SGI: Hardware

Dream came true - Tezro arrived, selfmade VRM/PSU probs

Hi guys.

At last the forums are up again, thx :D

So i finally found my tezro on ebay by the end of nov 2014, maybe some of you followed the auction .

Finally she came after a month shipping through half the world on time to christmas :D

The auction stated that there is a gfx config error on boot up and its got 2GB ram.
Prior to bidding i found the post on the forums about the error, that its just missing the monitor plugged in.

In fact the machine ran great, and also has 8GB ram, wohoo (of which 2gb are disabled, but hey :D )

Ran? Well, after booting her up for once and seeing everything is working, theres even the original irix installation on the disk.
I already got a plan regarding quieting the baby down. The fans were laying there for weeks to get built into the tezro, and so i started replacing the fans.

After 6 hours soldering i finally got everything back together. But something didnt work out quite well. Switched her on, wow, nice and almost unhearable....
But after ten seconds or so i got an error in the L1 display: "VRM not ok, vcpu #1...."
Damn...
After looking around i think i found the broken part: The error message says it: The vrm plugged between the DIMMs seem to be cooked, shit.

I then thought about the PSU fan replacement maybe failed and i killed the psu... (Its one pice of sh*t to work with)
Well the psu works, and the currents seem to be ok (im an electronic noob, but i can measure currents, i think ^^).
I then fiddled around with the psu, opening, looking what may could be wrong (im still living, no panic ^^), building back the original fan in, replug, run, vrm message still there.......

I found the right VRMs on ebay and ordered six of it (oO), i think they will arrive tomorrow or wednesday.

Now the question is if i could trust the psu. In case it fails i got enough VRMs left, but you know....

I also fiddled around with other psus to get the baby running. You know, i waited since ten years to get one, and now its broken and its Dez 23. evening and fuck.

So i got the tezro working with my standard enermax psu......
With that in mind that a standard psu could work in the tezro, im thinking about replacing the whole psu with a silent and efficient platinum psu.
Two problems there: The 10 Pin 12V (CPU) connector and the i-dont-know-what-it-is-cables (5 Pin, like the old cd-rom audio cables) that seem to handle environment monitoring (because i dont have the vrm errors with the enermax psu and the tezro boots to the maintenance menu, and yes i know im crazy as fuck to even try this out).
The first problem is none, because you in example just take the PCIe connector for the graphics card and plug it in reverse besides the 8Pin 12V CPU connector you have on the standard psu.

Trying to boot the tezro with the enermax psu freezes the machine, i think it has to do with the VRM.

I keep you updated with my experiments.


[E]
What id like to add:
The fans ive choosen are noctuas with 1800 rpm.
As i found the environmental specs of the tezro, telling that the warning rpm for the fans are at least 1500 - 1800 rpm, i thought they would be best.
Now i think the tezro dont give full power to them, because in L1 env they show around 1500 rpm.
Is there a way to tell her that she should always give full power to the fans?

Maybe i have to choose other ones that turn initially faster.

But guys i tell you - With all the noctua fans she is not really noticeable, except the whiny scsi disks, you know them :D
MrWeedster wrote: The fans ive choosen are noctuas with 1800 rpm.
As i found the environmental specs of the tezro, telling that the warning rpm for the fans are at least 1500 - 1800 rpm, i thought they would be best.
Now i think the tezro dont give full power to them, because in L1 env they show around 1500 rpm.
Is there a way to tell her that she should always give full power to the fans?


Does L1 actually complain the RPM's are too low shutting the machine down ?

Give my experiences with Fuel, it should adjust RPM's as temperatures go higher, so just keep observing L1 and RPM correlation with temperature.

Alternatively you *could* generate constant PWM signal that would always give full RPM... In theory a simple 555 timer circuit would do the job (and work even at 12V which some bigger fans need)
[click for links to hinv] JP: :Fuel: | :O2: | :Indy: || PL: [ :Fuel: :O2: :O2+: :Indy: ]
Yes it did until i shut env monitoring down.
Hmm, it turns L1 allows to change voltage margin levels, but you can't change RPM warning levels.
Then your options are to get different fan or fake the signal to ensure L1 reads a value within the specs...
[click for links to hinv] JP: :Fuel: | :O2: | :Indy: || PL: [ :Fuel: :O2: :O2+: :Indy: ]
To focus on the issue im posting my problems with the tezro here.

Yesterday i got the VRMs i thought that were faulty. After exchanging it the problem stayed the same:
If i want to boot into installation tools/OS the screen turns black and nothing happens.

The L1 reads "OS booting" to "OS running", but the screen stays black.

It took a while to figure out how to connect to the serial console/l1, but now i got it and made some screens.

The one is before switching on (I wonder why it onlys shows some currents of the (enermax) psu).
The other is right after switching on, and the last after shutdown - There you can see that the last line also complains about the VRM.

