The collected works of kubatyszko - Page 4

You're welcome Glad it worked, I think I need to become more active, seems a lot of people refer to my research

Announcement.

After many months of development (not to mention over 3 years since the initial pre-prototypes which I still use), I present you…
The snap-in ready-to-use Fuel ATX power supply adapter.

It’s a two-piece design - the main 24-pin converter and the 12V piece (it’s simply more efficient with PCB manufacturing).

It provides self-generated PSU FAN signal - roughly 1800 RPM which is well above 1600 RPM warning level.
It converts all necessary signals to use an ordinary PC ATX PSU in a SGI Fuel.

Availability - currently I have 3 fully assembled and tested PCB prototypes, those are available for sale at a promotional price of $60 plus shipping from Japan (should be roughly $5-$9 worldwide). Payment via Paypal (preferred) or EU bank transfer to Poland, bitcoin is also possible.

Users in the Bay Area - I will be in SF between Nov 17th and 30th - pick up in person possible, the beer is on you

The prototypes have the ATX socket mounted on the underside of the PCB - that means the board is at a certain angle and the ATX plug kind of leans on the CPU module’s debug connector. This is expected and I’m debating addressing that in the future.

Those 3 are on a First-come first-served basis.
One of the three is braced with more wire for increased current carrying capacity and has fully latched plugs - if you use 800MHZ+ CPU, V12 with DCD and many other peripherals or keep the Fuel in a moving environment I recommend that one. Otherwise I will assign the adapters to orders at random - https://xkcd.com/221/

All adapters are made in Japan by my very own self, fully tested and verified electronically to match maximum current capacity that ATX cables themselves can supply (hence very wide PCB traces and lots of solder all over).


If there is more demand I’m open to a larger production run - the price will then be $75 a set.
The earliest availability date for the next run will be mid-december at the latest, all orders placed beyond 3 prototype pieces will be considered up for the production run. A chance that *you* will have yours by Christmas


For the reference - the thread that started it all is here: http://forums.nekochan.net/viewtopic.php?f=3&t=16725574&start=45

Please be sure to read - it contains useful hints on how to buy an appropriate ATX PSU - it’s not that easy.

The easiest exclusion criteria is that it should provide at minimum 30 Amperes on each of 3v3 and 5v rails, most high-power supplies provide much less - those will likely not work. I recommend at least 500-550W total power.


Side note 1 - this is NOT an open source design at the moment and the schematics/PCB design will not be available.
If you want to make your own - all the information you need is on the thread above.
If you want a ready-made adapter - my offer is for you, you’re paying for a fully working and tested device and hundreds of hours spent on development and safety (believe me it was not easy), and obviously you're saving many hours yourself that you don't have to spend building your own.

Side note 2 - this is a new product and I will provide warranty for the workmanship and potential malfunctions within one year.
What the warranty DOES NOT include is:

• Bad ATX PSU choice including bad voltage regulation, the 3v3 sense line stays connected at the plug level not at the socket level - this may or may not result in variations in 3v3 voltage supplied by the PSU.
• Wrong installation
• Potential incompatibilities with *your* Fuel - it is possible that certain L1 or PROM versions detect PSU signals in a certain way that the adapter may not be able to subtitute. The adapters are tested on a L1 version “Firmware Image B: Rev. 1.24.11, Built 10/29/2003”

Obligatory screenshot (this was a test using 400W PSU on my designated motherboard for this kind of experiments, without GFX), not enough as you can see - the voltages are a tad too low.




And some pictures:

Fuel is fragile and ugly - perfect match, and besides, it's not built for the looks but rather out of necessity to save all those poor Fuel's from extinction.

If you look carefully (assuming you have a Fuel) - this is designed to fit into *VERY* cramped case of the Fuel (especially the plug area), there were a ton of compromises I had to make, such as the 12V module has to make room for the black front panel connector (hence the "L" shape), the main big board has a 45 degrees cut on the right because it has to go around 5.25 inch cage etc etc.

If you mean the assembly itself - I'm not sure if you realize how big currents ATX can produce - at full power that could even mean 5 cm thick traces - that's rather impossible - hence solder to ensure enough current capacity.

dexter1 wrote: A word of praise is in order.

Nice work and i am glad to see that someone took the time and effort to come up with a solution for maintaining a working Fuel without having to go through hard-to-find replacement parts. I honestly don't know if there even is a replacement market available for Fuel PSU's. I guess if there were, kubatyszko would not make the PCB in the first place

Thank you!.

Market for PSU's - probably not but there must be a graveyard somewhere...

Hi everyone.

Meanwhile I've been working on an improved version of my ATX adapter for the Fuel.

The changes are:

*) two fan signals generated:
1) one is by default jumpered to feed into the M/B - the jumper can be removed if one wished to supply their own (in case ATX PSU provided one).
2) second is just ready to use for any other purpose (say if one was in a need of fooling the env monitoring even more).

