Developing a analogue capture card with in built tbc

[johnmeyer]

As I read back through these posts, the problem I have is that it is not clear what problem the VHS-Decode project is trying to solve.

You obviously still need a VCR, so it doesn't solve the problem of VCR shortages, and VCRs which are not fully functioning.

I still need hardware to get the video/audio into my computer, so it doesn't solve the problem of obtaining quality capture hardware, and making sure I get something that works well.

The one, somewhat vague, promise is that I can get a higher-quality, better-looking video. The theory is very intriguing, but so far I haven't seen any solid test that shows this to be true.

Don't get me wrong, I am totally open to the idea, and do not in any way dismiss it out of hand.

The one other thing that is also a little off-putting is the workflow. Having to hot-wire my VCR is a little worrisome, even for me (i.e., someone quite comfortable with soldering), but then having to do all the extra steps to convert the "capture" into usable video (and audio which, apparently, requires some extra steps) makes it sound like it could create real workflow problems when having to capture 50 tapes for a project.

[Cornucopia]

https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://www.scienced...b4DOzkb_lR66PB

That was in 1999, so I expect improvements have been made since then.

With non-contact scanning as a possibility, I could envision a much simpler, more streamlined tape transport mechanism that just played the tapes linearly and scanned the entire tape linearly, then using a high enough bitdepth multiple oversampling of the tape, the rest should be able to be done in software. Track alignment & V + A + Control/TC separation, track segment stitching, the rf & fm decode, etc.


Scott

In fact, the transport & A/D could be one device, with perhaps a Thunderbolt 5 interface.

[dellsam34]

... unmolested SD will be the challenge, especially in NTSC. The SD is generally force-processed, be it deinterlace, scaling, or otherwise. Tapping into chips is also suspect, as some lock/encrypt code.

I was referring to some DVD recorders from the 2000-2010 era where the digital output from HDMI is conform to the standard, Not referring to any modern analog to HDMI scaler.

[dellsam34]

https://www.google.com/url?sa=t&source=web&rct=j&opi=89978449&url=https://www.scienced...b4DOzkb_lR66PB

That was in 1999, so I expect improvements have been made since then.

With non-contact scanning as a possibility, I could envision a much simpler, more streamlined tape transport mechanism that just played the tapes linearly and scanned the entire tape linearly, then using a high enough bitdepth multiple oversampling of the tape, the rest should be able to be done in software. Track alignment & V + A + Control/TC separation, track segment stitching, the rf & fm decode, etc.


Scott

In fact, the transport & A/D could be one device, with perhaps a Thunderbolt 5 interface.

Chances are nothing happened to that idea since 1999, There is nothing wrong with scanning the tape with a spinning drum, there is nothing to be gained doing otherwise, As a matter of the fact scanning the video tracks the same way they were laid down during recording may yield better results than any other scanning technology, There is no need to re-invent the wheel.

[mountaincabbage]

Perhaps a more realistic approach than these half drafted software solutions is to get few DVD recorders with S-Video input and HDMI lossless SD output and use them as capture devices, Multi HDMI recording is easier to achieve nowadays, and being digital it should be well supported by major software and hardware companies. If you have the knowledge you can be creative and see if you can remove the DVD drives and other unnecessary boards not needed for the purpose.

This could work in terms of optimising workflow and scalability, but we'd still be relying on high end VCRs which are dwindling. The goal here is to learn in depth different methods of analogue capture, and see if improvement can be made.

You obviously still need a VCR, so it doesn't solve the problem of VCR shortages, and VCRs which are not fully functioning.

If it can produce the same quality as a high end VCR, on a lower end VCR, then it helps by increasing the pool of VCRs available for transfer. The immediate issue is that the top end decks are now hard to come by. There is still plenty of normal VCRs out there.

I'll dig into the vhs-decode project and see if I can get some decent comparisons with a typical analogue capture setup.

