Cool, which LCD panels have you ended up with? I've got a Toshiba LTM10C209 and that's an industrial panel that's very tolerant […]
Show full quote

BitWrangler wrote on 2023-06-09, 02:50:

Thermalwrong wrote on 2023-06-09, 01:21:

Some years ago, my boss at work threw away a little 8" LCD tv from Maplins, which I kept. It could take VGA but looked terrible […]
Show full quote

Some years ago, my boss at work threw away a little 8" LCD tv from Maplins, which I kept. It could take VGA but looked terrible doing it with awkward scaling, but it had a 640 x 480 LCD in it, so I threw away the TV and kept the LCD.

Some months ago I made this PCB for the 44-pin LCD connector that you can find on many industrial PCs and single board computers. Initially I thought it was advantech specific but it seems to be standard across many of the Taiwanese integrators that were making x86 SBCs.
advantech-pca-6145-lcd.jpg
The LCD that I took out of that 8" LCD TV uses a 32-pin flat flex cable with I think 0.5mm pitch pin spacing that's way too small for me to solder, thus this PCB was necessary and it's worked for several other LCD panels but not this little Sharp 8" LCD and I had no idea why. It's been on my 'please fix this at some point' pile for a long time.
It would work with this board but it would show noise on any details on the screen like in the BIOS screen the red & white text would have some break-up here and there. In Windows any details like the start menu were illegible and when doing the Shut down Windows screen which does the cross-hatch thing, the display would just break entirely on displaying that.
Initially I thought it was down to the BIOS on this Advantech PCA6145 and other SBCs that had wrong details to drive this LCD panel since the LCD worked with its original board and the digital picture frame driver boards, where I'd sourced other 640x480 TFT LCDs from digital picture frames. I'm upcycling them 😀

Going through it all on the oscilloscope looking at signals I noticed that the picture frame boards did drive the screen at a higher frequency and was trying to figure out if there's a way to make the SBC's BIOS match this, but I couldn't see a way to do that. Then I had a read of yyzkevin's experiences with getting a VGA LCD working with custom wiring: https://www.yyzkevin.com/clock-signal/
I am so glad this information is shared, the frequences matched up to what my SBC was putting out at ~25mhz so maybe there's no way to change the frequency. Also the amplitude of my clock signal was all wrong, 8 volts peak to peak?

filtering-for-clock.png
I found that my SBC doesn't implement any filtering on the motherboard for the LCD signals so my PCB should have been doing that. To fix that I cut the trace for the clock signal and started bodging on new parts, to match up what the Chips & Tech CT65550 datasheet specifies for filtering. 220pf capacitor and initially I tried resistors in line, that was bad.
44-pin-lcd-connector.jpg
But then I stole this blue filter thing from a scrap PCB's USB data lines and tested again, now the display is 100% working! I need to figure out what this part is called so I can get a fixed PCB version eventually 😀
The clock signal looks cleaner on the scope, the display is great now, it even handles the cross-hatch in the Windows shutdown screen, pixel perfect. All it was was just how 'clean' the clock signal is when there are ~20 other signals in total.
little-sharp-lq080v3dg01-lcd-works.jpg
This means I can start making something to hold this screen and an SBC at last.

Thanks Kevin for sharing your experiences with LCD troubleshooting 😀

Nice stuff. I added a couple more photoframes to the pile to play with. Gotta source those ribbon connectors and hookup stuff.

Cool, which LCD panels have you ended up with? I've got a Toshiba LTM10C209 and that's an industrial panel that's very tolerant of the messy signal I was feeding it. But I don't like its colour reproduction.
Another interesting one I got that takes a flat-flex cable is the Sharp LQ104DG83 - these Sharp panels look great with nice viewing angles, like this little 8" LCD and like on my Toshiba Satellite 400CS that I upgraded to TFT. This Sharp panel came out of a Texet 10.4" digital picture frame and these are the best made ones I've seen so far. Most of the no-name ones were made as cheap as possible but this one has an ESS ES830FAA chip making the video and a separate TSUM16AK driver board that works like a VGA LCD display with screen controls - hope I can figure out how to get VGA into one of these.

