Jump to content
cubanismo

Jaguar VGA Cable Attempted

Recommended Posts

After @Cyprian_K and @Zerosquare detailed how a VGA cable for the Jaguar would be wired and what limited cases it would work in over in this thread:

 

I found the thread on the ST forums about LCD monitors that *should* be able to support the Jaguar's low resolution/frequency:

It turns out I have an old NEC MultiSync LCD 1980SX I use as my email monitor, and I couldn't help wondering if that would be close enough to the fabled 1970 models to work, so I murdered a VGA cable and got an edge connector from digikey to see whether I could see anything:

vga_jag_cable.thumb.jpg.3b923920fda5e913eb11abcd3240db42.jpg

 

This actually *almost* worked.  The NEC panel lights up, and states the signal is out of range on the OSD, but I  can click OK, that goes away, and I can see ~66% of the image on screen, though skewed up so ~30% of the image is "above" the monitor and some more is off to the "right" of the monitor.  It also regularly loses sync and blanks for a few seconds.  I figured I was close and just needed a supported monitor, so I went ahead and ordered an NEC MultiSync EA190M off Ebay since that is noted as being basically a 1970sx +  speakers, but when it arrived, I couldn't even get a signal on it.

 

Now I'm wondering if I am doing something wrong with the cable.  I wired up just the video R, G, B, HSync, and Composite/VSync signals as in the diagram Cyprian_K linked to, as well as ground/return (The cable seems to wire all ground pins together except HSync Ground, which I w wired up separately) to video ground pins on the edge connector.  It's a bit of a mess, as can be seen in the picture, but I didn't find any shorts or anything.  Seemed good enough for a prototype.  Do I need any resistors or anything here to get the monitor to detect a signal?  Do I need to power the I2C lines or anything for some monitors?  Am I correct in assuming monitors that support medium/low res on Atari ST machines should support Jaguar's 15KHz HSync video modes as well?

Share this post


Link to post
Share on other sites
41 minutes ago, cubanismo said:

After @Cyprian_K and @Zerosquare detailed how a VGA cable for the Jaguar would be wired and what limited cases it would work in over in this thread:

 

I found the thread on the ST forums about LCD monitors that *should* be able to support the Jaguar's low resolution/frequency:

It turns out I have an old NEC MultiSync LCD 1980SX I use as my email monitor, and I couldn't help wondering if that would be close enough to the fabled 1970 models to work, so I murdered a VGA cable and got an edge connector from digikey to see whether I could see anything:

vga_jag_cable.thumb.jpg.3b923920fda5e913eb11abcd3240db42.jpg

 

This actually *almost* worked.  The NEC panel lights up, and states the signal is out of range on the OSD, but I  can click OK, that goes away, and I can see ~66% of the image on screen, though skewed up so ~30% of the image is "above" the monitor and some more is off to the "right" of the monitor.  It also regularly loses sync and blanks for a few seconds.  I figured I was close and just needed a supported monitor, so I went ahead and ordered an NEC MultiSync EA190M off Ebay since that is noted as being basically a 1970sx +  speakers, but when it arrived, I couldn't even get a signal on it.

 

Now I'm wondering if I am doing something wrong with the cable.  I wired up just the video R, G, B, HSync, and Composite/VSync signals as in the diagram Cyprian_K linked to, as well as ground/return (The cable seems to wire all ground pins together except HSync Ground, which I w wired up separately) to video ground pins on the edge connector.  It's a bit of a mess, as can be seen in the picture, but I didn't find any shorts or anything.  Seemed good enough for a prototype.  Do I need any resistors or anything here to get the monitor to detect a signal?  Do I need to power the I2C lines or anything for some monitors?  Am I correct in assuming monitors that support medium/low res on Atari ST machines should support Jaguar's 15KHz HSync video modes as well?

Good luck with your project I've would love to see this working!

