Forums

Sega Master System / Mark III / Game Gear
SG-1000 / SC-3000 / SF-7000 / OMV
Home - Forums - Games - Scans - Maps - Cheats - Credits
Music - Videos - Development - Hacks - Translations - Homebrew

View topic - RGB - again

Reply to topic
Author Message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
RGB - again
Post Posted: Tue Aug 29, 2017 9:19 am
Hi,

there seems to be some confusion about how to correctly connect a SMS via RGB to a TV, especially with modern TVs.

When problems occur, people seem to sometimes quite randomly add resistors/caps or connect/change pins.

I tried to dig into it to do it right the first time when I finally use the RGB out. But the confusion even grew a bit :). Perhaps you can help me to clarify this:

The CX1145 has C-Sync out (pin 11) and C-Sync in (pin 10). According to the CX1145 datasheet (block diagram), C-Sync in is an essential input to generate the C-Sync out in the first place (and also Chroma out and Y on pins 15/16).

My questions to pins 10, 11, 15, 16:

    why is C-Sync out (11) not connected to anything? Seems to be the "better/cleaner" sync-signal when using RGB.

    C-Sync in (10) is connected to a logic circuit whose output is build from a signal from the CN1-connector (DIN connector), but this is labelled "C-Sync out" in the service manual. Shouldn't this be labelled as an input?

    Assuming the "C-Sync out" on the DIN-connector refers to an output from the TV, I would think it belongs to pin 19 of the SCART-connector (video sync out). Is this correct?

    C-Sync out (CX pin 11) is not used. The other pins which are dependent on C-Sync in (pin 10) are Chroma out and Y (pins 15/16) which seem to be used to generate the C video out (pin 20). So everything (correct sync for both RGB and composite video) depends on the C-Sync input (?) from the TV?


Some further questions arise from the suggested connections for SCART:


    the RGB switching voltage is taken from CX-pin 12 via a 100R. To set the TV to RGB mode, you have to supply 1..3V on SCART-pin 16. I assume, the 100R together with the TV-internal 75R creates a voltage divider which breaks down the 5V from CX-pin 12 to about 2V. But this means, you always have 30mA of current flowing through the voltage divider. I'm not sure if this is OK for the frequently blowing (*) 7805 voltage regulator...


(*) Many seem to have this problem. I'm not sure if this is the result from connecting the input voltage wrong (not using center negative connectors, NES power supply etc.) or if the 7805 works thermically in the red zone by design.
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
vertical stripes
Post Posted: Sat Sep 02, 2017 10:22 am
Now I have a RGB cable, but I have a display issue (present on 2 different SMS1, so the problem comes from the cable or my TV). See attachments. Does anybody know this problem?

Edit:
My (off the shelf) RGB cable connects the same way as the common SMS/Genesis cables do, but nevertheless I'm not sure this is ideal for all:

    Sync ist not connected => I don't have sync issues, Sync via Comp Video is fine, but for some people this may be a problem? See my initial posting.

    R, G and B have a RC circuit. Is this really neccessary since the SMS already has one on the RGB outputs (75 Ohm and 220µF)?

Stripes_HangOn1.jpg (3.52 MB)
vertical stripes with RGB (1/2)
Stripes_HangOn1.jpg
Stripes_HangOn2.jpg (3.3 MB)
vertical stripes with RGB (2/2)
Stripes_HangOn2.jpg

  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Sun Sep 03, 2017 2:31 pm
"Csync Out" is output to TV.

Anyway you are not supposed to use the Csync signal coming from VDP that goes into the CXA and AV out with a regular TV, because that signal is not meant to drive a 75 ohm load, in addition to this, it is an open drain signal. Half the problems with sync originate from this fact.

The Csync out of CXA is the much better option to use since it is nicely buffered and push-pull type, capable of much greater load without distorting itself. With MD it isn't an option as 32X will not be able to work with the Csync coming from CXA and won't produce correct output.

