👍

>I'm on the fence over whether I should make a proper display adapter PCB in place of this modified cable.

I wouldnt bother.

Well it does work without noise at this point but it's certainly a bodge 😀 And if anyone else in the world wants to try the same thing it would make it far easier to do.
If anyone's interested, I can share the derived pinouts from the 410CDT's lcd cable harness like where HSYNC, VSYNC and all the colour signals connect up

The extra connector for the TFT's wiring adds in 6 new signals, I'm pondering whether it would be better use the original cable and just add in the extra 6 wires. But that requires someone other than just me wanting to upgrade their 'CS' laptop.
It turns out that 640x480 TFT panels are actually pretty easy to get - in the UK I can get 10.4" panels from digital picture frames, or new 8.4" VGA panels from a cancelled project. And I've got 10.4" 'Ampire AM640480E' ex industrial use panel on its way to me that cost all of £6.50 and may work (but needs the pinout changed, hence my interest in adapter boards)

Today I put the new display cable into the 400CS's screen housing - it's quite a thick cable and really makes me appreciate how much LVDS improved things.

But it's all together now with the stand-in LCD temporarily mounted with some foam blocks at the bottom of the screen to hold it up - it's the wrong size and the LCD is damaged but it looks great!

Of course, there would be more problems.
That bundle of epoxy coated wires is causing issues because when pressed, something is shorting or disconnecting and the screen turns off. One time the whole laptop shorted out and that may have been my fault for not putting insulation on the LCD's 31 pin hirose connector.

I borrowed the LTM10C021 from my Toshiba T4900CT laptop since I wanted to see if the CCFL backlight was going bad on it, it's always pretty dim in the T4900CT but seems to be otherwise fine. This let me test the fit of the original 400CDT's LCD panel in this 400CS and I discovered a new problem...

The LTM10C021 fits in place nicely, all the screw holes line up as they should. I think Toshiba would have designed this laptop around the LTM10C021 panel, then got Sharp to make the DSTN panel to fit those specifications.

This means that Sharp's display fitment in terms of screws and LCD connector are the same, but the viewable area placement within that frame is different. I think Sharp would have adapted an existing panel design to fit Toshiba's specifications, then Toshiba would have made a new display bezel to fit Sharp's resulting LM64C389 LCD panel.

I'm not making a new display bezel for this thing, nor is it possible / worth it to source the 400CDT's bezel.
I can use a smaller-framed 10.4" LCD and adapt that. I got this Ampire AM6448 10.4" LCD that's basically got nothing sticking out of the sides so I'll see if this can work. I am going to make up an adapter after all, this Ampire display has a different pin layout from the Toshiba panel that I wired this up for originally and I don't want to do this over again with soldered wires, it's already barely working.

For anyone's future reference, with parallel LCDs like you'll find on all laptops up to around '96 there are the following factors to contend with when changing screens:

• LCD pinouts were barely standardised, there are 41-pin, 31-pin Sharp DSTN pinout, 31-pin Sharp TFT pinout and 31-pin Toshiba layout on this era of Toshiba laptops. Other manufacturers often use other connector types / pinouts too
• LCD form factors were not standardised at all, screw mount positions and LCD viewable area positions can be all over the place
• The TFT version of a laptop will usually have its specific cable
• The display bezel can also differ. On some the whole top housing could be different, my Kapok 6200AT has a display housing front & back specific to the LCD type & manufacturer.
• Selection between TFT / DSTN can be done without a different display BIOS on some laptops (like this one), but how that's done can be difficult to figure out. Just found that Toshiba have used the same Panel0 & Panel1 signals on the T21xxCS / CT laptops as well, so the T2130CS bios is the same as the T2130CT bios. But display selection is handled by detection of those two pins. Other manufacturers will have implemented their own solutions.
  Some laptops (usually smaller scale production) like my Kapok 6200AT just use different video BIOS files.

The easiest way to have a TFT version of a DSTN laptop is really to start out with that and find the TFT version, or use an external monitor. Next is to just swap the top half, like you can do on any of Toshiba's Satellite 400 to 430CDS/CDT and 200 to 210CDS/CDT models. Since all the differences are on the swappable video card, LCD plastics and display cable. From reading Creepingnet's posts, the NEC laptops are similar although require much less part swapping.

