TV Guide Channel Emulation Working!

[swest77]

Edit: Holy shit, swest. You got all that typed in while I was going back and forth between typing this post up, and editing the wiki. Time to soak all that in

If you really want to know, I composed the whole thing in Notepad in FInalH26f and then just pasted it into Ari's forum. (God, my eyes...)

Edit 2: Ok, I actually remember seeing the "dashed lines" glyph in FinalH26f when I was converting it to a .FON. I do notice that some of the letterforms are different, however (specifically from that screenshot, the M, 2, 3, and P look slightly different).

Same here. In the blue grid's version of the font, the middle point of the 3 is shifted by one scanline and the top and bottom points are less rounded... the M's middle point is trashed by the VHS resolution but you can still tell its further down... etc. 2 sports the biggest alteration of all, that I've spotted so far anyhow.

Perhaps one should try to compare to some of the other videos (unfortunately, 99% of EPG/Prevue videos are shit quality, although I think there were some ones linked from here that I snagged that were at full 480-line resolution, and hadn't even been deinterlaced! I converted them to MPEG-2 so that I could play them back with bad-ass 60Hz fluidity).

You can thank/blame me for that. I digitized my videos as 29.97 interlaced (field scanline interleaved progressive frames) MPEG-2 and just left 'em that way when converting to H264 for upload to Vimeo/Pootube. Any multimedia playback hardware connected to a TV that'll play them with interlacing enabled thus yields beautiful 60 Hz. Anyway, I'd put the original MPEG-2 files online somewhere except we're talking 400-700 MB a pop here. MPEG-2 does not like noisy VHS and running at 9.8 mbit/s during my captures was compulsory. I did however put all four videos in converted H264 form online at http://www.megaupload.com/?d=R3BKEDUB for anyone who wanted them. So feel free to get copies from there if you want to at least remove Vimeo/Pootube's generation of artifacts from your archival copies.

Oh, for the record, they did have a text editor - :C/UVed - so it'd be more like "ok, hold both mouse buttons while rebooting, click 'Advanced Options', disable startup-sequence, click 'Use', click 'DF0', type 'UVed s:uv-startup', and put the new select code after 'run >nil: esq' where the current select code is". Still much harder than "ok, hold both mouse buttons while rebotting, click 'Advanced Options', disable startup-sequence, click 'Use', click 'DF0', type 'sel new select code'".

Ah, nice. I can just see the spectacular success the former would've had given some of the cable company cercopithecoids out there.

To be fair, these probably aren't really "undocumented" functions, insomuch as they are just unknown to us. I guarantee you that 99% of these functions are detailed in the manual.

Maybe less than 99% though. Some of those functions seem to do things way too ... esoteric for the everyday cable technician. They sound much more like things UV tech support people would've instructed cable tech people to do over the phone during troubleshooting sessions. Ah well, it's all hair splitting anyway: the manual, if it's ever found, will reveal all.

Oh, and also for the record - the genlock means jack shit. I disabled "Genlock connected" in the WinUAE configuration and the software still booted and functioned properly. Doesn't surprise me, to be honest. The only thing I would guess that the software actually controlled was whether or not the Amiga graphics were overlayed over incoming video or not - it stands to reason that this feature (as seen in some of the earlier A2-Black videos) might be specific to the UVGEN cards. If my SuperGen 2000S were functioning properly in this regard, I'd test that. I bet this thing would run on a bog-standard A500+ with a sidecar holding at least 1MB fast RAM, with no other hardware necessary.

You're the Amiga expert. Though from a programmer's point of view, I'm still waiting for someone to successfully feed data to the software that makes it try to paint one of those overlays. Whether the software at that point simply does nothing gracefully, or guru meditates, will reveal for sure whether it "needs" a genlock -- and for that matter whether it needs a UVGEN genlock. Fortunately, even if it needs a genlock (or even a UVGEN genlock), we control the data stream and can prevent that need from ever arising. Just tell the software that every half-screen promo is for "The Ari Weinstein Channel", and then don't tell it that the local cable system carries said network.