Im so confused about this. What does she want from me and how can i find out?
Btw i took the original PSU to some guy who should repair it, maybe more luck with the genuine psu.

Do i have to replace the ip53 board? Or something else? V12? Maybe i try the V10 from the fuel.
Ok i think i found it, laying in the case.

Board broken :<

[Update]

It turned out the part was from the IO9 card, near the holder bar for the card.
I soldered it back in, changed nothing.
Update #2: It seems to be the V12 (maybe the whole machine ^^).
I exchanged it with the V10 from the fuel, now it works.

Installing now
Shes running really good now.

But i have to find out something about the fans. Is there a way to shut off autofan control and let them blow 100%?
Earlier in the thread, MrWeedster wrote: Ran? Well, after booting her up for once and seeing everything is working, theres even the original irix installation on the disk. I already got a plan regarding quieting the baby down. The fans were laying there for weeks to get built into the tezro, and so i started replacing the fans .

After 6 hours soldering i finally got everything back together. But something didnt work out quite well . Switched her on, wow, nice and almost unhearable....
But after ten seconds or so i got an error in the L1 display: "VRM not ok, vcpu #1...." Damn...

So it worked when arrived but not after you changed the fans?

Never messed with the fans in a Tezro, but other IP35 systems like the Fuel and the O350 *don't* use standard fan wiring. There's photo that illustrates the differences in the Fuels fan wiring at the bottom of the The IP35 (Fuel, Tezro, O300, O350) Hardware Aggregator . Might be worthwhile to take another look at how you wired the replacement fans.

Your Tezro, your call, but my unsolicited advice would be to reinstall the original fans and learn to accept the the Tezro fan noise as just one more over-engineered artifact from the golden age of SGI workstations. Even with the stock fans Tezros tend to run on the hot side, so if lowering the volume also lowers air flow, in the longer term the result might be a Tezro that's dead quiet..... permanently.
***********************************************************************
Welcome to ARMLand - 0/0x0d00
running...(sherwood-root 0607201829)
* InfiniteReality/Reality Software, IRIX 6.5 Release *
***********************************************************************
Right, i exchanged the fans and the trouble began.

But going back to the papst fans is no option. Instead, i cut the node fan cables today and soldered the +12v lines from the fans to an adapter i plugged directly to the psu.

Now the noctuas run at their desired speed of around 1900-2000 rpm, and even the tacho signal works now. Dont ask me why ^^

So, after around 30 minutes up in idle, the temperatures are as following:



The silence of the baby is awesome!

I now try to get all the tacho signals and try to get the pci #1 fan speed up, or exchanged, to finally turn on env monitoring again.
Ok, i redid some fans and put them on V12 (HD and PCI#1).
Now the RPMs are in the good range, so i turned env monitoring on.

After running about 30 mins in idle:



Well its not that quiet now as before, but at last she runs stable with cool temps and acceptable noise level.

Love it :D

[E] Some Pictures:


MrWeedster wrote: Ok i think i found it, laying in the case.

Board broken :<

[Update]

It turned out the part was from the IO9 card, near the holder bar for the card.
I soldered it back in, changed nothing.


If that cap became unsoldered by itself then its most likely shorted and you should leave it off. I've only ever seen a few ceramic caps go bad.
Today i played with the test command on the L1 a bit and found interesting things:

Code: Select all

001c01-L1>test exc mem
INFO: Cannot enable VRM: 9
INFO: Cannot enable VRM: 10
INFO: Cannot enable VRM: 11


****************************************
controller firmware panic!   resetting...
****************************************

IMAGE B: Rev. 1.36.17
[sys stack]
TR: fff6ce38 fff67066 fff6d53e fff3814a fff00fee
[thread ID 30004950 stack]
Memory Fault, PC=fff2eac8
TR: fff730ce fff10690 fff1100c fff37f30 00000000

(if you see this, please email [email protected] and include
the output from the 'log' command and a description of
what caused the problem)


