Capturing from old micro - Spectrum 128k +2 [SOLVED in a way]

[Arheos]

Hello,
I am trying to hook up my old Spectrum 128K +2 (grey) micro and running into the following problem: (hardware specs to follow)
AverMedia Hybrid Volar (A828) TV tuner (Analogue and DVB-T)
WinXP SP3
I can see a signal but it seems to have some kind of "horizontal-hold" problem if there's such a thing. It produces a fine picture on a normal CRT television. The TV Tuner also works fine for TV/DVB-T(aerial signal) and can also take signal from VHS which also looks good.
I made a mod to get the composite signal off of the micro but the same problem occurs.
This is how it looks:
http://img706.imageshack.us/img706/7316/128kcomposite.gif (http://imageshack.us/photo/my-images/706/128kcomposite.gif/)

Signal is PAL-B if it makes any difference, and I am emphasizing that pretty much everything else works, the tuner, the micro, the CRT as supposed to, the monitor, the VHS connection.
Any information on how to correct that, appreciated.
Thank you

PS. I tried it on another PC, same TVTuner, problem persists.

[Ghitulescu]

The CRTs are very warm-hearted when it comes to the signals they accept (think of macrovision).
AFAIK and IIRC, these home computers used special video modi, in particular a sort of progressive: they sent the whole image in one field while letting the other one completely blank.
Have a look on google, there are sites dedicated to such issues.

[TheSkiller]

Yeah, old hardware like game consoles (PS1, NES, SNES and the like) may output a special video mode where the vertical sync is altered so that a CRT screen projects every field at exactly the same physical position instead of alternating positions like it would do with a regular camera/TV/DVD source. If the old hardware wouldn't do that the fact that the vertical resoluton is only 288 and not anti-aliased would make the picture jump by 2 scanlines up and down 25 times a second in regular interlaced scan mode with alternating field positions (not nice to watch). Projecting both fields at the same position makes the TV effectively display a steady 288p50 video.

However, all that should not mean it is impossible to capture such signal. I have done it and the only thing that is a bit tricky is the slightly different refresh rate of such signal which can make it difficult to keep a/v sync. I don't know what kind of digitizer chip is implemented in your capture card but maybe the problem is simply that this chip cannot cope with this unusual signal.

[Arheos]

Yeah, old hardware like game consoles (PS1, NES, SNES and the like) may output a special video mode where the vertical sync is altered so that a CRT screen projects every field at exactly the same physical position instead of alternating positions like it would do with a regular camera/TV/DVD source. If the old hardware wouldn't do that the fact that the vertical resoluton is only 288 and not anti-aliased would make the picture jump by 2 scanlines up and down 25 times a second in regular interlaced scan mode with alternating field positions (not nice to watch). Projecting both fields at the same position makes the TV effectively display a steady 288p50 video.

Shouldn't that make it have vertical hold rather than horizontal?
I was suspecting the "problem" lies with the tuner but I thought there would be some solution.
In any case from what you're saying, this TV-Tuner probably can't cope with it and there's not much software-wise that can be done so I will have to test another tuner. Correct?
Thanks for the feedback.

PS. The weird thing is that I also have a 48K Spectrum (different modulator) from that era and that seems to have a stable signal. The +2 uses the TEA2000 IC if that tell you anything.

[Filker]

Have you tried to tune speccy's channel on the VCR and connect the VCR to the capture card?

I have a capture card (saa713x) that defaults to SECAM if the PAL-B video signal is unstable. On another card (BT848) I can set the signal type and it won't change it.

Record from an emulator :rolleyes:

[Arheos]

Have you tried to tune speccy's channel on the VCR and connect the VCR to the capture card?

I have a capture card (saa713x) that defaults to SECAM if the PAL-B video signal is unstable. On another card (BT848) I can set the signal type and it won't change it.

