USB Based Standards Convertor

[Dave Moll]

[Split from THIS thread]

As a by-product of the British Heritage Television Project, whose primary objective is to transmit a one-off 405-line production this November to celebrate 70 years since the start of 405-line television, there is the hope of setting up a 405-line webcast. It is thought by those who know about such things (among whom I do not include myself) that a USB device could be produced fairly cheaply to extract from this a 405-line signal that can be viewed on a vintage television with a modulator. My understanding is, however, that a designer is still being sought for this device. If this were to succeed, it would possibly provide a further impetus to the 405-line television community.

[channel405]

In response to Daves comment about the British Heritage project attempts to "webcast" 405 line TV, one of the main ideas behind this is to start a 405 line TV service over the internet which presently is the only legal way of "broadcasting". We would still love to get 405 on air from AP but this is subject to getting permission and may be a one off. An internet based service would make 405 line video easily accessable to anyone with a PC, albeit those with broadband connection. Look at it as a cable 405 line service over the wires of the internet!

The main obstacle is the technical aspect of getting analogue 405 line video off a PC. It seems this is possible by using a PIC device connected to the PC using USB 2 because of the data rate. The device would contain all the timing references needed to convert the digital output to a 405 line analogue signal (and sound) to a modulator/baseband video.

The 405 line video server would be streaming or "broadcasting" AVI/MPEG2 files resized to 503 x 377 pixels. This would also allow those without a 405 set or USB device to view on a PC monitor. We could even have 405 line colour in that case!

I have been informed by tubesrule that such a USB device is quite possible to make and the cost would be very reasonable, certainly much cheaper than a standards converter. Such a service may be subject to copyright laws as it would be publicly accessible, but then again it could be password protected and serve "members only" to make it more private or maybe that would defeat the purpose of 405 for the masses! But in anycase this would be a certain and cost effective way of bring back a 405 service, the coverage would be very good too! All we need now is someone with the right skills to have a go at designing such a device, any takers?

Tim

[Duke_Nukem]

Technically there is absolutely no reason that a USB device could not be made. However I don't think it necessarily offers the level of simplification that it first seems.

Using the PC an 8-bit "image" complete with sync pulses could be readilly produced, be it 405, 819, 441 or even something really weird like 525 . Then all thats needed is to stuff each byte in turn into a DAC et voila. Which is where the problems start.

The USB device would need to buffer the incoming data as it'll be coming down in bursts. Therefore some RAM plus associated address control is needed. The average data rate would be 6MB/s. The rate at which the buffered data got clocked out would need to be synchronised to the PC's speed (so a PLL makes an appearance). This is all well outside of the PIC's capability - which is a real shame since I have on my desk a PCB with a PIC micro with built-in USB that not only costs pence but also comes with something that would partly address big hurdle number 2 ...

... writing a USB device driver . I'm sure that most people find that as soon as they buy a new toy for their PC the first thing they end up having to do is download a new driver as the supplied one is cack/doesn't work. I used to slag of the S/W engineers that were clearly responsible. Then I started to read up on writing device drivers. Now when I switch the PC on in the morning a stare in wonder and amazement that the machine works at all.

All this extra hassle in effect just replaces Darius's two CCD's and an A/D.

Still, if anyone does suceed in the H/W plus a USB driver then I'm more than willing to write the software that'll take an AVI, DVD (provided the MPEG2 decoder you've got isn't crippled to only work with specific apps - most seem this way) or live video and create the data ready for stuffing into the USB driver. (Not sure about video from the web as I've never tried).

TTFN,
Jon

[ppppenguin]

Jon is right. You would end up with something of complexity not far short of the Aurora converter with the nice bonus of having to write the drivers

That PIC board sounds interesting. Up to now I've used the FDTI USB chips which come with simple drivers and interface to an 8 bit parallel bus.

[Panrock]

I hope all you digital specialists will continue to actively discuss how 405 video can be made available over the internet, both here and on the BHTP forum.

The ability to do this, along with offering a more conventional streaming feed for PC viewing, could well prove crucial for the success of our project.

When we get the chance to talk to the new AP developers about our ideas, it would doubtless help if we can convince them that our transmissions would be at least partly self-financing. An internet-based subscription service would be one way to do this ...

Steve

[Duke_Nukem]

Well, attached pic of the, er, PIC

It uses a PIC18F4550 which includes a built-in USB2 port. But with lots of these little chips you have to read the small print - it is USB2 _compatable_ but has a speed limit of 12Mb/s (we need a sustained 45Mb/s for a convertor).

