Current status of ld-decode / vhs-decode (true "backup" of RF signals)

[Titan_91]

My Dell Optiplex 7010 has a PCI port and I have no stability issues on Linux Mint. I'm using a 3rd gen i7 CPU but any i5 should process with vhs-decode reasonably quickly.

Here are a couple of eBay listing of this model. There are of course others with PCI slots but I know for sure this model has one. Probably one of the last series with PCI. I would install a 2TB HDD, you can get them cheap now.

https://www.ebay.com/itm/Dell-Optiplex-7010-Core-i5-3470-3-2ghz-8GB-DVDRW-NO-HD-MT-Tow...EAAOSw~jJcpqeF

https://www.ebay.com/itm/Dell-Optiplex-7010-Intel-Core-i7-3770-3-40Ghz-Computer-w-4GB-...ry!37771!US!-1

[outhud9]

Would a USB ADC eval board or maybe a USB oscilloscope do the job, if a PCI slot is not available?

They certainly meet the sample rate, bandwidth and resolution requirements, but I don't know if they would support capturing something like 400GB straight to HDD. (eg a 4hour VHS at 100GB/hr) Some of them might only capture a limited amount of samples locally before sending it over USB to the computer?

[fsquared]

You're criticizing everyone who's offering advice in this thread, but you're not offering your own. You seem to know what you're doing when it comes to this, so tell us what you think we should be doing. What equipment, setup, process, etc.?

[fsquared]

I don't even have a computer that has a PCI slot at the moment so can't pull that off. Also, nobody is calling themselves a professional in this thread.

[Titan_91]

Participation in this forum is voluntary. If you have nothing useful to contribute to the project you need to stop trolling and move on.

[Laserdisc]

I think these projects are fantastic and show how a small band of people can produce something that is nothing short of breathtaking.

Going back over a few points from above....

Would a USB ADC eval board or maybe a USB oscilloscope do the job, if a PCI slot is not available?

They certainly meet the sample rate, bandwidth and resolution requirements, but I don't know if they would support capturing something like 400GB straight to HDD. (eg a 4hour VHS at 100GB/hr) Some of them might only capture a limited amount of samples locally before sending it over USB to the computer?

Although most USB scopes have the bandwidth for short samples the basic USB specification either supplies Guaranteed bandwidth or guaranteed delivery of the data... not both. This is why there is an FPGA on the DdD capture board. To act as a buffer for the USB3 and to generate the clocks for sampling.

USB2 does not support the required bandwidth for raw capture of this sample rate of data if you take into account all of the overhead of the USB protocol and other devices on the same bus.
USB has a overhead for all devices (Handshaking and latency) so by using the buffer, the FPGA has the ability to pause and then 'catch up'

I know that people have managed to sample at about 30MSPS streaming to hard disc for an extended period of time using a USB3 PicoScope ( not cheap ) and the API that is available with it but I dont know the details myself. Most of the cheap USB scopes dont have the buffers or speed let alone without dropping samples during streaming to disc.

Using an old capture card although cheap and good for testing and provisional samples is not ideal for many reasons which im not going to go into detail here. (Sample rate, resolution, RF Amplification, filtering, impedance matching etc, All of these were worked out over many iterations for DdD for Laserdisc and although VHS may need a different configuration the process of understanding it is now well documented for all to read and scientifically tested with test and results available.)

When it comes to the whole recovery of the Domesday material if these guys didn't do it then it would have been lost in its original format within the next few years.

Back in 2002 there was an article about Domesday discs unreadable after 15 years https://www.theguardian.com/uk/2002/mar/03/research.elearning
Its now 18 years since that article and nobody else has come close to this level of recovery.

There have been efforts where samples of the images were taken from the composite video and or RGB output of the player and the original discs had the data sectors extracted with the original hardware but this approach missed some of the minor items of information such as the LeadIn/TOC on the disc that contains data to tell the LV-ROM drive where the disc data partitions are.

