I have three of these old orange gas plasma 386 machines and have made various mods and upgrades for them over the years and thought I should document them here in case anyone else is interested in doing the same upgrades.
Here is a brief summary of the mods:
1. Fitted CF cards (two different mounting options)
2. Made up adapter cables to fit ordinary floppy drives internally
3. Fitted Cyrix/TI 386 to 486 upgrade processors including adding hardware support where required.
4. Made up an adapter cable to use an external floppy drive on the printer port.
5. Made up an adapter to use ordinary 72 pin SIMMs for RAM upgrade
6. Fitted 2 extra ISA cards in the space where the Hard Drive was originally fitted giving 4 ISA cards total.
7. Patched the BIOS to support all the above
8. Made a custom build of XTIDE to integrate with the BIOS and support large CF cards on the standard IDE port
9. Used Grub4DOS to create a multiboot system booting an effectively unlimited number of operating systems from the same CF card (currently 20 OS)
When I first got them, the HDs in all three and two of the floppy drives were dead so the first things I did was to fit CF cards and make up adapter leads so that I could use normal floppy drives. The CF cards (master and slave) can be changed by just removing the keyboard instead of having to dismantle the entire machine.
I upgraded the CPUs using a Cyrix 486DRx2, a TI 486SLC2 and a TI 486DLC and modified the motherboard to add the A20 gate and Flush signals to the CPUs where required:
The T5200 has an option to connect a second floppy drive to the printer port which can be switched as either the A: or B: drive so I made up a cable and used it with a Gotek USB floppy emulator with FlashFloppy firmware.
One of the machines had 14MB of RAM, the second had 6MB and the last had only the standard 2MB so I modified that one by removing the proprietary SIMM slots from the RAM expansion board and making a small sub board with standard 72 pin SIMM slots. I then wired that to the RAM expansion board giving 14MB on that machine (2x2MB +1x8MB SIMMs).
On that same machine I mounted two extra ISA cards in the space where the hard drive was (A network card and a sound card) and used the two standard ISA slots for other ISA cards
This meant that I couldn't use the same space to mount the CF card on that machine so I came up with an alternative mounting which used a different type of CF adapter with a very short IDE extension cable bent around the adapter so that it could be mounted directly above the IDE socket.
I had to patch the BIOS to support some of the above changes:
1. Change one of the HD types to 528MB
2. Detect and initialise the Cyrix and TI 386 to 486 upgrades so no driver software was required in the OS (code skipped if 386 processor detected)
3. Detect a slave CF card on the IDE port and initialise the CMOS correctly
4. Change the external floppy type from 5.25" to 3.5" so that the external GOTEK drive on the printer port worked correctly (The original T5200 external floppy was 5.25" so the BIOS always assumes that when an external floppy is detected)
I also made a custom version of XTIDE to work with the BIOS to support large CF cards (The standard version of XTIDE made Windows 95 run in MSDOS compatibility mode and also stopped Windows 2/386 and Windows 3.0 from running at all)
Finally, to make use of the large CF cards I used grub4dos to make a multiboot system so I could boot virtually any OS that would run on the system.
If anyone wants any further details on any of these mods, please let me know.
Very Very Impressive Ian ..
Perhaps you could help me ..
I only have one T5200/100 and am struggling big time just to get it functional as a basic win 3.1 unit ..
The bios I guess is the main issue as it is so restrictive..
I'd like to be able to use any similar hard drive of that period ..rather than the rare Connor ..
I also have a dead floppy drive so need some way to fit a standard 1.44mb floppy ..
There are many other wants ..but that would do for now..
If it means using a CF then so be it ..but I dont really know how to do this ..
I'm pretty good at following instructions to get things done..
perhaps you would take a look at my thread ? A couple of retros I'm working on.
I tried booting from a scs1 card/drive ..but even this proved impossible.
Assume I know nothing ..cos basically ..it is true regarding a machine of this age..
can you help me please ..
Can you at least supply me with a copy of your modded bios ..?
Details of the adapter leads for a floppy would be useful ..and probably my first job.
The ability to use standard simms for memory upgrade would be amazing ..
I only have the proprietory 2 sticks fitted..
cheers..
VS
Details of the adapter leads for a floppy would be useful ..and probably my first job.
Here are the construction details for the internal floppy adapter:
You will need 34 way and 26 way IDC headers and a 26 way ribbon cable of suitable length
Mount the 26 way IDC on one end of the ribbon cable as normal with the red wire going to pin 1.