Finally got this LQ080V3DG01 behaving with the smaller SBC that I wanted to use it with, this one didn't go so easily. It was still giving noise on complex enough screen:

• In line emi filter on clock line = good improvement, went from unusable to noise on some screens
• Shorter cable = good improvement. This cable is too short but it gives 100% noise-free display
• Bypass capacitor for LCD power = no change
• Changing connector to top opening one = possibly some improvement though that may have been correcting a soldering error for the emi filter.
• Adding copper guitar tape to the outside of the 20cm flex cable, with tape around that for insulation = good improvement, clear & stable on windows shutdown screen at last

IMG_1440 (Custom).JPG
Very happy with this now, since the LCD is being driven directly by the graphics chip I think there should be less latency than if it was going through a VGA LCD display. A little LCD with a tiny PC104 SBC is my kind of 486.
Once I tidy up the design a bit I can share the PCB design for this adapter board, you'll notice it has a lot of bodges, but since I only had to worry about 5 of them it's easier to re-work what I've got.

Not sure if I mentioned the others when you posted before, only extra one I've got deets for here is Coby DP1052 photoframe that has a LG/Philips LB104 S02 panel in which details seem a little scant for but appears to be RGB 6:6:6 ... another one I picked up was a bust and had LVDS panel... but, just discovered a Johnstown Atom board I have might be able to drive it, so might get bodged to that for an XP system.

I have a few bus buffer chips, I'll have to look up if they are fast enough to clean up signals some. Without getting too analog and into the RF magick, I think it would be desirable to do low pass filters where the entire signal bandwidth is below it, and the glitches above it hopefully get blocked. I dunno if I'm making too much sense at the moment, having a bad brain day.

Ahh, that does look like an awkward situation for cooling if you want to use both RAM slots. Personally I'd wire tie a fan so it sits at an angle over the RAM and the top of the Voodoo 3000. It's a shame that these have become such valuable cards but I guess there's a limited number now.

BitWrangler wrote on 2023-06-10, 02:20:

Cool, which LCD panels have you ended up with? I've got a Toshiba LTM10C209 and that's an industrial panel that's very tolerant […]
Show full quote

BitWrangler wrote on 2023-06-09, 02:50:

Nice stuff. I added a couple more photoframes to the pile to play with. Gotta source those ribbon connectors and hookup stuff.

Cool, which LCD panels have you ended up with? I've got a Toshiba LTM10C209 and that's an industrial panel that's very tolerant of the messy signal I was feeding it. But I don't like its colour reproduction.
Another interesting one I got that takes a flat-flex cable is the Sharp LQ104DG83 - these Sharp panels look great with nice viewing angles, like this little 8" LCD and like on my Toshiba Satellite 400CS that I upgraded to TFT. This Sharp panel came out of a Texet 10.4" digital picture frame and these are the best made ones I've seen so far. Most of the no-name ones were made as cheap as possible but this one has an ESS ES830FAA chip making the video and a separate TSUM16AK driver board that works like a VGA LCD display with screen controls - hope I can figure out how to get VGA into one of these.

Finally got this LQ080V3DG01 behaving with the smaller SBC that I wanted to use it with, this one didn't go so easily. It was still giving noise on complex enough screen:

• In line emi filter on clock line = good improvement, went from unusable to noise on some screens
• Shorter cable = good improvement. This cable is too short but it gives 100% noise-free display
• Bypass capacitor for LCD power = no change
• Changing connector to top opening one = possibly some improvement though that may have been correcting a soldering error for the emi filter.
• Adding copper guitar tape to the outside of the 20cm flex cable, with tape around that for insulation = good improvement, clear & stable on windows shutdown screen at last

IMG_1440 (Custom).JPG
Very happy with this now, since the LCD is being driven directly by the graphics chip I think there should be less latency than if it was going through a VGA LCD display. A little LCD with a tiny PC104 SBC is my kind of 486.
Once I tidy up the design a bit I can share the PCB design for this adapter board, you'll notice it has a lot of bodges, but since I only had to worry about 5 of them it's easier to re-work what I've got.

