Sega SG-1000 AV mod

[Macross_VF1]

I recently added Segas first home console to my connection. I had a hell of a time getting it to work properly, though that turned out to be two broken traces to the cartridge connector.

Anyway, I figured I should add composite to it but the very few references I found on the subject was to add a 470 ohm pulldown resistor to pin 36 on the TMS 9918 and then pull the video signal directly from the same pin. This works but the picture quality was quite frankly worse than the RF-signal. 

Instead I looked to the sister system, the Sega SC-3000 home computer, which was released at the same time and was pretty much the same system. It featured composite out as standard so I dug up the schematics for it and straight up copied both the composite video and audio circuitry and this made a big improvement to the picture and sound output on my SG-1000. I'm pretty sure this will work with any system using the same SN76489 and TMS9918 chips though I haven't rested it yet. I've included a cobbled together schematic and the red marks are the parts I had to substitute due to lack of components. 

 

Now it's still only composite but when run trough a Retrotink 2X the picture quality is certainly decent. At the very least I hope this will stand as a better alternative to the crappy 470 ohm pulldown resistor method.

[carlsson]

On the distantly related VTech Creativision, there are at least two ways to get composite video: either tap it from the transistor 2SC9014 (labeled Q1 on the motherboard) or tap it from the video in signal on the RF modulator, where also is the place for the audio signal. I suppose the SG-1000 doesn't have a RF modulator box, but generates the RF signal integrated somewhere on the motherboard so there are no signals ready to tap? The method to tap the video in at the RF modulator is standard, sometimes enhanced with a transistor if the signal itself is weak.

 

I don't know how strong the signal is directly from the VDP. Adding a resistor sounds like you're dampening the signal, rather than amplifying it. Your design that copies the circuitry on the SC-3000 probably is more reliable.

[Macross_VF1]

The SG-1000 does have an RF modulator and I did try to get the composite and audio signals from there but again, I wasn't happy with the result. 

Now it should be possible to get RGB from this system. As stated earlier the VDP is the TMS9918 and it can only do composite but the later TMS9928 outputs a set of signals that can be converted to RGB. However they are not completely pin compatible, the 9918 provides both the CPU and sound chip with the clock signal on pin 38, something that the 9928 does not. In fact, pin 38 on the 9928 outputs one of the video signals necessary for RGB.

It should therefore be possible to build a separate clock circuit for the CPU and sound chip and swap the 9918 for a 9928, isolate the necessary pins on the 9928 and get RGB that way. However I don't know how to build a clock circuit with the proper MHz.

[phoenixdownita]

20 hours ago, Macross_VF1 said:

The SG-1000 does have an RF modulator and I did try to get the composite and audio signals from there but again, I wasn't happy with the result. 

Now it should be possible to get RGB from this system. As stated earlier the VDP is the TMS9918 and it can only do composite but the later TMS9928 outputs a set of signals that can be converted to RGB. However they are not completely pin compatible, the 9918 provides both the CPU and sound chip with the clock signal on pin 38, something that the 9928 does not. In fact, pin 38 on the 9928 outputs one of the video signals necessary for RGB.

It should therefore be possible to build a separate clock circuit for the CPU and sound chip and swap the 9918 for a 9928, isolate the necessary pins on the 9928 and get RGB that way. However I don't know how to build a clock circuit with the proper MHz.

... or piggy back the 9928 on top with only the necessary input wired in parallel to the left-alone 9918 and use the out from it (you may have to ground/10K-ground some pins of the 9928 rather than float them but I don't think the input signals would mind a second load).

[Macross_VF1]

14 hours ago, phoenixdownita said:

... or piggy back the 9928 on top with only the necessary input wired in parallel to the left-alone 9918 and use the out from it (you may have to ground/10K-ground some pins of the 9928 rather than float them but I don't think the input signals would mind a second load).

That would likely work as well, though I think I have found an easier solution, should I ever want to try it out in the future. I have spent some time looking into the schematic of the PAL SC-3000 and it solves the clock exactly as I suspected it could be solved. It features a TMS9929 (PAL version of the TMS9928) and has a separate crystal 3.58 MHz crystal and a 74HC04 chip to generate the clock for the CPU and sound chip. It should be easy to copy that part of the SC-3000 and insert it into an SG-1000. However I do find it a bit odd that the PAL SC-3000 runs on 3.58 MHz, same as the NTSC SC-3000, as that was primarily an NTSC frequency if I have understood it correctly.