Analog NTSC to RGB converter?

[Linards Ticmanis]

Hello, do you maybe know whether there is such a thing as a purely
analog NTSC to RGB converter that one could purchase, either as a
complete device or at least as a kit or a ready-made PCB?

I found a schematic here:

http://www.seekic.com/circuit_diagram/A-D_D-A_Converter_Circuit/NTSC_TO_RGB_CONVERTER.html

but it is pretty complex and I don't know whether this doesn't exceed my
abilities. I've never laid out (or etched) PCBs myself, and I think this
schematic is beyond what one would reasonably build on a breadboard.
Plus, I'm really too lazy to build it.

The reason I'm asking is that I've recently got a Commodore 1084 monitor
in very good shape and it is really a great video monitor as such, even
80 characters per line text is really legible on it - but it is a PAL
model and thus useless for NTSC sources such as my Apple. It can sync to
60Hz without problems but of course it can't decode NTSC, so it will
display only black-and-white with the Apple.

Thanks for any hints.

--
Linards Ticmanis

[Michael J. Mahon]

Such a converter is essentially the same as the video section of any NTSC
monitor (since color CRTs are all RGB).

TV-inspired integrated circuits have reduced the package count
considerably, but there's still a lot of analog lifting required--such as
the delay line to match the luminance delay to the filtered chrominance
channel.

I would expect there to be a lot of surplus NTSC-to-RGB decoders available
on eBay--some with professional specs.

-michael - NadaNet 3.1 and AppleCrate II: http://home.comcast.net/~mjmahon

[Linards Ticmanis]

Thank you Michael for the usual good advice!

On 10/22/2012 04:25 AM, Michael J. Mahon wrote:

> Such a converter is essentially the same as the video section of any NTSC
> monitor (since color CRTs are all RGB).

I know - I have modified another 1084 for NTSC in the past (then managed
to destroy it shortly after that). I might even be able to jury-rig some
kind of switchable solution (using some small relays maybe?), but I'm
kind of nervous whether I can pull that off without blowing this thing
again. That's why I'm looking for a nice-and-easy plug-in solution this
time...

> TV-inspired integrated circuits have reduced the package count
> considerably, but there's still a lot of analog lifting required--such as
> the delay line to match the luminance delay to the filtered chrominance
> channel.

Yeah, I figure that's what that schematic seems to hint at...

> I would expect there to be a lot of surplus NTSC-to-RGB decoders available
> on eBay--some with professional specs.

I found one, but they want $267 (sic!) for shipping to Germany:

http://www.ebay.com/itm/390381365254

There's also a PAL-NTSC format converter, which might do the job I want
as well, trying to find out:

http://www.ebay.de/itm/170926500970

And there's also lots and lots of the opposite - RGB to NTSC converters.
They seem to be in much higher demand.

--
Linards Ticmanis

[Linards Ticmanis]

Found something else which looks even better:

www.ebay.com/itm/300799535503

--
Linards Ticmanis

[Linards Ticmanis]

Ok, I know it is stupid to send follow-ups to your own posts... but... I
just found out that there is a multi-system decoder chip that you can
put into the 1084 PCB as well, as it's pin compatible with the NTSC or
PAL specific chip that's in there. Then all you need is a little switch
thing, and poof multi system compatibility.

I'll report on my results as soon as I have some - in case anybody is
interested.

--
Linards Ticmanis

[Alex Freed]

The 1084 schematics that I have shows a TDA4570 NTSC decoder chip. The
PAL equivalent is probably TDA4510 that is pin compatible.

Now the multi-system TDA4555 is "pin sequence compatible" but is a 28
pin chip while TDA4510 is 16 pin. The TDA4555 is a nice chip and it can
be either forced into a particular color system mode or can detect it
automatically.

I have just found this:
http://obsoletetellyemuseum.blogspot.com/2012/08/schaub-lorenz-itt-8838-i-chassis.html

A TDA4510 is populated for PAL only while the board is clearly designed
to accommodate a TDA4555. In addition to a second crystal for 2*3.58
=7.14 MHz one also needs to switch the filter from 4.43 to 3.58.

This is not hard to do for someone comfortable with a soldering iron,
but using an external decoder like the one you found in the previous
message is probably lower risk.

-Alex.

[Linards Ticmanis]

(Sorry for mailing this... Thunderbird is acting up again somehow.)

On 10/23/2012 12:22 AM, Alex Freed wrote:

> The 1084 schematics that I have shows a TDA4570 NTSC decoder chip.
> The PAL equivalent is probably TDA4510 that is pin compatible.
>
> Now the multi-system TDA4555 is "pin sequence compatible" but is a 28
> pin chip while TDA4510 is 16 pin. The TDA4555 is a nice chip and it
> can be either forced into a particular color system mode or can
> detect it automatically.