What you are supposed to do is use Cvideo as sync (because it is 75ohm load capable) and you are supposed to use a cable that has all the signals inside individually shielded, for example like how VGA cables work, this will remove any sort of signal crosstalks, so you have clean video and no buzzing in audio. All signals in one shield is about as bad as no shield at all.

Datasheets show exactly how the pins are supposed to be used. In all cases you need 75ohm resistor and 220µF or higher capacitor in series with each signal that goes to TV (RGB, Composhit and Csync).

The vertical lines you see are caused by nearby VRAM signals and color subcarrier frequency leaking into the RGB lines as they exist the VDP and find their way toward CXA. The signals are then conditioned into 0.7Vpp for conversion to compsohit and 75ohm capable RGB out.
Without buffering that the CXA does on the RGB lines you will be experiencing a whole lot more interfence and crosstalk in the video due to the signals being high impedance (weak).

To get rid of these lines you will have to trasnsport the RGB signals from VDP to rest of the machine in shielded lines or otherwise physically located away from other parts. It will also help to compactify the divider circuit, because smaller size = less area and loops to pick up interferences with.
MDs suffer from this problem also but to lesser extent, usually only problem is color subcarrier leaking into blue signal as they are right next to each other.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Sun Sep 03, 2017 4:35 pm
Hi TmEE,

thanks a lot for your detailed reply. Some questions are still open, though:

TmEE wrote
"Csync Out" is output to TV.

Then why is it connected to an input of a digital circuit? The signal goes through IC10 and finally reaches "C-Sync in" of the CX. (red markings in my attachment)

TmEE wrote
Anyway you are not supposed to use the Csync signal coming from VDP that goes into the CXA and AV out with a regular TV

The VDP's C-Sync is not connected to the outside world and I didn't want to use it anyway (green curcuit in my attachment).

If I understand you correctly, I have 2 options to solve my issue:
- using a cable with proper shielding (not sure if it has already)
- cut the RGB-traces between VDP <-> CX and rewire with shielded cables?

Using C-Sync-out of CX instead of Composite Video as sync would be "better" but only if I had sync issues (which I don't)?

TmEE wrote
Datasheets show exactly how the pins are supposed to be used. In all cases you need 75ohm resistor and 220µF

That's right and the SMS1 has this RC-circuitory. But why do all the off-the-shelf-RGB-cables you can buy have an additional R75/220µF?
C-Sync.jpg (303.92 KB)
C-Sync.jpg

  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Mon Sep 04, 2017 2:23 am
Quote
Then why is it connected to an input of a digital circuit? The signal goes through IC10 and finally reaches "C-Sync in" of the CX. (red markings in my attachment)

The VDP's C-Sync is not connected to the outside world and I didn't want to use it anyway (green curcuit in my attachment).


There's a little dot in the intersection of the red and green highlights, making both of them the same line.

Quote
If I understand you correctly, I have 2 options to solve my issue:
- using a cable with proper shielding (not sure if it has already)
- cut the RGB-traces between VDP <-> CX and rewire with shielded cables?


First will not have any effect if the problem originates from within SMS, and I'm not completely sure it comes from there. From the photos the lines seem to affect all fairly colors equally, suggesting the cable is to be blamed. VDP to CXA problem tends to manifest in blue color, as it is closest to the main pollutant which is color subcarrier signal. Inside unshielded cables the color subcarrier (as part of composhit) will end up leaking into RGB lines and causes similar problem, but on your photos it looks really really pronounced, cable crosstalk shouldn't be that pronounced. Do a test, while the machine runs touch the RGB signals between VDP and CXA with your fingers and see if the problem worsens, touching different parts with another finger can induce different patterns.
It could also be that you have sharpness filter turned way up and it can really make any artifacts really accentuated.

Quote
That's right and the SMS1 has this RC-circuitory. But why do all the off-the-shelf-RGB-cables you can buy have an additional R75/220µF?


The AV cables sold are for MDs, which do not have the resistor and capacitor on the RGB lines. I actually haven't seen any SMS with those resistors and capacitors, the two I have here do not have them. You should bypass the ones in the SMS, having the extra ones will result in image that is darker than it should be.