Share this post


Link to post
Share on other sites
12 hours ago, cubanismo said:

Am I correct in assuming monitors that support medium/low res on Atari ST machines should support Jaguar's 15KHz HSync video modes as well?

yes, the LCD should support 15KHz.

 

I would check whether it works with other 15Khz video signal, w.g with ST.

  • Thanks 1

Share this post


Link to post
Share on other sites
14 hours ago, cubanismo said:

Do I need any resistors or anything here to get the monitor to detect a signal?

No.

 

14 hours ago, cubanismo said:

Do I need to power the I2C lines or anything for some monitors?

No.

 

14 hours ago, cubanismo said:

Am I correct in assuming monitors that support medium/low res on Atari ST machines should support Jaguar's 15KHz HSync video modes as well?

Yes.

 

If you're sure your wiring is correct and that the monitor does support 15 kHz, your issue may be the fact the Jaguar's vertical sync pin actually outputs composite sync. Your monitor may not support the horizontal sync + composite sync combination. Try this instead:

5B Composite sync → VGA pin 13

VGA pin 14: not connected

 

If that doesn't work, try the opposite:

5B Composite sync → VGA pin 14

VGA pin 13: not connected

 
   
  • Thanks 1

Share this post


Link to post
Share on other sites
Posted (edited)

@Zerosquare, you sir, are a steely-eyed missile man!

9 hours ago, Zerosquare said:

If you're sure your wiring is correct and that the monitor does support 15 kHz, your issue may be the fact the Jaguar's vertical sync pin actually outputs composite sync. Your monitor may not support the horizontal sync + composite sync combination. Try this instead:

5B Composite sync → VGA pin 13

VGA pin 14: not connected

That was the working combination.  For the record, I actually tried 5B->VGA pin 14/VGA pin 13 disconnected first since it was an easier/incremental mod, but that didn't help.  When I moved 5b to pin 13, the monitor lit right up!  The picture looks great, but things are not 100% perfect:

 

bbb-lcd.thumb.jpg.f7dae9999d24e46294ad094d8d02222c.jpg

  • I made the mistake of selecting auto adjust in the monitor's menu and it successfully adjusted things such that the picture was indeed fully covering the monitor and centered, but then it constantly flickered out of sync every few seconds, similar to the experience I had with my 1980sx, except at least the image was properly centered.  This was very hard to recover from since when the monitor was out of sync, I couldn't interact with the onscreen menus and the monitor is quite smart and remembered its last saved settings even after I'd yanked the power coord out entirely.  Eventually I got it to let me in the menus while the Jag was powered off somehow and did a factory reset.
  • I then did a bunch of manual adjustments to get it about where you see in the picture: nearly spanning everything and well centered.  If I nudge it a bit wider horizontally, it starts to lose sync, so I'm inclined to not mess with it in this state.

Here's a close-up image of the NEC MultiSync EA190M displaying the timings it is detecting/using:

 

bbb-lcd-2.thumb.jpg.00ea766248d9e6551f5aab9297452d01.jpg

 

Just for fun, I also tried hooking it back up to the NEC MultiSync LCD 1980sx with the cable configured like this, but no dice.  It's an improvement in that the image doesn't constantly lose sync anymore, but it's still only showing ~66% of the image on screen, same as before, and when I try to access the monitor's menu to auto-adjust or mess with it manually, it just pops up the "out of range" warning in place of the menu.

 

While researching this a bit more last night in between my encoding sessions (As you can see, I'm messing with Cinepak stuff on the software side at the moment.  This is the existing Big Buck Bunny Jag CD image floating around the net (And available on Ebay for ~$50.  WTF?), but I've managed to get a toolchain set up to recreate this, and I'm using the existing CD image as a quality comparison to tweak my encoder/quicktime settings), I noticed there is a *lot* of info out there in the retro scene about composite sync and its hsync/vsync components.  A couple of good write-ups I found for future reference:

I'm wondering if sticking that TI chip in there would improve things any and let me generate a real HSync/VSync that made the monitors happy.  Digikey has them in stock for ~$10 in small quantities, so I'm thinking about putting one of those on a little prototyping PCB and seeing what it spits out when hooked up to the Jag's component video or component sync pins.