This project was stared by Happycube on the LDDB forums 5+ years ago and then other people with different experience such as analogue electronics, hardware design, video engineering and PAL Specialists have helped massively. There have also been some people with a massive amount of professional and technical experience pointing out errors and guiding us in the right direction.

For myself its not about the money its about saving the history of early interactive videodiscs. If we waited for the companies to produce anything even remotely like this then the discs would be a good chopping board at best or the hardware/software would be so expensive that the general public could never afford to archive anything properly.

And in fact we have been waiting for the last 18 years!

[fsquared]

Moving back on topic again, I thought I heard someone say before that I need to put it into the FM coaxial port, but I'm reading an older part of this thread and it says something about the luma pin of the S-Video port. Which one of these should I use?

[oln]

Yes , I criticize near all in this thread, because its a strange to see how somebody that don't understand given matter are willing to give advices to others... But this frequently happened on amateur forums.

I don't get why you have to act so negative, you already got booted out of the ld-decode facebook group for your behaviour there.

As for impedance matching, it is an issue worth looking into. LD-decode and by extension vhs-decode is mainly designed using the domesday duplicator hardware for capture. It's designed to work with the output characteristics of a laserdisc player, I have not looked into to what degree a VHS rf out differs impedance-wise, but it does seem they have lower output compared to ld. The cxadc solution is a low-cost alternative, but even without accounting for impedance mismatch we are already able to decode things captured with it and there is MUCH more that can be done on the software side yet. For that matter the cxadc card works for laserdisc decoding, that use a much wider frequency band, though not as well as the Duplicator . The cxadc inputs are presumably designed for 75 ohm video, which may not be entirely optimal for rf. I'm not personally by any means an expert when it comes to electronics so I don't much about what one would have to modify to improve this, but hopefully someone else can be of help here.

[fsquared]

I wish I could get a Domesday Duplicator, but I just can't find anyone who's willing to build it and get it shipped to me, so I'm just going to use the cx card until then (however I'm going to need a PC that has a PCI port so this is going to be a bit more expensive than I thought, but nothing too bad)

[oln]

Yeah it's a bit hard to get one if you don't have the tools to put it together yourself, I was lucky to pick up one that had the components done. Hopefully there will be more people making batches eventually.

A core2 or early core I computer with PCI should hopefully not be too hard to pick find, at least around used ones seem to be quite cheap. Wonder if it would be possible to mod drivers for the pcie version of the conexant chipset.

A USB scope may work, but you may need an USB3 one to have enough bandwidth for realtime capture. Not sure about the picoscope, I have an analog discovery 2 and it can't do continuous data capture at a high enough sample rate.

[Zcooger]

https://www.youtube.com/watch?v=Yq85NdvrvDQ

I assume this is a "regular" VHS capture from the VCR composite output?

O no no, this is current VHS-Decode capability of color-under track recovery with software ( ͡° ͜ʖ ͡°)

So apparently we don't have any computers that support PCI so that might push back this quite a bit. I'm scouring around for any computer that has a PCI port but as a last resort, I'm on PCPartPicker working on a build that has PCI. Would really help if this was a PCIe card, but it's fine, we're dealing with legacy hardware here after all.

I think there two solutions for lack of PCI socket - #1 use an bridge:

[Attachment 51411 - Click to enlarge]

Another solution you mentioned is #2 - buying modern motherboard with PCI socket - currently I am using one exactly for this purpose, we're lucky that Asus made one for such powerful Ryzen platform (it's obvious they are becoming obsolete). Personally I can recommend you Asus Prime X370-A. It has two classic PCIs.

[fsquared]

O no no, this is current VHS-Decode capability of color-under track recovery with software ( ͡° ͜ʖ ͡°)

Very cool! I can't wait until this comes to NTSC!

I think there two solutions for lack of PCI socket - #1 use an bridge:

[Attachment 51411 - Click to enlarge]

That seems much easier. Do you think it would have an impact on performance or add noise though?