[dellsam34]

This could work in terms of optimising workflow and scalability, but we'd still be relying on high end VCRs which are dwindling. The goal here is to learn in depth different methods of analogue capture, and see if improvement can be made.

You would be wrong, The sole purpose of using such DVD recorders is to take advantage of their TBC capability because low end VCRs lack it. The DVD recorders I referred to have HDMI out so no need to an analog to digital capture card, you just need a HDMI acquisition card or device.

[lollo]

The DVD recorders I referred to have HDMI out so no need to an analog to digital capture card, you just need a HDMI acquisition card or device.

Nice! You start to think about the "German" approach has alternative (with the drawbacks we all know, I personally do not use it).

Just as curiosity, what made changing your mind about the HDMI output from such machines not being strictly compliant with Analog SD specifications ?

[dellsam34]

I've never changed my mind about proper HDMI method, maybe it was a misunderstanding on my end about that German website. On the TBC side I still think a DVD recorder does not replace a proper VCR with line TBC and a capture device built in frame sync. But for the OP's purpose of running a business it might be just ok as long as he properly process the lossless files.

[lollo]

On the TBC side I still think a DVD recorder does not replace a proper VCR with line TBC and a capture device built in frame sync

Sure. (A specific device) Just replaces a DVD-Recorder used as external TBC in a workflow (example ES10 or ES15) and the chained capture card

[pandy]

NP - not sure what kind of price is on Amazon but probably cheapest Pluto is on ADI site (195$).

Wow, that is a LOT cheaper than the $495 on Amazon for a "kit."

Pandy, thanks for the SDR and ADALM2000 links. I just recently bought a cheap Hantek digital oscilloscope, and while it has a few functions found in the ADALM, it sure doesn't have them all, and it is not directly programmable. I have used the Hantek as a frequency counter and, through its very crude FFT, as a spectrum analyzer, but both are extremely marginal and barely usable. Something like the ADALM might be very useful. Thanks!

ADALM is suggestion from Sharc (btw it is a good candidate for VHS decode - 12 bit, 100MHz ADC is more than enough). I had this kind of suggestion earlier in thread - some DAQ supporting at least two channels (preferably 4 simultaneous sampling so Y, C and L, P - 12 bit or more, by adding large amount of noise to 12 bit over usable audio signal bandwidth and later decimating easily 90dB++ SNR can be achieved)

[pandy]

if you sample signal with sufficient sample rate and sufficient accuracy then you can later do whatever you want

If, if, if. Archivists, orgs, studios, and even "normal people" deal in known knowns. Not "let's spend time and money on this now, and then hope software works magic later".

Nope this is math i.e. science - something exactly opposite to what are trying to do - we know haw VHS signal is stored, there is no magic there - if you have 3..4MHz signal bandwidth then sampling it 4..8 times more (oversampling) is more than enough (but you can oversample more - 16 times if you wish - we are still on safe, sane <100MHz sample rate). 10..12 bit provide around 50..60dB SNR - way more than you need to demodulate FM - one of many examples https://www.nsf-shrec.org/sites/default/files/2024-03/On-FM-demodulators-in-software-d...sing-fpgas.pdf

There is lot of science there despite some nuts claiming best

Yes, pseudo-science is always "crazy + science". If it were a sundae, science scoop on bottom, nuts of top.

Well... you sounds like nutty audiophile...

I'm also quite old so you can call me conservative one but still i observe how truly amazing progress is in electronics

I agree, but we didn't get Star Trek. We largely got inventions and progress that were not planned. So to say "this will be big someday" is naive. It may be small, or nothing/vaporware, or not even needed later. That's the natural course of science, lots of "eureka!" moments, not "next, we will do this".

We not talking on WARP drive but we talk about real life solutions - nowadays you have ADC with SFDR 80dB+ and bandwidth around GHz - this is science (read once tape, demodulate in HW/SW to avoid old analogue electronics).

This is not emulation

It's also not an aardvark. But it was an example of tech not always being able to recreate the past. As another example, the Antikythera device took 100 years to be recreated, and people are still not 100% in agreement on how it works.

