VBXE Sync issues and potential solution(s)

[morelenmir]

Over the weekend I fitted a new VBXE to my 130XE, after trying any number of other ways to improve the native video. It was inevitable really, but the considerable price of the addon board forced me to look in every other direction first.

 

Attaching the mod to the XE is actually easier in my opinion than the XL - which itself was not hard by any stretch of the imagination. On the later machine most patches run to the 'Freddie' IC and only one, very bulky and easy to identify component has to be pulled from the motherboard entirely. However one initially intractable snare I encountered was the screen would collapse to black and then flick back at random, despite the machine now functioning perfectly in every other way. In trying to track this down I spent a good few wasted hours reworking the various connections, but nothing did any good. Eventually I reached out to FJC and he instantly (correctly!) identified this as a sync problem between the VBXE output and the specific television I was using as a VDU. To sort this out he firstly suggested I slightly modify the 4050 display buffer chip so it would provide 'gated sync'. Sadly this did not cure the problem, however his second piece of advise hit the nail on the head; to put a variable resister/potentiometer somewhere along the sync wire and tweak its setting until the picture was solid. This proved a little tricky as the only variable resistor I could lay my hands on was a 500K PCB-mount unit, so making small adjustments to its sweep took a great deal of precision and patience. However I finally got it to just the right place and like magic the display has (almost) not collapsed to a black screen since!

 

So I wonder if some of the more technologically minded chaps could explain why altering the resistance along - therefore I suppose altering the current flowing through - the sync line should cause the picture to become stable?

Edited August 22, 2016 by morelenmir

[phoenixdownita]

https://ianstedman.wordpress.com/2016/07/06/synchronise-your-video-engines/

 

Maybe the signal is TTL level and it ought to be reduced.

[Frankie]

I don't understand 1/2 of that article but it appears I need to do something with my VBXE output that I send to a GBS-8200. It looks fine, but it seems it won't stay that way for long.

 

Frank

[morelenmir]

https://ianstedman.wordpress.com/2016/07/06/synchronise-your-video-engines/

 

Maybe the signal is TTL level and it ought to be reduced.

 

That sounds pretty categorical - 330ohms for a 1V SCART CSYNC.... Annoyingly I don't have a 330ohms resistor spare!!!

 

Although that is of course only for the known resistor value inside the converter box he names. In so far as I have no idea what the value inside my television is it would be better to build a crude potential divider circuit that allows as near as possible to 1V to be tapped off from inside my atari before it even gets to the tv at all. That is basically what I have been doing with a variable resistor - by accident probably finding the right value that taps off 1 volt.

 

Ideally I would like one of those arduino cards he shows, but they are only available from china on the UK ebay at least.

Edited August 26, 2016 by morelenmir

[phoenixdownita]

 

That sounds pretty categorical - 330ohms for a 1V SCART CSYNC.... Annoyingly I don't have a 330ohms resistor spare!!!

 

Although that is of course only for the known resistor value inside the converter box he names. In so far as I have no idea what the value inside my television is it would be better to build a crude potential divider circuit that allows as near as possible to 1V to be tapped off from inside my atari before it even gets to the tv at all. That is basically what I have been doing with a variable resistor - by accident probably finding the right value that taps off 1 volt.

 

Ideally I would like one of those arduino cards he shows, but they are only available from china on the UK ebay at least.

A TV usually has a 75ohm termination for Composite and SVideo and usually RGB (I don't know about components), the sync over RGB usually comes from the composite (unless you use CSYNC obviously) hence why the expected range.

If you don't have 330 you can try something bigger (say 470) or smaller (say 220) to see how it goes and yes the variable pot you have does the same thing.

With regard to VGA it says usually VGA monitors are terminated with a 2K omh so you need to adapt the signal accordingly.

 

So told we are assuming the CSYNC you have right now is at the wrong level, and at that that it is TTL, not sure that to be the case with VBXE you should ask the guy that made them so you have better info to go off of.