Split the ribbon cable at the other end between wires 11 and 12. (red wire is pin 1)
The piece of ribbon with wires 12 to 26 should be inserted into the IDC header so that wire 12 makes contact with pin 18 and wire 26 makes contact with pin 32 leaving pins 33 & 34 at the end unconnected.
This connects the following pins:
Clamp the IDC as normal then remove the top of the IDC header to reveal the unconnected IDC pins.
Then connect each of the following wires to each IDC pin as follows:
wire 2 >> pin 8 !INDEX >> !Index
wire 4 >> pin 12 !FDSELA >> !Drive select B
wire 6 >> pin 34 !DSKCHG >> !Disk change
wire 8 >> pin 5 !READY >> GND (Ready is not a signal present on IBM type floppies but needs to be grounded to work)
wire 10 >> pin 16 !MONA >> !Motor on B
wire 11 >> pin 2 !LOWDNS >> !High density
The easiest way to make the above connections is to drape a wire over the required IDC pin then refit the top of the header and press it together by hand.
Remove the top again and repeat for the next wire. You may break the latches on the top with this continual removal but it can still be used to push the wires on the pins, after which it should be replaced by an intact one from another 34 way header.
Finally, the first four remaining unconnected wires (1,3,5 & 7) which are all +5v can be soldered together and connected to the +5v line of a floppy drive power connector (0v comes from the GND on pins on the 34 way connecter).
Note that this can only be used if the floppy drive is +5v only, many early drives used both +12v and +5v. In this case, the +5v lines should be insulated and a molex splitter used to split the supply to the hard drive.
Wire 9 (!NOTCH) is the status of the HD hole in the floppy. This is used by the BIOS to determine the size of floppy inserted and if you try to format a floppy of the wrong size you get an error "Parameters not supported by drive". Unfortunately standard floppy drives don't have this signal so you either have to ground it (for 1.44MB) or leave it floating (for 720K) which means you can't format one size of floppy although you can always read & write correctly formatted 720K & 1.44MB floppies irrespective of the state of this signal. The BIOS call that generates this error is not used by the formatter in DOS 3.3x so you can always format any size disk by booting to DOS 3.3x first. (Some T5200s with original floppy drives also generate this error when trying to format 1.44MB drives so that signal might not be available on some original floppy drives but they were supplied with DOS 3.3 so it didn't cause a problem at the time)
To avoid these problems, my patched BIOS versions for both the T5200 and T3200SX have removed the check of this status bit so they will format any floppy with any DOS version. (Obviously you can still only format a floppy according to it's capacity)
If you are running my patched BIOS, the state of Wire 9 doesn't matter, if you are not using my BIOS, then Wire 9 should be grounded or left floating depending on what size floppies you want to format most and then boot to DOS3.3 to format the other size (gnd = 1.44MB, float = 720K).
If grounding wire 9, connect it to another ground pin such as pin 7 although if you have a floppy drive with ground pins removed (see below), you will have to join it to the wire going to pin 5.
Warning: Some drives (e.g. more recent Mitsumi drives) don't have all the GND pins fitted on the 34 way socket. This means that the !READY signal might not make contact with GND if pin 5 is missing and the drive may not work. Pin 5 is present on the Mitsumi drives which is why that one is used but if other manufacturers did the same thing and didn't have pin 5 then you might have to move that wire to another pin or connect it directly to GND.
Also as there are fewer GND pins, there may be insufficient GND return for the +5V supply in which case the molex splitter should be used. I suggest avoiding this type of drive if you have a choice.
Here are some pics:
(Wire 9 is left floating in these pics but should probably be grounded as mentioned above)
Edit:
Strictly speaking, grounding the !READY line is not an ideal solution as it is then grounded for all floppy drives connected to the bus, although doing it this way is a simple solution which doesn't require any soldering if you use a molex splitter for power. I don't think this will have any effect on the T5200 although it might cause problems if this cable is used on another system with other drives. It might also cause issues with the original Toshiba 5.25" external floppy drive if that made use of the !READY signal. (There won't be any problems with 3.5" external drives as they don't have the !READY signal anyway). In the unlikely event that this is an issue, it may be fixed as follows:
Remove wire 8 from pin 5 of the 34 way IDC and solder it to the unbanded end of a high current (1A or so) schottky diode and solder the other (banded) end of the diode to wire 10 (!MONA/!motor on B) while leaving that wire connected to both IDC connectors (normal diodes don't seem to work with all drives due to their voltage drop - maybe a more complex transistor circuit would be better). To do this means scraping away the insulation of wire 10 to expose the bare metal for soldering. This change makes the !READY signal follow the motor on signal.