The method I'm using here, taking some parts from a TFT laptop and adapting those to this model, is a lot of work. It does show what's involved well, there are extra wires needed to drive a parallel TFT compared to a parallel DSTN screen, extra parts needed to send those signals and the BIOS needs to know which type of screen it's supposed to be driving.

Another cool solution though really hacky, is to just get a small VGA monitor and stick that in the display housing - these guys in Japan actually do that! https://pc-otasuke.jp/result/detail.html?id=13757
(eventually, that will be the way to go, all these LCDs are liable to fail in a few years when the polariser material breaks down, but that's many years away yet and LCDs can have new polarisers fitted)

Something I'm wondering if it would even be possible, would be to have some kind of microcontroller or other means decoding the DSTN data and using that to drive a modern LVDS/parallel TFT, but that would cause a bit of latency like the Pi's RGB to HDMI project. Kind of unnecessary when the video controller can be fairly easily told to drive the panel the correct way though.
I haven't mentioned it in this thread I think, but I've been following Kevin's & this site's work on upgrading the IBM PC110. Which only ever had an STN screen, to a TFT screen. These two sites have been very useful to understand what's involved, along with the Chips & Technology datasheets. They use a program or BIOS update to tell the C&T video chip to drive a TFT instead of DSTN.

Ah, I tried making a PCB to connect up the LCD but the cost, trouble of routing and wondering what I'd do with the remaining 4x PCBs put me off.
So I have re-made the wiring and connected up a 31-pin Hirose DF9 with Sharp pinout instead of Toshiba, since they are of course different. This time I used more tape and was more careful to keep the little wires separated near the parts that I was soldering - because the insulation can be damaged in the area that gets heated up and I can't deal with heatshrink at this scale / quantity.

I found out that the LCD I bought, an Ampire AM640480 from 2008, requires 3.3v to run instead of 5v. That sucks, it means I need a voltage regulator but it uses only 350-500ma so I can use a little AMS1117 3.3 regulator I have on hand.
Hopefully it won't overheat but this is a linear regulator so it will give off a bit more heat than a switching regulator.

This 10.4" LCD has no sides that stick out so it can be adapted to fit in the right space within the laptop's screen housing. It was cheap though and cheap comes with caveats, this was untested and shipped in a poor box so I discovered that both of the screen's glass CFL tubes were smashed and I had to fish broken glass out of the LCD so it can't get damaged. I'm still finding bits of glass on my desk.

I have an LED backlight retrofit kit that I've cut down to 10.4" long, which is now fitted in the screen. Sadly unlike old LCDs, the screen is *black* when fully backlit and shining a flashlight in the front I couldn't tell if it was smashed or not.
For this test run, bench PSU #1 is powering the screen so that I can test out how much pwoer the screen draws. Bench PSU #2 is powering the LED backlight in a janky way but it works.

I hate this part, finding out the fast way whether lots of things work all at once:

Thankfully VGA TFTs are all quite similar in terms of timing and how they run, it worked with this screen too. I'm stuck in DOS right now because for some reason I nuked the hard disk last time I was using it, but for like 1/2 a second I saw the Windows 95 boot logo in full colour.
I'm really glad I didn't spend out on the original LTM10C021 screen now, this Ampire LCD is supposed to have great viewing angles and it's so bright white with these LEDs. Gotta work on hacking in the LED backlight driver now, always more to do 😀

Here it is all test fitted, with the regulator installed and the screen fits into the housing. The colours on this are so bright and clear thanks to the LED backlight that it looks amazing:

It's running with the backlight powered externally, now I'm thinking of how to control the brightness.
With another LED retrofit on a Toshiba that had a software controlled backlight and 4/8 backlight brightness settings, I used a microcontroller and that was a lot of work.
This time I'm feeling lazy - why not use the Contrast wheel from the side of the display as a variable resistor to control the backlight brightness?

pretty nice really like the attention to detail and professional look of the finished product. I wanted/want to swap the 1024x768 panel out of my toshiba 2805-s402 and replace it with a 800x600 or a 1600x1200 one day for 1:1 scaling.

Haha, of course just a few days after I did all of this, I found and purcahsed a very cheap Toshiba Satellite 400CDT that seems to be functional. Quite looking forward to getting that.