Edited by cubanismo
Fixed the 2nd image attachment
  • Like 3

Share this post


Link to post
Share on other sites

Just a quick update: I dug through my bin-o-parts after dinner and found an old 3.5mm audio cable to butcher and hooked that up to the appropriate edge connector pins as well.  Now I have both audio and video working great on the LCD monitor.  Pretty happy with how this is working out so far.  Thanks again for the help!

Share this post


Link to post
Share on other sites

I don't think a sync extractor would be useful here, since the sync signals are already available on the video connector. You could try extracting the missing VSYNC signal using  74HCT86 XOR gates:

image.thumb.png.88b9633eb2759671434e6fcf99961445.png

 

There's some power available from pin 11A on the video connector, but it's unregulated 9 V, so you'll also need a 5 V regulator to power the chip.

 

Share this post


Link to post
Share on other sites
5 hours ago, cubanismo said:

Just a quick update: I dug through my bin-o-parts after dinner and found an old 3.5mm audio cable to butcher and hooked that up to the appropriate edge connector pins as well.  Now I have both audio and video working great on the LCD monitor.  Pretty happy with how this is working out so far.  Thanks again for the help!

that's great

Share this post


Link to post
Share on other sites
On 6/15/2020 at 2:14 AM, Zerosquare said:

I don't think a sync extractor would be useful here, since the sync signals are already available on the video connector. You could try extracting the missing VSYNC signal using  74HCT86 XOR gates:

This isn't working out so far unfortunately.  I got a voltage regulator (UA78L05ACPK) and XOR x 4 chip (CD74HCT86E) from digikey, and wired them up as outlined above.  I rigged up a switch on the HSync line on the VGA cable so I can toggle it between the Jaguar's HSync line and its composite sync line, and fed the output of the circuit above into my VGA cable's VSync line.  Having the VSync wired up to the VSync extraction circuit doesn't seem to hurt anything if I leave the cable's HSync line hooked up to the Jag's VSL/composite sync, but if I toggle it over to send HSync down the HSync line, I get no signal from the monitor.

 

Could this have anything to do with the mismatched voltages from the Jaguar edge connector for HSync and Composite Sync?  5B/VSL/Composite Sync states it's +5V, TTL levels, while 6A/HSync states it's 3.0V peak-to-peak.  Alternatively, or in addition, do I need to worry about HSync not being a 0->3v (e.g., -1.5v->+1.5v or something)?

 

I also didn't add any capacitors to the circuit, while you seem to be indicating I should have a 100nf on the XOR gate power supplies?  Pardon my very amateur circuit/datasheet reading abilities, but a few questions there:

  1. Is inserting a cap on the vcc line of the XOR chip indeed what your diagram is indicating?
  2. The datasheet also contains a 100nf (0.1uf) capacitor in its "simplified schematic" diagram (Link).  I can't tell if this indicates it is included in the package, or suggested to be used in circuits containing the package.

Thanks again for the help.  I suppose the next step is really to go ask my neighbor if I can borrow a scope from him for a few hours 🙂 All I was able to verify electrically was that the voltage regulator is indeed putting out a steady 5V when wired up to the 9V pin on the Jag edge connector.

 

And just for fun, a picture of the franken-cable in its current state, switched to use composite sync as HSync:

franken-cable.thumb.jpg.a9a52493f2ad2897d6c22f35827feb86.jpg

 

Such a pain to test in this ghetto form.  Almost every time I solder something onto the edge connector, I accidentally desolder or snap off one of the neighboring wires, hence getting aggravated enough to insert the switch.

  • Like 1

Share this post


Link to post
Share on other sites
On 6/14/2020 at 9:45 PM, cubanismo said:

This is the existing Big Buck Bunny Jag CD image floating around the net

Big Buck Bunny is available for free , don't pay 50 bucks on Ebay for it !

