I'm honestly surprised at how well it survived the first couple generations, being that you weren't using any TBC in between the machines. Obviously not the _correct_ way to dub these tapes, but nevertheless a cool experiment!
I had always heard that the F1 was designed to work with Beta rather than VHS recorders due to the higher bandwidth and quality of the original Beta machines and tapes. At the CBC we used the units in this manner until they were replaced by DAT recorders and later digital workstations (Pro-Tools, Pyramix, Soundscape). We didn't however do multiple transfers from tape to tape unless the signals were decoded and re-encoded in the process.
Sony at least made the unit look and feel similar to their beta machines. But I seem to recall that it was u-matic that was used as the videocorder when making a master for CD pressing, am I correct? And you could use NTSC or PAL, but one of the resulted in 44100 Hz and the other one slightly less. Which one was used for CD? NTSC or PAL?
I could be wrong, but I think NTSC versions of the PCM-F1 used a sample rate of 44.056kHz, and PAL used 44.100kHz. The system commonly used for mastering CDs was a slightly different format from that used by the PCM-F1, and was NTSC-only, and operated at the CD sampling rate of 44.100kHz, typically recorded on U-Matic. I'm sure someone with more knowledge will be along to correct me very soon...
@@timf-tinkering OK so the reason that NTSC PCM-F1 processors run at 44.056kHz is because of the horizonal frequency of NTSC color video, which is reduced by 0.1% for color video. This is why NTSC video is 59.94 Hz / 29.97 FPS instead of 60/30. Since Betamax is a color recording format and never was optimized for black and white recording the frequency had to be slightly reduced. However, the PAL model runs at exactly 50Hz / 25FPS. The professional U-Matic PCM processors and recorders such as the PCM-1600 series used black and white NTSC video, so the frequency was exactly 60Hz / 30FPS which allowed the sample rate to be the standard 44.1kHz)
I think the recording method represents kind of a worst case scenario. You probably would have gotten better results if you had played back the tape on the deck that recorded it rather than swapping tapes between decks...Or at least used 2 identical decks. Decks of different brands had slightly different tracking and calibration. Playing a tape in a different deck the tracking differences degrade playback slightly over using the original record deck to playback...Playing the tape in a different deck from the record deck each time in a multi-generational copy experiment amplifies the problem with each generation. Also those decks look like bottom end consumer decks. Something like a pair of high-end consumer/low end prosumer S-VHS deck with an internal TBC and or possibly digital video noise reduction even in VHS recording mode likely would have done better.
I completely agree, this is probably close to a "worst case scenario" - but that was kind of one of the aims of the experiment. I had (wrongly) anticipated several generations of copy before the sound started to deteriorate, so didn't want to start with high-end gear and high-quality tapes as I feared the video would end up being 4 hours long :-) I aim to repeat the experiment with a 14-bit recording at some point, but will deliberately use exactly the same setup so that a direct comparison is possible. Also I wouldn't want to put all that wear on my nice VHS decks :-)
me too. When there was no input, the right side shook by itself. This may be a problem with the 10V power supply of the LINE input amplifier circuit next to the microphone amplifier circuit.👍
Brilliant! Thank you for making this. Just to be clear, I was indeed interested in hearing how generation loss affected the tape outside of the "Copy Out" feature. What's the fun of bit-perfect digital anyway? Lots of interesting artifacts in that second copy pass! Almost as if some white noise had been sidechained to the music. Maybe someone with more technical knowledge can chime in on how specifically the audio is degrading between generations.
Indeed the degradation of the audio was slower than I expected. Generally speaking, pcm with bit errors tends to crackle until it turns into white noise (as you introduce more and more errors). What I heard in the early generations sounds more like quantization noise, which would imply that the bit errors tend to be in the less significant bits. In human terms, it's as though the machine tends to make errors in the ones and tens place, but never in the thousands or ten thousands place. There could be a few reasons for this, and it could even be an intentional feature of the format, but without more information I can only speculate on the mechanism behind this. One possibility is that the designers of the format reserved more of the video bandwidth for the high order bits (for example making them a little longer) and as the vcr corrupts the signal everywhere, the ability to read the "fine print" of the low order bits decays first. Another possibility is that the high order bits come at the start of the line (immediately after the horizontal sync), meaning that there's very little ambiguity in their position. Then as the code word continues, the decoder has to count how many bits it's seen and as ambiguity accumulates it's more likely to introduce mistakes at the end of the number. This would imply that the data are layed out in a big-endian format. I would be very interested to know the answer to this, but without access to this equipment, I would need someone to send me uncompressed video and audio files of the various generations to analyze. It shouldn't be hard to figure out what's going on if I had access to the raw data.