 

[as an extreme example, I have an XRGB mini and over its RGB input some consoles gave me all sorts of trouble wrt syncing, in specific the NwoGeoAES, in the end I mounted a sync stripper (LM1881) over composite and I can now switch among original composite as sync or sync stripped (which from that link appears to be TTL as I used an LM1881) .... needless to say the sync stripped version is what I use the most and saw no issue .... and no extra resistor there either .... I even tried it on a Scart based TV I have around and it works there too .... but maybe as the article says it's a disaster waiting to happen]

[morelenmir]

A TV usually has a 75ohm termination for Composite and SVideo and usually RGB (I don't know about components), the sync over RGB usually comes from the composite (unless you use CSYNC obviously) hence why the expected range.

If you don't have 330 you can try something bigger (say 470) or smaller (say 220) to see how it goes and yes the variable pot you have does the same thing.

With regard to VGA it says usually VGA monitors are terminated with a 2K omh so you need to adapt the signal accordingly.

 

So told we are assuming the CSYNC you have right now is at the wrong level, and at that that it is TTL, not sure that to be the case with VBXE you should ask the guy that made them so you have better info to go off of.

 

[as an extreme example, I have an XRGB mini and over its RGB input some consoles gave me all sorts of trouble wrt syncing, in specific the AES, in the end I mounted a sync stripper (LM1881) over composite and I can now switch among original composite as sync or sync stripped (which from that link appears to be TTL as I used an LM1881) .... needless to say the sync stripped version is what I use the most and saw no issue .... and no extra resistor there either .... I even tried it on a Scart based TV I have around and it works there too .... but maybe as the article says it's a disaster waiting to happen]

 

Yeah - going from that chap it sounds like SCART really demands the 1V input.

 

In the case of the VBXE install, the mod itself does not provide a CSYNC signal itself and you need to tap it from Pin 15 of the 4050 buffer chip. If I - carefully! - put my multimeter across that pin to ground I get 4.45V which clearly is not going to do my TV any good!!! Somewhat worryingly I used it like that for almost 6 months with my old 800XL with a VBXE, although not every day nor for 24 hours... Hopefully I have not permanently toasted anything!!! However I do think this too high voltage was to blame for a really intractable seeming problem I encountered with my old hardware.

Edited August 26, 2016 by morelenmir

[flashjazzcat]

This is probably why Candle said the optimal hook-up point for CSYNC is the composite video pin of the legacy video jack. I doubt 4.45v will toast anything, though, otherwise he would also have stipulated a resistor when describing to me the gated sync taken directly from CD4050. I have a bunch of older LCDs here which sync solidly with unattenuated output from CSYNC, but a couple of slightly more modern displays (including the LG M227WD) which appear to be a bit more picky and respond well to an inline resistor.

[phoenixdownita]

 

Yeah - going from that chap it sounds like SCART really demands the 1V input.

 

In the case of the VBXE install, the mod itself does not provide a CSYNC signal itself and you need to tap it from Pin 15 of the 4050 buffer chip. If I - carefully! - put my multimeter across that pin to ground I get 4.45V which clearly is not going to do my TV any good!!! Somewhat worryingly I used it like that for almost 6 months with my old 800XL with a VBXE, although not every day nor for 24 hours... Hopefully I have not permanently toasted anything!!! However I do think this too high voltage was to blame for a really intractable seeming problem I encountered with my old hardware.

 

 

This is probably why Candle said the optimal hook-up point for CSYNC is the composite video pin of the legacy video jack. I doubt 4.45v will toast anything, though, otherwise he would also have stipulated a resistor when describing to me the gated sync taken directly from CD4050. I have a bunch of older LCDs here which sync solidly with unattenuated output from CSYNC, but a couple of slightly more modern displays (including the LG M227WD) which appear to be a bit more picky and respond well to an inline resistor.

 

As FJC states (and as in the manual for VBXE wrt SCART connection), use the composite instead of CSync and see what happens, the image may not be as clear or it may be depending on how clean the composite is to begin with. You can always add an LM1881 circuit on that signal to clean it (it's a pretty trivial circuit really) and if you live in Scart land they sell them premade to be wired inside the SCART connector (http://retrorgb.com/syncinscart.html) albeit without the extra resistor to attenuate the signal (which I have not had a problem with insofar)

[morelenmir]

Both excellent points chaps!!! If I recall correctly I think I did use the composite signal when I first installed my first VBXE, but then swapped to the 4050 output as I was trying to eliminate all the potential sources of patterning - which eventually turned out to be down to something else entirely! I might give it another try on this motherboard though, just to see if it obviates the need for an inline resistor altogether.