Record from an emulator :rolleyes:

Yes tried that too, VCR plays, Speccy-through-VCR doesn't. Same issue, maybe it wasn't clear on my 1st post.
Even switching to SECAM has same result only it's B/W.
The property windows seem all to be correct, ie PAL-B etc..
In the end I think the tuner is just too picky with the signal. Will try an older one if I find it and see how that goes.
Of course in the meantime any suggestions are welcome

PS. Btw it's not about record from emulator, sure that's possible :) but I want to use the real speccy for something on the monitor.

[pandy]

Most of older computers generate something like 624 or 626 lines video without interlace - vertical frequency is different than for 625 lines video, most of the older TV can accept such video without problems displaying it as something that was mentioned by TheSkiller i.e. 288p50 (in fact 312 - 313p50). Most of modern digital front-ends can have problems with such signal unless they are reprogrammed to deal with such non interlaced video. (AFAIK first computer that generate correct interlaced video was Amiga however most of the games and demos on Amiga use non-interlaced video with different V frequency)

[TheSkiller]

@pandy If you include the lines used for VBI (teletext) then yes it is a 625 lines signal or in this case half of that, but as usual in PAL, the actual picture is always made up by 574 + 2 half lines or half of that without interlacing.
574 + 2 half lines (625) is always regular interlaced scan, it cannot be progressive because that would require twice as much bandwidth.

To compensate for overscan there is often a black border around the picture with old hardware (especially game consoles) so the actual area used to carry the picture is even less than 288 lines – around 256 but that's irrelevant for capturing.

[GrofLuigi]

I remember I pulled out a composite video out from ZX48's edge connector and it wasn't fully kosher, something was missing (some sync?). It was picky about the devices it was connected to (or vice versa) and often produced B&W or rolling picture.

RF was always good and passed through VCRs without much problems (I am trying hard now to remember about recording, but I can't :o). The only problem with VCRs was that Spectrum's RF out was on the same channel (36?) as VCR's RF out, so it created interference or something (was resolved by adjusting VCR's RF out with a little screw). But that surely isn't the case here, so I'm with pandy on this.

And the overscan border was quite big, and also beautiful while loading programs (all the blue-yellow lines, you could tell apart blocks of 0's or 1's by their width)...

GL

[hanfrunz]

I bet there are software emulators for your "old" computer. Look for one with video export functionality.

[Arheos]

I bet there are software emulators for your "old" computer. Look for one with video export functionality.
As already mentioned, not interested in emulating here, for what I want to do. I use them for everything else.
The only problem with VCRs was that Spectrum's RF out was on the same channel (36?) as VCR's RF out, so it created interference or something ....
And the [s]overscan border was quite big, and also beautiful while loading programs (all the blue-yellow lines, you could tell apart blocks of 0's or 1's by their width)...

GL
That 36 channel was the usual problem, I wish it was the case here, it'd be much easier.
Border was there for a reason as mentioned here as well, but you could do some nifty tricks with it and even see scrolling messages on it, interesting times when the hardware had it's limits.

Trying to fiddle with DScaler, and it does have a LOT of options, but I can't find any to rectify the problem.
Different tuner report to follow..

[Arheos]

So.. I found my old EM2820-chip Tuner (analogue only) and it seems to work fine, ie, the signal is stable.
The older the better then?
Appreciate all the answers trying to shed some light. Apparently the AverMedia (A828 based) is NOT good for older machines and from what I saw there's no way to fix this via software.
Thank you and happy retro-ing!

:thanks:

[pandy]

@pandy If you include the lines used for VBI (teletext) then yes it is a 625 lines signal or in this case half of that, but as usual in PAL, the actual picture is always made up by 574 + 2 half lines or half of that without interlacing.
574 + 2 half lines (625) is always regular interlaced scan, it cannot be progressive because that would require twice as much bandwidth.

I can agree but only when we talk about video accordingly to BT.601 and BT.656.
Computers (especially old ones) don't care about very correct BT.470 signal due obvious simplification for internal video circuitry - so for old 8 bit and most of the 16 bit there is no interlace video generated (even if some trick offer so called interlace usually this is very fast changing for two progressive fields thus subjective feeling that this is interlaced video).