Not sure if I mentioned the others when you posted before, only extra one I've got deets for here is Coby DP1052 photoframe that has a LG/Philips LB104 S02 panel in which details seem a little scant for but appears to be RGB 6:6:6 ... another one I picked up was a bust and had LVDS panel... but, just discovered a Johnstown Atom board I have might be able to drive it, so might get bodged to that for an XP system.

I have a few bus buffer chips, I'll have to look up if they are fast enough to clean up signals some. Without getting too analog and into the RF magick, I think it would be desirable to do low pass filters where the entire signal bandwidth is below it, and the glitches above it hopefully get blocked. I dunno if I'm making too much sense at the moment, having a bad brain day.

I think I got lucky getting 640 x 480 ones, it's worth checking up the specs or online manuals first. Mine all seem to have all either 32-pin flat cable, and Sharp or Toshiba pinout 31-pin hirose connectors. In my opinion there's not that much to worry about with RF for TTL LCD panels; it's only the clock signal that's running at a higher frequency over 20mhz. Everything else is pretty much in the hz or khz range so signal integrity of the digital data isn't much of a problem.

It gets easier with LVDS too, that has built-in noise filtering for all the clock and data signals on just 4 wire pairs. You just need to make sure each wire pair is twisted together and it should work without issue. I've even seen bodges of LVDS cables without the proper twists working quite happily at 1024x768.

Thermalwrong wrote on 2023-06-10, 01:40:

Is that just a generic ATX breakout board? That looks great. I've gotta try something similar with a DigiPOS computer that has a crazy external PSU.

No, a French seller on eBay was making them two or three years ago. There's a website address silkscreened on them but it fails to load now. I hadn't seen them listed again so they may not be selling them anymore.

I feel like I've seen similar at Universal Solder, Tindie, or Temu or somewhere like that.

I've been using Gotek floppy emulators exclusively on my test bench. I tried moving one into a real case today and found that the screw holes are at a level that won't work with my case. My real floppy drives line up just fine. Anyone else ever have this problem?

Took the heat sink off my Voodoo 3 to better figure out the best way to get a fan on it, and replace whatever heat sink compound was on there and, well:

Yes, it was bone dry under there. I felt a little ill. Again, I'm new to Voodoo 3s, is this a common thing I should have known?

Anyway, glad I'm getting it sorted now. Got the ram clearance licked too: 😂

ubiq wrote on 2023-06-11, 04:49:

Took the heat sink off my Voodoo 3 to better figure out the best way to get a fan on it, and replace whatever heat sink compound was on there and, well: IMG_5026.jpeg

Yes, it was bone dry under there. I felt a little ill. Again, I'm new to Voodoo 3s, is this a common thing I should have known?

Anyway, glad I'm getting it sorted now. Got the ram clearance licked too: 😂 IMG_5028.jpeg

If the heatsink had never been removed before then it'll still have the thermal epoxy glue they used, so yes it would appear to be bone dry which would be perfectly normal for epoxy. But now you have it off you can throw some superior TIM on it with a nice fan and itll run far cooler than it ever did stock.

nVidia and ATI also used the evil Thermal Epoxy .. Geforce 4 cards were a pain in the arse to replace the cooling on because of it.

Trashbytes wrote on 2023-06-11, 05:02:

ubiq wrote on 2023-06-11, 04:49:

Took the heat sink off my Voodoo 3 to better figure out the best way to get a fan on it, and replace whatever heat sink compound was on there and, well: IMG_5026.jpeg

Yes, it was bone dry under there. I felt a little ill. Again, I'm new to Voodoo 3s, is this a common thing I should have known?

