Hi everyone! I have been trying to install windows xp (win98se was installed before), to upgrade the motherboard with the uGuru thing, since my AV8 motherboard is not marked with "3:rd eye". I found out that there's an update for that functionality to kick in. But since it seems i can't install windows xp due to file copy errors, this problem might be deeper than the uGuru panel itself, and might be a problem with the motherboard. So I've ordered a replacement board. Meanwhile i'll try some different cables and hdd/ssd and cd/dvd units and burn some different iso:s. I've already swapped the ram sticks with no difference. I'm using IDE only, since Sata on this board is really early and wonky. WIth that said, i did try sata once with the same results. Anyway, I'll try to solve this problem first to find out if it's related to the problem with the panel.

Edit: Nope, I'm being dumb again. XP install works, just needed MORE ram.

Edit 2: Ok, so what i've learned from installing xp and the guru utility, is that everything installs fine, and the guru utility in the activity tray can launch every utility but the "guru clock" utility, which i suppose is for setting up the display device. Supposedly that means that, since the guru panel can receive the cpu temp, but not do anything else, there's a problem with the panel's data output, and the motherboard can't receive it to confirm that it's connected. So as someone here mentioned, i should probably check some more fuses on the guru panel's main PCB.
Another interesting observation is that, between the 3+1 pins for the guru link, it can do fine with only the 3 grouped pins, without the separate 1 pin. Connecting or disconnecting the separate 1 pin doesn't do any difference. So maybe the problem has something to do with that. It still needs the ground from the usb connector, though.

snufkin wrote on 2022-12-14, 19:08:

Ok, so I found http://abit.ws/upload/products/guru-panel.htm which sounds like your description. It shows the 3-pin side of the link connector going to the CMOS clear. The other side shows a 4 wire cable going from the GURU1 header to the combined connector on the GP-02 board. Looks like it's 4 wires arranged as 3+1 with a gap probably to make sure the cable is plugged in the right way around. Without knowing anything more I'd guess that it's an i2c like link, in which case the wires might be Power,ground, clk and data, but no idea about the order. Which would mean that that USB ground connection shouldn't be needed to make the display work.

Looking at a photo of the cables here: https://c1.neweggimages.com/ProductImageOrigi … -999-405-04.jpg (taken from here: https://www.newegg.com/abit-guru-panel/p/N82E16813999405 ) I'm not sure that the connection from motherboard to guru is directly 1-1, with the gap lining up the same. You can see the motherboard header going "White-Gap-Black-Green-Red". You can't see all of the motherboard connector, but you can see that it goes "White-something-something-Gap-something".

Might be worth going back to the motherboard and (with the PSU unplugged from the mains) check the resistance between each of the 4 pins on the GURU1 header and +5 and GND pins on the ATX power connector. That should show if there's power and ground on the header.

You are completely right about the wires not being 1-1. I totally overlooked that. It's very clear that the white cable is the separate 1 pin on the mobo side, and the first of the 3 grouped pins on the panel side. Now i just have to figure out the rest. I'm surprised i haven't smelled magic smoke yet.

HOLY MOLY! YESSSSSSSS! IT'S ALIIIVE!!

I'm so glad this all came down to user error!

A super big Thank You to everyone who helped me! This forum is truly the best!

And just to not leave anyone who was wondering hanging, here's a drawing of the pinout difference.
Edit: Apparently I'm so happy i can't even spell "motherboard".
Edit 2: And for anyone who thinks my terrible drawing skills are confusing - it's the same order but in reverse. Should have used numbers instead of lines.

can you measure pins 6 - 8 and 6 - 9 again now? something tells me it will read 5V

rasz_pl wrote on 2022-12-18, 08:53:

can you measure pins 6 - 8 and 6 - 9 again now? something tells me it will read 5V

I've considered it, but measuring those chip legs was really really slippery, and for the sake of Schrödingers Guru Panel, might we just assume that you're right on this one? 😀 I really don't want to mess this up now.

Nice. I'm going to make some wild guesses and then ask if you can make a few measurements to check.

