So this will likely read like the ramblings of a madman, but It is everything I have had bouncing around my head this weekend and I need some room up there.

I have disassembled my DW200 and compared it to the pictures of the NC200 repairs I have seen online. 

Comparison of PCB:

NC200 (Front) (Back)

DW200 (Front)(Back)

Datasheets:

Intel 82077SL Floppy Controller  

Intel N82077AA essentially the same as the SL

- This is definitely the same plastic casting as the NC as the stress points and injection nubbins are in the same locations. This also explains why the HD floppy on the DW has a bit of a gap. If you made your own moulds you would have corrected that gap.

- The keyboard construction is completely different. The plastic key frame is different and the membrane end point location is different. This is odd unless this is some standard layout or the NC was purchasing their keyboards from a vendor and NTS was unable to continue use when they changed to their layout.

- The conceptual layout of the PCB is identical. The core chips are in exactly the same locations (ROM, ASIC, RAM) and the part numbers on the things that "just work" are the same. (CFL backlights circuit, resistors, caps, inductors and placements are identical. The power barrel jack is still inverted centre negative which is very odd for north american built DC jacks so it was a known part that they just kept using. The RTC is also the same chip. The same components have the same solder mask ID as well and that's not a coincidence that can happen.

- The power layout is the same but completely different. The DW has a rechargeable battery pack in the form of NiCads taped together in the same C-Cell compartment. Instead of locating the battery wiring to one location like a new design would have done they just trailed some wire from the existing battery compartment springs to the battery pack connector. This was an existing layout and they kept the power points. The power regulation/charge area is in the same physical space, just different and newer linear components. The decision to use NiCads and a charger is obvious as they were targeting a school market. The NC was notorious for not being able to use the floppy drive after 30 min of use and that would just not be acceptable if you wrote for a class and nobody could save! Why Amstrad did not do this day 0 I have no idea.

- There are two 8x128k RAM chips unlike the NC's stock single ram. The NC has the footprint for TSSOP & SIP likely because of a current price tipping point and they likely chose the cheapest footprint as TSSOP. As memory became cheaper it was likely an easy addition to the DW as you only would need the OE brought out to the ASIC. This does bring up something with the ASIC. The NC would re-map the RAM and ROM is nearly any configuration and was pretty neat. The larger 1 meg address space of the 8088 makes the z80's 64k address space issues moot. I wonder if the MMU in the ASIC has been re-spun or if they just don't use it on the DW at all??

- The NC has the 720k floppy controller baked into the ASIC and the control IP written by Ranger Computers. NTS would not have received the IP for this or maybe decided it was the early 90's and that HD floppy was king in north america (it was). The easiest way to control a floppy drive is a dedicated chip and that's exactly what the DW has. An Intel N82877SL. I need to dredge up a data sheet to find out it it provides the FAT filesystem translation on board or if it was simply a hardware interface chip.

Here are some random musings:

- The DW has an NEC v20 CPU on the back of the board directly under the ASIC. The v20 is an unofficial souped up 8088 processor so its x86_16 ASM instructions. This also increases the memory addressable to 20 bits (1 Meg). This is completely incompatible with the z80 and seems like a odd choice as you would need to write your new OS from the ground up. Now if NTS bought the rights to the NC200 injection moulds, the PCB layout and ASIC its a pretty safe bet to think they got the original firmware in ASM as well. All the z80 code could easily be ported to x86 if you have sources.

- So the NC has no Z80 dedicated IC, this mean that they ASIC in the center is a complete custom fab job with the Z80 IP baked in. Now if Amstrad sold the NC200 IP to NTS this would NOT include the rights to the Z80 IP inside their ASIC but it WOULD include all the ASIC functions and masks. Now a chip fab is not going to make new masks for the same chip functionality, the costs would negate not just rolling your own ASIC in the first place. From the layout and support circuitry is definitely looks like the same Amstrad ASIC. What is the chances the DW ASIC actually has the Z80 IP still baked in but not checked off during QA/QC?! A simple check would be to roll a custom ROM that makes the speaker beep in x86_16 ASM then write the same I/O call in Z80 and see if it works. I my need to lift the V20 power pin to disable the chip though. From above, if they re-spun the MMU in the ASIC they would have removed the Z80 mask guaranteed. Any ASIC changes would have required the chip fab to remove the IP, I am just banking on the cost savings of them playing cheap if it wasn't modified.