Also, I can't imagine they'd use A4000s, given the incompatibility I've found with anything other than 2.x ROMs at this point. AGA hardware works, but 3.x ROMs do not. I can see them using modified A3000s though, given that the later ones came with hardware 2.04 ROMs (and the early ones softbooted 2.0 ROM images unless you held both mouse buttons and made them boot 1.3). No way they're going to expect cable headends to not only deal with the hackish software that is A2-Blue, but also deal with softkicking 2.0 on top of this - this would require either a hard drive, or taking up every bit of extra free space on the disk for a 2.0 ROM image.

No, I'd say not. Anyway, it has been more than 10 years since I consulted for UV, so whether they said "A3000s" or "A3000s and A4000s" has become a little fuzzy for me. At this point, I've given about all I can give informationwise, and anything you folks discover to the contrary, I'll treat as truth over my own fuzzy memories.

And now I have spent almost 45 minutes typing up this post and editing it about five times to add something. I guess that's to be expected when software one has been dreaming about playing with for some 15-odd years finally shows up.

It does have that effect, yes. Speaking of which, Ari, when you read this, please do e-mail me that disk image. I'd like to have a go at it now myself!

@tin -

Just wanna say swest, that was a great post I also think you're on to something

I guess the software functions on our emulators cos it tells it cards to do something (by whatever means) and doesn't care if the right response comes back.

My theory is, more like no response would be expected to come back. The Prevue Guide software would more or less be "flipping a light switch on and off" and expecting some unknown card to see it and reliably know how to react.

Regard the multiple F-connectors on the cards, surely that's for looping through a single signal from the sat-recieve through the several cards?
*EDIT* - aha the stickers on the back reveal all - the baseband was looped through the serial and then the audio card. I guess there's an out on the audio card for if you want to use the signal somewhere else too.

Blast! Again, I had the truth right in front of me and didn't see it. I swear, I'm losing it in my old age. Okay, see http://img230.imageshack.us/img230/8553 ... ignalf.jpg (and forgive the horrible MSPaint grade photoshopping, I'm just too tired to do "quality work" right now ). So yeah, with that other photo of the other half of the label in my sights now, it's all starting to make sense. No need for a three-way splitter between the cable company satellite receiver and the A2000. Just one direct coax between the receiver and the A2000's audio card's input F-connector, and use jumper cables from there to take it everywhere else...

The input for the genlock is labelled video - so unless that's just a lazy person not thinking, that suggests to me that the input to the genlock is the video from the sat-reciever, not the baseband. Of course that's more conjecture....

Actually, when you're dealing with a demodulated multiplex signal from a satellite receiver, it's both baseband video and subcarriers. Think of the output of the satellite receiver (demodulated multiplex) as being exactly the same as a VCR's "line out", but with audio signals "pitch shifted" way up to 5.8, 6.2, and 6.8 MHz -- well above the upper frequencies of the baseband video -- as a way of transporting them on the same wire. It's just that on the A2000's labels, instead of them saying "demodulated multiplex", they say "BASEBAND" ... 'cause in the broadcast world, "baseband" is what you call a multiplex that's no longer at the 3 GHz satellite transponder downlink frequency.

BTW I guessed the amigas with the BNCs were for Sneak - but what was the one that looks like it's got 7s-video connectors in the same slot (one of your pics from before)?

This one? http://img10.imageshack.us/img10/906/bv12.jpg ... Yeah, I noticed that one too and ... I honestly don't have a clue what it is! My best guess: it's an EPG Sr. Because it has no audio demod card, just a VBI decode card. And one of the VBI decode card's F-connectors (top one) is terminated as if once the C-band signal goes into it there's no further need for it to exit it. (Though strangely the one that's terminated is the one that's the input on the A2000's VBI decode cards.) That said, EPG Sr. boxes also wouldn't have had genlock cards for video, so perhaps the video card they did have was a Zorro S-video output card -- but why they would choose an S-video output card with seven friggin' S-video connectors on it is beyond me! (And why does the video in/out label indicate 2 ports for a 7 port card?) And ... why do they have an RS-232 serial expansion card ("SERIAL CARD DIN CONN / BAUD RATE N8?1 2400")? Did the A1000s not have their own?!

and.... damn Commodore for being so short sighted, surely they could have just kept manufacturing A2000s for PG? It would be more expensive of course, but I am sure UV would have preferred to carry on rather than struggle, given that they were probably priced at a real premium to the cable cos.