SGI SN1 L1 Controller
Firmware Image B: Rev. 1.36.17, Built 06/01/2005 14:43:15


001c01-L1>rst
001c01-L1>log
01/14/15 14:00:50 ChiWS IP53
01/14/15 14:00:53 CONTROLLER FIRMWARE PANIC!
01/14/15 14:00:53 IMAGE B: Rev. 1.36.17
01/14/15 14:00:53 [sys stack]
01/14/15 14:00:53    TR: fff6ce38 fff67066 fff6d53e fff3814a fff00fee
01/14/15 14:00:53    <bad stack FP @ 301017f0 => 71727374>
01/14/15 14:00:53 [thread ID 30004950 stack]
01/14/15 14:00:53    Memory Fault, PC=75767778
01/14/15 14:00:53 USB0: waiting on open
01/14/15 14:01:47 PANIC: exception.c line 1244 ; Memory Fault
01/14/15 14:01:48 L1 booting 1.36.17
01/14/15 14:01:48 ChiWS IP53
01/14/15 14:01:50 CONTROLLER FIRMWARE PANIC!
01/14/15 14:01:50 IMAGE B: Rev. 1.36.17
01/14/15 14:01:50 [sys stack]
01/14/15 14:01:50    TR: fff6ce38 fff67066 fff6d53e fff3814a fff00fee
01/14/15 14:01:50 [thread ID 30004950 stack]
01/14/15 14:01:50    Memory Fault, PC=fff2eac8
01/14/15 14:01:50    TR: fff730ce fff10690 fff1100c fff37f30 00000000
01/14/15 14:01:50 USB0: waiting on open
01/14/15 14:02:21 PANIC: test.c line 305 ; panic test
01/14/15 14:02:21 L1 booting 1.36.17
01/14/15 14:02:21 ChiWS IP53
01/14/15 14:02:23 CONTROLLER FIRMWARE PANIC!
01/14/15 14:02:23 IMAGE B: Rev. 1.36.17
01/14/15 14:02:23 [thread ID 30004950 stack]
01/14/15 14:02:23    TR: fff6ce38 fff67066 fff2eb0e fff730ce fff10690 fff1100c fff37f30
01/14/15 14:02:23    TR: 00000000
01/14/15 14:02:23 USB0: waiting on open
01/14/15 14:03:59 PANIC: exception.c line 1244 ; Memory Fault
01/14/15 14:03:59 L1 booting 1.36.17
01/14/15 14:03:59 ChiWS IP53
01/14/15 14:04:01 CONTROLLER FIRMWARE PANIC!
01/14/15 14:04:01 IMAGE B: Rev. 1.36.17
01/14/15 14:04:01 [sys stack]
01/14/15 14:04:01    TR: fff6ce38 fff67066 fff6d53e fff3814a fff00fee
01/14/15 14:04:01 [thread ID 30004950 stack]
01/14/15 14:04:01    Memory Fault, PC=fff2eac8
01/14/15 14:04:01    TR: fff730ce fff10690 fff1100c fff37f30 00000000
01/14/15 14:04:02 USB0: waiting on open
01/14/15 14:04:36 reset (COMMAND)
01/14/15 14:04:36 Node 0 XTalk clock 88


Which part do i have to change? Node board?
After playing around more i found out some interesting thing:

The V12 works, but only at the PROM and the Desktop - Between the screen stays black (Which also means you cant install the OS).
My problem was that it somehow killed the OS disk that came with the tezro - So i wanted to install new on a new disk, but the screen stayed black.

So i built in the fuel V10, installed the OS, switched back to V12, be patient to let her boot, and as soon as the login screen appeared, the picture returned on the monitor.

For the PSU i took a cheap used Corsair HX520, which works perfect. Also i got a new front panel (Thank you Ian), so shes perfect:

MrWeedster wrote: After playing around more i found out some interesting thing:
The V12 works, but only at the PROM and the Desktop - Between the screen stays black (Which also means you cant install the OS).
My problem was that it somehow killed the OS disk that came with the tezro - So i wanted to install new on a new disk, but the screen stayed black.

That's how my Fuel behaved - since upgrading from V10 to V12 I can no longer see boot messages or use a "single" console, so keep fingers crossed until the login screen appears. Not much consolation but you're not alone (and maybe it's normal/intended behaviour!)
Fuel ; Indigo2 ; RiscPC Kinetic-StrongARM/448MB/RISCOS4.39 or Debian-etch; EspressoPC ViaC3/900MHz/256MB/Debian-testing; RPi B RISCOS5.21 or Raspbian-jessie; A5000/33MHz/FPA11/8MB/RISCOS3.11; A540/25MHz/FPA10/16MB/RISCOS3.11 or RISCiX1.21; R140/35MHz/4MB/RISCOS3.11 or RISCiX1.21
Sounds like a PROM flashing may need to be done - but I'm no expert at the newer SGIs.
SGI:
:Fuel: R16000A@900MHz 3GB V10
L2 Controller
Non-SGI:
HP C8000
HP EliteBook 8560p [email protected] 16GB Youmu FreeBSD 10.1/Windows 8.1
IBM IntelliStation 265 Dual POWER3-II@450MHz Jigoku-Karasu ( Hell Raven )

Incoming/On bench for repair/not in service:
2x :O3x0: Origin 300
:A3502L:
For Sale: O2 DIMMS, Octane and O2 caddies, Octane2 maxed out.
Well at the installation of 6.5.30 it says PROM is up to date, no flashing required.

But without console/boot messages how should one be supposed to install the os? (Because that screens doesnt work either)