Briefly: I experienced the same. But here are the details...
Yesterday I finished my sync stripper, and experimendet a bit.
I modified a standard SCART cable, so built this circuit small enough to fit into a SCART plug's housing, as it's recommended to put it as close as possible to the video signal's source.
This is a more or less universal solution, I can plug the other end into the above mentioned SCART->VGA adaptor if I want to connect it to the BenQ, and leave it out to use it with sandard SCART input devices.
I tested it with my 13-years-old Yamada DVX-6700 as signal source. It was one of the first DivX-compatible standalone DVD-players. Actually this is my only device at home which outputs RGB + composite video instead of pure csync.
I used basically the standard schematics, with minor modifications (picture attached to this post).
1. The power supply:
On some forums they recommend to feed the LM1881 IC directly from SCART pin 8, which should have work theoretically, as the standard voltages of this pin are 5-8V at 16:9, and 9.5-12V at 4:3 aspect ratio, and the IC needs 5-12V. I tried it, but didn't work: On my main TV, a Samsung LE40B650 the picture was too bright, distorted and unstable. The BenQ's screen remained black. Instead of the former mentioned manner giving 5V to the IC using a standard USB phone charger solved the problem.
2. The 75 Ohm resistor between the composite video input and ground:
Some sources recommend it: "It should include a 75 ohm (or thereabouts, the value isn't critical) resistor between the input and ground, before the coupling capacitor. This sets the optimum video level and combined with the lower line impedance and reduced noise, give the LM1881 a much better chance of seperating the sync from a video signal which is changing from bright to dark rapidly." I tested it mostly with static images, but it's absance or presence didn't do any visible difference in regard of picture quality. I left it there because I was too lazy to remove it, and I think it won't harm anybody
3. The value of the resistor on composite sync output (R1):
Different authors - different recommendations again. Some say it's not necessary at all, some to use a 330-470/270-510 Ohm impedance in case of SCART, and 680-820 Ohm in case of VESA monitors. I tried the circuit first without this resistor, and the picure of BenQ had a strong greenish tint. Next I tried a 750 Ohm impedance, the result was better, but still green shaded. After it used 1K value, again strong green tincture. The best result was with using 470 Ohm, that effected quite good white balance, but the red is still a bit weak (I would call it strong orange).
I tried the DVD-player with PAL settings, and had to choose the RGB colour format manually, as the monitor identified it as YPbPr.
On the Samsung TV there wasn't visible difference in aspect of picture quality using different values of R1 resistor.
After it worked acceptable with BenQ (a bit blurry picture, but it's not surprising, the source is an 576i signal), I tested the separator on the TV, and the only visible effect was that the picture got darker if using it, otherwise the image looked the same. Don't know if on older CRT's would be any major improvement or not, didn't try it so far.
"I get vertical banding, just like with my real AMIGA..."
Did you try to set the pixel clock manually? On other forums members recommend values near 90, i.e. 94, as they say this will completly eliminate vertical banding. The exact value may vary using different Amiga models, i.e. on the Amiga 500 they had to adjust the pixel clock to exactly 88 and Phase to 30 and got a crystal clear image with no vertical lines. On A1200 on the other hand they ended at pixel clock 92 and Phase around 30. this will vary a bit from model to model also: Some had to put their pixel clock to 88 and Phase clock to 30 + to get a perfect picture, even though 90 was the recommended pixel clock for Amiga use. Also seen others needed other values like 92. I think it worths to try different values.
I dont' have an Amiga (yet), so cannot test it by myself.
You do not have the required permissions to view the files attached to this post.