 

  • Like 1

Share this post


Link to post
Share on other sites

@cubanismo: you're right, I never noticed that the HSync signal is only 3V peak-to-peak, not TTL-level like the VSYNC/CSYNC one. Atari's fondness for weird hardware quirks strikes again... In theory it could be a problem, in practice I'm not sure. I'd have to do a simulation and/or get out the oscilloscope to find out for sure. I'll try to do that later.

 

The 100 nF capacitor is not included in the package, and should indeed be in parallel with the power supply, physically close to the integrated circuit.

 

 

  • Thanks 1

Share this post


Link to post
Share on other sites
2 hours ago, Fredifredo said:

Big Buck Bunny is available for free , don't pay 50 bucks on Ebay for it !

Indeed.  I burned my own.  It was rather hard to find the cinepak image though.  The only one I found that wasn't a dead link was some all-jaguar-cd-images torrent of all the Jag CD commercial releases plus a bunch of homebrews, so I won't link it here.  I've not successfully encoded a cinepak file I can play from CD myself yet.  Still struggling to find the right combination of params that results in a CD-friendly bitrate and a non-garbage quality video.  They play great from RAM in 6-secomd clips though 🙂

Share this post


Link to post
Share on other sites
25 minutes ago, cubanismo said:

non-garbage quality video.

It's Cinepak we're talking about. Don't get your hopes up too much :D

Share this post


Link to post
Share on other sites

Couple other misc. notes on the VGA stuff from experiments last night:

 

* The MultiSync 1980 monitor I have seems happy enough with the 3v Hsync/extracted Vsync, as in it's no worse than using raw composite sync as Vsync with the 3v Hsync (out-of-range message, but stable ~1/3 of the image visible).  If I yank the Vsync wire off the sync extractor circuit, everything goes unrecognizable, so the signal must be ok-ish at least.

 

* I added a 50/60Hz switch while doing the BJL mod soldering on my console, and the EA190 doesn't like 50Hz (Unfortunate, since my main motivation was to have a monitor I could quickly test both modes with).  It appears to work visually, but displays an out of range message and then generally goes black but remains synced.  Sometimes it stays visible longer, but usually it's just a few seconds.

Share this post


Link to post
Share on other sites

So when you attach a cat box or similar, what is being output from the DB9 video port?

 

I guess I just always incorrectly assumed it tapped into RGB & output 15Khz VGA as RGBHV. Looks like that's not the case.

 

Related...  can the Jag output CSYNC over SCART natively?

Share this post


Link to post
Share on other sites
1 minute ago, Jagosaurus said:

So when you attach a cat box or similar, what is being output from the DB9 video port?

Yes, I'm curious about that myself.  I don't have any of those devices to take apart and look at, so I don't know.  It might just be wired straight through.  However, there are varying levels off circuits you could put in between to try to get more VGA-like sync signals.  If the 3v peak-to-peak HSync is a nice 0-3v wave, that's technically a valid 5V TTL signal anyway I believe.  Anything over 2v should be considered  "high".  I have a bunch of ideas I want to experiment with, but I need to get a scope to really test any of them, and I'm dithering between just borrowing one, buying a cheap one, or buying a good one.  Plenty of other projects to work on, beers to drink, and games to play as well, so I haven't made any real progress here in a while, but I've not given up yet.

 

I also haven't looked into what CGA signalling looks like or how it differs from VGA in relevant areas here, if at all, yet.  A DB-9. more properly a DE-9 connector, is what CGA used.  I don't know why these boxes would use a CGA connector, given it doesn't look like CGA could support high-color or modes larger than 320x200.  It looks like some very old VGA monitors used a "VGA9" connector, rather than the standard 15-pin connector used by later monitors (and the Jaguar controller port as well of course :-)).  I'm new to the world of Atari, but maybe some Atari monitors or commonly-used replacements had VGA9 cables/ports?

