of 30 ndf, thus the blinking "Video" warning indicating that the 7030 is not using external video sync, but instead reverting to its internal crystal. With Sync nrr "off," the limits within which the 7030 locks to an external video source has widened (I think it's either 0.2% or 12.5%, but see the 7030 manual for exact specs). In cases 6 and 8 the external video signal falls within these specifications resulting in the solid non-blinking "Video" indication. But by brute-forcing the 7030 to sync a 30 ndf rEF-tcF to an external 29.97 sync source, we've pulled down the 30ndf timecode to 29.97 ndf and the audio down by the same ratio as well, justifying the automatic "-0.1%" indicator on the display. Hence this is one successful way of pulling down the audio of a timecoded DAT striped with 29.97 ndf timecode while NOT changing the rate of timecode. This particular case is also a good example of how misleading the 7030 can be. It displays "30 ndf" but outputs 29.97 ndf. Also, in this case rEF-tcF DOES alter audio rate but NOT timecode rate, contrary to the other cases we tested. To complicate matters even more, the pull-down indicator "-0.1%" may not always be shown on the screen, depending on what display mode the 7030 is in.

In the common interest of this article, one related issue is that of sampling rate. If the 7030/50 is referenced to NTSC color video and set at Fs=44.1kHz, is it sampling at 44,100Hz or 44,056? And what about 44,144Hz and the various 48kHz sampling rates? Once again Timeline has developed a clever way of dealing with this issue by introducing Sampling Rate Ratio (SRR). See the MicroLynx manual for a rundown. For example the Avid AudioVision samples at a true real-world 44,100 samples/second referenced to our black burst which runs at 29.97.

[Appendum: Since this article, I've written another article, to be published in TV Technology in 1997, entitled "0.1% Caveats" and it expands further on this confusing timecode/sampling rate issue. Also, Brainstorm Electronics, Inc. has developed the SR-15+ Time Code Distripalyzer which actually measures timecode rate. It would have been a valuable asset in this experiment. Finally, case 6 does allow one to pull-down the audio by 0.1% while maintaining the original timecode rate. Unfortunately, switching rEF-tcF also changes the start time of the timecode on the DAT (except in one situation which very rarely occurs, if ever). This is covered in "0.1% Caveats." Nowadays, while posting for film in the video domain, the Sony PCM-7030, 7050, Tascam DA-60, and a few other timecode DAT recorders are commonly used in their 30ndf pull-down modes, referenced to video, while recording. This allows the audio to be easily transformed back into the film domain, running at real-world film speed. - Evan 4/12/96]

[Note: The successor to the now discontinued Sony PCM-7030 and 7050 is the Sony PCM-7040 Timecode DAT Recorder, and although these tests have not been verified on this newer deck, chances are the results would remain the same. - Evan 4/22/00].

Click here for Part I of II.

Copyright © 2010 by Evan T. Chen. All rights reserved.