Right... and the board already has the necessary traces to put in a
resistor for pushing it into either NTSC or PAL mode. Auto-detection
isn't really useful, as you have to switch between the two quartzes
anyway. And I'm quite happy to leave SECAM to the French.

> I have just found this:
>
http://obsoletetellyemuseum.blogspot.com/2012/08/schaub-lorenz-itt-8838-i-chassis.html
>
> A TDA4510 is populated for PAL only while the board is clearly
> designed to accommodate a TDA4555.

Exactly, that's what I found out after a while of staring at the PCB and
the schematic. All the solder holes are already there. And the chip is
easily available on eBay.

> In addition to a second crystal for 2*3.58
> =7.14 MHz one also needs to switch the filter from 4.43 to 3.58.

You mean the filter that removes the chroma frequency from the luma,
right? To be honest I didn't even think of that. Oh well time to take
another look at the schematic I guess...

> This is not hard to do for someone comfortable with a soldering iron,
> but using an external decoder like the one you found in the previous
> message is probably lower risk.

Too late... I've already started soldering. ;-)

So far I've only re-designated the leftmost pot on the front for hue
control, put in another pot in the back to serve as the new horizontal
centering, beefed up the main caps in the power supply circuits (no more
crushed or expanded scanlines in bright parts of the screen - or at
least almost none), and put in the missing 6-pin DIN socket and - most
important of all - the GREEN switch (surely, being able to turn
everything garishly green at the flip of a switch is absolutely de
rigueur in every monitor...).

I think I'll use a little relay to switch between the quartzes.
Connecting them to ground through long wires and a switch probably
doesn't help the frequency stability, I'd guess.

--
Linards Ticmanis

[Alex Freed]

Forgot to mention yet one more filter: the notch filter tuned to 4.43
MHz in the luma signal path. For the best results it should also be
re-tuned to 3.58 in NTSC mode.

-Alex.

[Linards Ticmanis]

On 10/22/2012 05:18 PM, Linards Ticmanis wrote:

> I'll report on my results as soon as I have some - in case anybody is
> interested.

OK, the main part is done. It even does auto-detection of PAL or NTSC
mode, which I didn't know was even possible until Alex Freed rightly
suggested that I read the data sheet more attentively. All in all it's
working really well.

There are two remaining issues: First, the filter that removes the
chroma info from that signal which goes down the luma path still has to
be re-tuned by hand for either PAL or NTSC. Not too convenient as it is
deep inside this beast. As Alex has suggested, this needs to be switched
between two differently tuned filters just like the two quartzes are
switched; but currently I'm having a hard time finding out what exactly
to get for the second filter. According to the schematic the filter is a
tunable inductor and a capacitor in series (inside a grounded metal can,
as a single component), but no pF or 焙 values are given and I don't
understand the inscription on the component itself either.

Second, no matter how you tune the filter, the Apple II signal seems to
be a bit too "violent" for it. With the beam moving from white or black
into a colored area, it takes about 2-3 chroma cycles until the
filtering really gets going. No such problem when watching DVB TV, a DVD
or VHS through it, so I guess the fact that an Apple color signal is
more or less "pure" color carrier, highly saturated and instant-onset,
defeats the circuit until it's had time to adjust. I guess it's a trade
off with the fact that the monitor has a much higher bandwidth than a
normal video monitor - 80 char text is quite legible on it.

I'll stop replying to myself now.

--
Linards Ticmanis

[Michael J. Mahon]

Linards Ticmanis <ticm...@gmx.de> wrote:
> On 10/22/2012 05:18 PM, Linards Ticmanis wrote:
>
>> I'll report on my results as soon as I have some - in case anybody is
>> interested.
>
> OK, the main part is done. It even does auto-detection of PAL or NTSC
> mode, which I didn't know was even possible until Alex Freed rightly
> suggested that I read the data sheet more attentively. All in all it's
> working really well.
>
> There are two remaining issues: First, the filter that removes the
> chroma info from that signal which goes down the luma path still has to
> be re-tuned by hand for either PAL or NTSC. Not too convenient as it is
> deep inside this beast. As Alex has suggested, this needs to be switched
> between two differently tuned filters just like the two quartzes are
> switched; but currently I'm having a hard time finding out what exactly
> to get for the second filter. According to the schematic the filter is a
> tunable inductor and a capacitor in series (inside a grounded metal can,

> as a single component), but no pF or µH values are given and I don't

> understand the inscription on the component itself either.
>
> Second, no matter how you tune the filter, the Apple II signal seems to
> be a bit too "violent" for it. With the beam moving from white or black
> into a colored area, it takes about 2-3 chroma cycles until the
> filtering really gets going. No such problem when watching DVB TV, a DVD
> or VHS through it, so I guess the fact that an Apple color signal is
> more or less "pure" color carrier, highly saturated and instant-onset,
> defeats the circuit until it's had time to adjust. I guess it's a trade
> off with the fact that the monitor has a much higher bandwidth than a
> normal video monitor - 80 char text is quite legible on it.
>
> I'll stop replying to myself now.

