So the only SMD inductors I have at the moment are 10uH...which I'm sure is inadequate...but I just wanted to see if the Clock chip actually worked..
..and yes it does ..though I'd hardly call it pretty..
uneven output and very spikey..
but it does run.
CK signal ugly but present at the CPU ZIF.

Might order 20X 100uH from china ...as that appears to be the only place I can get them cheap..
They will take a while to get here sadly...but looks promising.
unless I can find something suitable I can harvest from an old pcb.
rgds

So have you got +5 on both DVdd and AVdd now? What does the supply look like? If it's smooth then the inductor value wouldn't be why it doesn't look right.

It's possible the MX8315 might have been damaged if the supply voltage was missing. The clock input XI (pin 3) is probably driven directly by the clock generator (I wrongly called it a crystal before). If that was working then it would have been switching at +5V (can you check that?). The inputs on the MX8315 are only allowed to go to 0.5V above Vdd. So it may have had +5V on an input with only 1V on Vdd.

May not actually be a problem if the clock generator is powered with the same 5V supply (after L1) as the MX8315.

[looked up on stackexchange about isolating analog and digital supplies and someone had an Intel reference design which used a .68uH, so 100uH might be a bit high:
https://electronics.stackexchange.com/questio … og-power-supply ]

A lot of variables here ..
The bench power supply is adaquate but not excellent..
The scope is ok but not brilliant...but then I've never leally liked digital scopes..I much prefer analogue.
This one might be struggling a bit at high frequencies.
I may try hooking up to an AT psu tomorrow.
I did however get frequency readings on the MX clock pins ...and the scope correctly identified 14.318 on pin 4 Xtal clock out
locked at 24.004 mhz on pin 5 peripheral clock ..
locked at 33mhz with some slight frequency drift on pin 8 CPU clock.. ???
But then no cpu in the system so no load atm..
There does appear to be an underlying ripple of some sort on all signals ...but again ..could be the scope ..could be the probes..could be the psu I'm using.
OR ..could be as you say that the MX chip took some damage.
Tomorrow I shall fit a 14 pin base ..so I can swap it out if neccessary.
I guess some progress is made ...but may be more to do yet

....and yes the bench psu is only set to 4.6v atm ...and both Analogue and digital supplies are also 4.6v.
I may as well turn it off for tonight ...as I want to go watch the movie 'Nobody'
catch you later buddy.
VS

Fitted a 14 pin socket for the MX chip and ordered these :-
https://www.ebay.co.uk/itm/114390315286
No idea if these are suitable..but they were cheap and actually in the UK ...so took a chance.
Either that or a long wait on a slow boat from China..
Also found a used MX8315 on fleabay so added that to the shopping list.
fingers crossed.
rgds
VS

Another curiosity ..
mod or upgrade perhaps
a tantalum at TC3 is connected directly to both supply rails on the MX chip .presumably to provide rail smoothing.
pins 9 and 14 appear to be connected together so share the same 5v supply.
I referenced Necrowares video on youtube where as part of his mobo repair marathon this motherboard is covered.
On my board TC3 is a 10uf 16v tantalum....on his board it looks like a 330uf electroytic he changed.
So regardless of it being suspect or faulty I'll dump the tantalum and fit a 220uf 16v electroltic cap I think.
It can only help imo.
...and if the tantalum is responsible for poor smoothing or damage to the MX chip ...
...well it can't any longer...cos it is gone.

Probably won't do any harm, but the difference could also be that the board in the video has a crystal fitted (2 pin can + 2 surface mount caps) rather than the clock generator on yours. I would guess that TC is meant to be a tantalum. The picture on UH19 ( http://www.win3x.org/uh19/public/motherboard/ … 9e231847758.jpg ) has the same clock generator and that has a tantalum fitted. It's possible that L1 and TC3 form a deliberately tuned LC filter, so changing the values might matter. Probably won't though.

You make a fair point Snufkin...maybe I should just order a few tantalums....to be safe..
or at least use a smaller value electrolyic..
As you say ...it may not make any difference ...but perhaps not worth taking a chance.
Curious that boards with the same model number have such variations ..but then there are revisions to consider.
rgds
VS

Now that's what you call digging and finding gold..
just been up in my loft where a few boxes of mixed components given to me years ago were dumped.
In the second mixed box of bits I found a plastic bag containing some tants ..10uf 25v
nice..
but hold the bus..
spotted a bag of mixed frequency 4 pin clock generator packages.
mostly 10 and 20mhz...but in amongst them was a lonely 14.318 mhz..
I guess this motherboard will get a replacement...which may or may not help.
I really must unpack these cardboard boxes....and see whats in them.
rgds
VS

A quick test reveals no change ...but then is there still a problem...?
A different ..smaller L1 taken from an old TV mainboard...
Replaced the tantalum TC3..
fitted a socket for The MX chip ...for now the one from this mobo has been fitted ...replacement on order.
Also replaced the Xtal clock generator.
A base was fitted for this also..
Turned pin 14 way with middle pins removed both sides ..so just the end 4 left in place..
A small slice of gecko tape under the Xtal to keep it in place .
Pretty impressive stuff that tape.
I can actually lift the mobo off the bench by the crystal ..without it pulling out of the 4 pin socket.

Might be worth giving it a whirl with that AMD DX4 with the possibly dodgy L1 cache, see if it comes out of reset.

A bit late tonight Snufkin...but I will test with the DX4 tomorrow..
Fingers crossed for a +ve result.

Good work ! You must have a love for that board to put that much into it : )

I just Love old Tech in general Horun ..
VCRs from the 80s ...PCs and Retro Computers in general ..Reel To Reel tape recorders etc etc..
but then I am a child of the 60s and 70s..
My motto..if it can be fixed..it should be fixed.
..and to prove that love...
Here is the first snapped image of this motherboard booting again.
still reporting the wrong cpu.(must check all the jumpers)
maybe the new MX chip (when it arrives) will help but I doubt it...🤣
rgds
VS

snufkin wrote on 2021-10-02, 22:05:

Might be worth giving it a whirl with that AMD DX4 with the possibly dodgy L1 cache, see if it comes out of reset.

Looks like a win Snufkin..see my reply to Horun.
rgds

Good work, and good thinking to check the clocks. Any chance of a photo showing just the motherboard with CPU fitted so we can see all the jumpers?

I'll do that in a short while..
But first ...This might confirm our suspicions about this DX4 and it's internal cache.
Obviously this mobo needs an interface card for floppy..
So I fitted one and hooked up a floppy disk 1 of MS DOS install..
Total refusal to boot ..
Process starts but hangs as it did on the SOYO.
enter bios and disable CPU cache..
result you can see below.

it may be why the DX4 is wrongly identified by the bios ??

Uncluttered image of the Gigabyte mobo

As an aside to the main projects.
Thought I'd take a look at the faulty from new POST diagnosis card .
It does make you wonder at the quality control IF ANY of these guys throwing these together in China.
Nuff said...FWIW be aware when buying such stuff.
It may be cheap ...but could cost your dear.
Take a look at what I found in the photo..
It actually looks like a chip leg trapped in there and is definately shorting 2 pins together.
Worth a try I guess ...I'll desolder this chip and attempt a repair.
rgds
VS

Yep, worth a try.