Actually earmarked the board for that PIC-controlled modulator design that was presented here a while back, with the intention of adding some primative test cards. The only use of teh USB port in this context was that you can download program code down it. But guess what, not got that bit working properly yet - seems to be (yes, you guessed it) a driver issue

Think PIC's really are ruled out of any convertor, at least as regards USB interface and buffering are concerned. Not quite sure how much of a buffer would be needed, but I'm pretty sure it'd be more than this PIC's 2 measly K).

The one other issue I should have mentioned regarding the PC side of things, I guess you'd not be happy if half way through the 1937 corronation broadcast the whole thing stopped cos Windows has chosen to do one of those mysterious now-and-again tasks that thrash your hard disk at the expense of the apps _you_ are trying to run.

TTFN,
Jon

[channel405]

I acknowledge all the comments said above by people more experienced in this field than me. I hope tubesrule doesn't mind me quoting his comments-

"Using a PIC controller with built-in USB and a D/A this should be a fairly
simple matter, and very low cost"

Part of the appeal of such a device is that it can be connected to a PC (Windows comments fully undertood, this could be the "weak link") to allow the downloading of 405 line video over the internet.

The original idea was that this would need only half of a converter as the signal was already in the digital domain and just needed to be taken out of the PC and converted back to analogue 405 again. Another suggestion was to use firewire, but this uses another form of compression and so would therefore add complexity. The end result would allow enthusiasts to share 405 line video "files" which I am sure is an appealing concept!

And of course a webcast of the forthcoming AP broadcasts this November

[Duke_Nukem]

PIC's with built-in USB are indeed cheap. The 18F4550 I mentioned is less than 4 quid ... in one-offs !

However 6MBytes/sec is well fast, too fast for a PIC to process and well above its maximum USB data rate. The PC will also send the data in bursts which would need buffering - 2K of memory that is on-board this particular chip is roughly enough to store three 405 lines. Not good enough, so an external RAM chip is needed ... which the typical PIC can't address directly, so we now need to implement a FILO buffer. It's now getting more complicated.

The USB interface, at the driver level, can be configured for various styles of operation. Bearing in mind the target level of a PIC device I doubt that the base driver that is part of the dev kit will support the streaming mode that'd be needed.

And we still haven't addressed locking the rate of clocking data into the final DAC to a multiple of the video source's frame rate within the PC. It might be possible to write a driver that acts as a "Pull" device i.e. it is the component that will dictate the frame rate but this wouldn't work if for example you were trying to drive the convertor from a TV card in your PC since that'd dictate the frame rate (=broadcast rate!).

...the signal was already in the digital domain and just needed to be taken out of the PC...

This is the key problem. Lets take a simple example of flashing an LED.

In around 1980, I had an Acorn Atom. One minute to build the hardware, 30 seconds to write the software.

Late 80's, I had a PC. One minute to build the harware. Half an hour to write the software.

5 Years ago. Oh dear. Only accessible interface was the serial port. 2 days writing the PC software, two days builing a processor board with a UART and writing the appropriate S/W (ever tried debugging two lots of S/W running at the same time when you don't know which bit has the bug?).

Today ? How many PC's still have serial ports. Oh dear :-( USB it is then. Probably still only take a couple of days to build/write s/w for the target device, but writing a stable PC driver etc ? How longs a piece of string ...

But the offer still stands. If someone else does the bit that takes a buffer full of 8-bit DAC-ready data and stream it down a USB interface then I'll write a DirectShow filter around it and the software to create that data.

TTFN,
Jon

[Sam]

This is starting to sound like a final-year project kind of thing now. And before anyone asks: NO!

Is there anyone out there who is at Uni and doing electronics? I know there are a couple of Uni-age people here as well as me. It seems to me that the USB converter could be developed into a viable project. However, this would mean it probably wouldn't be ready too soon, and probably not in time for the 70th anniversary deadline .

Sam

[Duke_Nukem]

The chap that sits next to me here at w*rk was until a few years ago lecturer in electronics at Birmingham University and there were two students there whose project involved USB.

Neither of them got any further than writing characters to a display and not particularly reliably.

If a USB standards convertor is to be viable as a hobby project then we'd need a device that comes with a driver that already supports transfers of large blocks of data using the appropriate mode of USB operation.

There are a number of USB-to-Paralell devices about (someone just mentioned Crystal Semiconductor?) but be sure to check that it can handle a sustained 48Mbit/sec - USB2 compatable doesn't necessarily mean something supports the full USB2 data rate.