Pure sophism - we talking about FPGA - it is not emulation, this is HW using software description to perform data processing - in theory you can build discrete circuitry (for example using SRAM) to act like FPGA, from electronic perspective it is (FPGA) just smaller and more flexible approach between hard logic (like ASIC or VLSI) and pure software - you can reprogram HW (FPGA) multiple times but it is not emulate anything - it perform exactly same algorithm as real digital gates and registers made from transistors.

[pandy]

As I read back through these posts, the problem I have is that it is not clear what problem the VHS-Decode project is trying to solve.

From my perspective it is to replace old (deteriorating) analogue circuitry - VCR signal processing is quite complex so naturally analogue processing is affected by series of compromises - even latest generation of VCR's using very complex, partially hybrid (like discrete time analogue processing but still not digital signal processing where signal is sampled in time and voltage axis) IC's are affected by technological limitations - analogue memory (like CCD/BBD) allow to implement some algorithms in more efficient way but still are unable to provide sufficient signal to noise ratio - let say TBC can be made as analog delay line but probably you will not get SNR higher than 40dB, also you will be limited by "memory size" i.e. how many samples are stored and can be later used. In software this is not problem - you can store many frames so perform statistical processing over huge amount of samples and perform highly sophisticated processing not possible in typical VCR.
So use mediocre mechanics to deliver highest possible quality limited only by VCR system itself.

[johnmeyer]

ADALM is suggestion from Sharc ...

Oops! Sorry Sharc. I edited that post so it reads correctly now.

[dellsam34]

The problem with these high frequency sampling software schemes is that they are not able to make it to a functioning, hassle free, easy, and ready to use product. we can fantasize about how great their capabilities are but cannot be used out of the box to replace a ready made capture card or device. I also read a lot of books about how to make a car that runs on water but does that mean I will include that option on my shopping list when I go buy a car next time?

[pandy]

The problem with these high frequency sampling software schemes is that they are not able to make it to a functioning, hassle free, easy, and ready to use product. we can fantasize about how great their capabilities are but cannot be used out of the box to replace a ready made capture card or device.

This is just beginning - what do you expect from project driven entirely by enthusiast sometimes with relatively low know how especially for old designs - they may excel in modern tools but old designs use some shortcuts known to older generations and lost in modern times - this is sometimes lot of effort to get something just work. But i'm sure that there is learning curve and additionally to learning also bit of luck.
Ready to use solutions are affected by compromise - they are designed in particular goal like consumer market and their are just fine for consumer but not for broadcast quality. Just check audio / video encoding tools and solutions - 20..30 years ago this was really knowledge of few people nowadays this no longer elite know how. I recall my first plays with jpeg or mp3 formats - on 386SX 20MHz decoding 800x600 jpeg by Image Alchemy took 26 seconds, i recall encoding audio wav to mp3 - Fraunhofer L3enc DOS encoding on 386DX 40MHz was like 8..10 times slower than real-time, decoding approx 2 .. 3 times slower - today cheap IC from China doing this in real time and it cost less than 2$ .

[dellsam34]

Not when you are running a business and want something better, faster and convenient, which what the OP is trying to achieve.

[pandy]

Not when you are running a business and want something better, faster and convenient, which what the OP is trying to achieve.

So i wrongly (obviously) get idea to create (develop) analogue cart to capture signal - all ready to use solutions are against idea to be "novel concept" and most of them is directly limited by series of compromises as they targeting two principal areas - consumer video and CCTV - so they goal is to satisfy mostly "price per capture port" with acceptable but not superb quality.
Also TBC is good question mark - real TBC is able to deal even with lost frame i.e. assumption is to store whole video frame in memory and output it with perfect timing (mostly for broadcast purpose - this is not so important for software encoding as computers may work with asynchronous stream of data usually).

[mountaincabbage]

Reading between the lines, it appears the general consensus hardware wise is that all the ICs developed for A2D have stopped progressing. VHS-Decode offers exciting potential as it's all software side.