In deed.

PS - Ari, you should raise the attachment file size limit here. I tried to attach (rather than link to) the signal flow image so it'd be safely archived here, but forum wouldn't allow it.


@edit - About the genlock card having 3 connectors.... I'm assuming one is output (obviously), another is input of PG C-band (again obviously), but the third MAY be ... input of the cable company's local commercials' video. That way, when the VBI decode card cues their VCR to begin playback (via contact closure), the genlock card can simultaneously switch to the local cable compnay video input while keeping the listings grid overlaid in the bottom half of the screen on the genlock's output connector. This would also be the most sensible way to do it, because you wouldn't want the cable company's VCRs to "override" the c-band input itself, which would knock out the incoming VBI data stream from the PG software's perspective for the duration.

Hmm. Now I've just realized what the "extra' wire from the A2000 RS-232 port to the VBI decoder card's GND is for. When PG software wants to fire local cable company commercial, it signals the VBI decoder card to contact-close via that wire, and simultaneously tells the genlock card to switch to the cable company commercials input. Hey hey! Now it all makes sense...

[AriX]

PS - Ari, you should raise the attachment file size limit here. I tried to attach (rather than link to) the signal flow image so it'd be safely archived here, but forum wouldn't allow it.

Done

GREAT post, by the way... Very interesting stuff. I did actually notice that the H26f font was the older EPG Sr/Prevue List font, but that was before this community here was really alive :p

[LocalH]

You're the Amiga expert. Though from a programmer's point of view, I'm still waiting for someone to successfully feed data to the software that makes it try to paint one of those overlays. Whether the software at that point simply does nothing gracefully, or guru meditates, will reveal for sure whether it "needs" a genlock -- and for that matter whether it needs a UVGEN genlock. Fortunately, even if it needs a genlock (or even a UVGEN genlock), we control the data stream and can prevent that need from ever arising. Just tell the software that every half-screen promo is for "The Ari Weinstein Channel", and then don't tell it that the local cable system carries said network.

The good thing about genlocks on the Amiga is, while they can be software controlled, it's not generally a requirement - most normal genlocks have controls to enable color 0 keying, and (if it's a good enough genlock) the degree to which Amiga graphics are overlayed (from transparent to opaque). My SuperGen has two sliders on it's control box - one for background and one for graphics. My thinking is that, since this is a purpose-built genlock, there was no need for that, so it operates in two modes - full Amiga graphics, or color 0 keyed out - and this is software controllable, but not fatal if it doesn't happen (since it would be feasible to treat this like the "split-screen EPG Sr" and only run graphical and/or text ads), and thus why it runs even with no genlock connected. WinUAE doesn't actually emulate anything more than the presence of a genlock, btw.

[curtjr4]

Hello there everyone. I havent posted in a few days, but I have been lurking around on the forum everyday. I now have a question regarding 9.0.4's DST time data. Everyone knows that the time is 2 hours behind normal time. I was wondering if this is something to do with DST. If not, how would I fix this?

[AriX]

Hello there everyone. I havent posted in a few days, but I have been lurking around on the forum everyday. I now have a question regarding 9.0.4's DST time data. Everyone knows that the time is 2 hours behind normal time. I was wondering if this is something to do with DST. If not, how would I fix this?

You can't until we figure out some sort of Amiga-specific serial data extension.