Last edited by IanB on 2018-03-15, 20:04. Edited 5 times in total.
I have three of these old orange gas plasma 386 machines and have made various mods and upgrades for them over the years and thought I should document them here in case anyone else is interested in doing the same upgrades.
Ian, this is brilliant! Apparently, you are a very advanced player when it comes to hardware, but as a result this thread is golden for all T5200 users. If you could share modified BIOS and details of some other mods too, that would be great.
Furthermore, a good chunk of these mods could be used with other Toshiba machines of that line - I'm specifically thinking about T3200SX, but overcoming limitations of Conner drives and proprietary FDD can help with T3100 (T3100e/T3100SX), T5100 and T3200 too. There is a similar thread but focused on T3200SX here: Toshiba T3200SX BIOS? - I will crosslink this one from there as well
It is also interesting to see that we have come with similar, but different solutions to the same problems. For example, jaZz_KCS solved the proprietary FDD issue by using ISA I/O controller, while in my T3200SX the CF card is accessible via opening for ISA cards on the far left:
I had T5200, but prefer T3200SX myself (visually I like classic plastics more than this strange coating on T5200). Who knows, though, maybe I'll get another one some day.
Furthermore, a good chunk of these mods could be used with other Toshiba machines of that line - I'm specifically thinking about T3200SX
It looks like the external floppy cable will work on the other models as well so I'll do that one next. However it needs a BIOS patch to be fully functional.
As standard, if any of these Toshiba models detects a floppy drive on the printer port it sets it up in the CMOS as a 360K 5.25" drive which was the only option available from Toshiba. It also rewrites the CMOS on every restart so using a utility to edit the CMOS won't work. With the setting incorrect, the drive may work from DOS but will generate an error if you try to boot from it. I made a simple patch to the T5200 BIOS to fix this so if an external drive is detected the CMOS is set to 3.5" 1.44M
Looking at the various T3200SX BIOSes posted in the thread you linked above, they are very similar to the T5200 ones (i.e. the Award BIOS versions will boot any drive and the Phoenix versions won't) so it is very likely that I should be able to patch the T3200SX Award one and maybe other BIOSes to work with the external drive but they will need testing by others.
An external 3.5" floppy isn't that useful on its own but if a Gotek USB floppy emulator or similar product is used then hundreds of floppies can be stored on it for OS and program installations etc. I did look at fitting one internally using the above adapter cable but as it is side mounted, it's not easy to see which floppy is selected so an external one connected when required seemed a better option.
OldCat wrote:
It is also interesting to see that we have come with similar, but different solutions to the same problems. For example, jaZz_KCS solved the proprietary FDD issue by using ISA I/O controller, while in my T3200SX the CF card is accessible via opening for ISA cards on the far left
I looked at using a ISA slot CF adapter but I was using both ISA card slots in the machines so the internal solution was preferable.
I looked at using a ISA slot CF adapter but I was using both ISA card slots in the machines so the internal solution was preferable.
I actually like my retro rigs NOT having network or internet access. When I can whip out CF card and add any software or files needed from my regular machine, I don't really need it. Hence - easily accessible CF card.
IanB wrote:
Looking at the various T3200SX BIOSes posted in the thread you linked above, they are very similar to the T5200 ones (i.e. the Award BIOS versions will boot any drive and the Phoenix versions won't)
BIOS is a bit of a mystery to me. I knew there were versions that accepted only Conner drives and some that could use others. But I never knew they were made by two different providers. How do you even begin looking into BIOS of these machines, let alone modify it?
After some googling, I could see that there are indeed two different versions of BIOS screen that look nothing like one another (see below). Is one Award and the other one Phoenix?
After some googling, I could see that there are indeed two different versions of BIOS screen that look nothing like one another (see below). Is one Award and the other one Phoenix?
Yes, the first one is the Phoenix BIOS and I assume the second one is the Award BIOS although it looks a little different on the T5200
Can you please post your custom BIOS? Does it have XTIDE integrated?
Up until today, XTIDE wasn't integrated and was running from a network ROM socket.
However, while I was looking at the various BIOSes for the T5200 and T3200SX I noticed that there was an unused space in the Award BIOS versions for both types of machines of approx 5.5K and so I made a stripped down version of XTIDE with all the unused options removed and it just fit in that space.
After patching the BIOS init code, I now I have XTIDE integrated into the BIOS EPROM itself and I should be able to do the same for the T3200SX.