Anyway, glad I'm getting it sorted now. Got the ram clearance licked too: 😂 IMG_5028.jpeg

If the heatsink had never been removed before then it'll still have the thermal epoxy glue they used, so yes it would appear to be bone dry which would be perfectly normal for epoxy. But now you have it off you can throw some superior TIM on it with a nice fan and itll run far cooler than it ever did stock.

nVidia and ATI also used the evil Thermal Epoxy .. Geforce 4 cards were a pain in the arse to replace the cooling on because of it.

you just twist before you pull the radiator , so you don't pull the chip with the Thermal Epoxy.

ediflorianUS wrote on 2023-06-11, 10:29:

Trashbytes wrote on 2023-06-11, 05:02:

ubiq wrote on 2023-06-11, 04:49:

Took the heat sink off my Voodoo 3 to better figure out the best way to get a fan on it, and replace whatever heat sink compound was on there and, well: IMG_5026.jpeg

Yes, it was bone dry under there. I felt a little ill. Again, I'm new to Voodoo 3s, is this a common thing I should have known?

Anyway, glad I'm getting it sorted now. Got the ram clearance licked too: 😂 IMG_5028.jpeg

If the heatsink had never been removed before then it'll still have the thermal epoxy glue they used, so yes it would appear to be bone dry which would be perfectly normal for epoxy. But now you have it off you can throw some superior TIM on it with a nice fan and itll run far cooler than it ever did stock.

nVidia and ATI also used the evil Thermal Epoxy .. Geforce 4 cards were a pain in the arse to replace the cooling on because of it.

you just twist before you pull the radiator , so you don't pull the chip with the Thermal Epoxy.

Never had any success with that method, freezing the cooler with an air blaster does work though as it makes the epoxy brittle and itll generally come away in full with the cooler.

Id rather not twist old coolers in any case, the BGA chips don't like stress like that and it can damage the BGA solder balls.

Big Pink wrote on 2023-06-10, 15:04:

Thermalwrong wrote on 2023-06-10, 01:40:

Is that just a generic ATX breakout board? That looks great. I've gotta try something similar with a DigiPOS computer that has a crazy external PSU.

No, a French seller on eBay was making them two or three years ago. There's a website address silkscreened on them but it fails to load now. I hadn't seen them listed again so they may not be selling them anymore.

Ah, good to know. If they're not generally available then I'll probably just solder stuff onto one of those ATX 20 > 24pin cables and use a similar picopsu. Your upgrade is very tidy and professional looking even if there's a lot of wiring 😀

ubiq wrote on 2023-06-11, 04:49:

Took the heat sink off my Voodoo 3 to better figure out the best way to get a fan on it, and replace whatever heat sink compound […]
Show full quote

Took the heat sink off my Voodoo 3 to better figure out the best way to get a fan on it, and replace whatever heat sink compound was on there and, well:

IMG_5026.jpeg

Yes, it was bone dry under there. I felt a little ill. Again, I'm new to Voodoo 3s, is this a common thing I should have known?

Anyway, glad I'm getting it sorted now. Got the ram clearance licked too: 😂

IMG_5028.jpeg

Good solution, going to EDO might have a bit of a performance hit though??? Probably worth testing which is faster.
Thanks for sharing the pictures too, I don't know that I've ever removed the heatsink on my V3 3000 so I hope that one at least has epoxy rather than no thermal interface at all.

Cleaned the case from the 386dx40 (well it's an Intel 33 overclocked for 20+ years) I got earlier this year. Back then I tore out the motherboard to remove the starting-to-leak Varta battery, now the case got a good damp cloth, and I reassembled the computer. I foresee some trial and error with the connectors to the case, though 🤣

And also concluded that never having build Mini AT systems back in the day, were no greater loss for me 🤣 wow, they're compact and counterintuitive to work with ...

Thermalwrong wrote on 2023-06-11, 19:49:

Thanks for sharing the pictures too, I don't know that I've ever removed the heatsink on my V3 3000 so I hope that one at least has epoxy rather than no thermal interface at all.

Well, I can say for certain that mine showed no evidence of any sort of thermal coupling agent whatsoever. Just the ugly marks of aluminum abrading on ceramic. The heat sink could be wobbled by hand on the mounting springs before removal, so def no adhesive.

Thermalwrong wrote on 2023-06-11, 19:49:

Good solution, going to EDO might have a bit of a performance hit though??? Probably worth testing which is faster.

