My last repair turned out to be not done yet. The Packard Bell PB430 mainboard didn't work well, let's see how PB410 would behave. Music by Model Povedeniya modelp.bandcamp.com/ Patreon: / necroware
Waiting for a video from you is like waiting for a girl to get ready. It might take some time but when they arrive you forget all about the wait and just enjoy the great moment. Can't wait for what you have in store in the next one!
@@necro_wareBrother, you are a machine! I just rewatched the first video before watching this one since your last update was a while ago. The amount of patience and diligent work that went into this is insane. I know that I would be burned out by the end of this adventure, but for you it seems to just be good for! Keep up the work! Looking forward to more of your videos!
@@necro_ware Of course you can investigate how to change the thing that was annoying to take turbo off to use the keyboard and instead use a button on the case for example. I had back in those days when i was 16 years old one of these packard bell concretly the PB430 (i remember it because of the cirrus logic CL-GD5428 which works out of the bok on slackware 3.0) and it was annoying specially if you want to use OS/2 Warp without hanging every 2 times 3, without knowing what is causing the hang.
Necroware, the king of all cliffhangers! :D I had a problem with a TMC PET48PN once, where 2 pins were connected inside a VLB slot. Data line 4 was connected to ground, which caused all sorts of unexpected behavior. One of such was the computer randomly hanging, always seemed to be running in de-turbo mode, but able to enable turbo nonetheless, which made it run even slower. Eventually, I found the pins inside the slot. I was working on it for months! Rabbit hole anyone!? Anyhow, thanks for the awesome video. I can’t wait to see the next one. :) Stay safe!
Thank you for repairing that PB410 board. That was my first PC in 1993. Still have fond memories of it. It's also emulated in PCem and 86Box if anyone is interested.
I like how you bend the repair wire to follow the shape of the original trace routing instead of a bodge wire just connecting point A to B. Much cleaner and superior attention to detail! Cool video.
Yes, that was the initial problem. Would I have tested the board with doom last time, I'd realize, that it has an issue much earlier. Of course I tested it this time running Doom, just didn't take it into the video ;)
Thanks for working on this, as I said in a previous comment, I have 2 of these that I need to do these repairs on as well and this is extremely helpful.
Great to see you back! Awesome work. It always amazes me just how much damage those batteries can cause. It's also very interesting how various manufacturers implemented the turbo functionality. I had always assumed it was L1 cache being disabled on most 486s and didn't know about the halt signal that some used too.
Always love seeing your repairs. Your channel name is quite fitting given the kind of stuff you've repaired before. Your trace repairs are always top notch looking.
Thank you very much for the video!! It was a wow moment i view two o three times or more this video. Two weeks ago I found two computers next to the trash. It turns out that I got excited and I said to myself, well, nothing's wrong, 90% of the problems that caused the computers to be thrown away can be solved. One had two capacitors (3300 microfarads 6.3 Volts) in bad condition. I changed them (I had a hard time removing them, but I replaced them with 3300 microfarads 16 Volts and the box was bad so I changed it. The other one had the box to throw in the trash again, I took advantage of it. To my surprise the motherboard is an Intel S3200SH a server board with an intel xeon x3220, curiously it works perfectly.
This takes me back! Back in the early 90's I had a 486 Packard Bell PC with the 410 mainboard! Even with all of the limitations of that board I was still able to modify and upgrade that machine well into the socket 7 days. The only thing I didn't manage to upgrade was the video RAM as I left it at 512KB. I added 128 KB of L2 cache, upgraded the CPU to a AMD 5x86 133MHz running at 120Mhz with a 5V-3.3V voltage adapter. Upgraded the RAM to 12MB total. Added a SB clone soundcard, 2x CD-ROM, a 2nd HDD, 28.8 US Robotics modem, external SCSI Zip drive, 10 Mbit NE2000 compatible 10base-T NIC. Loads of learning and good times were had with that machine!
23:55 "that was one of those wow moments" i bet you were jumping up and down in excitement, shouting "YES YES,YES! its working! Amazing. I can't believe it. Its only gone and bl00dy worked! Yes!
@18:09 That's incredibly neat. I'm doing exactly this repair on an identical board from a Zenith Z-Select 100 3x3 and I know how much patience and practice this takes! Excellent and inspiring :)
Great video! Sometimes I wish I could do what you do and other times I’m also glad I can not 🤠 Your videos are easy to watch and you make them enjoyable.
Thank you for this. Packard Bell machines deserve love too. The Pb in my username is kind of a nod to the brand and also to my love for heavy metal music, hehe.
Yay!!! You're back. And what a great video too! That was art. That music feels nostalgic now ;) Perfect intro, catch-up and bringing up to speed too. I look forward to the next one.
yeeeeey, finally!! i started to think you wouldn't post anymore. so relaxing to watch after a 60km bike ride into the mountains. my knees hurt like hell :)) amazing video and repair, awesome
I screwed up trying to swap out SIPP slots with SIMMs on a rare board... I have the stuff to repair it, but had my confidence shaken... your videos are great for building that back up!
