• Please review our updated Terms and Rules here

NuXT - MicroATX Turbo XT Motherboard - 9.55MHz, UMBs, XT-CF, SVGA, HD Floppy, Serial

nztdm

Experienced Member
Joined
Mar 17, 2018
Messages
180
Location
New Zealand
Not sure whether this should go here or in marketplace, as it's sort of a show n tell n sell.

Here's the result of a project I've been working on for a few months.

Shout out to Sergey Kiselev for his schematics which I've used in a few functional blocks on this board, such as his Micro 8088 SBC.

80's is coming back with the NuXT!
MicroATX sized Turbo XT motherboard.

Can be used as a full featured XT without any cards.
Could add Sound, Network Card, or more storage/connectivity.

Features:
  • 8088 and NEC V20 support
  • 8087 co-processor support
  • 4.77MHz, switchable to 9.55MHz with hotkey
  • 2x64K ROM, switchable with DIP switch, writable
  • 800K SRAM
  • - 640K conventional
  • - up to 160K UMBs (128K max when using EGA/VGA)
  • - Configured DOS 6.22 uses only 10K conventional!
  • Highly-compatible SVGA, up to 512K VRAM
  • HD Floppy controller, with FM support
  • Bootable CF Card Interface
  • Option ROM Socket, supporting 8K and 32K ROMs, writable
  • 16C550 Serial Port
  • PS/2 Keyboard Port
  • ATX Power Input. Generates own -5V
  • 4 x 8-bit ISA Slots
  • Front-panel connector for modern cases
  • Highly configurable with DIP switches. Can disable onboard peripherals.

Now I just need to find a place that cuts out I/O shields in low quantities... I can get blank ones.
I think they provide a professional finishing touch.

Click the images for higher res.






*Ugly rear bodge wire only on my prototype

 
I have made two. One is my bodge prototype seen here, and the other is for someone who bought one. They are in my little vintage PC parts shop in my signature.

I have enough parts to make 3 more. If they prove popular, I'll source more parts.
 
Any chance a PS/2 mouse port can be wedged in? They did exist on the XT class PS/2 Model 25 and 30 although BIOS support might be an issue.
 
Any chance a PS/2 mouse port can be wedged in? They did exist on the XT class PS/2 Model 25 and 30 although BIOS support might be an issue.

The XT chipset has XT keyboard interface built in. I'm using an AT2XT microcontroller here. This is based on Sergey's Micro8088.
Sergey's Xi8088 is an older design, that uses a full XT chipset instead of a single chip. This allowed using a PS/2 KB+M controller from Pentium systems. He also had more IRQs on the Xi8088 like an AT.

I only have the 8 IRQs.
There is one way I can add PS/2 mouse support. I need all of the following:

Stacked mini-DIN 6 connector
Figure out 16C552 instead of 16C550
PS/2 to Serial mouse microcontroller
We'll only have IRQ 2, 5, and 7 left.
The 16C552 has parallel port too. I can add that if I can find a raised DB25F connector. Then only IRQ 2 and 5 is left for sound cards etc.

I do plan on sorting out a PS/2>Serial mouse converter box some time. With the option of coming with a new Microsoft Optical Mouse that supports PS/2 mode (can still get them).

Cheers
JD
 
Excellent! If it's based on Sergey's work, you'll be releasing the design files in accordance with the GNU GPL v3, correct?
 
Nice!

$400 CDN is a bit out of my budget range for a toy (looking to get a 3D printer sometime soon). But it's comparable, if not less than what a fully loaded XT can go for, and you can slap this together with any random ATX chassis/PSU you happen to have lying around (we've all got those)!

Any chance of getting this as a bare board or kit format at a reduced price? Building it is half the fun anyways.
 
Nice!

$400 CDN is a bit out of my budget range for a toy (looking to get a 3D printer sometime soon). But it's comparable, if not less than what a fully loaded XT can go for, and you can slap this together with any random ATX chassis/PSU you happen to have lying around (we've all got those)!

Any chance of getting this as a bare board or kit format at a reduced price? Building it is half the fun anyways.

I can offer bare boards for $20 + postage.
I can offer a kit for full price minus $40.

I don't think I'll offer kits yet, as I have no way of testing the Trident SVGA chips, and don't want to send untested vintage chips. I am working on making up a QFP-160 socket board for testing them.
 
nztdm: FYI, that the Ad Lib clone card you're selling is based on my open source clone design which is licensed under CC BY 4.0 (Creative Commons Attribution 4.0 International). According to the terms of the license you need to give appropriate credit, provide a link to the license, and indicate what changes you made, if any. I'd appreciate it if you followed those terms by adding that information to your website and your eBay auctions. Thanks!
 
nztdm: FYI, that the Ad Lib clone card you're selling is based on my open source clone design which is licensed under CC BY 4.0 (Creative Commons Attribution 4.0 International). According to the terms of the license you need to give appropriate credit, provide a link to the license, and indicate what changes you made, if any. I'd appreciate it if you followed those terms by adding that information to your website and your eBay auctions. Thanks!