Good point, going with this now:

😛

Just bought the geforce 2 mx 400 64mb, and I'm waiting for it to arrive ^^

I definetely didn't get up to do that. But using my W95 HDD on an IDE->USB reader corruped the system files.
Now getting it fixed the lazy way.

I was trying for ages to fix up my Pentium 1. It was acting all erratic for ages , wouldnt run any games for long before rebooting or bluescreening.

I tried everything , but in the end, I think its a power issue. I think the PSU is going down slowly. I hope the Mobo is fine though. Not all was lost though as I found a 5 dollar computer at the thrift store, which had an assus socket 7 motherboard with an AMD K6 266 chip and a voodoo 2, 3000 AGP card in it.

I tried it out and didnt like it, its ran slower than my P1 266 MMX.
So I just took out the K6, and replcaced it with my 266 MMX and had to change some jumpers becuase the K6 was 2.2 volts while the P1 is 2.8 I also had to change the FSB jumpers and boom. Works great. Though I think the Voodoo3 is overkill and MAY replace it with one of my voodoo 2 cards, or even a voodoo 2 sli setup and put the voodoo 3 in a P3 I have.
Anyway, I'm happy to have a nice working P1 machine again and the first game I want to beat is Fallout.

I also started a game of Unreal on it as well and it plays just like I remember. Loving it.

Also im going from 32 megs of EDO RAM to 128 of SDRAM

Some pics
The old board on the right with the new Asus board on the left

The old case

The two voodoo 3 cards

The new case

My helper

Convert SIMM RAM to SIPP.
What better way to relax, than doing small welds 😀

Thermalwrong wrote on 2022-11-03, 18:06:

Today I had to do two completely unwanted repairs on my Zenith Z-Note GT 120mhz Pentium - which I was putting back together, giv […]
Show full quote

Today I had to do two completely unwanted repairs on my Zenith Z-Note GT 120mhz Pentium - which I was putting back together, giving its screen back (borrowed for tests) only to find that it wouldn't turn on. It was behaving like there was a short.
I tested between a component labelled BCxx near a component and both sides connected to ground, oh no. There was a dead short on the 5v power to ground. Had to take it all apart which is tough considering I was never able to find a service manual for this thing, here's how it looks in one piece - not mine but very similar: https://wiki.preterhuman.net/Zenith_Z-Note_GT
After probing around for about half an hour, I looked more closely at this metal shield that blocks the PCMCIA slot from the inner case, it screws and solders onto ground points but at the corner it's going over a trace that connects to a fuse... I desoldered that shield and the short was gone. What a relief, I don't want to do component level troubleshooting on this old thing. Resoldered that back into place with some Kapton tape underneath to give some insulation.
z-note-gt-150-shield-shorted.JPG

It now powers up though I'm not running it until the casing is back together since the crazy design means the top case is half of the heatsink and there's no fan. The way it's designed is pretty bad in my opinion, the heatpipe has to be carefully guided through a narrow hole and from my last work on it, the aluminium shield/heatsink got separated from the plastic top case. I broke all the plastic weld points - having these as two parts makes it even tougher to put back together, so I've fixed all of the plastic pegs with superglue and baking soda. The edge of the keyboard area on the top case is a thin part and that was broken in half, so I used some 2.54mm header pins and melted those into the ABS plastic after supergluing.
Beware if doing that, the fumes from superglue melted are quite bad for you, I had a fan running to avoid those.
z-note GT-150-topcaserepair.JPG

When back in one piece, it's a pretty nice albeit quirky laptop. Similar in spec to the Toshiba Tecra 500CDT, or the Satellite 430CDT but with L2 cache. Mine has a smashed screen but one day I'll resolve that too.

It's been a while since I've posted about my Zenith Z-note GT laptop, which had a broken screen. It's been sitting around the flat and on my desk as a pile of parts.
When received a while back, it came with a smashed LCD - it's an NEC 11.3" (NL8060BC29-01) and that's really hard to get replacements for. Another dead laptop that I disposed of had a Sharp LQ11S31 11.3" 800x600 TFT and I thought it would be a simple process to get the Sharp working in place of the NEC panel.