The chroma trap is, as you've observed, a series-tuned circuit. If there is
a PAL/NTSC level generated by the decoder IC, you can use it to control a
relay that would switch in an (adjustable) additional capacitance in
parallel with the trap capacitor. Hopefully, the signal goes to the
inductor and the capacitor goes to ground, but it will work either way.

Align it by tuning for PAL chroma rejection with the relay contact open,
then adjust the additional trimmer capacitor for NTSC chroma rejection with
the contact closed.

Alternatively, if you don't mind seeing the pixel structure of colors (like
most RGB monitors), just remove the chroma trap entirely. That will
improve the bandwidth even more. (This option won't work if the chroma
signal is taken from a second winding on the chroma trap.)

The only time that it can get annoying seeing the chroma signal in the
luminance is in fairly large areas of relatively saturated colors (like an
Apple II display ;-)--but RGB monitor users seem to get used to it.

[Kevin]

I cant wait to see what happens to a fairly high speed analog signal as it passes millimeters away from a giant inductor though creappy contacts...

really? a relay for a no current 1 volt signal?

gets popcorn =)

[Michael J. Mahon]

Well, the "signal" we're talking about is the unneeded chroma channel being
shunted to ground, and a low impedance path is important.

A small relay should work well in such a case, while providing relatively
little stray capacitance when it's open.

If the "giant inductor" you refer to is the relay coil, no worries--the
coil is driven by DC and is well isolated from the contacts.

[Linards Ticmanis]

On 10/26/2012 07:04 AM, Michael J. Mahon wrote:

> The chroma trap is, as you've observed, a series-tuned circuit. If there is
> a PAL/NTSC level generated by the decoder IC, you can use it to control a
> relay that would switch in an (adjustable) additional capacitance in
> parallel with the trap capacitor. Hopefully, the signal goes to the
> inductor and the capacitor goes to ground, but it will work either way.

Well... the trap is a single metal-can component which according to the
schematic contains both inductor and capacitor, so I could only put
something in parallel with the whole thing, which I hope will work too.
It doesn't go directly to ground either, but through a 470 Ohm resistor,
which I hope won't matter.

I've tried with a 27pF adjustable capacitor, but that was clearly too
low a value - adjusting it didn't seem to make any difference
whatsoever. Unfortunately adjustable capacitors seem to have gone out of
style, particularly when you don't want SMD parts. You can still order a
few types on-line but electronics parts stores have stopped carrying them.

> Alternatively, if you don't mind seeing the pixel structure of colors (like
> most RGB monitors), just remove the chroma trap entirely. That will
> improve the bandwidth even more. (This option won't work if the chroma
> signal is taken from a second winding on the chroma trap.)

Since this monitor doubles as a TV for me, and the color grain in TV
images is pretty strong without the trap, I guess I'll try to find a
switchable solution.

Is the relay necessary? Alex suggested a simple transistor switch with a
standard BC548 or other small NPN.

--
Linards Ticmanis

[Kevin]

I would do a transistor, or if your switching a few things around a mux like the CD405(0) series

[Michael J. Mahon]

Linards Ticmanis <ticm...@gmx.de> wrote:
> On 10/26/2012 07:04 AM, Michael J. Mahon wrote:
>
>> The chroma trap is, as you've observed, a series-tuned circuit. If there is
>> a PAL/NTSC level generated by the decoder IC, you can use it to control a
>> relay that would switch in an (adjustable) additional capacitance in
>> parallel with the trap capacitor. Hopefully, the signal goes to the
>> inductor and the capacitor goes to ground, but it will work either way.
>
> Well... the trap is a single metal-can component which according to the
> schematic contains both inductor and capacitor, so I could only put
> something in parallel with the whole thing, which I hope will work too.
> It doesn't go directly to ground either, but through a 470 Ohm resistor,
> which I hope won't matter.

I'd carefully open the can to get to the internal node. A little delicate
soldering and you've got it.

> I've tried with a 27pF adjustable capacitor, but that was clearly too
> low a value - adjusting it didn't seem to make any difference
> whatsoever. Unfortunately adjustable capacitors seem to have gone out of
> style, particularly when you don't want SMD parts. You can still order a
> few types on-line but electronics parts stores have stopped carrying them.