 

Some references I've used so far for this one, in case others are also learning or re-learning EE on the fly so they can do a better job of cheating off ZeroSquare's exams... err... generously provided suggestions, and the massive body of existing misc. Jaguar hacks and circuit diagrams floating around on the web:

 

https://info-coach.fr/atari/hardware/interfaces.php#EGA_DB9_Connector

https://en.wikipedia.org/wiki/D-subminiature

https://en.wikipedia.org/wiki/Color_Graphics_Adapter

https://en.wikipedia.org/wiki/Video_Graphics_Array

https://en.wikipedia.org/wiki/VGA_connector

https://learn.sparkfun.com/tutorials/logic-levels/ttl-logic-levels

Share this post


Link to post
Share on other sites
Posted (edited)
On 7/10/2020 at 7:02 AM, cubanismo said:

Some references I've used so far for this one, in case others are also learning or re-learning EE on the fly so they can do a better job of cheating off ZeroSquare's exams... err... generously provided suggestions

😛

 

I finally dug out the oscilloscope and measured the signals.

 

Hsync is a 3V high, 0V low signal (so it's valid TTL indeed) :

image.thumb.png.a68a53bd12c36b7246415e232e5ad074.png

 

Composite sync is a 5V high, 0V low signal:

image.thumb.png.6e761626e255b1630caf7359c3759b1c.png

 

So... it should work fine with the 78HCT86. I don't know why your monitor doesn't like it, and I don't have a similar one so I can't test it here. Sorry.

 

Regarding the "CGA" converters, I don't know why they're advertised like that. CGA was a digital RGB video standard which supported only 16 colors ; those converters are mostly used with analog RGB sources. There were some non-CGA analog monitors with the same connector (Atari and Commodore computers, among others, used them in the 1980s), so maybe that's the source of the confusion.

Edited by Zerosquare
  • Thanks 1

Share this post


Link to post
Share on other sites

Yeah, I don't know what's wrong with that monitor.  Besides getting a scope on the output of the sync extractor circuit, I also want to try just running the hsync signal through one of the spare XOR units, or maybe both spare ones, with ground as the other input(s) to see if that shifts the level up and matches the (tiny) propagation delay introduced into the Vsync signal by the extractor.

 

I'll have to fix the cable again first though.  I've managed to yank one of the wires off the connector yet again.  😑

  • Like 1

Share this post


Link to post
Share on other sites

I suspect they are referring to CGA in the analog RGB arcade monitor sense not the digital PC monitor sense. Very confusing to have two different monitor standards with the same acronym.

 

Mitch

Share this post


Link to post
Share on other sites
On 7/12/2020 at 6:43 PM, cubanismo said:

I also want to try just running the hsync signal through one of the spare XOR units, or maybe both spare ones, with ground as the other input(s) to see if that shifts the level up and matches the (tiny) propagation delay introduced into the Vsync signal by the extractor.

I gave this a try last night, but it didn't improve things.  The NEC MultiSync EA190M wasn't any happier with it and still wouldn't even light up with extracted VSync + leveled-up HSync.  However, I did "verify" it levels up the 3V HSync to 5V by cheating and probing it with my multimeter.  Probing the raw HSync from the Jaguar measures ~2.83V DC, Probing the signal coming out of the XOR gates after combining Jag HSync + Ground measures ~4.75V, about the same as what I get when I stick the multimeter on the composite sync or extracted VSync signal.  Measuring the inverted HSync from the VSync extractor gives me ~0.3V, so it seems like a valid method of measuring the rough amplitude of the wave.

 

I was able to get a much more stable signal on my NEC MultiSync LCD 1980SX with the extracted VSync + leveled-up HSync.  It's still wrong, in that it's squished onto the upper 66% of the monitor and the right side of the image is forced offscreen with no way to adjust it, but it didn't flicker in and out anymore.  This worked with both 60Hz and 50Hz output from the Jaguar.  The monitor lost sync when I unplugged the wire feeding it the VSync signal, so it must not be an entirely invalid VSync signal.  The EA190M just doesn't like it for whatever reason.  I'm wondering if it just assumes these low resolutions will only come from a composite source or something and doesn't handle a real separate VSync+HSync source for them.  I'm still not clear why feeding it composite sync on its HSync line works at all, but I'll take it.