@@timf-tinkering yeah, the other experiment should be boring - perfect copy number 1, perfect copy number 2, ... perfect copy number 4734, perfect copy number 4735, ... Of course sometimes my expectations are violated and in that case it might be super interesting to see what flaws make it into the recording. Anyway, thanks for the really interesting and provocative experiment. I want to see part two with some analysis of what went wrong on those early generations.
Kinda wonder how far you would've gotten with a higher bandwidth format like S-VHS or Hi-8 or even just Video8. It's already been mentioned it was probably made for Beta.
This would be challenging for me to set up, as while I have quite a few SVHS decks, I have very few SVHS tapes, and the inverse problem with Hi8 - loads of tapes, but no decks (that can record from an external video source). Would love to see someone else take on this experiment.
Direct VCR to VCR. The "Copy Out" connection on the F1 didn't do what I expected - it is just a duplicate output from the ADC, so the Digital out and Copy out carry the same signal. It doesn't output a re-generated copy of the digital input, so cannot be used to make a "perfect" copy.
@@timf-tinkering Sony prescribes to use the copy function on the F1. Direct tape to tape copies get progressively worse as multiple generations are copied, as you have proven. Using the copy function on the F1 activates an error correction and restores the timing of the video sync signal. This way, in principle, unlimited copying is possible. When using the line inputs carrying record out and copy out connections indeed the same signal when the copy switch is set to off.
@@1lennetje Ah yes, of course you are right! I had not set the "Copy" switch on the front panel to "On." With that switch set correctly, the Copy Out socket does indeed carry a re-generated digital (video) signal, so it would in theory be possible to make an infinite number of perfect copies through the F1.
@@timf-tinkering This doesn't work of course. You are not copying the digital signal, but a Pal movie to another Pal movie. It's just like copying videocassettes.
I have been buying boxes full of VHS tapes from eBay, in order to recover the Teletext data off them. The recovery can be patchy from an original recording, but even a first generation copy (both SP) loses quite a lot of information. SP to LP, and 2nd-3rd generation copies are practically unusable for teletext recovery - and I suppose this is why!
This is really interesting. I remember “back in the day” I used to use the teletext on VHS recordings to work out when the recording was made, and from what channel. The pages were almost readable in many cases, but with a lot of errors. Very interesting that this can now be captured. SVHS tapes captured teletext data almost perfectly.
Teletext requires higher bandwidth than regular VHS can manage, so the results will be scetchy even with a first generation tape. But teletext survives SVHS quite well because that format has full broadcast bandwidth. PCM audio is a different cup of tea because its bandwidth requirement was designed with the limitations of early 80s home video formats in mind. So it is perfect at least in the first generation. Of course like any analogue copy it's bound to degrade after a few generations. The "proper" way to achieve lossless generational copies would be to have two VCRs each with its own PCM decoder, and connect the two digitally (if they come with digital inputs and outputs which I don't know) , thus "regenerating" the digital data in each step. Conceptually that would be the same procedure as when copying digitally between two DAT machines, or copying files on a computer, both of which can done with unlimited generations.
@@skinnypotato4452 Just play a VHS recording from an analogue era broadcast, connect it to the AV input of a TV with teletext decoder, and activate teletext on the remote. As I said VHS doesn't have enough bandwidth for this to work properly, but most of the time you can get spurious pages with lots of errors. Sometimes waiting a few refresh cycles gives the TV the chance to assemble the page almost error free, but it depends on the individual decoder. As you said, teletext is still possible with the European DVB standard, which is backwards compatible with analogue teletext. Most digital set top boxes for any of the DVB standards come with a feature called "teletext insertion". That means their analogue output contains teletext data in the old format which can be decoded by any vintage TV set (provided it has teletext). Teletext data can't be recorded by digital recorders like say a DVD recorder or a computer TV card, those recorders completely omit the sync pulses of the analogue signal, where the teletext information sits.