Ganz viele Daumen hoch! Wieder brilliante Arbeit! Die Retrocommunity ist froh, dass nach 4Monaten wieder ein Lebenszeichen in Form eines Videos auftaucht.
Wonderful repair work. It honestly looks a little sloppy or cumbersome in magnification, but the final macro shot looks unbelievably tidy. I've resoldered pins to a CPU before and that was bad, but I can't begin to appreciate whether I would have the patience for this work. Great music choice for the montage too 😊
Beautiful repair work as usual, it's a shame you couldn't repair the other board in this video, but no doubt you'll get both running perfectly in due time
Hi Necro, nice video and extremely nice soldering. Regarding the speed just check your jumpers. Some of these boards behave weird if you have it set to pentium P24 of smth. Just google PB430 Processor Upgrade. Some boards also used Jumpers for maintenance modes which interupts the cpu. Greetings from Nurnberg.
These are godlike skills in repairing old motherboards. For me the best I can do is to change the bad caps. I saved a couple of motherboards that way. But I don't have the equipment, nor the skills for advanced repairs. Best regards.
That's what I call a pro repair, ie, wire soldered on the full length of the original trace - saw a bad repair vid today lol - I wonder if a mini sandblasting gun would be easier, but then again you would probably need a spray can of UV solder mask because it removes a larger area.
I had a similar issue on a Commodore A2386SX pc card for Amiga, where the turbo was always off. It was a chipset pin that was loose, normally connected to a pin on the KBC. But I had access to a clickable board view, schematics and chipset datasheet, way easier than the issue you're facing here! Looking forward to the next part.
Thank you for the very good video and your search and research. Happy to see that the old school skill can do miracle. So far I suppose 486 DX2 will be good for the board without turbo on - good so that all whistles and dongles as Cirrus video and so on can do perfectly. Very good thanks once more.
Great video. I've been tinkering with a PB410 board myself. I just checked and I do have working turbo function. If it would help, let me know if there's anything I can check on my board for you.
I have a PB410 with the same issues as your DRAM trace repair. Though I don't get any beeps. Only thing I get on the post code analyzer is 66 --. Going to go investigate now!
I really really dont want to sound condescending here because you are literally the GOAT of these retro repairs.... But.. My Dell of similar vintage has a BIOS option for boot speed Fast/Slow. Did you check if the PB has similar?
I currently have the 430 board running a dx2-66 in a PB... It runs as well as any dx2-66 ive had.. mine had a leaking battery also, and I had to remove the serial port buffer ic's.
I had a PC with the same mainboard with the Headland graphics chip. No onboard RAM though and only a 2MB SIMM. Was so upset I had to throw out the old module when I upgraded the RAM as it was so expensive at the time; there had been a fire at a semiconductor factory in asia which meant supplies were limited. The VLB Headland graphics was pretty good in DOS as far as I can remember. Only 8 bit though, so 256 colours max.
Sometimes almost looks like tinning is enough to restore the traces (if the gaps are not too wide). But even though you use a wire. Is it necessary? Great video, glad you are back!
Thank you. I had bad experience with tining only, where after reassembly the cracks reapeared. Probably because of bending the board, temperature changes etc.
3:42 That is false. Majority of 286 and 386 have fixed crystal oscillators and they cannot change their speed. De-turbo is actually done by PWM signal to HLDA pin of the CPU (duty cycle reduces the speed proportionally - for example at 33% duty cycle of the signal 16MHz processor runs at 12MHz. On-the-fly clock modification requires a PLL circuit which started to be a thing on the very end of 386 life. Boards like PC Chips had these chips. PLL was more of 486 and newer boards thing and that was also very much the end of the turbo button.
Yeah, you are right. On the earlier CPUs the signal HOLD send to the CPU indeed. The clock manipulation started with Am386 Enhanced CPUs, which had power management features and could properly react to changing clock. My bad.
I thought the earlier stuff (286 / 386) was usually a clock divider? E.g., 33MHz / 4 = 8Mhz, or / 2 = 16MHz. 25Mhz / 3 = 8Mhz. 16MHz / 2 = 8Mhz... that kind of thing. I know those old LED speed displays were configured via jumpers, so it didn't actually mean anything at all, but they were usually programmed for 8 or 16MHz on non-turbo mode, and it seemed like that wasn't *just* a coincidence. Like there was precedent for why they were set like that. At least, in theory.