Quote
Using C-Sync-out of CX instead of Composite Video as sync would be "better" but only if I had sync issues (which I don't)?


It is a thing to try, CRT TVs are not gonna care but LCDs and other things might. You should try the Csync output on CXA, if it cures the vertical lines then it is 100% certain that cable is the problem. If the situation improves but some lines remain then those are caused by VDP <> CXA path.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Mon Sep 04, 2017 7:19 am
Hi TmEE,

TmEE wrote
There's a little dot in the intersection of the red and green highlights, making both of them the same line.


Oops, you're right! I looked for the dot several times explicitely but I couldn't see it. Probably my printout absorbed it ;). In my defence the VDP line seems discontinious as well around the intersection which led me to beleive it goes under the other one... Anyway, that clears up a lot, thanks :).

I'll have to do some experiments (and probably I'll have to make my own cable)... I'll report back!

[Edit: I'll start with shorting out the RCs in the current cable. If they lead to a 50% drop of the signal strength and my TV compensates with some AGC (because I don't see any darkening), it amplifies any noise as well]

Bye
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Tue Sep 05, 2017 6:30 pm
Removing all RCs and connecting all grounds didn't change a bit unfortunately.

Next step is building a complete new cable with shielded RGB...

I wonder, what everybody else uses to connect the SMS to the TV via RGB.
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Sat Sep 09, 2017 2:17 pm
Here some further results:

creating my own cable

I used a cable where all signals are shielded separately, all grounds are connected, Sync through Composite Video, no additional RCs. => Result: not the slightest change (compared to off-the-shelf-cable w/ or w/o RCs)

using the cable from retrogamingcables.co.uk

This should create a clean CSync from the comp. video, so I don't have to rewire my SMS. The image still looks exactly the same (vertical stripes).

As this occurs on two different SMS1 and most other people don't get these stripes (regardless which cable they use), I'm quite sure my TV is the problem. @TmEE: do you aggree?

To confirm this, I will try to loop the signal through other components inbetween (I don't have another TV). Hopefully, I find a way to do so...

Argh, I'm not concerned about any noise, but these stripes pretty much destroy the whole experience...
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Mon Sep 11, 2017 7:29 am
Since the stripes have not disappeared the problem does seem to be coming from the TV. If it still does the same thing with other devices there's a chance the VDP is faulty.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Mon Sep 11, 2017 7:43 am
As I mentioned I tried 2 different SMS, so I'd rule out the VDP.

Unfortunately I didn't find any other device with SCART/RGB input. Have to ask a friend.

Maybe I also hook up my (cheapo) oscilloscope and have a look, if each line has these peaks between the pixels? Because they're bright, they should appear on any of R, G and B - assuming, the source of these stripes is the SMS in the first place.
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Mon Sep 11, 2017 12:11 pm
The scope will be able to show the lines if it has 10MHz or more bandwidth, but even on a weaker thing you should see wiggles on top of the lines. Solid colored screen should be the best way to test this, then the scope can trigger on any video line and still show same waveform.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Mon Sep 11, 2017 12:48 pm
I've got 20MHz, that should suffice.

But I kind of doubt to see anything, because removing the RC circuit didn't change anything... Well, we'll see :).
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Tue Sep 12, 2017 12:10 am
I made a small program, where I can cycle through several background colors (black, dark gray ($15), light gray ($2A), white) by pressing button 1.

With black and dark gray background the stripes are visible on the screen just as in my Hang Out pictures above. Light gray and white both look maximum white, too bright to see stripes (but they are still visible in the black part of the screen (border)).

I used the cable without RC... maybe the signals are too high now so I don't see any difference between light gray and white? The screen changes somehow, but not in brightness.

Anyway: I attached my scope to Red and Blue of the AV-Connector while still connected to the TV (thinking about it now I better should have measured the sync). See the measurements attached to this post (channel 1: blue, channel 2: red).

Black: 1.8V with noticeable ripple on blue and way less ripple on red.

