As you describe it it seems very probable that you have a memory issue. But because 1. failing to even start memtest usually means worst symptoms than just "weird instabilities" and 2. various versions of memtest do crash on some computers, I'd try another version just to be sure. 4.10 is the last version that runs on a 386 and you can also try an earlier one.

I assume you have already cleaned, reseated and switched the two simms.

konc wrote on 2024-06-19, 12:44:

As you describe it it seems very probable that you have a memory issue. But because 1. failing to even start memtest usually means worst symptoms than just "weird instabilities" and 2. various versions of memtest do crash on some computers, I'd try another version just to be sure. 4.10 is the last version that runs on a 386 and you can also try an earlier one.

I assume you have already cleaned, reseated and switched the two simms.

Yep I've swapped their seating and used contact cleaner just to be sure.

I've also installed checkit V3 on the hard drive and done a memory test with that, but it doesn't pick up on any issues

The weird behaviour in windows is that as soon as I try and open some programs like file manager, it'll start loading but just kick me out to a dos C:\ with no error reported.
It's weird

Also I don't own any suitable ram for this board. This is the first time I've actually seen the pin type SIMM modules rather than the better sort that just clicks into place.

This is another example of an error I've had when I swapped to a different hard drive.

Note that the bios is rather limited and doesn't give me any options in terms of ram access times. It's a Phoenixbios A386

The ram modules in question. Look to be 80ns.

Memory might be misleading.
It's 100 years old-with all the problems that entails.

Personally, I'd run it through the dishwasher without soap, or wash it manually with soapy water and a toothbrush. Let it dry, and reseat everything that can be removed +reseated.
Then see if the behaviour changes.

If not, I'd be grabbing a multi meter and doing basic voltage sanity checks.
Eg, I wouldn't expect to see any voltages below 3.3v.

I'd also try flexing it gently with all the sockets empty and listening. Old solder might need reflowing.

Also, the Dallas rtc battery has to be dead by now. They can cause issues. There's a guy on YouTube with a guide on replacing the battery without damaging it.

myne wrote on 2024-06-20, 01:00:

Memory might be misleading. It's 100 years old-with all the problems that entails. […]
Show full quote

Memory might be misleading.
It's 100 years old-with all the problems that entails.

Personally, I'd run it through the dishwasher without soap, or wash it manually with soapy water and a toothbrush. Let it dry, and reseat everything that can be removed +reseated.
Then see if the behaviour changes.

If not, I'd be grabbing a multi meter and doing basic voltage sanity checks.
Eg, I wouldn't expect to see any voltages below 3.3v.

I'd also try flexing it gently with all the sockets empty and listening. Old solder might need reflowing.

The only unusual part that's raised a flag to me so far is that during a CPU speed test, it is reading 19.2mhz or so instead of 16.
It's got a 32mhz crystal so I'm not sure how that happens!

Perhaps I need to look at the oscillator on the cro

Just tried memtest86+ v4.10 and same thing

Found the manual?
Jumpers set wrong isn't impossible

myne wrote on 2024-06-20, 01:24:

Found the manual?
Jumpers set wrong isn't impossible

https://theretroweb.com/motherboards/s/chaintech-lynx-ct-201

This is the only info I can find on the matter. I'm not sure how a jumper setting could change it's speed though, when it's fixed by the crystal

Actually, I see a crystal for the ISA bus, but also a 32 MHz crystal and a 40mhz crystal. One is probably for the trident video?

Edit: Yes, the 40mhz is for the video

Jumpers can select dividers/multipliers and sources.

Eg 32 /2 =16
But if you choose the same source as the trident with a jumper then it's 40/2=20.
19.2 is close. The 40mhz crystal could be running at 38.4.

No idea, but it can't hurt to switch jumpers around and see if anything changes.

myne wrote on 2024-06-20, 02:15:

Jumpers can select dividers/multipliers and sources. […]
Show full quote

Jumpers can select dividers/multipliers and sources.

Eg 32 /2 =16
But if you choose the same source as the trident with a jumper then it's 40/2=20.
19.2 is close. The 40mhz crystal could be running at 38.4.

No idea, but it can't hurt to switch jumpers around and see if anything changes.

There's also a dip switch block. I'm hesitant to go flicking stuff not knowing what it's all doing.

BurntOutElectronics wrote on 2024-06-20, 00:26:

The ram modules in question. Look to be 80ns.

That is a 30 pin SIPP. Single Inline Pin Package.

https://en.m.wikipedia.org/wiki/SIPP_memory

It is electronically identical to a simm.

A simm with some little legs soldered onto the fingers should work, as long as it is FP mode.

(Dont quote me on it, but I think a normal IDC header strip has the correct pitch. Clip it in place and solder one on.)

BurntOutElectronics wrote on 2024-06-20, 03:09:

myne wrote on 2024-06-20, 02:15:

Jumpers can select dividers/multipliers and sources. […]
Show full quote

Jumpers can select dividers/multipliers and sources.

Eg 32 /2 =16
But if you choose the same source as the trident with a jumper then it's 40/2=20.
19.2 is close. The 40mhz crystal could be running at 38.4.

No idea, but it can't hurt to switch jumpers around and see if anything changes.

There's also a dip switch block. I'm hesitant to go flicking stuff not knowing what it's all doing.

PXL_20240616_070404467~2.jpg

It's unlikely they would do anything truly harmful. Those old chips are pretty tolerant.
They're also far more likely to be set wrong by some idiot messaging around.

Me? Since there's no manual to be found I'd just change them and see what changes.
If nothing appears to change, then it's possibly voltage related and you'd have to check with a multimeter.

Okay I'm on pin 15 here, and I did find a jumper which changed the clock frequency. It looks to be very close to 16mhz, but the effective speed is lower on landmark, seemingly due to these random flatlining in the clock.
I have no idea if that is apart of normal function for that to happen or not.

It seems to be switching between nothing, 16 and 32mhz on that pin.
Is that normal?
Does the CPU do the halving in clock speed?

And it should be noted, windows still is doing it's weird thing, but no other application I've used on it has anything weird going on. It's just memtest and windows.
I've run wolf3d, benchmarks, and installing do's and windows went down without a hitch.
Windows just closes to the dos prompt for no apparent reason

Weird thought.
Is the fpu enabled in hardware but doesn't exist?

myne wrote on 2024-06-20, 12:08:

Weird thought.
Is the fpu enabled in hardware but doesn't exist?

There is nothing in the bios, and checkit reports no FPU so doesn't test it.
Again, no idea if there is a jumper for it, but programs detect no fpu

Dunno man. I'm going back to my original suggestion.
Wash it. Reseat everything.

I've fixed completely dead GPUs with a quick run through the dishwasher.