Assume that in that photo of the official connector that Red is +5, Black is Ground, White is Clock and Green is Data, and that both Clock and Data have a pull-up resistor connected to +5. From the photo, that would mean order at the Guru end is Clock-Gap-Ground-Data-+5. You've found that the order is reversed on the motherboard, so that would make it +5-Gap-Data-Ground-Clock. So if you previously had them connected up one-to-one then the display had the Clock signal connected to its +5, Ground connected to its Data, Data connected to its Ground and +5 connected to its Clock.

If that's correct then it would mean there was no voltage difference between Clock and Data (assumed both have pull-ups), so nothing on the display would work. When you connected the USB ground then that pulled the display Ground from +5 (via the Data pull-up resistor) to 0V. The display could then be powered, badly, by current flowing through the pull-up resistor on the Clock line from the motherboard (connected to the display +5 input), through the display and back through the USB ground. Current drawn through the resistor would cause the voltage to drop, which might be why you measured that 3.7V. Separately the Clock input was directly connected to +5, so no data getting clocked in, but not actually causing any problems (unless the display tried to drive the clock). In the photos of the half working display there doesn't look to be any information that would be sent from the motherboard, like AGP voltage or CPU frequency.

The only one that'd worry me a bit would be the Data input connected to Ground. For the times before you connected the USB ground then that input would be significantly below the Ground input on the display (which was connected to Data, and pulled to +5 through a resistor), which would normally be outside allowed input voltages. But it's possible that those pull-up resistors limited the current enough to prevent any damage.

I think that would explain why the display only started working a bit when you connected the USB ground, and also why nothing smoked when it was connected up wrong. Also, pretty sure rasz_pl will be right about the display voltage.

Might be nice to check some of that for anybody looking for a pinout in the future. You can safely measure some resistances with the board off. I think, again based on that photo and what you've found, that the order of pins on the motherboard is White-Black-Green-gap-Red. So try measuring resistance from Red to a +5V point (spare molex power connector might do, or +5 on the motherboard power connector), then from Black to a Ground point (again, molex connector, or any d-sub surround). If those aren't 0 ohms, then try the other pins. If you can't find any that are 0 ohms to Ground and +5 then I'm wrong. If they both read close to 0 ohms then check the resistance from Green to Red and White to Red and you might find something around 2k ohm.

yes, parasitic power thru pullup would totally explain previous behavior

snufkin wrote on 2022-12-18, 22:14:

Nice. I'm going to make some wild guesses and then ask if you can make a few measurements to check. […]
Show full quote

Nice. I'm going to make some wild guesses and then ask if you can make a few measurements to check.

Assume that in that photo of the official connector that Red is +5, Black is Ground, White is Clock and Green is Data, and that both Clock and Data have a pull-up resistor connected to +5. From the photo, that would mean order at the Guru end is Clock-Gap-Ground-Data-+5. You've found that the order is reversed on the motherboard, so that would make it +5-Gap-Data-Ground-Clock. So if you previously had them connected up one-to-one then the display had the Clock signal connected to its +5, Ground connected to its Data, Data connected to its Ground and +5 connected to its Clock.

If that's correct then it would mean there was no voltage difference between Clock and Data (assumed both have pull-ups), so nothing on the display would work. When you connected the USB ground then that pulled the display Ground from +5 (via the Data pull-up resistor) to 0V. The display could then be powered, badly, by current flowing through the pull-up resistor on the Clock line from the motherboard (connected to the display +5 input), through the display and back through the USB ground. Current drawn through the resistor would cause the voltage to drop, which might be why you measured that 3.7V. Separately the Clock input was directly connected to +5, so no data getting clocked in, but not actually causing any problems (unless the display tried to drive the clock). In the photos of the half working display there doesn't look to be any information that would be sent from the motherboard, like AGP voltage or CPU frequency.

The only one that'd worry me a bit would be the Data input connected to Ground. For the times before you connected the USB ground then that input would be significantly below the Ground input on the display (which was connected to Data, and pulled to +5 through a resistor), which would normally be outside allowed input voltages. But it's possible that those pull-up resistors limited the current enough to prevent any damage.

