Toshiba T3100/20

[pippy]

I bought a Toshiba T3100/20, I was attracted to it because of the unique orange gas plasma display. I hoped to play some ms-dos games on it.

The listing claimed it could boot to dos, however it turned out that this was only the BIOS. Neither the hard drive nor floppy disk worked, and after going through the service manual to try and diagnose it, it seems to indicate a faulty motherboard.

So what should I do with it? is there a way to track down a replacement motherboard?

I was thinking that potentially I could gut the device and use a raspberry pi to drive the internals. This however would mean using customizing a CRT driver board. After digging through the manual, the plasma input uses seven pins, standard 5v signals. Two vertical/horizontal signals, four for signal intensity and a mysterious last one that seems to be labeled 'video signal'. The plasma is 640 x 400 at 60MHz which seems like a standard out the box resolution. The VGA666 seems to have the flexibility I need.

has anyone done a hack like this before?

 

[eswan]

I looked into it with the T3100's big brother T5200. The 5200 has a vga 640x480 screen though. The signals are pretty simple, 4-bit greyscale, horizontal sync, vertical sync, clock?, and display enable. Theoretically a Raspberry Pi should be able to drive it, but when I tried the lines shut down unless the frequency was an even division of something. I heard they may have fixed it in later revisions of the 'pi.

https://github.com/raspberrypi/firmware/issues/734

 

[eswan]

Wouldn't want to use a CRT driver board, it's all TTL digital. Might be able to sweet-talk a VGA card to drive it over the VESA connector.

 

[Eudimorphodon]

The VGA666 seems to have the flexibility I need.

So far as I can tell the VGA666 is basically just a resistor ladder to convert parallel output from the GPIO pins into the +0-0.7v voltage range expected for VGA so, yeah, you definitely don't need the actual hardware widget to drive a TTL display. You *might* get away with just hooking it straight up to the GPIO outputs, although running it through a voltage shifter/buffer would probably be safer.

A more complicated problem might be that it seems like the DPI output format is kind of hardwired for color? If the display only supports 4 bit grayscale it might be kind of non-trivial to make it sensibly match up to any of the available output modes. (My incredibly vague understanding is the Raspberry Pi only natively supports 16 bit color and up framebuffers, so chunking that down to one bit of resolution for red and blue and two for green is going to make for some interesting display challenges.)

Here's a guy who came up with a couple different methods for driving a single-bit Macintosh CRT from a Beaglebone Black, I wonder if the PRU approach he describes would be extendable to a four bit display. (The virtual framebuffer it sources is set up for 8-bit and there's dithering built into the output routine, which presumably could be adopted for grayscale?)

 

[MikeS]

If the BIOS comes up and the display & keyboard are working I'd spend a little more time trying to get it working; why do you think it's the motherboard?

 

[bifo86]

If the BIOS comes up and the display & keyboard are working I'd spend a little more time trying to get it working; why do you think it's the motherboard?

Yes, the hard disk being dead is no surprise, but have you tried swapping the floppy?

 

[pippy]

Yes, the hard disk being dead is no surprise, but have you tried swapping the floppy?

Both the hard drive and floppy are not receiving power. I looked into swapping them out, but sadly both the connectors are non standard and have their own pin layout. I managed to find a thread on this forum where someone else managed to bodge wire a SCSI card made for desktops to connect to standard IDE, but that's an awful lot of work. The woes of using a specific Toshiba-made hard drive controller card.

After following the diagnosis steps in the manual, it winds up getting to a fantastic step labeled 'replace motherboard'. hence my suspicion that the motherboard is faulty.

 

[pippy]

So far as I can tell the VGA666 is basically just a resistor ladder to convert parallel output from the GPIO pins into the +0-0.7v voltage range expected for VGA so, yeah, you definitely don't need the actual hardware widget to drive a TTL display. You *might* get away with just hooking it straight up to the GPIO outputs, although running it through a voltage shifter/buffer would probably be safer.

A more complicated problem might be that it seems like the DPI output format is kind of hardwired for color? If the display only supports 4 bit grayscale it might be kind of non-trivial to make it sensibly match up to any of the available output modes. (My incredibly vague understanding is the Raspberry Pi only natively supports 16 bit color and up framebuffers, so chunking that down to one bit of resolution for red and blue and two for green is going to make for some interesting display challenges.)

Here's a guy who came up with a couple different methods for driving a single-bit Macintosh CRT from a Beaglebone Black, I wonder if the PRU approach he describes would be extendable to a four bit display. (The virtual framebuffer it sources is set up for 8-bit and there's dithering built into the output routine, which presumably could be adopted for grayscale?)

I was thinking about the DPI and grayscale leveling. According to the manual there's only four levels of intensity, and I was wondering how the different color channels were to map to the values. If I could write my bit filter for the color output I could do some nice things like setting the 'contrast' ratio of the output. I'm imagining each color channel maps to one bit of the intensity level.

thanks for that Mac SE link, I'll have a through read through it.

I took screens of any part of the T3100 manual which seems to indicate anything in relation to the plasma display output. Table E-6 seems to be the most important: https://imgur.com/a/qf36Ujb

https://i.imgur.com/DJSVy0K.png

 

[eswan]

I _think_ the floppy drive in that series of Toshiba clamshells is an oddball with power in the data cable. Might be easier to rig up an external drive. Toshiba's of that era usually can switch the parallel port over to an external floppy connector and assign it either 'A:' or 'B:'. In fact, if it's not booting from the internal floppy, that switch might be set to assign floppy 'A:' to the external connector. Check and see if there's an 'A/B/PRT' switch set to 'A'.

 

[pippy]

I _think_ the floppy drive in that series of Toshiba clamshells is an oddball with power in the data cable. Might be easier to rig up an external drive. Toshiba's of that era usually can switch the parallel port over to an external floppy connector and assign it either 'A:' or 'B:'. In fact, if it's not booting from the internal floppy, that switch might be set to assign floppy 'A:' to the external connector. Check and see if there's an 'A/B/PRT' switch set to 'A'.

You're dead right, the proprietary ribbon cable is the reason why i can't swap out the hard disk or floppy drive. There was another user on this forum who managed to bodge wire together a desktop SCSI controller to get another model of hard drive working for this device, but that's an awful lot of effort to get something that might just be the motherboard (or my suspicion is the hard disk controller daughter board).

You raise a good point with the A/B/PRT switch, i've tried booting it 100 times with different combinations, but i'll see if it will boot to FDD on B again.

 

[Removed0]

I have a working 26 pin 20 MB JVC hard drive and 26 pin 720k floppy that I've pulled from a Toshiba T1200 if you're interested.