[flashjazzcat]

The primary reason I never use the composite pin is that I routinely remove the RF modulator, and on the XE line I believe this kills composite video.

[Dmitry]

It may be unrelated to this discussion but I have a monitornthat if I hook it up to a Tandy coco 3 which is 15khz rgb, it works great. Also works with an amiga. Will not work with vbxe.

 

I have a scan doubler that is for hd15 rgb out. Works just fine with the vbxe and the output is beautiful. Still, what about the vbxe versus tandy coco could be different? Both are ntsc ...my understandingnis the pal might have 15.6 versus 15.75Khz. I dont know exactly what vbxe does but that monitor is usually quite forgiving

[morelenmir]

Okay guys, not to reopen the endless 'what is causing my crap display' debate - trust me, I am done with that torture! - here is the screen that using Composite Video for SYNC gives me:

 

- Very clear zigzag corruption and a very familiar text effect.

 

Needless to say its back to gated CSYNC for this machine!

 

One interesting point I will mention; clearly the superior capabilities of the VBXE allow the underlying corruption which has plagued this machine since I bought it to be finally seen. The effect those zigzag lines create under the text is clearly the same as what was corrupting text and graphics produced by the native 'Monitor' composite video output.

 

In regards the matter at hand, I'm next going to try a simple potential divider to get the CSYNC output down to SCART-compliant ~1V levels and see what that gives me.

[flashjazzcat]

What's the make and model of the TV, so we can blacklist it?

[morelenmir]

What's the make and model of the TV, so we can blacklist it?

 

Its a so-called 'curry's essentials 16'' HDReady LED iDTV' with the more informative model number 'C16LDIB11'.

 

On the whole it isn't a terrible little unit, but it does lack an SVideo input and of course has really struggled with the 130XE display signals - if it is to be blamed for some of the problems I have encountered. It didn't really give me any trouble at all with neither the 'Monitor' nor VBXE RGB output from the 800XL - although I think even back then it didn't like the high-voltage CSYNC.

[phoenixdownita]

In regards the matter at hand, I'm next going to try a simple potential divider to get the CSYNC output down to SCART-compliant ~1V levels and see what that gives me.

Last would be an LM1881 on the composite, I was surprised of how clean the sync is after that.

I actually questioned why the XRGB mini does not have that part integrated (it's a <1$ chip) given it costs 300US$.

 

I'm afraid a few consumer electronic takes the sync signal (composite or otherwise) and do subpar cleaning/filtering on it, so much so that when they recombine it with the chroma/intensity info from RGB there's crosstalk and some of the crap in the composite (used as sync in this case) bleeds over.

 

The sync signal per se should have zero impact on luma/chroma as it is just a series of pulses to signify start-frame/end-frame start-line/end-line and little else. Clearly you see an interference pattern, likely your TV expects a much cleaner signal, not sure if the gated CSync of VBXE is that clean, but worth a try with the voltage divider.

[morelenmir]

Last would be an LM1881 on the composite, I was surprised of how clean the sync is after that.

I actually questioned why the XRGB mini does not have that part integrated (it's a <1$ chip) given it costs 300US$.

 

I'm afraid a few consumer electronic takes the sync signal (composite or otherwise) and do subpar cleaning/filtering on it, so much so that when they recombine it with the chroma/intensity info from RGB there's crosstalk and some of the crap in the composite (used as sync in this case) bleeds over.

 

The sync signal per se should have zero impact on luma/chroma as it is just a series of pulses to signify start-frame/end-frame start-line/end-line and little else. Clearly you see an interference pattern, likely your TV expects a much cleaner signal, not sure if the gated CSync of VBXE is that clean, but worth a try with the voltage divider.

 

I will certainly look that device up and if there is a through-hole version maybe give it a shot on some strip-board.