I think that would explain why the display only started working a bit when you connected the USB ground, and also why nothing smoked when it was connected up wrong. Also, pretty sure rasz_pl will be right about the display voltage.

Might be nice to check some of that for anybody looking for a pinout in the future. You can safely measure some resistances with the board off. I think, again based on that photo and what you've found, that the order of pins on the motherboard is White-Black-Green-gap-Red. So try measuring resistance from Red to a +5V point (spare molex power connector might do, or +5 on the motherboard power connector), then from Black to a Ground point (again, molex connector, or any d-sub surround). If those aren't 0 ohms, then try the other pins. If you can't find any that are 0 ohms to Ground and +5 then I'm wrong. If they both read close to 0 ohms then check the resistance from Green to Red and White to Red and you might find something around 2k ohm.

Wow! That's some Sherlock Holmes-level deduction you've done. I think i'll have to read your comment at least a couple of times to understand it, and get back to you. Sure, I can measure some resistances with the power off. That's safe stuff.

It's so great you got it sorted out! The panel is so rare it took me several years to get a few of them; more common is the guru clock GC-02; also the final version - guru clock 03 GC-03 is exceedingly rare, I have never seen one for sale, same for the Abit postman which took me years to find one- it's one of the last Abit items missing from my collection!

DJonekill wrote on 2022-12-19, 17:47:

Wow! That's some Sherlock Holmes-level deduction you've done. I think i'll have to read your comment at least a couple of times to understand it, and get back to you. Sure, I can measure some resistances with the power off. That's safe stuff.

More spinning a tale to try and fit what we know, then it has to be tested. Plus, pretty sure Holmes wouldn't have got the colours for the motherboard end wrong like I did, especially after you'd already included that information and attached the photo of the cabling that I found. I mixed up the two ends and should have said White-gap-Black-Green-Red at the motherboard end. So that means the resistance checks should start with checking if the middle pin is connected to Ground and the pin furthest from the gap connected to +5V. If they are connected, then check resistance between the two end pins (White to Red) and between two pins at one end (Green to Red).

nd22 wrote on 2022-12-20, 07:46:

It's so great you got it sorted out! The panel is so rare it took me several years to get a few of them; more common is the guru clock GC-02; also the final version - guru clock 03 GC-03 is exceedingly rare, I have never seen one for sale, same for the Abit postman which took me years to find one- it's one of the last Abit items missing from my collection!

Cheers! Yes, this truly made my day, and week!
Then i figure you have the original cables for all your panels? I mean, otherwise I'm sure you'd run into the same problem, seeing as the connectors are wired in the opposite order than what's intuitive.

Also, Abit postman, what connector does it use? Looks like the 3+1 guru link, amirite? Or is it for earlier motherboards?
Edit: Abit postman looks like the perfect thing to modify if you want to have a newer motherboard in one of those 486 AT cases with a Mhz display.

Abit postman uses the system management bus header present on early Abit boards up to AT7/KX7/BE7. It is a 2 digit display present on later Abit motherboards which was manufactured in extremely small numbers.

Found some pictures of one:

(just for the record, on this one the pinout colour order is the same at both ends).

Looks like one sold on German Ebay in October this year (which is where I found the pictures). I guess the motherboard sends the POST codes out over the SMBus, so it might be possible to replicate the PostMan by watching the bus, figuring out the address the codes are sent to, then programming a uC to read and display the codes.

[edit to add: based on the PCB routeing, I'd hazard a guess that white and yellow are the clock or data, so red/green might be +5/Gnd.]
[and now I'm wondering about the SMBus header on my KA7 and adding this as a possible project once Real Life(tm) gets out of the way]

yes, replication could be trivial after owner of one of those boards puts logic analyzer on this header or we disassemble bios. No need for building custom hardware, you can buy $4 with free shipping 4-digit LED voltmeters based on STM8s003 http://smokedprojects.blogspot.com/2013/08/i2 … -voltmeter.html