[fsquared]

What I'm concerned about here is that I'll buy all of the stuff I need for the cx card only to eventually figure out that it's inferior to another method such as the Domesday. My whole idea for getting involved in the VHS-decode project was to get some high quality caps of my rare tapes (which obviously won't happen for a while but I'll help out until then), so this concerns me a little. Should I just splurge and go with the cx or wait and save up for materials to assemble a Domesday Duplicator?

[Titan_91]

For my needs the CX card works great. The 35.8MSPS sample rate is very close to the Duplicator's 40MSPS rate. But the Duplicator does have the advantage of 10-bit precision. I would think the Duplicator is more important for Laserdisc captures though as the format uses both a higher frequency carrier signal and modulates full composite video. I remember reading about Simon's decision to use a 10-bit ADC with the benefits being measurable but not very significant.

[fsquared]

For my needs the CX card works great. The 35.8MSPS sample rate is very close to the Duplicator's 40MSPS rate. But the Duplicator does have the advantage of 10-bit precision. I would think the Duplicator is more important for Laserdisc captures though as the format uses both a higher frequency carrier signal and modulates full composite video. I remember reading about Simon's decision to use a 10-bit ADC with the benefits being measurable but not very significant.

What I'm concerned about is that Oln just said that cx is made for 75Ω which may not be right for analog video. Should I be concerned about that?

[dellsam34]

If this to be finalized after all the experimentation stages, Would the final product will have 3 duplicators, one for Luma. one for Chroma and one for FM audio? Would a FM demodulator can work for HiFi stereo or FM radio is completely a different beast?

[fsquared]

If this to be finalized after all the experimentation stages, Would the final product will have 3 duplicators, one for Luma. one for Chroma and one for FM audio? Would a FM demodulator can work for HiFi stereo or FM radio is completely a different beast?

Because it's color under, I don't think we would need three duplicators unless you want to do some mess separating the chroma and luma signals, which just alters the original signal more. The luma/chroma would just go through one Duplicator. Audio can be handled by one.

[Titan_91]

What I'm concerned about is that Oln just said that cx is made for 75Ω which may not be right for analog video. Should I be concerned about that?

It’s designed for analog video, what he said was it may not be optimized for RF. I see where he’s coming from and I’m no electronics expert either, but that’s not entirely true. Unmodulated NTSC video contains up to 3.58MHz as the highest frequency component. PAL video contains up to 4.43MHz. The FM video carrier of VHS ranges between 3.4MHz and 4.4MHz. Assuming the same chip can also process PAL video, the frequency response it can accept is spot on for a VHS RF signal. This explains why my captures look as good as they are, and running my signals through a spectrum analyzer shows no signal frequencies appear to be lost (the entire luma and chroma signals are present in my captures). Oln has also captured and demodulated luma and chroma just fine using his CX2388X card, check the first page with the black and white Daffy Duck and color Cheshire Cat screenshots.

In short I wouldn’t worry too much over using a CX2388X card right now. Yes Domesday Duplicator will be better in terms or sample rate and bit depth, but again that solution is designed to handle a Laserdisc FM signal, which is a higher frequency and has wider bandwidth. What I’m saying is it seems a Domesday Duplicator wouldn’t make a tremendous difference in your VHS captures. In terms of how it’s modulated vs. VHS Laserdisc encoded a full composite video signal, while VHS has a cut down version (mainly in terms of chroma at 3.58MHz/4.43MHz for full bandwidth video vs. only 622kHz for VHS).

In terms of impedance matching and signal level, no I didn’t do any impedance matching. However, the level I am getting is quite good and exceeds what Oln has been able to get with his Sony machine, which he successfully decoded from. Also, it appears that riser card only works for low profile PCI cards, as now you have less height to work with.

[Brad]

I think this question was skipped over.

I thought I heard someone say before that I need to put it into the FM coaxial port, but I'm reading an older part of this thread and it says something about the luma pin of the S-Video port. Which one of these should I use?

Neither, IMO. Just use the composite RCA input of the capture card. I think the guys who originated the "luma pin of S-Video" method either thought that the composite input would be negatively filtered in some way (notch, comb) or were simply unaware of the vmux option for choosing the correct input in cxadc.