[LocalH]

This one? http://img10.imageshack.us/img10/906/bv12.jpg ... Yeah, I noticed that one too and ... I honestly don't have a clue what it is! My best guess: it's an EPG Sr. Because it has no audio demod card, just a VBI decode card. And one of the VBI decode card's F-connectors (top one) is terminated as if once the C-band signal goes into it there's no further need for it to exit it. (Though strangely the one that's terminated is the one that's the input on the A2000's VBI decode cards.) That said, EPG Sr. boxes also wouldn't have had genlock cards for video, so perhaps the video card they did have was a Zorro S-video output card -- but why they would choose an S-video output card with seven friggin' S-video connectors on it is beyond me! (And why does the video in/out label indicate 2 ports for a 7 port card?) And ... why do they have an RS-232 serial expansion card ("SERIAL CARD DIN CONN / BAUD RATE N8?1 2400")? Did the A1000s not have their own?!

I don't think that's a video output card - those are 8-pin mini-DIN connectors, not 4-pin Y/C. I think that is some sort of serial card, given that the label says "SERIAL CARD DIN CONN". Given the card configuration (and the fact that it doesn't appear to have the same SCSI/RAM card found in the other machines), I think this might be the inside of that particular Amiga (interestingly, it's the only one with disks [!] hanging out of the drives). From that view, it appears that the top-most card (not counting the video slot) might have a row of mini-DIN connectors at the bracket end.

In fact, I can do you one better - I am quite certain I have just ID'd that card - it's a Commodore A2232. 8 serial ports, with an set of miniDIN8 to DB-25 cables.

And, if you look at the IV4 image I reposted, you'll clearly see there is a card in the video slot, but I can't tell how many BNCs it has. I'm going to guess, for sake of cost efficiency, they used the same genlock cards across all their products. They may have had the headends connect blackburst to the inputs on the Sr. machines, for timing purposes.

Edit: Scratch that. I was looking at the Sneak Prevue entry on The Big Book of Amiga Hardware site, and it shows a card labeled "UV CORP - AMIGA GENLOCK 2". This card has a blue DIP switch block in the upper left hand corner, and two BNCs. This genlock probably only has a single input and a single output, and it matches what is seen in IV4. GC1 shows a 3-BNC genlock card labeled "UVGEN" - and I think by now that we have indeed deduced that the Prevue systems (whether running A2-Black or Blue) had the capability for a second video input, given the references to starting tape in the diagnostic screen. Thus, I think that out of the images that were posted here (by swest, I think?), IV4 is the only one I can virtually "guarantee" is an EPG Sr. IV2 might be, but it has no cards apart from what looks like a RAM/SCSI card in the processor slot, which IV4 does not have. It would also stand to reason that EPG Sr. would not necessarily need the extra memory that a A2-Black/Blue would need. For the record, we can start calling the hardware platform "AP2000" based on the label that swest reconstructed.

My guess is that the headend would connect blackburst to the Sr. genlock input, so that in case the machine needed rebooting or power cycling, the genlock would pass through blackburst instead of there suddenly being no signal there - much less jarring if someone is watching the EPG and the system loses power or needs rebooting.

I just had a thought that might help us further with regards to collaboration - we need an IRC channel somewhere

[swest77]

Regarding 8-pin DIN vs. 4-pin Y/C, yeah, you're right. Also thanks for pointing out BV12 and IV4 are the same machine. I surmised BV12 was an EPG Sr as well, and now that I can that see it does in deed have a genlock card, I wonder if the reason for its input isn't house reference black but instead that "upgrade option" UV offered for split-screen EPG Srs. Remember my writing about that on Wikipedia? Maybe the split-screen EPG Srs could not only do graphical ads in their top halves but also switch cable company VCR-sourced commercials into their top halves as well. (That would possibly explain why on this EPG Sr, there's a label bothering to point out that the green punchblock on the VBI decoder card will "cart switch," like on the AP2000s' VBI decoder cards.)