Dark Gray: rectangular signal 1.8V-2.4V with quite noticeable ripple on blue, better again on red.

Light Gray: rectangular signal 1.8V-2.7V with quite noticeable ripple on blue, better again on red (but with some clipping?).

White: rectangular signal 1.8V-2.8V, the ripple is substantially cut off at 2.8V (not the 1.8V-part of the rectangle), signal seems the be maxed out...

Any enlightments so far :)? It seems the ripple is really on the line.

I'll try to redo the measurements with a different cable (Retro RGB) as well as disconnected from the TV (to exclude some feedback from the TV which I still suspect to be at least part of the problem).
0_black.png (78.69 KB)
Blue/Red with black background
0_black.png
1_dark_gray_a.png (80.22 KB)
Blue/Red with dark gray background
1_dark_gray_a.png
2_light_gray_a.png (81.41 KB)
Blue/Red with light gray background
2_light_gray_a.png
3_white_a.png (78.06 KB)
Blue/Red with white background
3_white_a.png

  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Tue Sep 12, 2017 1:44 pm
For next test measure output of the VDP directly. If that is relatively clean then maybe the problem is power supply capacitors, you can scope the power rail and see if there's any ripple on top of it also.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Tue Sep 12, 2017 2:31 pm
Thanks for your suggestions!
I already changed all the capacitors, but in my next test session I'll check these two things as well.

Any thoughts on white/"light gray" being to same brightness on the screen?
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Tue Sep 12, 2017 5:21 pm
Next measurements are done...

First: the additional RC seems to be needed, because using them lets me distinguish between light gray and white again. I understand that the RC lowers the output signal, but why is the internal RC not enough? How were the original cables build back then?

Now to the measurements:

- Voltage is stable on output of the 7805, very clean, no ripple.
- Blue signal on AV port has the ripple no matter if I connect any cable. Turning off TV and other nearby appliances didn't change it either.
- The Blue from the VDP seems much clearer, see my attachment (yellow: blue on AV, green: blue from VDP).
Blue_on_AV_and_VDP.png (64.96 KB)
Blue on AV (ch1) vs. VDP (ch2)
Blue_on_AV_and_VDP.png

  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Tue Sep 12, 2017 5:21 pm
If you have resistor and cap in the console you don't want one in the cable, or if they're in cable (preferred) you don't want them in the console.
Having none at all will explain the while and grey being crushed, as the signal is way higher than it should be and will overload the input and TV will clip. If that's not the case then TV settings are wrong...
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Tue Sep 12, 2017 5:46 pm
The RCs are in the cable (now), but I didn't put any additional RC in the SMS, according to the schematics they're already there (*). That's why I removed them from the cable in the first place, but obviously this made everything worse. Also, I didn't hear of anyone removing their RCs from the cable...

(*) Checking again, I must have seen different (wrong?) schematics, the one I have now (from the service manual) only has RC on the Composite Video, not on RGB. Well, at least that question is solved now.

Back to the main topic :): The Ripple. I will check again directly on the CX's "R/G/B in" and "R/G/B out", but as there's nothing inbetween VDP and CX, the ripple is generated inside the RGB encoder (CX)?
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Wed Sep 13, 2017 7:34 am
Most SMS do not have the resistors and caps in them on RGB lines. What is the PCB number (171-xxxx) and model (big text somewhere) of yours ?

You'll want to measure VDP output too. VDP out, CXA in, CXA out. There are some resistors and capacitors between VDP and CXA. I can imagine one failure mode, the RGB signals have some resistors connecting to same spot, and if the spot goes to something like a via and that via is damaged you can get huge crosstalk that could possibly cause those bars. Also when you measure, do it with AV cable plugged and unplugged, there might be a difference.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Wed Sep 13, 2017 11:07 pm
Just measuring again, some interesting findings occured!

Regarding your question:
I've got a Power Base VA2/PAL, PCB number 171-5672

Layout and part names are almost spot on to the service manual, I found only an additional ecap so far.

To my measurements:

Ripple appears on RGB in and out of the RGB encoder, not on VDP out.