[Titan_91]

Correct, in the case of my HP branded card it only has a coax input and s-video. We do not use the coax input at all. I just use vmux=2 as described on the CXADC GitHub page and this works. I use an s-video to double RCA breakout cable as pictured in my other post with photos to connect my RCA input cable to the luma input on the adapter.

[fsquared]

Excuse me, but I'm asking questions because I don't know what I'm doing here yet. You call them "dummy questions" because that's exactly what they are: dummy questions because I have no idea what I'm doing yet, which is why I'm asking said "dumb" questions in the first place. You walked into this thread expecting the wrong kind of people. We're not RF professionals here. What did you expect?

[Titan_91]

Fsquared, thanks for the interest. I'll see what PCs I have available.

[radomarkov]

For my needs the CX card works great. The 35.8MSPS sample rate is very close to the Duplicator's 40MSPS rate. But the Duplicator does have the advantage of 10-bit precision. I would think the Duplicator is more important for Laserdisc captures though as the format uses both a higher frequency carrier signal and modulates full composite video. I remember reading about Simon's decision to use a 10-bit ADC with the benefits being measurable but not very significant.

CX also have 10 bit mode.

[happycube]

The 10bit mode makes the cxadc output 16-bit words at half frequency (14.4msps) - that is probably just fine for VHS however, and certainly good enough for composite if you want to do something like make your own TBC/composite decoder with the output. (that might be very nice for PAL if it works ok upscaled to 4fsc/~18msps, since you could with some work make .tbc(.json) files compatible with ld-chroma-decoder and it's 2d/3d transform decoder)

My favorite card (a Hauppauge) doesn't have native composite input, which is where the luma thing comes from. Some CX cards are better than others for LD RF.

The cards are certainly cheap enough to experiment with, so there's no reason not to have one around if you have a box with PCI to use it with.

[Titan_91]

Thanks Chad for the input on that. I was aware of the 10-bit mode but assumed the lower sample rate would be detrimental to the capture accuracy of the signal. If it is not, that's nice to know. We were concerned that 28.6MSPS was adequate but not ideal, hence the experiments with the 35.8MSPS rate and discussion on using the Domesday Duplicator for VHS. Would you say to spend the data on 10-bit 14.4MSPS instead of the other two 8-bit modes in terms of signal to noise ratio? Given your experience with this I would say you're the most qualified person in this thread to answer that.

[Brad]

Assuming vhs-decode can handle the other bit depth & sample rate configurations, I would suggest capturing a selection of test patterns in all modes and doing the comparison ourselves.

For NTSC VHS, I'll do this myself as well, but I need to get my Linux PC set up again.

I don't think Chad has used cxadc since the formation of the Domesday86 project, and he has always been focused on LD, not VHS.

[oln]

Made some progress today, turns out I was looking at the wrong spot, and it was the final chroma band-pass filter that was too narrow. Thought I had checked that earlier but apparently not. Anyhow this makes colour look much nicer:


[Attachment 51451 - Click to enlarge]

Also added a --cxadc3 option for 35 mhz cxadc, and --othertrack to swap which track has phase rotation (try this if you get no color).

[fsquared]

Made some progress today, turns out I was looking at the wrong spot, and it was the final chroma band-pass filter that was too narrow. Thought I had checked that earlier but apparently not. Anyhow this makes colour look much nicer:


[Attachment 51451 - Click to enlarge]

Also added a --cxadc3 option for 35 mhz cxadc, and --othertrack to swap which track has phase rotation (try this if you get no color).

Damn! That's pretty good, especially seeing that you only recently began working on chroma! Great job!

[fsquared]

So I have a few creased tapes that my deck refuses to play and I searched google and found this: https://itstillworks.com/how-to-correct-wrinkled-vhs-tapes-10141.html

Does this sound like a reliable method for fixing creased tapes or should I use something else? I really do not want to resort to splicing but I will if I have to.

[dellsam34]

Looking good OLN, I see a lot of dot crawl though in a form of checkerboard, Isn't that a problem of capturing from the composite output? It shouldn't exist in the RF stage, Could it?