No problem!
I swear I did this, but looks like I had forgotten. I even had the README.md ready with the license and changes. I have updated my site listing (eBay listing will be able to be updated in a couple days).
The changes I made are:
> add footprint for Alps RK09K potentiometer. It's a 10th of the price of the large blue ones.
> add part values, for helping kit-builders

You have great attention to detail, right down to the shape of the pads :)
 
Hello how is the video card compatibility of this great motherboard? i would like to check several monochrome cards and monitors.
 
Hello how is the video card compatibility of this great motherboard? i would like to check several monochrome cards and monitors.

I have an IBM MDA, IBM CGA, Hercules, Trident VGA. All work fine.

However I'm working out an issue. Currently, if you disable VGA, you can only use another VGA.
Only boards without the VGA parts onboard will work with other video cards.
The next version will likely have this fixed, or not include VGA.
 
I have an IBM MDA, IBM CGA, Hercules, Trident VGA. All work fine.

However I'm working out an issue. Currently, if you disable VGA, you can only use another VGA.
Only boards without the VGA parts onboard will work with other video cards.
The next version will likely have this fixed, or not include VGA.

The VGA chip buffers all the bus lines and has lines to ram etc, pullups, pull downs etc. I dont see anywhere you detach all the pins on all these devices.

I bet the monolithic VGA IC has reverse protection diodes and its activating it, making it 'sit on the bus' and mask its IO address etc.

Dave has an interesting video on reverse protection diodes across IO pins powering an IC that has no power/gnd connected.

https://www.youtube.com/watch?v=2yFh7Vv0Paw
 
The VGA chip buffers all the bus lines and has lines to ram etc, pullups, pull downs etc. I dont see anywhere you detach all the pins on all these devices.

I bet the monolithic VGA IC has reverse protection diodes and its activating it, making it 'sit on the bus' and mask its IO address etc.

Dave has an interesting video on reverse protection diodes across IO pins powering an IC that has no power/gnd connected.

https://www.youtube.com/watch?v=2yFh7Vv0Paw

Thanks for that. That was very interesting!

I have since solved the VGA disable function. There is a pin in the datasheet that can disable it. It just wasn't clear. Will update the github with fixed schematic soon.

Just one more issue to fix, which will hopefully be solved in the "rev 1.1":
The serial port is unreliable. And a few factors can affect this for good or bad. It's a mystery to me at present.

Testing method:
Switch to 9.55MHz
Open Windows 3.0
See if mouse works and doesn't drop out after a few seconds
Run checkit test with loopback connector
Run CTMOUSE and then Lemmings

Order of unreliability, from least reliable to most reliable:
Windows 3.0 mouse (ultimate test it seems)
Lemmings
Checkit loopback test
Text-mode mouse (works almost always)

Seems like perhaps the VGA going into graphics mode makes it worse?

Things tried:
NI 16550 instead of TI 16C550
Crystal instead of DIP oscillator
Standalone serial card with same schematic
No VGA onboard, but same VGA as standalone card
Test voltages

Things not tried yet:
75C185 instead of GD75232 transceiver
74ACT bus logic instead of 74F

I wouldn't think trace lengths, ground return path length, noise, separate ground planes, etc would be too important at these low speeds, and as such i haven't paid much attention to these factors. I could be wrong.

Notable things:
>Installing another serial card as COM2, makes the onboard COM1 work perfectly (?? bus capacitance/termination)
With the above, the COM2 serial card will now have the unreliability issue
>Running at 4.77MHz improves things sometimes
>With loopback test in checkit, when a port is unreliable, sometimes the registers test will fail, and sometimes the main test will fail at higher baud rates.
>When creating an almost-NuXT out of Sergey's cards in a backplane, I can recreate the unreliability by moving the serial card to the last two ISA slots, furthest from the CPU SBC card.
The serial circuitry -is- on the other side of the PCB as the bus/CPU circuitry on the NuXT. Maybe laying the board out again with the serial circuitry close to the bus will solve it. That would be nice, but wouldn't help me understand why. Also, the serial card being unreliable, while also fixing the onboard with identical schematic, makes me thing this won't work, because that second serial card, is now right next to the bus/CPU area.
 
how noisy is the bus? have you put an oscope on it? com port requires IRQs. wonder if its a bus timing issue.

What should I probe specifically to find a noisy bus?

I've checked some things.

> The UART stops generating IRQs whenever the mouse drops out.
> The UART CS line is always active. This makes sense since it will go low whenever 3F8 is on the address bus, regardless of any address lines above A9. Forgive my noob question but I've never figured that part out regarding chip-selects on ISA. If the CS logic is not looking at the higher address lines, wouldn't all the false chip-selects cause issues? Or is this related to IOR/IOW? IOR/IOW can only be active with addresses within the first 10 address lines?

IBM 5150 - A7:
5150 A7.jpg

NuXT - A7:
NuXT A7.jpg

NuXT - UART CS:
NuXT UART CS.jpg
 
Back
Top