About IV2, I think that may actually be a Sneak Prevue. In the AP2000s, the SCSI/FastRAM card always appears in the 5th-from-the-right slot. But in the Sneak Prevues, the configuration appears to be that they instead put the satellite/laserdisc switcher card in that slot, and move the SCSI/FastRAM card further left, to the the processor slot. And Sneak Prevue boxes also had one-input one-output genlock cards. So, with IV2 having a one-in one-out genlock and a SCSI/FastRAM card in its 6th-from-the-right slot, my money goes on it being a Sneak.

Anyway. Now that I know the EPG Sr had 7 serial ports instead of 7 video ports, I'm still no less confused. Why seven serial ports, is now my question.

It would also stand to reason that EPG Sr. would not necessarily need the extra memory that a A2-Black/Blue would need.

From what I can tell, the BV12 / IV4 machine does have a FastRAM/SCSI card. Compare IV4 to XX1. The card has a vacant back half and white RAM DIMM sockets at its front, and that's what I see in IV4. The card also has a DB25 port on its backplate, and in BV12, I see that too.

Re: IRC channel, sounds like a good idea for the experts here. I wouldn't be of much use however, what little Amiga hardware knowledge I had 10 years ago having nearly all faded away.

[swest77]

Some things noticed while tooling with prevue.adf in WinUAE:

1) You cannot select the forbidden scroll speeds on the Change Scroll Speed menu with any key ... except the up arrow key.
2) Shift-right_arrow and shift-left_arrow are undocumented for the Edit Attributes interface, but do work there for switching between ad attribute screens.
3) This one is just for Tin: Press Ctrl-[ and watch what happens.

Also, I searched for something like, and found, unADF and was able to dump the A2-Blue disk and inspect its files. Interesting in deed that the ESQ, Installer, and sel executables truly do have no visible strings in them. I'm almost positive now: they have to be compressed executables. And on that note, I looked around and found some interesting possibilities:

http://www.kyzer.me.uk/pack/ -> "I'm interested in data compression algorithms and file formats. In particular, I like to reverse engineer proprietary, unknown file formats and make decompressors which allow anyone to unpack these formats." (This guy's remark is inclusive of Amiga software.) He also discusses something called XAD and XFD which can recognize and unpack "thousands of Amiga compression formats" including those used to crunch executable files. See introduction at http://www.kyzer.me.uk/pack/xfd/ ... and http://www.dstoecker.eu/programs.html#xfdmaster.library ... and http://www.aminet.net/package/util/pack/xfdmaster

The Amiga experts here might want to check these out, and see if those UV executables can't be decompressed after all. (If the executables are compressed with a proprietary packing method, then I wonder if this guy might find Prevue Guide a novel enough target to vanquish by hand.)

Anyway, just a lead for you all.

[LocalH]

I'm almost certain that they are compressed - if you boot the system at standard 68000 7MHz speed, you can see the classic decompression effect of tweaking one of the palette entries, specifically the red color on the mouse pointer. Testing this by running some stuff manually from the initial CLI shows me that PD, sel, and esq are definitely compressed. Possibly PowerPacked? I thought I tried depacking esq without success. Installer does not twiddle the palette, but it does take a few seconds to run once it reads from the disc. Other than that, however, Installer is a standard Commodore tool, so nothing interesting would come of looking into it. Time to see if we can figure out what format these executables are compressed with...

Quick note - I looked at PD, sel, and esq in a hex editor. The first 0x2a5 bytes are nearly identical across all three binaries. Decompression code maybe? There are a few differences that I think may have to do with filesize, will investigate Amiga executable packer formats later and see if I can match anything up.

Edit: Looked at Installer, it has some similarities to the others, but it looks like different code, possibly. Haven't tried to disassemble this to see if it's actual code or not. All four files end with the 5 bytes 00 00 03 f2 00. Can not determine if this is a proprietary format as of right now, trying to find as much information as I can on Amiga packing formats. It's not PowerPacker or Imploder, as the former has a defined header (and would depack with the tool I'm using), and Imploder did not twiddle the palette on depacking.

[rudyvalencia]

I'm almost certain that they are compressed

I kinda figured that was what it was doing flashing the cursor - unpacking the application.