Repairing my poor NMS 800

Champion (316)

20-11-2022, 10:44

Hi all!
I'm trying to repair an NMS 800 I bought, in quite a bad state unfortunately. I'm sharing some of the things I'm doing here, perhaps someone can give me some feedback.
First, I cleaned up the motherboard and specially the keyboard. I saw in one of the video of Noel's Retro Lab that they keys can be cleaned quite efficiently if submerged for a while in water with tabs for dental prosthesis (!). And oh boy... it does work. You put them for a about 1/2 hour and they look like new! What a discovery

However, the computer has still a lot of problems.
- The keys get stuck in the plastic support. The mechanism is quite basic, the plastic has a shape than simply pushes the keys up. I guess that after more than 30 years it's lost its elasticity. I'm thinking of making a custom PCB with high-quality switches to replace it, since at this moment it's quite unusable.
- The video has some moving jailbars and it's noisy. Surprisingly the interferences correlate with the sound being played! So I guess the SCART cable might need to be replaced (the other end is connected directly to the board). Or perhaps the capacitors are in such a bad condition that are causing this? Has anyone ever seen something similar?
- The sound is horrible especially when mixing 300 Hz and 1 Khz in two channels. Again, I suspect of the capacitors of the amplifier after the AY chip. I think the amplifier is the 7407 chip near the AY chip. I'll have to use the oscilloscope to see. I wish I had at home those wonderful labs such as The 8 Bit Guy or Noel, instead of doing this at midnight in the kitchen XD
- There are a lot of "write-too-fast" VDP artifacts when playing games. Operation Wolf or Bounder are quite destroyed! I made some tests, like counting how many tick happen (with the TIME variable) during a fixed period in BASIC (just a for which increments a counter until TIME arrives to 5000). Quite interestingly, I see that the emulator counts 14481 ticks, and the real thing 16496, My interpretation is that probably the capacitors are not working correctly and the clock that arrives to the VDP is slower (or without sharp edges, which might not be detected). It wouldn't be that the CPU is too fast, but the VDP too slow, giving the same effect. Just a guess.

If anyone has some ideas or feedback, it'd be quite welcome!

By Wlcracks

Hero (566)

20-11-2022, 11:22

Interesting machine. I couldn't find the schematic of it, but using the vg8020/00 as a reference. I don't see a x-tal on the nms800 board picture for the z80. Only the VDP. So how is the z80 clock generated? Looks like the Z80 is running too fast, but could also be a ram/vram issue. You need an oscilloscope and a decent probe for checking the clocks, event better a logic analyser.
Recapping is alway a good start, with a low budget board as this unit.
Fun project, but hard to debug without the rom slots. I like the non engine MSX hardware, they are always a bit different.
Start with the supply voltages on a oscilloscope.

By mcolom

Champion (316)

21-11-2022, 21:07

Wlcracks wrote:

Interesting machine. I couldn't find the schematic of it, but using the vg8020/00 as a reference.

Yes, I'm afraid the tech manual doesn't exist :/

Wlcracks wrote:

I don't see a x-tal on the nms800 board picture for the z80. Only the VDP. So how is the z80 clock generated?

It's got a x-tal, it's right on the left of the 9129 VDP.
In the pictures I made I can't really see any numbers on it, so I'll have to look again to see if I can identify which one it is exactly.

Wlcracks wrote:

Looks like the Z80 is running too fast, but could also be a ram/vram issue. You need an oscilloscope and a decent probe for checking the clocks, event better a logic analyser.

The VRAM chips seem to be OK. I made some software tests and they look fine.
I'll see if with my oscilloscope I can measure the frequency of the x-tal, and also at the Z80 and the VDP.

Wlcracks wrote:

Recapping is alway a good start, with a low budget board as this unit.

The caps looks fine, but I'll see if they get any hot. It could be, if there's any internal shortcut.
There are plenty of caps, so I'll try to leave this as a late option.

Wlcracks wrote:

Fun project, but hard to debug without the rom slots. I like the non engine MSX hardware, they are always a bit different. Start with the supply voltages on a oscilloscope.

Yes, I had to use the cassette interface to load tests programs. It's doable, but slow.
Thanks for your advices!