At present I'm going to:
- Wait for the PCB of the TW6865 chip I've had designed, see if I can get that working, learn what I can and see if I have any additional control over the quality via parameter tweaking
- Learn about VHS-Decode, get it up and running and see how the quality compares with typical methods

Potential hardware route if TW6865 fails/deadends is the Analogue Devices chip mentioned.

Potential software route would be to investigate vhs-decode, improve processing speed and hardware so it is usable.

PS - thanks for everyone's comments so far

[pandy]

Reading between the lines, it appears the general consensus hardware wise is that all the ICs developed for A2D have stopped progressing. VHS-Decode offers exciting potential as it's all software side.

At present I'm going to:
- Wait for the PCB of the TW6865 chip I've had designed, see if I can get that working, learn what I can and see if I have any additional control over the quality via parameter tweaking
- Learn about VHS-Decode, get it up and running and see how the quality compares with typical methods

Potential hardware route if TW6865 fails/deadends is the Analogue Devices chip mentioned.

Potential software route would be to investigate vhs-decode, improve processing speed and hardware so it is usable.

PS - thanks for everyone's comments so far

Hypothetically there is software video decoding so between VHS decode and hardware video decoder - also TBC can be implemented at the software level easier. NTSC/PAL software decoding should be less complex and offer some benefits (as you have access to all lines so you can build statistic and decode more robust way even noisy video).

[lordsmurf]

Reading between the lines, it appears the general consensus hardware wise is that all the ICs developed for A2D have stopped progressing. VHS-Decode offers exciting potential as it's all software side.

That's the wrong takeaway.

The "software only" approach has already failed, because they keep adding in bits and pieces of hardware. Bandages, patches.

It's not so much that ICs have "stopped progressing", but rather that there's nothing really to progress. It's now mature and evergreen. It's like going to the grocery store, looking at the ears of corn, and asking "what new progress has been made?". Nothing, it's corn. Nothing, it's VHS.

[dellsam34]

The OP should know better, He worked for SingMai, If Daniel could not pull it off with all his knowledge and access to hardware and software chances are no one can, I hope I'm wrong.

[seabasstin]

@mountaincabbage This is a late bit of encouragement, but I only found this thread today while researching something else, but I figured I should add my 2cents as well as chastise some of the lameness in this thread...

This thread is hard to read due to the overwhelming amount of uninformed mansplaining. It seems like some people are only focused on attacking your project with various strawman arguments. It's disappointing when people use old, limited information to attack someone who's inspiring and driven.

Your project is very exciting! It shows how easy it is now to use rapid and individualized prototyping and manufacturing tools. This change has transformed how we create and build.

I teach at an Ivy, design, fabrication, and maker-related theory and practice courses.

After 30 years of practice and 13 years of teaching, I can say we no longer face the same limits in hardware and software design. We have made significant strides in prototyping, development, and manufacturing over the last 20, 10, and even 5 years. Many barriers we faced up to 2015 were erased during the pandemic. Generative AI has greatly reduced tedious and frustrating design and development tasks. It has also removed various manufacturing limitations.

Keep at it, and if you follow modern iterative, design, prototyping and development methods you'll create great hardware. This hardware will meet your basic needs and become even more useful and fulfilling with further iterative software development and integration. With ongoing iterative development, you will be able to add even more functionality and features to enhance your workflow.

I want to praise you for your creative efforts to tackle the challenges in your business. I can’t wait to hear the results!

Some points I do want to make against the uninformed but loud voices here, that rely on dated knowledge:

1: Development of analogue video chips continues. While consumer devices are less needed, they are vital in science, engineering, real ai, robotics, aeronautics, and military use. A lot of the information presented here about limitations of the technology is dated and limited. Sure there is not much new development for consumer or even Provideo uses of these technologies, but it continues if you look in the right places.