I'll post the BIOS as soon as I've finish testing it although there are now so many patch options I may have to write a config program so you can create a custom patched version tailored to your needs.
I wonder if such bios would work also with T3200SXC (the one I have).
Mods I did here:
replaced original, faulty TFT color screen (it is almost 100% white all the time, and that's a common issue; by the way it's really thick) with a new one, much newer, intended for use with industrial machines. Old screen had 9bit interface, and the newer one had 18bit interface, so I just left half of the R/G/B connectors unconnected and it worked perfectly fine. First I just soldered the wires, but that was a ugly solution so I etched a PCB board with pin header for the LCD cable coming from the mainboard and pads for soldering a flex cable connected to the new screen. New screen had much smaller external frame, so I 3D printed some adapters.
Second mod: that's a simple one, i used hot glue to mount a set of small speakers inside (taken from modern laptop), and soldered the wires to an ESS sound card which had a volume potentiometer 😁 so now I have a fully functional 1990 "laptop" with active screen (nice contrast and response time) and sound; modifications are not visible from the outside.
Third mod: I desoldered built-in ram chips, soldered DIP sockets and tried to replace them with different ones - because sometimes I was getting "PARITY CHECK" error, but now I think that it might have just been a Parity_Boot.A virus 😜
I also think about making new power supply because the original one is big, heavy and noisy (the fan is loud), but the loudness is a "feature" of this machine and trying to change it would remove some authenticity 😜 anyway, it would be really easy to make a PSU which would fit there, off-shelf buck converter modules could be used, and a -5V and -12V rails can be made with mc34063 converter. I already made a really simple prototype some time ago Re: ATX to AT psu converter diagram?, but I would also like to add some protection, like a crowbar circuit or a microcontroller which would measure temperature/voltages and shutdown everything (with a relay?) if something goes wrong.
I wonder if such bios would work also with T3200SXC (the one I have).
Post a dump of your BIOS and I'll have a look.
adalbert wrote:
so now I have a fully functional 1990 "laptop" with active screen (nice contrast and response time) and sound; modifications are not visible from the outside.
Nice mods! I wired my internal sound card to the existing "beep" speaker but it's a little quiet.
I'm adding this info here Ian ..as it might get buried on my thread..
rgds
VS
---------------------------------------------------------------
My 2x NEC D27C1024D chips arrived today ..
Claimed erased/blanked but both were attempted to flash the modified bios..
both failed at around 30% ..
12.5v VPP - 100 us delay - VDD 6.25v ..
So out came the UV eraser ..and both chips subjected to 20 mins UV ..
This time both chips appeared to program and verify ..
No time to test them tonight ..but hopefully over the weekend ?? 😉
Tested the chips this morning ..
neither will boot the 5200.
I do not believe it is the bios firmware ..but rather the chips at fault..
I erased one of them and flashed with a copy of my own bios ..thus cloning it..
This cloned bios will not boot either yet the original does..
All chips were flashed and the verified ..and passed..
The only other thing I can think of is that I have to ignore the chip IDs to use them in my TL866 ..
I wonder if somehow the bios actually checks the chip ID before booting it ..and if it fails ...refuses to boot ????
I am certainly no expert in this regard.
I've since ordered 5x more chips for 5 dollars from AliExpress ..so at least 20 days to get here..
These chips are exactly the same number as the original bios ..made by Atmel AM27C1024-205DC
Wheras the 2 I recently purchased were NEC D27C1024D-15
Ian ..?? ..could this small difference in chip type be an issue ?
Or could it be because my programmer does no recognise the chip ID and has to be ignored during read/write operations ??
I'm at a loss here
rgds
VS
Tested the chips this morning ..neither will boot the 5200. I do not believe it is the bios firmware ..but rather the chips at fault..
I erased one of them and flashed with a copy of my own bios ..thus cloning it.This cloned bios will not boot either yet the original does.
Not sure what the problem is there, I'm using the same chips (D27C1024D-15) from the same supplier so there shouldn't be any fundamental problem with them.
Are there any beeps on startup? Make sure you are plugging them into the right socket as the keyboard controller in front is in the same ceramic package.
Also I suggest clearing the CMOS by removing the battery lead for 30 secs or so when changing between the two types of BIOS (Phonenix & Award).
I've finished incorporating XTIDE into the BIOS now so you can use any size drive on the built in IDE port. It also supports a slave drive although that is only practical with CF cards due to lack of space for a second drive.
I've also made the same patches to the T3200SX BIOS although that is completely untested at the moment (any volunteers?)