It can't be usefully paralleled with the trap--you have to parallel it with
the internal capacitor.

>> Alternatively, if you don't mind seeing the pixel structure of colors (like
>> most RGB monitors), just remove the chroma trap entirely. That will
>> improve the bandwidth even more. (This option won't work if the chroma
>> signal is taken from a second winding on the chroma trap.)
>
> Since this monitor doubles as a TV for me, and the color grain in TV
> images is pretty strong without the trap, I guess I'll try to find a
> switchable solution.
>
> Is the relay necessary? Alex suggested a simple transistor switch with a
> standard BC548 or other small NPN.

No, a saturated transistor or analog switch will do as well--particularly
with a 470 Ohm resistance to ground.

[Linards Ticmanis]

On 10/27/2012 10:33 AM, Michael J. Mahon wrote:

> I'd carefully open the can to get to the internal node. A little delicate
> soldering and you've got it.

Bingo - the internal connection was even available externally on one of
the pins of the component (normally unconnected in the 1084). There was
also a convenient, unused space for a transistor on the PCB. I used a
BC548 and connected the base to the decoder's NTSC output pin through a
10K resistor. Everything works fine now, time to close the housing and
be done with it.

Again thanks to all who chimed in with valuable advice and ideas.
Without you I couldn't have finished this.

--
Linards Ticmanis

[Michael J. Mahon]

Very nice, Linards!

[Alex Freed]

> Linards Ticmanis <ticm...@gmx.de> wrote:
>> 10K resistor. Everything works fine now, time to close the housing and
>> be done with it.

Good!

I only want to add that some 30 years ago I had an Apple connected to a
TV and discovered that the color saturation was way too high. Not a big
surprise: consider a 01010101... pattern. It is a 3.58 MHz square wave
rail to rail. The first harmonic of that is about 63% of that - a lot
more than the normal chroma signal. So I added a shallow notch filter -
a low Q LC tank in series with the signal and it actually looked better.
That's an LC tuned to 3.58 and shunted with a resistor. I definitely
don't remember the values after all these years :) If the colors are
oversaturated you may want to give it a try.

-Alex.

[Linards Ticmanis]

On 10/30/2012 01:27 AM, Alex Freed wrote:

> I only want to add that some 30 years ago I had an Apple connected to a
> TV and discovered that the color saturation was way too high. Not a big
> surprise: consider a 01010101... pattern. It is a 3.58 MHz square wave
> rail to rail. The first harmonic of that is about 63% of that - a lot
> more than the normal chroma signal. So I added a shallow notch filter -
> a low Q LC tank in series with the signal and it actually looked better.
> That's an LC tuned to 3.58 and shunted with a resistor. I definitely
> don't remember the values after all these years :) If the colors are
> oversaturated you may want to give it a try.

Thanks for the hint, but the 1084 has a fairly standards-conforming
saturation when the knob is in the middle position, which means it's
lower than on most TV sets (which tend to be over-saturated from the
factory, since - as I've read somewhere - this supposedly boosts sales
in the showroom). So even with the Apple II the colors are not all that
garish, and with a C64 connected they're even somewhat dull.

One more chapter to the saga: I didn't notice until a bit later that
when the multi-standard decoder chip can't detect any color encoding at
all - such as when an Apple II in text mode is connected - it will
continuously cycle through its four color systems while trying to find a
subcarrier. That means that the NTSC output line gets activated for
short time about twice per second, resulting in an ugly "pulsing" effect
of the external LC filter, with short drops in sharpness. I solved it
with a large capacitor and two resistors, now the NTSC setting of the
filter doesn't kick in until the NTSC output pin is stable at high level
for about one seconds.

Also I had a rather unpleasant fight with "Hanover Bars" in orange or
greenish-blue screen areas in PAL modem (around the U = -V points).
Correctly adjusting the PAL delay line is quite tricky in this monitor
as the setting doesn't seem to be entirely stable. I got it into a state
where at least at standard saturation the bars are no longer visible.

--
Linards Ticmanis

[nyder]

The Commodore 520 Adaptor has the circuitry to display pal or ntsc, might be easy/cheap to make an apple to amiga adaptor for it. It's made for use with TV's and composite monitors. Only thing that sucks is the weird 21pin plugs amiga use. Thought I had one, was going to open it up and take pics of it's pcb, but it seems I don't. Not surprised though, I manage to keep 3 commodore monitors (1902, 1080, 1084) and a Sony EGA Monitor (CPD-1302) over the years.

Anyways, those adaptors used to be pretty cheap, quick check on ebay put one at $40ish USD, but maybe you can find one there in germany pretty cheap. Probably can hack a s-video single out of it.