At the beginning I only selected the 1:1 ratio on my probes. This had the effect, that the ripple disappeared noticeable in the region of the CX's RGB in. Aha!

With 1:10 ratio, I continued measuring R15 and C17, presumably R and C on the Blue input of the CX. Result: Signal is good on VDP-side of R15, but gets bad on the other pin (the point between R15 and C17).

Forgot to check R16, though (pull-down to gnd).

Bad soldering? Bad Cap? It's not an ecap, so I didn't change it when I "refurbished" the board.
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Thu Sep 14, 2017 8:29 am
This is good info, very much narrows down the search area. Can you take a photo of both sides of the board around the CXA ?

I attached an image that shows the special attention areas. The GNDs in red must have a nice connection, if something like a via doesn't conduct very well (some Sega boards use glue vias which are absolutely terrible) you can get the problem you experience. In that case you'll want to run an additional GND via a wire to a known good location such as near 7805 or AV out or DC input.

The red cap on Vref pin will also make sure that CXA works properly, if the board lacks it you will have to add one. Higher value will not cause problems, you can even reuse the removed capacitors from RGB lines, and while at it stick one on the VCC rail too. Vref is the internal reference, if it moves around even slightly you'll see it on the image, and the capacitor makes sure it cannot move around as much.
CXA.png (135.7 KB)
CXA.png

  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Fri Sep 15, 2017 6:22 pm
Hello again,

did my next measuring session, which was some kind of rollercoaster ride :)...

[In both attachments, the yellow signal always is the Blue output on the AV connector.]

I'm quite confident my grounds are solid. Measured them and connected some manually.

Vref on the CX is fine as well, additional ecap didn't change anything. Checking the scope output (see Vref.png, green signal), it's not perfectly smooth, though.

Going back to the Blue output of the VDP (see vdp28.png, green signal, but you see it in one of my older posts as well), you see a ripple there already, it's not so noticeable because of the scale (yellow/green have the same ground level in my scope window!).

Both Vcc on the VDP are rock solid and also the grounds.

Went back to the circuit at the RGB-inputs of the CX, again. Playing around with some capacitors resulted in some strange behaviour which was not quite reproduceable. Sometimes just measuring voltages in that area with my multimeter changed the picture at bit.

Maybe I should replace C11/C14/C17?
vdp28.PNG (78.6 KB)
VDP pin 28 (Blue)
vdp28.PNG
Vref.png (75.42 KB)
Vref on CX
Vref.png

  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Sat Sep 16, 2017 1:07 am
update:

I replaced C11/C14/C17. The problem still persists, but my measurement is a bit more consistent. To sum the situation up:

VDP RGB-out looks fine, on VDP-side and R9/R12/R15 (coming from VDP). Signal starts to get messy immediately after these resistors.

The ripple appears on almost any CX-pin except grounds and Vref. It looks very similar to the clock input - maybe this signal interferes?

On my screenshots the ripple seems to have a quite low frequency, that's just a display issue, when zooming in you see it's quite similar to the clock input (about 4,6 MHz). Maybe the clock is interfering?
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Sat Sep 16, 2017 5:35 pm
Yes, the subcarrier (4.433619MHz) is what you see as the interference. Why it is everywhere is the question and which is why I was thinking of the grounds/power supply and asked for a photo of the CXA and surrounding area. Disconnecting the signal will most proboably remove the interference but you also lose color in the composhit signal. RGB will be unaffected if the TV isn't in some intermediate state where it tries to show both RGB and composhit at the same time (can happen when RGB enable voltage isn't right).
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Sat Sep 16, 2017 10:18 pm
Sorry, completely forgot the pictures.

Top picture:
Alle the e-caps are new.
C11/14/17 are new as well.

Bottom picture:
For orientation: 7805 on the left, CX is in the right half of the picture relatively close to the center.