2: The debate between software and hardware hasn't been significant in engineering for a long time.
Nowadays, we can abstract most hardware in software, and our daily compute needs are based on this fact --a great majority of our "hardware" is just emulated on generic processors to reduce the need for specialized hardware.
We can easily, cheaply and quickly create integrated circuits (ICs) that replicate any software algorithm, and the cost to do so has been greatly reduced from what it was just 10y ago, so you hardly need to be a major corporation to take on this work, and make customized/individualized ICs.

If you believe software can't match the functions of any hardware that's 10 to 50 years old, you're very mistaken, again even a cheap $5 general purpose processor Arm chip, is as powerful as Intel's most powerful Xeon of 2010.
Software can replicate most functionality of this old specialized hardware while running on wide variety of generic hardware platforms.Today, through abstraction software can be optimized to outperform most hardware devices of the past 3 decades, be customized to any set of needs, and be more efficient in doing that work.
Today's software, paired with modern processing power, will always be significantly more useful and flexible than any analog video processing hardware from a decade ago. So statements on this board claiming that processing SD analogue video will be impossible without incredible research in hardware development, are just wrong, and misinformed. This has also always been the case for digital technologies, and will continue to be the case. (until the robots kill us all) The $300k SGI Onyx super computer I worked on to create 3D animations on Softimage in 1993 had the same processing power and graphical abilities (down to the instruction set) as a PSP 11 years later.

Therefore claiming that software can’t match the functions of old hardware video processors, is at best, misguided, at worst misleading.

3: The need for analogue conversion services is not disappearing; in fact, it’s increasing. Demand is rising, especially with the rise of GEN AI processing opening new possibilities which were not even a dream 5years ago. A growing community of young analogue video enthusiasts is emerging. They aren’t concerned with the complaints from us older graybeards who lived through the "end" of consumer analogue media. They aren't held back by old beliefs about the limitations of analogue video. Instead, they have access to technologies we couldn't even dream of when we were trying to make do with the resources available to us. They approach these issues from fresh angles, and are finding new solutions for problems we once thought were impossible to resolve.

4: Stop putting down people putting their best foot forward with FUD.
For years, Microsoft, Novell, BlackBerry, IBM, Intel, DEC, Nortel, Sun, and other tech leaders spread fear, uncertainty, and doubt (FUD) about Ethernet, RISC, Linux, x86, Google, Apple, the Web, and many other emerging technologies and companies.
Their tactics aimed to undermine these competitors and create doubt in their solutions. Look at where those companies are now in relation to the targets of their misinformation and disinformation. (some might still exist, and be even still dominant (MS) but their standing in tech and culture in general is greatly diminished.


Support people engaged in interesting and new endeavors by providing real information and researched resources.
Avoid criticizing their efforts or warning them unnecessarily with phrases like, "you are going to fail," or "why are you doing such a silly project?"
Such behavior portrays you as a bitter old man, resentful of the drive, resourcefulness, and excitement of others.

Encourage their pursuits instead.
Go go go, @mountaincabbage can't wait to hear how your solution has come together, and how you will be improving it!

[Sharc]

.....I'll keep posting updates here on how things go.

A year has passed since. Any news? Experience? Just curious

[aramkolt]

Not exactly related, but there is supposedly a version of the GV-USB2 that corrects line timing least:
https://www.youtube.com/watch?v=amWY9CW6NBA

Not sure if it reduces frame drops or not, but that'd be the first modern USB capture device to have line TBC-like properties that I'm aware of.

The main use-case for that one would be VCRs that don't have line TBCs and wouldn't require the use of a passthrough DVD recorder. Advantage of a DVD recorder though is that you could still put a frame TBC after it, whereas having the line TBC in a USB capture device won't allow you to put a frame TBC "after" the line TBC in the chain.

I did take a stab at ordering one for testing, but that version obviously did not correct line timing, so I've ordered a couple of other ones to test.

