Origin3200 IBrick voltage problems

[rusti]

@jan-jaap: Found a shop in Leuven who says they can do it. So I will try to get the chips now and then lets see. I just hope he knows what he is doing because now the IBrick is working with env monitoring off. So if replacing the chips doesn't fix it, so be it, but if it really gets destroyed in the process then I will have an Origin 3200 with no IBrick. And I have seen that these IBricks are relatively rare and expensive, so getting a new one will probably cost me more than the whole system.

[rusti]

Yes one Cx Brick would really be a nice replacement for my two CBrick(s).
It has 4U, so it is 1U higher than a CBrick but:
- It can contain 16 Processors as opposed to 4
- It can contain 32 GB of memory as opposed to 8 GB
- It contains the functionality of an RBrick, so it has 4 NUMALink and 4 XIO Ports (external)
- It works with all existing Bricks (C, I, P, X, N, G, V,...)

Anybody has one he wants to get rid of?

[bri3d]

viewtopic.php?f=4&t=16723644

The prices are pretty... serious, though.

[hamei]

http://forums.nekochan.net/viewtopic.php?f=4&t=16723644

The prices are pretty... serious, though.

Ummm ... I paid $3500 for a Fuel five years ago. Rothers paid more. Give it another couple years.

[rusti]

I jut paid 1800,- € for a dual processor Tezro without any flame or similar stuff. Only to see, that others offer them for 900$ two days later. I am even paying this in multiple pieces. (I don't know how you say that in english correctly).

Yes prices are serious but that's what you should expect if you are looking for "relatively modern" stuff.

But hey, where is the fun in collecting this stuff if you can get anything you want right now for free.

[SAQ]

I jut paid 1800,- € for a dual processor Tezro without any flame or similar stuff. Only to see, that others offer them for 900$ two days later. I am even paying this in multiple pieces. (I don't know how you say that in english correctly).

"installment plan"

Yes prices are serious but that's what you should expect if you are looking for "relatively modern" stuff.

CX will probably be pricey for a long time. Last-generation SGIMIPS and lots of compute power squished into a small box. Couple that with the numbers sold (not many, comparatively) and things are looking difficult for the hobbyist. Same thing will probably be the case with Tezro quaddies.

[rusti]

So I got the replacement DS1780s. Thanks jan-jaap for the tip. Had to pay customs. Damn.
Next week I will hand in the IBrick board and three chips to the shop that does the replacement. What I want you to do is:

a) keep your fingers crossed that at least the system is still running after the replacement

b) even better would be that this really fixes the false alarms

Will keep you posted.

[rusti]

Hmm. Nothing very positive to report.

Had the DS1780s replaced. First result was, that the -12V line was still reported as -31V. On top of that I now have one fan reported at 1400 rpm which makes them all spin up like hell. From just looking at them and holding my hand next to them I can tell that all (including the slow one) run at high rpm.

Checked the soldering job and found that some pins of the newly inserted chips did not have proper connection

Turned it back in. Pins were resoldered (partly).

Success story is that now the -12V show as -12V again and remained like that for several hours. But the new fan issue is still there.

Result from my point of view:
1) My confidence in the shop's ability to solder SMD is very low by now. Will not attempt to make him have another go at it because I am afraid he might destroy something else.

2) The voltage thing makes me believe, that I am looking at the right place, though.

3) Will try to find an alternative place to get the chips replaced again.

Not very satisfactory so far. Will keep you posted.

[tjsgifan]

One of the other benefits of customers having O3000 series over O300 was upgradability - the PIMMs in O3000 were upgradable, so the first machines that shipped with 400Mhz CPU's, 24 months down the line could be upgraded to 600Mhz if required. O300 was non upgradable. Of course with Ebay prices 5 years later upgrading doesn't enter the equation, you just buy a new machine (I only have Onyx2 machines running at 400mhz now, the O3000 series have had PIMM upgrades to 600Mhz!)

[dhjj]

The board mounted power supply under that big heat sync is most likely the problem, I once found a supplier for them, but at $100 each with 100 unit minimum.

[rusti]

The board mounted power supply under that big heat sync is most likely the problem, I once found a supplier for them, but at $100 each with 100 unit minimum.

I hope you are wrong. That's what I tried to rule out by measuring the Voltage at the PCI slot. Once I have these 1780s replaced by a more professional institution I will know soon enough I guess. You might still be right. maybe I measured at the wrong spot and maybe the problem is just intermittent. Let's hope I get rid of the "home-made" fan problem as a minimum and then we will see what the voltage is and whether it is stable. Will keep everyone posted but it might take some time before I get the chips replaced by a pro.

[joerg]

My O3400 showing the same problem since 2 days. I lost 2 cBricks in the last 2 years and now the iBrick going to die. Looks the end of my SGI/MIPS time is comming.....

Regards
Joerg

[rusti]

My O3400 showing the same problem since 2 days. I lost 2 cBricks in the last 2 years and now the iBrick going to die. Looks the end of my SGI/MIPS time is comming.....

Regards
Joerg

This could be rather "interesting". Interesting in quotation marks because I feel with you. I did not buy my Origin3200 to rip it apart and never be able to use it, obviously. But what I mean is: Can you give more details on the symptoms your cbricks showed and your ibrick is showing. So far I did not actually loose my ibrick. It ran fine with env monitoring off. The only problem was that I wanted it to run with env monitoring on...

On a more "grave-raping" note: I know of two cbricks that would happily take care of your R14k-600 PIMMs from your dead cbricks. That is if I get my ibrick back and it is working.

[joerg]

Well... the iBrick shows the exact behaviour like yours....

Code: Select all

Jul 21 02:10:58 4A:o3k unix: |$(0x163)WARNING: 001i21 ATTN: power down aborted, environmental monitor reset
Jul 21 02:10:58 4A:o3k unix: |$(0x157)WARNING: 001i21 ATTN: -12V high fault limit reached @ -14.614V.
Jul 21 02:11:01 3A:o3k INFO: The system is shutting down.
Jul 21 02:11:01 3A:o3k INFO: Please wait.

The messages from the cBricks are similar but just like CPU related in the message. I cant remember to the details.

Regards
Joerg

[rusti]

The interesting question will be if it is the environmental monitor failing (in some are even all cases) or if there are true faults for instance in the voltage regulators.

As we have seen in the Fuels it was a very common failure of the DS1780 env monitors. When I asked the question whether this is something that was also observed on the O3xxx the answer was no, probably not, because that would have been known by now. I am beginning to wonder if that was a valid assumption. I will keep investigating and let you know.