The soldering mess was not done by me :D, I just did the caps. Somebody really "had fun" there :-/...
CX_area_top.jpg (3.93 MB)
CX area top
CX_area_top.jpg
CX_area_bottom.jpg (3.84 MB)
CX area bottom
CX_area_bottom.jpg

  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Sun Sep 17, 2017 6:02 am
Ok, that board has good layout and proper vias, none of the glue crap.

Disconnect the subcarrier from the CXA and see what happens. light gray and white being same is still not right though...
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Sun Sep 17, 2017 9:01 am
The light gray/white issue is gone as I wrote before, this was because of the missing RCs in the other cable.
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Sun Sep 17, 2017 9:12 pm
I removed R20 which connects the clock to the CX pin 6. No change :( (on RGB output anyway, the XO in is clean now). I've chosen R20 because I didn't want to cut any traces.

But maybe on thing is interesting:
Pin 5 (XO out, which is not connected at all) of the CX still has the noise...

Find attached 2 screenshots:
Pin 5 and 6 of the CX. Yellow signal as always Blue out on the AV connector (this time the time scale is smaller to see the noise much better).

Additional note:
The problem with the stripes is also visible when the SMS is completely assembled.
All my troubleshooting/measurements were done with the bare PCB, though, for convenience. Maybe the shielding of the original case has some influence as well? Perhaps I should keep that in mind as well.


  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Wed Sep 20, 2017 1:36 pm
No ideas?

My other SMS1 has exactly the same problem, so I wonder if anyone else has this problem as well? Or is it faulty by design and I really have to mod the PCB (separate, shielded connection from VDP to CX)? I'd prefer not to do that.
  View user's profile Send private message
  • Joined: 01 Feb 2014
  • Posts: 849
Reply with quote
Post Posted: Wed Sep 20, 2017 3:59 pm
Knaeckebrot wrote
My other SMS1 has exactly the same problem, so I wonder if anyone else has this problem as well?

I can see the same effect ever since I use an RGB connection for my SMS. I suspect it's always there, but gets blurred out if you connect the SMS via composite.
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Wed Sep 20, 2017 4:34 pm
Thanks for this info! So I definetely no longer have to look for an error (bad soldering/caps), but have to completely redo the connection between VDP and CX.

Since disconnecting the 4.4MHz from the CX without any effect, I wonder, if the 4.4MHz really are the source of the interference, here's why:

- I didn't measure 4.4MHz, but more something like 4.6MHz

- the stripes are exactly aligned with the pixels (256px per line => 256 peaks/52µs => 4.9MHz => plausible?)

So it's interference of VDP-RAM-access or something like this?
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Wed Sep 20, 2017 6:59 pm
Pixel clock is 53.203424MHz / 10 = 5.3203424MHz. Color clock is 53.203424MHz / 12 = 4.433619MHz, and CPU clock is 53.203424 / 15 = 3.546895MHz.

XO is color clock output from the internal oscillator of the CXA, but the internal oscillator is not used in SMS, instead external clock derived from master clock is fed into XI so that there are no moving patterns on the image.

I don't get anywhere near as strong jailbars in any of my SMS, barely visible over RGB even on a good TV. Shielding the RGB lines cures them completely. The RGB lines can pick up noise from VRAM activity nearby, and it has pixel clock frequency to it. Put a finger on pins of one of the VRAM chips and another on the RGB lines and you'll get exaggerated view of them.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Wed Sep 20, 2017 11:11 pm
Hi TmEE,

by the way: thanks for your many tips, you're a great help :-). Hopefully, soon I won't have to bother you and the forum with this problem any further :).

You say you don't have so strong vertical bars on any of your SMS, but you also said you shielded the RGB lines. Without shielding (as originally produced, no mod at all), would you encounter the same issue?

What's the best/easiest way to shield the RGB-lines (is there a tutorial?)? Do I have to cut traces or does removing R9/12/15, R10/13/16 and C11/14/17 do the job (while of course adding new, shielded connections between VDP/CX including R/Cs). I'm not sure if the open traces will pick up noise again.

Bye
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Fri Sep 22, 2017 8:51 am
Without shielding there's very faint jailbars that are only visible in blue channel. With shielding the problem is completely gone.

