This is a weird one for sure. There's a memory map In the Z386SX-20 manual I have that shows only two ranges marked as "slushed," attached for reference.

Sometimes my Z386/20 will not want to boot. All diagnostic leds just stay on and never turn off as they should. To fix it I've normally just lifted one corner of the machine (the back left) and dropped it about 1 cm. That normally does the trick and it works flawlessly again. If I have the cover off I dont need to drop it, then I normally just tap the BIOS I/O board.

Its not exactly the same behavior you are experiencing, but it just goes to show that mechanical problem exist that is very marginal.

Taking out socketed chips one by one and using deoxit on sockets is something worth a try. But I dont see much in the way of sockets. Bios, password eeprom, Keyboard controller aka SCP, maybe jumpers? I would mainly suspect SIMM sockets. Maybe try with ram only in second bank, and different ram sticks altogether.

Hi all. Sorry for my late reply. Life gets in the way of hobbies 🙁

vetz - I modified your technique - I would lift one side of the motherboard while there were some screws left in the case. This would cause the mobo to bend slightly. I did this in all four corners but it didn't help.

rasz - I can't get the CPU out and I'm afraid to try. It's in a weird socket and I'd need some kind of tool to press in those 4 bumpers at the same time I think to release the chip. Other than that, I did deoxit the sockets including the SIMM sockets. I did try the RAM only in the second bank but that did not work. The manual indicates this is not a valid RAM configuration. I did try three different sets of RAM and all had the same issue.

I'm not disregarding your notion that it is a mechanical problem, don't get me wrong. I took the motherboard completely out of the case again and gave it a good look. If something is wrong it's very hard to spot. Given that, I thought maybe the next thing I could do was to use my oscilloscope and probe the pins on the SIMM socket. I figured that if there was a broken trace for example, or a broken component, maybe the scope would show me some strange readings on the pins. The problem with my idea is that I have no idea what I'm really looking for in terms of a waveform. I will share a link to some pictures I took while using my scope on the first RAM socket. By the time I got to pin 29 (a day or so after the rest of the pictures were taken) I realized that perhaps I had the time set too low and I couldn't see the whole waveform. That is why the pic of pin 29 looks different than the rest. That one look different than all the rest even taking into the account the timebase thing I mentioned. I don't suppose you know if that looks normal for data parity in on a 30 pin SIMM?

Is my approach narrowing down the mechanical issue sound? Flawed?

Rather than sharing all those pics of waveforms in this post, I'll just link to the album on imgur below.

https://imgur.com/a/wKTM8bg

P.S - I am almost done scanning the manual for my Z386SX. Only a dozen ish pages left to go!

Datasheets for ICs or modules often have timing diagrams, so you can assume that's what the manufacturer was trying to achieve.

To make SIMM socket readings make any sense they must be performed when BIOS is actually trying to detect ram (that means resetting with probe on each pin), because after ram detect fails it will just loop in same spot somewhere in ROM = there is no reason for any activity on RAM bus after very short initial burst.

There should be activity on pins 2 27, probably all data lines 3 6 10 13 16 20 23 25 and at least few Address lines so 4 5 7.

Whatever is happening on pin 29 looks weird, but then again Ram controller wasnt properly initialized so I would dismiss that.

geiger9 wrote on 2024-09-23, 19:05:

I'm not disregarding your notion that it is a mechanical problem, don't get me wrong.

I mentioned my boot issues on my Zenith caused symptoms similar to yours. I actually identified the issue. It was one of the bodge wires that had cracked in the solder and making interim connection. Tapping it with a plastic tip made the system pass boot when the problem occured.

Looking at the board pictures you provided to TRW I'd say its very likely you could have the same issue as you have bodge wires running very close to the ISA riser slot on the bottom, but also on other parts of the mainboard.

Sorry for my delay responding, vetz. I don't regularly check the site. I'm so thankful you followed up on this and I will try to dig out my motherboards soon to look into the bodge wires.

Well I finally got around to it and checked out the bodge wires. They all had solid continuity when checked. I did try pressing down on some while booting just to be sure but no dice.

Someone on Discord recently told me to check all my 74F logic chips. I guess I can look into that.

I think my best bet though is to use the scope on the SIMM pins, see if I can spot something out of the ordinary, then trace from there to see where something is failing. That's going to be a lot of work I think because I have to look at it just as the system is powering on for a pin, then switch pins, then reset, then do it again. Would I have to look up the datasheet for my RAM and see if it has waveforms in there so I can compare it to what I'm seeing?