In the early days of DVB transmissions in the UK, many digital channels carried traditional teletext (DVB-TXT), although I remember the BBC channels just carried a single page directing viewers to use the new "Digital Teletext" service instead. So I can concur, it definitely is possible, and was used for a while. PVRs that essentially save the video transport stream to disk (rather than "record" the video) may capture the DVB-TXT as well, but I don't think any UK channels carry traditional teletext any more.
I wonder if a video editing device which can fix time based errors and a device which can increase the sharpness of the video signal to some degree could be used to copy the digtal audio recording. Maybe you could copy it a few more times before the PCM-signal gets noisy. I use such a device to digitize VHS tapes because when the H- and V-sync passes out because a dust particle on one of the video head interrupts the video information on the VHS tape, the tape is bent because it got entangled in the mechanism of a VHS-recorder at some point, the tape was already copied from another VHS tape and has time based errors or there is noise from bad reception when the tape was recorded via analogue TV (e.g. via analogue sattelite TV when there was a thunderstorm blocking the signal from the satellite) the video grabber would skip frames and have problems re-synchronizing the video recording. The MPEG2-recording would take up to 1 or 2 seconds in "slow-motion" and the audio is out of sync before it will continue playing normally. It's the Panasonic Digital FX Generator (Model WJ-AVE3/C) as a Time Base Corrector in my case and a device to readjust sharpness, contrast and brightness of the video signal. I managed to digitize a VHS recording with family recordings from the early 90s on it. It was obviously copied from a VHS-C tape onto the VHS-tape and at some point there's a lot of noise on the tape for about 20 minutes, probably because there was a dust particle on the video heads inside the VHS-C camcorder from which the tape was played. I first had problems recording it via the video grabber only but when I used it with the video editing device from Panasonic I was able to record it without problems from beginning to the end (except the noise on the recording, but the H- and V-sync is fixed).
I have two questions: 1) What resolution (RES) was selected? 14-bit mode provides much more error-correction data than 16-bit mode. 2) Can you capture those tapes into any video files? I want to try those on a software decoder.
1) This was a 16-bit recording. Now I'm tempted to repeat the experiment in 14-bit mode! 2) The best I could do would be a capture to consumer DV format, or from a cheapo eBay USB video capture unit.
@@timf-tinkering this format uses lines in the inactive region and also PCM-F1 can recover up to 16 consecutive damaged lines in a 128 line span, while STC-007 can recover twice as much - up to 32 lines. Combined that means that STC-007 most certainly can be fully recovered even when using cheapo USB capture that does not capture inactive region. For PCM-F1 special capture hardware may be required to perform full recovery. But even USB captured one will be interesting to play with. If you can attach link for video captures in your next video, that whould be great. Even short (>10 sec) fragments whould be fine (PCM captures demand high bitrates, thus big file sizes).
@@lawrence.porter It's the black and white bars and dots you see in the video. The Sony F1 PCM Decoder reads this black and white signal, decodes the line code and sends the payload to DAC.
I wouldn't say it's impossible to get a perfect copy without disabling the DOC. Error correction may work better with it disabled, but if the video recording is of sufficient quality for the PCM-F1 to be able to recover the original PCM data without errors, any copies made through the F1 would be "perfect." If there is a dropout on the tape, then the data that should have been there is lost. The only difference a DOC would make is that the missing data would be replaced with something else rather than noise. In both cases, the F1 would just see this as corruption, and attempt to apply error correction as normal, so I can't see disabling the DOC would make a significant difference.
@@timf-tinkering I remember I used to make a small cube from tape to put behind the tape as you had done so that it wouldn't stretch out and fail in the middle of a recording.
It's just a cheap no-name set. The brand is "Currys Essentials" (Currys is a large chain electronics retailer in the UK) but I don't know who the actual manufacturer is. I like it because it's small, very light and easy to move around, has all the different input types you could possibly need, and supports NTSC (and PAL obviously).
@@timf-tinkering I've heard of curry's essentials, I believe the company who makes the tv is vestel or a random chinese, but curry's owns the brand logik as well