It wasn't that easy though - my adapter made from the LCD board that connected to the NEC LCD, which was wired up using those thin copper wires from an 80-pin IDE cable, wasn't good enough. I spent ages making this Sharp LCD some new standoffs in the screen housing and all it would display was a rolling noisy picture on the LCD. I thought it was a video BIOS problem and this laptop has sat on the backburner.

In the last couple of weeks I've learned about the importance of signal quality for high frequencies (40mhz) from other LCDs I've been messing about with on industrial PCs. And how to improve it without getting into the physics stuff, with ferrite cores and shielding.

I found a nicely made cable that had the same LCD connector in my parts bin(s), a 41-pin Hirose which had lots of shielding tape and little ferrite cores covering a few pairs. I re-wired this "ASKA rev.0" cable that may have possibly come from a Toshiba 1620CDS, to fit the 41-pin Sharp pinout. The wire colours were picked so that red/yellow was colour data and white/black were grounds, which made figuring out which wire is what much easier.
The Clock and DE signals each go in their own shielded wire pair with ferrite beads.

The connector on this cable I'd use instead was totally different from the connector used by this Zenith Z-Note GT laptop, which is a generic Taiwanese made laptop with a Chips C&T CT65548 graphics chip. There are three connector headers:
LCM - 15 pins - This I think means 'LCD Mono' with the minimum viable number of pins to run a monochrome LCD. That would be odd on a Pentium
LCD1 - 14 pins - These are I think the extra pins for DSTN and TFT screens,
LCD2 - 8 pins - Extra signals for TFT screens, adds 6 colour signals. Toshiba do the same thing

To figure out what the signals were on the three connectors, I worked back from the known pinout of the original NEC LCD panel.
39 wires and lots of heatshrink. It barely all fits into the space that the original cable took but it's working and the covers fit back into place.

It's not 100%, I didn't place the screen quite level with the new standoffs I had to make for the LCD. And it does seem there is a minor difference in the LCD panels, because the display although working correctly is shifted down by ~3 pixels. I might fix that in future through potentially modifying the video BIOS, but for now it's working and the LCD is actually pretty fresh and bright.

The laptop itself is nothing all that special hardware wise, a fairly generic Pentium with ES1868 audio and Chips VGA. The Z-note GT is a Zenith rebrand of a Taiwanese laptop chassis which I've seen other brand names on.
But I need to fix things when I receive broken stuff, so I'm happy at this point, and I learned some lessons about signal quality along the way - the mess of wires in this thing helps me to understand why the industry pivoted to LVDS all at the same time.

Repaired HP version of motherboard's SiS 760L north bridge's heatsink missing hoop by making new hoop from stranded wire reduced in number wires to fit into the through hole, soaked the wire in solder for rigidity, and soldered it in. When board is hot, I removed stuck on heatsink then cleaned off the famous pink phase change TIM from both northbridge's heatsink and chipset while both are hot. Finished with low odor mineral sprits to get the stickiness off, 99% alcohol wouldn't touch that.

pink phase change TIM is strongest bond due to different type of carrier glue. If you have this type, do not attempt to remove since bond is very strong, unless heated some.

Cheers,

OSkar000 wrote on 2023-06-08, 20:55:

Tried to get some more life signs from my HP Vectra XU 5/90 today.. not much to report unfortunately. […]
Show full quote

Tried to get some more life signs from my HP Vectra XU 5/90 today.. not much to report unfortunately.

received_790597562572372.jpeg
Signs of life!

I would say its orange and yellow. Not mentioned in the manuals that I have found so far.

The cpu gets warm
Chips on the motherboard gets warm
Keyboard blinks at start
Soft power on/off works most of the time.

IMG20230608223838.jpg

I have tested with one and two known good Pentium 133 and two sticks of FPM memory, also tested and working.

Any ideas 😀

Maybe check the internal of the power supply (capacitors, etc).
Keep the move, HP Vectra XU 5/90 it's a very nice machine.