I do also think that some of the Cloner Alliance standalone capture devices have line TBC-like properties, but most (or all) of those won't let you capture lossless or interlaced and forces upscaling. Those should still produce a lot better results than the elgato video capture since they aren't starved for bitrate (I think they upscale to 1080P60 at a bitrate of 18Mbit/s). I have a Cloner Alliance box pro for testing, main other limitation with that one is there's no S-Video input and it zoom/crops the edges (roughly 10-15% horizontal/vertical image cropping) for the composite and I presume component input, but you can sort of get around that by going through an upscaling DVD player first (that accepts S-Video) and outputs HDMI which the cloner box can still capture. It will not crop HDMI input sources.

[Sharc]

Not exactly related, but there is supposedly a version of the GV-USB2 that corrects line timing least:
https://www.youtube.com/watch?v=amWY9CW6NBA

Not sure if it reduces frame drops or not, but that'd be the first modern USB capture device to have line TBC-like properties that I'm aware of.

I have my doubts. His video doesn't really show/prove much. Looking forward to your test results though .....

[The 14th Doctor]

Quickly tested an IO-Data without a frame TBC nor built in camcorder's TBC enabled. Wasn't much but did straighten out the image. Granted a Pinnacle 710 did as well.

Way better than the Elgato card, doing absolutely nothing on that front.

Funnily enough the IO-Data kept inserting frames when capturing while I switched the camcorder's TBC on and off from the menu. That ain't really useful as.. why would you even do that during capture?

[Sharc]

Way better than the Elgato card, doing absolutely nothing on that front.

Comparing with Elgato's lossy encoder or with a lossless capture using the Elgato? Comparing apples with apples?

[aramkolt]

It's definitely only certain newer versions of the GV-USB2 that can supposedly do the line TBC part, but I don't believe these versions have ever been sold in the USA as hinted at by the video. At least some versions of newer Japanese marketed cards come with different software, so could be that those software deals aren't allowed in the USA, or perhaps USA sales are strong enough for the regular GV-USB2 that they'll milk that until releasing a new product. What they really should have done is changed the model number to something like GV-USB3 and that'd get rid of the confusion of which is which.

[Sharc]

It's definitely only certain newer versions of the GV-USB2 that can supposedly do the line TBC part, but I don't believe these versions have ever been sold in the USA as hinted at by the video. At least some versions of newer Japanese marketed cards come with different software, so could be that those software deals aren't allowed in the USA, or perhaps USA sales are strong enough for the regular GV-USB2 that they'll milk that until releasing a new product. What they really should have done is changed the model number to something like GV-USB3 and that'd get rid of the confusion of which is which.

There is the GV-USB2 and the GV-USB2/HQ. From the Q&A of their Japanese website:
https://www.iodata.jp/support/qanda/answer/s32350.htm

Differences Between the GV-USB2 and GV-USB2/HQ

The two main differences between the GV-USB2 and GV-USB2/HQ are as follows:

1. Included Software
<For the GV-USB2>
・Video recording software: “LightCapture” by I-O DATA
・DVD authoring software: “CyberLink PowerProducer 5 for I-O DATA”

<For GV-USB2/HQ>
・Video recording software: “LightCapture” by I-O DATA
・BD/DVD authoring software: “CyberLink PowerProducer 5.5 for I-O DATA”
・Video editing software: “CyberLink PowerDirector 12 for I-O DATA”
・Label printing software: “CyberLink LabelPrint 2.5”

2. Saving to DVD and Blu-ray
<For GV-USB2>
You can save to DVD using the included software.

<For GV-USB2/HQ>
You can save to DVD and Blu-ray using the included software.

Translated with DeepL.com (free version)

Added:
The original TW9910 chip has always claimed
- Digital Horizontal PLL for synchronization processing and pixel sampling
- Advanced synchronization processing and sync detection for handling non-standard and weak signal
Maybe that gives a certain advantage which may be experienced as "line TBC like" or similar.

[lollo]

There is only one (hardware version) of the IOData GV-USB2. I bought the last one (out of many,) 2 months ago.

It does not feature any significant time base correction, just some more resilience to unstable signal, compared to USB-Live 2 for instance.