 

I also think you are right about the 'unclean' sync to some extent as well. As I mentioned above I have struggled for seven months or more with terrible display on my 130XE and discussed it at length on several threads. The VBXE was my last dice throw - a fairly safe if pricey one! - to finally sort it out. Luckily for the most part it has and this sync issue is the only problem I have left to fully smooth away. However given the SYNC signal is generated purely from the Atari itself and not the VBXE, therefore no doubt to some extent afflicted with DRAM noise/bad capacitors/poor HF motherboard design/whatever the hell else the Tramiel-era Atari engineers managed to screw up in designing the XE's I was still a little bit cagey. Thankfully the variable resistor has largely cured the issue and hopefully reducing the voltage fully - via a crude PD divider or LS1881 - will finally put the top hat onto this fix.

[phoenixdownita]

 

I will certainly look that device up and if there is a through-hole version maybe give it a shot on some strip-board.

 

Even if you don't find a thru hole (they make it) you can use a SOP8 to DIN adapter:

http://www.ebay.com/itm/10pcs-SOP8-SO8-SOIC8-TO-DIP8-Adapter-PCB-Converter-Board-Single-Row-Pin-Header/401045515775

 

Actually soldering the SMD version of the LM1881 onto the adapter isn't that bad (I use a normal 15W fine point soldering iron and small movements).

Keep in mind that even the output of the LM1881 may require rescaling.

[morelenmir]

A good point! In fact that LM1881 circuit will make an interesting little project all by itself.

 

TI LM1881 Video Sync Seperator - Datasheet.pdf

 

Moving back to the sync problem for a moment; I have managed to tweak the variable resistor with sufficient precision to almost totally remove the blank screen at 40 column resolution. However I have noticed that at the same setting, but with "RC_VBXE.SYS" and "CON" providing the fantastic 80-column display which the VBXE allows I do get much more frequent blank screens. Sadly - perhaps because at 500kohms the resistor I am using is so comically over-size, therefore very difficult to adjust with precision - I cannot quite set the sweep to the point that will elliminate the black screens at this resolution while still allowing the display to appear at reboot. By the last comment I mean if I set the resistance too high at any resolution and then power-cycle I find I get no screen at all and most readjust the resistance down slightly - again tricky with my setup - until it works again. Obviously if the potential divider does not do the job then I will need to get a smaller variable resistor. I find the system works almost perfectly at ~8.5-9.5kohms - so the 500Kohm unit I am using is totally unfit for purpose. I think a 20kohm potentiometer would be perfect.

Edited August 27, 2016 by morelenmir

[morelenmir]

Another bit of experimentation returns an interesting result.

 

While I have had a pretty solid 40-col display with ~8.5kohms in series with CSYNC, I could not dial this down any further for a stable 80-cols. However this resistance is far, far higher than that suggested by the data presented by phonixdownita above, that is to form a 1V potential divider with the built-in termination resistor of my television at likely 75ohms. While I don't have the ideal 330ohms this model would require, I do have a very big bag of - literally! - five thousand 200ohm units that I 'inherited'. So, if we run with the 75ohm standard for the TV and having measured my XE's CSYNC at 4.43V that means using 200ohms will tap off a reduced figure of 1.63V - if I am remembering how to do these calculations properly! So, not the ideal 1V, but probably a lot better for my television nonetheless. Moreover if I put this in practice and solder a 200ohm resister in series with CSYNC I get.. A perfectly stable screen at both 40 and 80 cols for the last half hour!!!

 

Weird to say the least! I don't understand therefore why the huge 8.5kohms setup ever worked at all.

 

Update:

 

Scrub that. Just with a little continuing suspicion on this I thought I would check the resistor value with my multimeter... I think now I know why my father got such a large job lot of components for practically nothing; the supposedly 200ohm resistor actually measures 4.7kohms! What's worse I'll have to check a good sampling of the others in the box as well as I don't think they are all miscoloured or miscoloured to the same value... Groan...

 

Anyway, what I can actually state is that with a fixed resistor of 4.7kohm the display seems stable at both 40col and 80col settings. What that in turn says about the termination resistor on the SCART CSYNC pin of my TV I don't know. Other than perhaps it is a ~1kohm unit to give a little under 1V divided!