 

All I have left to try at this point is to get a scope and look at the signal to see how cleanly VSync is being extracted. 

 

Incidentally, for my original goal of finding something that would accept 50Hz output for testing PAL-frequency modes, I found that my LG OLED TV will surprisingly accept that mode over regular composite input.  The image is ever-so-slightly shaky, and it takes it a good long while to sync to it, but it works and displays all the pixels.  Good enough for testing things now and then, and not so much worse than the terrible way it handles composite in general (Seems to have some temporal filter that makes pixels just blink in and out on alternating frames and look like a blurry mess when viewing fine, sharp images like the BJL ROM boot screen).

  • Like 1

Share this post


Link to post
Share on other sites
1 hour ago, cubanismo said:

Seems to have some temporal filter that makes pixels just blink in and out on alternating frames and look like a blurry mess when viewing fine, sharp images like the BJL ROM boot screen

The BJL ROM isn't very composite-friendly in the first place: it uses bright red and blue, which are fuzzy on composite, and the "waiting for upload" banner is especially hard to read.

I assume @42bs developed it using a RGB monitor or TV.

Share this post


Link to post
Share on other sites
Posted (edited)
9 hours ago, Zerosquare said:

The BJL ROM isn't very composite-friendly in the first place: it uses bright red and blue, which are fuzzy on composite, and the "waiting for upload" banner is especially hard to read.

I assume @42bs developed it using a RGB monitor or TV.

Yepp, Commodore 1084 
 

Edited by 42bs
  • Like 1

Share this post


Link to post
Share on other sites

After a long email chain with my EE neighbor and much time on the bay, I'm proud to say I have an old Agilent Infiniium 54832D hooked up to this circuit, and it's now pretty clear why it's not working.

 

During the main part/active region of the frame, the XOR circuit is pretty successful at canceling out the HSync pulses from composite sync.  It still ends up with an edge in there that could be causing problems if monitors use edge trigger to detect VSync, but I'd have to revisit that later if the main issue below can be resolved.

 

The main problem is that around the beginning and end of VSync, composite sync goes all funny and starts sending half-length HSync pulses at twice the line rate, then inverts to generate the VSync pulse, continuing to send these half-length HSync pulses at twice the usual rate, then inverts again to end the VSync pulse, and repeats the whole thing again as a sort of lead-out into the normal HSync pulses.  The end result is that the "filtered" VSync generated by the circuit ZeroSquare suggested doesn't really look much cleaner than the original composite sync signal anywhere near VSync.  Of course, this is all outlined in the HDRetroVision article I linked above in post #5, and looking at that article again to verify it matches what I'm seeing on the scope, I found it even included a sample circuit to extract VSync from composite sync:

 

vsync_recover.png

 

Should be easy enough to find the resistors and capacitors to try this out using the XOR gate IC as a comparator, but if i can't find them lying around, I might have to wait until my next digikey order.

  • Like 1

Share this post


Link to post
Share on other sites
3 hours ago, cubanismo said:

The main problem is that around the beginning and end of VSync, composite sync goes all funny and starts sending half-length HSync pulses at twice the line rate, then inverts to generate the VSync pulse, continuing to send these half-length HSync pulses at twice the usual rate, then inverts again to end the VSync pulse, and repeats the whole thing again as a sort of lead-out into the normal HSync pulses.

Yes, those are called equalization and serration pulses, and they are a feature of standard TV signals ; standard VGA signals don't have them. If your monitor doesn't support those, you indeed need a more complex circuit to suppress them. It looks like the LM1881 sync extractor could fit the bill: https://www.ti.com/lit/ds/symlink/lm1881.pdf

  • Like 1

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...