As far as shielding goes you cut the RGB traces as close as possible to the VDP and CXA and run short shielded cables from one end to another.
The unconnected traces can be wired to GND for additional protection.

This is what I did in one MD clone using a CD ROM audio cable from old computer : http://www.tmeeco.eu/BitShit/JailbarsFix.jpg
(I don't have any photos of SMS for the time being).

On VDP end the shields you connect to the GND pin nearest to RGB outs on VDP and on CXA end you connect it to the resistor network or CXA's GND pin.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Thu Sep 28, 2017 11:25 pm
I cut traces and rebuilt the connection from VDP to CX including shielding, not the slightest change unfortunately :-(.

I always wanted to leave the traces uncut, but now it even was in vain...

Very frustrating.

I might redo the troubleshooting from scratch using a good oscilloscope at work, but that's not that easy to arrange.

On the pictures from my original post the stripes don't look that bad, but in reality they're quite disturbing (depending on the game) in case you wonder why I bother so much.
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Mon Oct 02, 2017 12:16 pm
That is really strange...

Maybe it is easier to arrange a test at some other TV ? i.e make new neighborly relations ? lol
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Thu Oct 05, 2017 10:53 pm
My neighbor actually has a modern TV with SCART (a newer generation than mine), different manufacturer (mine is Samsung, his is LG). Today I made a short test:

I expected the same stripes or no stripes at all, but I got this:

Half of the pictures has stripes, the other half doesn't. More precisely: An alternating pattern of around 20 stripes and the same area without stripes (didn't make a photo)

my Samsung:
||||||||||||||||||||||||||||||||||||||||...
||||||||||||||||||||||||||||||||||||||||...
||||||||||||||||||||||||||||||||||||||||...
...


his LG:
|||||     |||||     |||||     |||||     ...
|||||     |||||     |||||     |||||     ...
|||||     |||||     |||||     |||||     ...
...


The SMS is cursed. Whenever there may be answer A or B, I get AB :-). Still don't know if it's a modern-TV-problem. Don't know anyone with a CRT to repeat the test.
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Fri Oct 06, 2017 1:15 pm
That very much suggests TV incompatibility...
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Fri Oct 06, 2017 2:02 pm
Perhaps a filter will do the business?

Composite works fine. I'd even use that if there wasn't that ghosting...
  View user's profile Send private message
  • Joined: 31 Oct 2007
  • Posts: 853
  • Location: Estonia, Rapla city
Reply with quote
Post Posted: Fri Oct 06, 2017 4:05 pm
You can try to add some capacitance in parallel to the RGB signals but you will start getting smear towards right at some point. Tens to few hundreds of pF range is worth to try.
  View user's profile Send private message Visit poster's website
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Fri Nov 03, 2017 4:48 pm
I finally managed to test both of my SMS1 on a CRT TV (Samsung CB-5373T). The picture looks beautiful, even composite signal is not that bad.

Seems that the old TV is not susceptible to the high frequency ripple on the video signal?

@TmEE: Where exactly should I put the pF-caps, RGB out to ground?
  View user's profile Send private message
  • Joined: 03 Aug 2017
  • Posts: 61
  • Location: Munich
Reply with quote
Post Posted: Mon Nov 13, 2017 2:27 pm
Hello,

I just connected SMS1 number 3 to my TV via RGB and guess what? Picture ist actually quite fine!

SMS #1 (3005-18-B, w/ Hang-Out) and #2 (3010, w/ Snail Maze):
bad on modern Samsung (even after hours of soldering), "half" bad on modern LG, perfect on old Samsung CRT.

SMS #3 (3005-18-B):
Fine on modern Samsung (didn't test the rest).

So we have both: "compatibility" issue with modern TV (too sensible to higher frequencies?) and a bad signal (which probably didn't matter back then).

Should compare the layout of both 3005-18-B models. But at the moment I prefer not to touch it and just play some games :-D.
  View user's profile Send private message
Reply to topic



Back to the top of this page

Back to SMS Power!