Edited August 28, 2016 by morelenmir

[phoenixdownita]

....

So, not the ideal 1V, but probably a lot better for my television nonetheless. Moreover if I put this in practice and solder a 200ohm resister in series with CSYNC I get.. A perfectly stable screen at both 40 and 80 cols for the last half hour!!!

 

Weird to say the least! I don't understand therefore why the huge 8.5kohms setup ever worked at all.

 

You can get a 300 by doing 1 200 series with 2 200 parallel like this:

 

|--200--|

>--| |--200--> (the 2 200 in parallel amount to 100 which series the 200 gives you 300)

|--200--|

 

(you can get an extra 33.3 ohm via 6 200 parallel but at that point I believe it is overkill, 3 parallel 100 does the same 33.3 btw).

 

Wrt to the 8.5K it is possible the TV decided to amplify the signal, some TVs have complex conditioning input stages so you never know.

The worse is when TVs have sensing circuits that sometimes decide to do pass-thru and sometimes not.

 

How about the interference pattern you saw with the composite as sync, is it gone now with CSYNC + voltage divider ?

 

EDIT: I have an LCD TV that can stomach the composite out of a 7800 LHE mod just fine (little blurry), but the SVideo from the same LHE mod makes it lose sync because on the SVideo the TV decides to do almost no conditioning and obviously the 7800 signal is slightly out of spec, obviously the SVideo looks stunning and I mean for those 5 secs that it can display before losing sync and then again and again like stuttering

[Rybags]

A bit of a *Bump* on this one.

 

I ordered a GBS 8200 and it arrived today. Getting sick of having the Amiga 1084 take up so much space despite just being a dust-collector most of the time and the video board was all of about $20 so why not?

 

So far I've tested it with a couple of my old DTV tuner boxes and old DVD player into component. Somewhat amazed that it seems to work fine with PAL devices and in turn it works on my oldest LCD monitor. I was expecting trouble in that regard but it looks like this thing frame-buffers regardless of what's coming into it, then outputs at 60 Hz.

 

Not overly impressed with the video quality though I was using a 12V 1 Amp PS but have since made up an adaptor which will let me use my 5V 2 Amp PS with it. Additional to that, I just have the bare board sitting on an open exercise book exposed to all the RFI going on around it.

 

 

I am concerned though about the CSYNC voltage, I read in a blog elsewhere with an Amiga implementation that a 680 Ohm resistor in series is suggested so might try that. Ref: https://ianstedman.wordpress.com/gbs-82xx-experiments/

 

I remember from way back in my VBXE install, I deliberately used a 15 pin VGA type output on the Atari since I have a cable ready made for it. Haven't checked yet though if I have the CSYNC going to the required pin, I did my install to suit the 15-9 pin cable that I had so it'd work on the 1084S.

 

Once I get this up and going I'll put up comments + pics.

Edited December 13, 2016 by Rybags

[flashjazzcat]

I now use a 330 ohm resistor straight off the sync pin of CD4050 in all installations and it appears bullet-proof.

[Rybags]

OK... I've got 2 obstacles to overcome. My current setup has the CSync going to pin 8 of the 15 pin output which in the cable I used is fed to the proper pin on the 1084.

Problem with that is that if I use a straight up VGA cable it gets sent to B-GND which would be a bad thing.

 

So the solution I should think would be to just build an adaptor which takes the signals off the 9-pin end of my 1084 cable then just feeds it to the CGA/EGA header on the board, and put the required resistor in it as well.

Edited December 13, 2016 by Rybags

[flashjazzcat]

That's the best way. I have a GBS-8200 here and when I boxed one up for someone, I wired a DB9 to the CGA/EGA header on the board. This way, the thing worked with a stock 1084S cable (in my case, ST-compatible DIN13/DB9).

[Rybags]

Yep, agreed. Found my VBXE -> 1084 cable and rather than messing with stuff that's already done and making something that'll work on one thing but not the other, just go the proper way.

I'm hanging out to give this a try, dug my proto pegboard thingy out so might just see if I can do a test run that way.