Fuel fixer uppper

[Hombre71]

Hmmm, I just noticed this indeed

Code: Select all

001a01-L1>ALERT: Error initializing DS1780 0, no acknowledge

SGI SN1 L1 Controller
Firmware Image B: Rev. 1.10.12, Built 02/01/2002 14:40:22


Do based on the number 0 can we identify which one is likely to be toast?

Or is there another way to double check both(?) DS1780 chips through L1 command or other test?

Cheers

Thing is that all those are simply MONITORING issues, disabling env monitoring (env off) should let you go further and boot the machine. If that doesn't happen then you do have bigger problem with either the PSU or some voltage regulation onboard.

In this thread:"

viewtopic.php?f=3&t=16206

There is another user where disabling env monitoring did not allow the machine to boot :

User Pinball_O stated "thanks pentium... I already tried env off to turn off the monitoring.
but that wouldn't allow this motherboard to come up. it was reporting bad voltages too... but i measured the voltages. they are correct. I checked powersupply it is good."

He then goes on the replace the DS1780's (overheating parts of the mobo as part of the work) but then
does get it booting again. I will look for a repair shop to swap out all 3 on the motherboard as I suppose
the others will fail at some point anyway.

Perhaps the chips need a small heatsink or such?

Anyway, whilst the board will be out, anything else I should replace? Timekeeping chips perhaps.

[mazzinia]

I'm doubting that these chips fail due to temperatures. I know of some that developed quirks by having the board not receiving current for more than 6/12 months.

You said you didn't turn on the Fuel for a while, since before the move. Was it being kept connected to the wall outlet, or totally isolated ?

I'm musing ... these chips had a production or design flaw, and maybe it shows once they are not getting powered for enough time after an initial operation ?

[Hombre71]

Yes it was probably not connected to power for 6 months or so.....could be. Hopefully the new chips will not have similar issues.

[mazzinia]

given that those chips are widely used in modern hw, I think it was just a bad batch / bad quality control / bad stroke of luck.
Anyway it may be a safe bet for anyone having a Fuel with an older revision board to at least keep it connected to the power outlet even if just stored (if that could be enough to avoid an i2c issue, it would be a very small price to pay, isn't it ?)
This is all empirical anyway, unless more people that suffered these chip failures provide a bit of background related to their Fuels

[Hombre71]

Warning: naked Fuel below

[Hombre71]

This is really at the limit of what I can do by hand, but I'll try sometime soon (I have the chips).

I guess you could remove one and apply power to see which one "disappears" from the L1.

Jan-Jaap I found a repair place in Holland that can do it :

Het vervangen van deze chips is geen probleem maar door de vrij grote thermische massa van de print moet deze op een correcte wijze (langzaam) worden voorverwarmd en afgekoeld (wij gebruiken hiervoor een nauwkeurige softwaregestuurde preheater en een warmtebeeldcamera) en het professioneel vervangen kost daardoor wel enige tijd. Na vervanging word de pcb uitvoerig gereinigd en onder een microscoop geïnspecteerd, alles uiteraard in een volledig ESD veilige omgeving. Het zal nauwelijks zichtbaar zijn dat de ic's zijn vervangen.
Ik schat in dat het vervangen €60,- a €70,- gaat kosten ex de ic's a €7,- p.st. maar deze mag u natuurlijk ook zelf aanleveren.

With shipment costs etc. and the chips it will be close to a 100,- euro to fix the board, but at least the board should
presumably be good to go for some time after this and it seems at least nice to keep a Fuel going rather than turning it
one yet another box that has been raided for spare parts

[mazzinia]

It seems a good price for that kind of service ( seems quite a quality one, if they do all they mention ) and to replace all 3 chips. Maybe it would be worth to service also the V10 ( has one i2c too ) and the cpu module ? (assuming it would be a marginal extra if done at the same time vs as a separate job)

[Hombre71]

It seems a good price for that kind of service ( seems quite a quality one, if they do all they mention ) and to replace all 3 chips. Maybe it would be worth to service also the V10 ( has one i2c too ) and the cpu module ? (assuming it would be a marginal extra if done at the same time vs as a separate job)

These are parts I would prefer to upgrade anyway, assuming I could find replacement parts at some point in the future.

I prefer to test just the mainboard with as little as possible parts left in it, and see if that works.

Then at least I know where to turn if I would need to fix any of the other parts in the future.

Also a matter of balancing cost of repairs vs the final value of the machine in question (already this does not add up
but it's a labour of love / interesting project)

[mazzinia]

You do have a point.

I do wonder if there are alternatives to changing the chips.
By example, would it be feasible to place a chip on top of the existing and ensure contacts at least temporarily, to identify the culprit ?
Then fixing it permanently on top in parallel, and cut the power pins to the faulty one.

Maybe is easier ( no heat gun , no xray etc ) and more feasible even with less skills ?

[kubatyszko]

Stacking chips in a very cramped setting (especially between the PCI slots) is going to be harder than replacing the chip completely.
I've soldered a couple of http://kubatyszko.com/2016/06/13/minimig-amiga-clone-hand-soldered/Minimig Amiga clones (BY HAND) including stacked RAM with a CS (chip-select) signal bent outwards and driven from an extra I/O port - it it's not easy... And those RAM chips had much bigger pitch than the DS1780 chips do...

[mazzinia]

Ah gotcha.... was just an idea.
If only having the L1 sources,,,

[kubatyszko]

I do have it in my backlog to create a map of DS1780 chips to the signals they monitor (voltages AND fans), haven't had time for that yet - it will eventually happen

[Hombre71]

I do have it in my backlog to create a map of DS1780 chips to the signals they monitor (voltages AND fans), haven't had time for that yet - it will eventually happen

From what I could find online, the problem with the bad ds1780 chips is mainly with
the earlier Fuel models, so I suppose the 500 and 600 mhz variants.
While useful to identify which chip is problematic, once you go through all
the trouble of taking the board out, you may as well replace all three.

[Hombre71]

“Lazarus” lives!

Yesterday I picked up my board from the post,
spent today putting it all back together.

Started it up, machine shut itself down...mhhhh.
Quick forum search, ahhhh stupid, forgot to connect rear fan.

Try again...strangely no L1 connection, but via rear serial 1 yes...
and...Prom screen!
Reconnect hdd and....boot!

This feline does seem to have at least 3 lives

[Hombre71]

The company said it wasnt an easy job to replace them, cost were 70 eur plus vat
plus shipping back to me in my box in a very pink antistatic bag.

So not cheap, but the board looks very clean plus it has a nice green sticker now
And a Fuel has been saved so far.

I recommend the company Jan-Jaap, if you dont want to do it by hand.