For most of the computers, video is made or as 624 (slightly higher rate than 50Hz for fields) or 626 (slightly lower than 50Hz for fields) lines - and both fields are exactly same (except tricks when there are two different fields).

Check datasheet for some sync extraction IC (LM1881, EL4581)



To compensate for overscan there is often a black border around the picture with old hardware (especially game consoles) so the actual area used to carry the picture is even less than 288 lines – around 256 but that's irrelevant for capturing.
Usually there is even less lines than 256 - most of video circuits are same for NTSC and for PAL - this means that number of lines vary between 160 and 240 (usually around 200 - AFAIR Atari XL, ZX Spectrum 192 active lines, C64 200 lines - in many cases number of lines is hardwired and can't be changed easily as for modern graphic hardware) - of course this is nice area to discover tricks about opening left, top, right, bottom borders to make so called overscan and similar effects - thus recommendation to capture ALL lines and BOTH fields ie at least 704x576i25 with assumption that sampling frequency is 13.5MHz.

This especially valid for various demo's where "288p50" is produced and simply capturing only 25 fields will loose some data.

[TheSkiller]

I can agree but only when we talk about video accordingly to BT.601 and BT.656.
Computers (especially old ones) don't care about very correct BT.470 signal due obvious simplification for internal video circuitry - so for old 8 bit and most of the 16 bit there is no interlace video generated Yes, that's what I did say when I said "half of that" because that essentially it – a video signal that consists of only half as much vertical resolution compared to regular interlaced scan but in return it's "progressive". I didn't use the term progressive in this case although one could say that, I just find it odd to call it that because technically in the video signal there are fields only, always. They just happen to present the same spatial positions for both fields due to the low resolution (=both fields are the same spatially) and the altered vertical sync makes sure they are all projected at the the same place in a CRT to eliminate interlaced scan. From that point of view one could call it progressive I guess. Yes, the Hz is somewhat slightly different as well, it's a pain it the butt though when capturing because you wanna end up with a 25.00000 fps video.


thus recommendation to capture ALL lines and BOTH fields ie at least 704x576i25 with assumption that sampling frequency is 13.5MHz.Absolutely. Because there are 50 fields a second that may contain 50 different discrete steps in time, capturing any less than 576 interlaced lines will discard information.

[StainlessS]

Cant say for sure but, I seem to recall having to fit an internal diode inside the spectrum 48K
to fix the sync signal, no longer have the ZX Spectrum hardware manual though.
(For use with composite monitor).

EDIT: Google for

zx spectrum sync composite

brings up a number of hits.

EDIT: Yep, I think I had a cable and plug hangin out the back of the machine with an
internal mod for the composite signal, perhaps later versions had built in composite, dont know.

EDIT: a load of manuals/hardware manual schematics etc here:
http://www.worldofspectrum.org/documentation.html

[Arheos]

Yep you do remember something..
All this I know, the hardware stuff, worldofspectrum.org etc..and the composite is there on the first post :)
Both composite and aerial have the same problem and I'm 99% sure if I had RGB->Scart it would have the same issue so I'm not fiddling with the Spectrum electronics anymore, being a sensitive machine and all.
I thought here I would find some more expertise on such a problem and the answers show it, that's why I came here to ask.
It seems the problem lies somewhere in this tech-explanation and apparently some modern tuners just don't like to cope with certain signals as they don't expect them.. anymore.
From my part I don't know what other tweaks to do with DScaler, which I may add is an excellent configurable piece of software and I think unique in the win32 world. If anyone happens to come across this problem it may help them or they may add their input should they manage to solve it in other ways than using a more "flexible" tuner.
The EM2820 chip-based tuner worked for me in this case.

PS. Editing Thread title to [SOLVED in a way] so as for people to see.