This project is not done yet sadly, I found that the display does some crazy dithering in 256 colour mode but works okay in 16-bit mode. 24-bit just makes the display go completely broken 😀 I think it's due to the Ampire screen I'm using, which I've since found is a rebranded Chi-Mei G104V1-T01. That's an MVA panel which is why the blacks and colours look so good. It was made for retrofitting in industrial applications so the display itself contains a level shifter, along with a parallel to LVDS converter, since the Chi-Mei display controller chip actually just uses LVDS directly. Oh also, it does actually support 5 volts operation so the little regulator is gone.

An interesting thing I discovered, is that the Ampire LCD works perfectly with the T2130CT, even if it doesn't fit - no weird dithering problems:

However back on the 400CDT, It's so odd, I get what looks like dithering on only some colours, particularly gradients or dithered gradients in games. That's static though rather than jittering all over the place, so it's a consistent display problem that doesn't appear to be noise related.
If I set the Windows 95 display to 256 colours then the whole display ends up looking very strange, like the whole LCD has a grid filter over it.
I think it's a timing problem where the time period where the graphics chip sends certain colours or colour bits, ends up rolling over into the next pixel's timespace. My theory is that the LVDS conversion delays things enough that it can't work properly with the 400CDT's display timings.

The VBIOS on this thing is programmed for a Toshiba VGA LCD, I'm connecting up a weird retrofit Sharp-type VGA LCD and it seems the timings are off just enough that it's causing colour problems. But now that I've got the T2130CT, which has a Sharp LCD - when I try that on the 400CDT, it works without issue. Which is great, because that rules out a cabling problem from my special adapter cable.

However I don't want to take the T2130CT's screen and the Ampire LCD is the only screen that's going to be easy to fit, no trimming or breaking of LCD housing screw posts required.
I think what I want to do is figure out how to get the T2130's VBIOS display timings into the 400CDT's display timings. Kinda hard but they're both CHIPS video cards. I should really bust out the oscilloscope or logic analyzer to work out what the timing differences are, that's quite some lot of work though.
Using the Ampire display is what I'd like to do, the MVA display looks quite unique - though somehow the redraw of platformers ends up being a bit flickery, it looks good in games and the colours on it are much better than a TN panel.

Oooor, I can forget all that and just get the original LTM10C021 display, making a 3d printed copy of the bezel, but that also requires work to go back and do that.

I decided to finish this up - the arrival of the Satellite 400CDT and the repair of that offset things somewhat.

I got a Sharp LQ104V1DG52 LCD screen in a 10.4" DVB TV. It turned out to have none of the timing / weird display problems that I was seeing with the smaller screen. BIOS modding a Toshiba laptop is hard and I don't wanna.

Initially I thought it was just too big but with some trimming it was made to fit and upgraded to LED backlighting. This Sharp panel was meant for LCD TVs and industrial applications so it's pretty thicc. The trimming and removal of the dual CCFLs means it juuuust fits:

Some more trimming and some 3d printed 'measured by eye' brackets made to fit the screen in place:

I'm rather proud of what happens with the contrast wheel - last time I did a CCFL to LED retrofit I used a microcontroller and PWM to adjust the backlight's ADJ pin, which goes from 5v at lowest brightness to ~2 volts at top brightness at which points its power consumption passes the original CCFL inverter and the LEDs get hot. The contrast wheel being a variable resistor was perfect, just needed the resistors adjusted to limit the max brightness / lowest ADJ voltage:

White wire = ADJ wire, which further up has an electrolytic capacitor between ADJ and GND because the brightness flickers if the ADJ's 5v voltage moves about, like with hard disk activity, the capacitor smooths that out.
Red wire = 5v
Black wire = GND but notice it has a 10k resistor to limit what it can do, otherwise the backlight goes to crazy brightness

The long thin green PCB is a generic cheapy LED driver which I've servo taped into place. The blue PCB above boosts the 5v supplied to the original inverter, up to 12v. The screw mounting it to GND works nicely 😀
To connect the power wires into the original inverter connector, I found some really thin legs cut from a capacitor / LED and bent them over. There are 2x VCC (pin 6&7) and 2x GND pins (4&5) on that original connector and this handles it nicely compared to finding an original connector to fit.
There are three other wires that I'm not using from the inverter, those are the FLV0, FLV1 and FLHIGH signals. I can't really figure out what FLV0 / FLV1 do but FLHIGH is how the inverter switches from regular to high brightness. I like the analogue adjustment more 😀

And there it is, done and for now working. I like the Sharp LCD more than the 400CDT's Toshiba LCD. Both are TN panels but the viewing angles are better and with the LED backlight it looks really nice.