I'll post them both sometime later today if tests on the T5200 one go OK
IanB wrote:Not sure what the problem is there, I'm using the same chips (D27C1024D-15) from the same supplier so there shouldn't be any fun […] Show full quote
Vipersan wrote:
Tested the chips this morning ..neither will boot the 5200. I do not believe it is the bios firmware ..but rather the chips at fault..
I erased one of them and flashed with a copy of my own bios ..thus cloning it.This cloned bios will not boot either yet the original does.
Not sure what the problem is there, I'm using the same chips (D27C1024D-15) from the same supplier so there shouldn't be any fundamental problem with them.
Are there any beeps on startup? Make sure you are plugging them into the right socket as the keyboard controller in front is in the same ceramic package.
Also I suggest clearing the CMOS by removing the battery lead for 30 secs or so when changing between the two types of BIOS (Phonenix & Award).
I've finished incorporating XTIDE into the BIOS now so you can use any size drive on the built in IDE port. It also supports a slave drive although that is only practical with CF cards due to lack of space for a second drive.
I've also made the same patches to the T3200SX BIOS although that is completely untested at the moment (any volunteers?)
I'll post them both sometime later today if tests on the T5200 one go OK
Some news re the bios chips ..
My own cloned bios is now working ..I left the rtc batt disconnected ..and re-seated the chip ..and now it is at least beeping and booting to the point where it asks to strike F2 run setup..
The modded bios however still wont even get to the beep point..
I'll try erasing and flashing again ...though If you post a revision soon I may wait for that ..simply because of the time needed to UV erase..
..a question if I may Ian ..
are you testing on a basic unmodded 5200 as is mine ? ..or on your beautifully modded unit ?
btw ..I may well be able to help test 3200s soon ..as the cracked screen unit I bought will hopefully arrive next week..
providing that is ..that the vga is initialised at bios access ..and doesn't require reset by the plasma screen first.
(a sort of chicken and egg situation ..🤣)
Now I'm just plain confused..
I have keyboard unplugged
Floppy unplugged
Connor HD unplugged
RTC battery unplugged
I can boot the bios with my phoenix original ..or the cloned phoenix bios chip ..
I erased the 1024 and this time flashed with the unmodded award bios ..
Still no boot as with the modded award bios..
Something about my 5200 doesnt appear to like the Award ??
rgds
VS
Last edited by Vipersan on 2018-01-21, 12:21. Edited 1 time in total.
Tried again ..after pulling the simms ..
Still nothing from the award ..
but I still can get this far with the cloned Phoenix chip.
sorry about the camera flash
Something about my 5200 doesnt appear to like the Award ??
Edit:
It turned out that the machine had a later 32K VGA BIOS instead of the previous 24K one and that required the 3.xx Phoenix or later 5.30 Award BIOS to work
Here is a zip with all the main BIOS and VGA BIOS versions found so far including patched versions:
This contains BIOS versions 1.20, 1.30, 3.00, 3.10 & 5.30 and the 24K and 32K VGA BIOS
All versions except for 1.20 have been patched and there are two different versions of each patch, one with XTIDE that uses workspace in low memory and one that uses workspace in high memory. See the readme.txt file in the zip for more details.
I'm currently using "T5200 BIOS V5.30 (Award) patch 1.2 with LOMEM XTIDE 640K.BIN" but the others are included in case of any compatibility problems.
As the T3200SX is very similar, I've patched some of the BIOS versions for that as well:
These BIOSes have XTIDE Universal BIOS built in, support for a 3.5" second floppy, slave IDE and they will initialise 486 upgrade processors on the T5200. No guarantees, use at your own risk.
I will update this post with any new versions if required and the current patch update is 1.2
The changes from the previous 1.1 patch posted here are:
1. Enable both A20 and Flush mods if A20 mod detected (previously only A20 enabled)
2. Add HIMEM version of XTIDE
The only reason to upgrade from the 1.1 patch is if you intend to install an upgrade processor and do the A20 & Flush mods for improved performance or you find some compatibilty issue (1.1 only had LOMEM XTIDE)
Last edited by IanB on 2018-02-27, 01:01. Edited 6 times in total.
I will get on that when my 3200 arrives ..
I'll help if I can ..
and a patched Phoenix may well be what I need ..
I really do appreciate your help Ian..
..as to the clock ..that may also need a fix ..
it wraps around when set to 2018 ...and results in 1918 being displayed in win3.1
millenium bug ..???lol
..then again ..the bug could be in win 3.1
rgds
VS