But you'll still need a FILO buffer, the logic to support it and some mechanism to support linking the clocking rate of the DAC's data to the PC's video stream frame rate.

TTFN,
Jon

[ppppenguin]

Originally Posted by channel405

I hope tubesrule doesn't mind me quoting his comments-

"Using a PIC controller with built-in USB and a D/A this should be a fairly
simple matter, and very low cost"

I've just exchanged emails with Darryl (tubesrule) and he admits that his thoughts on the subject were fairly speculative. It's still likely to involve some kind of programmable logic device (FPGA). He reckons (and I agree!) that the hardware would be fairly simple. The software would not.

[Duke_Nukem]

Techically I totally agree that a gate array would be a logical choice. But this raises two issues.

Firstly, any design that involves a gate array, or indeed a programmable micro immediately puts it out of reach of most home builders. I have a test card genny on my website - I'd be the first to acknowledge the address counter works on a wing and a prair. But the only emails I've had about it have been askeing wether I can supply eproms. So more specialist micros and gate arrays would be worse.

Secondly, it sounds like we just re-invented the Auroa convertor but with the A/D removed and replaced by a USB connection. So would it actually save any money ?

TTFN,
Jon

[David_Robinson]

Originally Posted by Duke_Nukem

any design that involves a gate array, or indeed a programmable micro immediately puts it out of reach of most home builders.

I'm not so sure about that, when I built my standards converter, the programmer I used for the Altera PROM (that holds the code for the FPGA) consists of a single 74HC244 and a few resistors. I built it myself. Hardly out of reach. The software to drive this is a free download. My PIC programmer cost £50.

A bigger issue with today's devices, is the physical difficulty of mounting devices with minute pin spacing onto a board. The Altera FPGA I used has 100 pins spaced at 0.5mm.

As to using a PC as a means of simplifying standards converter hardware, this kind of thing has been proposed many times before and no-one to my knowledge has actually done it. The fact is, a stand alone converter is just as easy and much more practical. Streaming from the web with reasonable quality (even on 405) would need a much higher bit rate than most of us have available. I think the practical way to distribute programme material, copyright permitting, is on DVD.

[humorist2751]

Greetings to all.

Enlarging on Duke_Nukem's earlier point about the potentially disastrous effects of MS Windows (or whatever else) deciding to "do its own thing" and creating "hiccups" in the data throughput, I am afraid that is the price paid for running an operating system that is not specifically designed with real-time tasks in mind.

While operating systems that are so designed are available (the ex Red Hat "eCos", for example), it is a fair bet that considerable work would be needed even with such as that. But to indulge in a bit of kite-flying, if a CD-ROM were to be produced (by whoever) containing the eCos operating system and a dedicated "405 streaming video" application, all the prospective user would need would be the external hardware box of tricks and the ability to run up their PC from the CD. The usual everyday uses of the system would be unaffected, provided sufficient main memory was available in order to prevent the need for use of HDD as backing store... which would probably be too slow to be any practical use anyway!

If any reliance were to be placed on operating systems not specifically aimed at real-time tasks (and Unix/Linux is no different to MS Windows in this respect), then certain system housekeeping tasks may well cause glitches. A case of between the devil and the deep blue sea...

[channel405]

"it sounds like we just re-invented the Auroa convertor but with the A/D removed and replaced by a USB connection. So would it actually save any money ?"

Well, this started out by asking Darryl if it was possible to add this feature to the Aurora before the D/A converter stage. He said the design was now signed off but in principle it would be quite easy to build a device to do the job. Its not just about saving money, this may be the only realistic way of getting somekind of 405 "network" going again if we can find a practical way of getting the video out of a PC into a telly.

"Streaming from the web with reasonable quality (even on 405) would need a much higher bit rate than most of us have available."

Streaming from the web of course would invlove compressed video, typicaly using MPEG4 which would produce reasonable results on a broadband connection. Dial up would of course struggle and would have to buffer first. But even with buffering, this could be played out from the HDD into the USB2 converter. Also, the content doesn't have to be streamed, file download would work, just wouldn't be "live". But this is hardly the biggest problem to be tackled first.

[DoctorWho]

If this could be realised at a reasonable cost it would re-invent 405 line television, but this time on an intrnational scale. If Heritage transmissions are taking place on a regular basis people may well 'look in' world-wide. This could be just what is needed to secure the long term future of the studios at AP.

Peter.