@@nickwallette6201 Nope. All 286 have clock divider /2 hardwired in them. The clock is still technically the same, just the NOP instructions are inserted on selected interval. Late 386DX and 486 had the real clock dividers, but only when the PLL chip was there. If the oscillator was fixed, it was slowed down through the HLDA and HOLD signals
Hrm, why would all external ramdacs produce vertical lines on a LCD? Having developed display drivers for linux for 20 years, i am sure that this is not an issue with ramdacs being external. Especially with the symptom of LCDs showing lines, as LCD are much less likely to show any variation in clocking. These days, spread-spectrum is often enabled on the dotclock to reduce EMI, as LCDs don't really care (for a value of "these days" I spent several days chasing this in 2007 on the RadeonHD driver). And i am still keeping some old CRTs around as those are just as good or better than a scope for showing instability in the PLL, an LCD controller either syncs, or not, and shows a valid picture, or not. You are likely seeing an issue with the drivers (which in my world view includes the VGA/INT10 BIOS) for a specific external ramdac on a specific graphics card of yours, not doing proper pll value calculations for a specific dotclock (it might depend on a table of dotclocks and "tested" pll values), and a specific TFT monitor of yours that might be unhappy with a dotclock being slightly off. Similarly, differing track lengths between display chip and ramdac are not that critical for VGA/SVGA resolutions, and if pixeldata reaches the ramdac at differing times, you should see the off colours just the same on a CRT as on an LCD. And this difference in track lengths, and or any interference on this short stretch is very board specific again. Then, between graphics chip and ramdac there was usually either an 8 or 16bit bus, and the driver (again, i count the int10 bios as a driver), decides how to use that bus depending on the colourspace, colourdepth and frequency. All of which is specific to an intricate combination of: display chip, ramdac, board design, driver, and the chosen display resolution and bitdepth/colourspace. Again, likely a combination of 1 specific graphics card, 1 specific lcd based monitor, and 1 specific display mode. Also, from its general shape and the shape of the print on it, that looks like a Chrontel CH9398. I think that that was the one ramdac i actually did end up re-implementing in xf86-video-tseng after i cleaned up the X.org/Xfree86 clock handling code, back in late 2005 early 2006. The other ramdacs are still listed TODO today, but I've had no complaints :)
My statement was probably misleading. You are right, this is not necessarily an issue of external RAMDACs, but what I wanted to say is, that solutions with external RAMDACs more often have that issues, where Cirrus Logic with its integrated RAMDAC is guaranteed always stripes free. Regardless, very interesting comment, thank you very much!
@@necro_ware Again, not guaranteed. It depends on what combination of things actually caused the stripes you saw, once or multiple times. And if it was multiple times, then was it with the same combination of hardware? When you run into this again, try taking note of what combination of: * graphics card * monitor * display mode (you should be able to let a reasonably modern monitor tell you what resolution and refresh rate it thinks it gets) * display driver caused this phenomenon. A picture of the effect on the monitor would also be helpful. If it is vertical stripes, then it could even be that the driver got confused with the colour space or bit ordering needed to feed the ramdac correctly, and then messed up how the pixels are stored in display ram, and how it reads them out of ram. In this case, the issue would entirely be on the graphics chip side, but it was triggered by the driver incorrectly dealing with the pixel bus for ramdac. External ramdacs do not have to be bad, they just add a bit of driver complication which then leads to more driver bugs.
@@luc_libv_verhaegen Your last statement is what I actually wanted to say. With external RAMDAC a probability of stripes is higher. They are of course not a guarantee for a bad picture quality. For example Trident 8900 cards have external RAMDAC, they always have stripes, no matter which monitor, who made the card, which RAMDAC is used. Trident 9000 have integrated RAMDAC and never have stripes, just as Cirrus Logic. I tested dozens of TFT monitors and cards. Tseng ET4000, well eh 50/50. Newer PLCC packaged RAMDACs, like the Winbond on that PB410 mainboard usually are good, I saw them on multiple WD90C31 doing stripes, but on ET4000 never. However older RAMDACs especially in DIP28 form factor quite often produce stripes even on the ET4000. If I have a choice between Cirrus Logic CL-GD542x and something else and I want that it just works, I'd always go with the CL. Regardless of that, of course I have a lot of ISA VGA cards and I like to experiment with all of them, so Cirrus Logic is not always what I use eventually. It depends.
This has always puzzled me. (Well, in the 5 or so years since I got back into playing with retro computers on modern displays.) What is it about LCDs that makes these vertical lines visible, where totally analog displays didn't? I had assumed maybe it was a sampling thing, where the color signals aren't actually steady-state for the entire period, and just pulse on for a time in the middle of their window. So, sampling that for conversion to digital might miss the color value, or just capture the decaying tail of it. I suppose, if it happens at a high enough frequency, on a CRT, it would blur into the phosphors on either side of the pixel. It's something I've been meaning to put a scope on.
@@nickwallette6201Hrm, you too? What graphics cards are you seeing this with? Is this with any mode with a specific driver, is this with a specific mode, is it always like this with a given card?