*) "kickstart" button - a small pushbutton that forces the ATX PSU to start, in case Fuel won't (it may refuse to power up if components are missing or env chips are fried). This allows one to read all the voltages (not only standby) and verify L1 output even if the Fuel refuses to boot. The rationale was that since you're buying the adapter because your main PSU died - it might have broken something else in your Fuel. This helps debug the issue.

*) standby (yellow) and power on (green) LED's - indicate whether the standby power is good (for L1), and whether the PSU is on providing power on 5V rail. This helps prove that your ATX PSU is connected correctly.

*) There are NO changes to the 12V adapter (small board)

The original thread announcing the prototype is here: http://forums.nekochan.net/viewtopic.php?f=3&t=16730132

The price is $75 a set plus shipping from Japan.

Pictured is an actual product, without added solder and wire brace for more current capability.
The units by default have upright ATX connector (until I figure out solder amount on the traces and heat exchange) - this means your ATX cable may protrude into secondary 3.25 inch cage if you have any drives there. I have a SATA disk in mine and it fits just fine, albeit it required some work to wiggle the cable in. On special request I am able to install a 90 degree connector.

don't know if that helps but I'm pretty sure the word you're looking for is "stripped" not "striped" - maybe that makes the search easier.
Stripping binaries (often via 'strip' command) or OS means removing all non-critical stuff, essentially making it lighter which is likely what you want to achieve...

There is no need to shoot with dictionaries, I understand the difference between the words.

I was referring to this sentence - that kind of looked like a name of an OS image you're looking for.

ivelegacy wrote: .... striped-ram-rootfs with an MIPS3-BE-softfloat-uclibc....

Anyway, looks like an interesting challenge, too bad it's not in my current line of SGI-related work, I'm wresting with my Fuel's and bringing them back to life, two motherboards had missing components that came off due to harsh treatment. Both are alive now.

Thread lives on!

Small PSA for those experimenting with ATX and Fuel, I just tried a 400W PSU with one of my adapters, and despite voltage level issues it works okay, the Fuel has a V10, 800MHz PIMM, 2G RAM and one IBM 18GB HDD - I'd say it's a rather good configuration, except V10.
The PSU has the following specs:

3v3 : 28A
5v : 40A
12v : 16A (only one rail)

According to our research from this thread, it's fine on the 5v but definitely lacking on the 12v and 3v3 rails - and that's clearly visible in L1 output...

It's clear the PSU has difficulty regulating the voltages, 12V at boot and then 3v3 during operation, but even then the machine runs fine.
I have not tried loading the machine up with any software and I won't (because I'm also missing that blue air-duct - using a regular household fan to improve the heat exchange ).

WARNING - this is just an experiment (because I used up my PSU's and this was the only I had left), and I highly advise against using underpowered PSU in *your* Fuel...

toasty wrote: Giving this a bump because I bought one of these adapters (version 1) from kubatyszko a few months ago. And I was finally able to use them for the first time last night when I replaced my Fuel's dead motherboard. I went with the safe choice and got an ATX version of the Sparkle PSU (FSP460-60PFN). Installation was easy and I'm pleased to say my Fuel has been running for several hours now without any problems. Many thanks to kuba for his work on those adapters! I highly recommend it.

My env output. Voltages all seem to be within normal limits:

Code: Select all

rhaegar 2# l1cmd env
Environmental monitoring is enabled and running.

Description    State       Warning Limits     Fault Limits       Current
-------------- ----------  -----------------  -----------------  -------
12V    Enabled  10%  10.80/ 13.20  20%   9.60/ 14.40   12.19
12V IO    Enabled  10%  10.80/ 13.20  20%   9.60/ 14.40   12.25
5V    Enabled  10%   4.50/  5.50  20%   4.00/  6.00    5.04
3.3V    Enabled  10%   2.97/  3.63  20%   2.64/  3.96    3.32
2.5V    Enabled  10%   2.25/  2.75  20%   2.00/  3.00    2.46
1.5V    Enabled  10%   1.35/  1.65  20%   1.20/  1.80    1.48
5V aux    Enabled  10%   4.50/  5.50  20%   4.00/  6.00    5.07
3.3V aux    Enabled  10%   2.97/  3.63  20%   2.64/  3.96    3.30
PIMM0 12V bias    Enabled  10%  10.80/ 13.20  20%   9.60/ 14.40   12.25
Asterix SRAM    Enabled  10%   2.25/  2.75  20%   2.00/  3.00    2.52
Asterix CPU    Enabled  10%   1.44/  1.76  20%   1.28/  1.92    1.61
PIMM0 1.5V    Enabled  10%   1.35/  1.65  20%   1.20/  1.80    1.49
PIMM0 3.3V aux    Enabled  10%   2.97/  3.63  20%   2.64/  3.96    3.29
PIMM0 5V aux    Enabled  10%   4.50/  5.50  20%   4.00/  6.00    5.07
XIO 12V bias    Enabled  10%  10.80/ 13.20  20%   9.60/ 14.40   12.12
XIO 5V    Enabled  10%   4.50/  5.50  20%   4.00/  6.00    5.04
XIO 2.5V    Enabled  10%   2.25/  2.75  20%   2.00/  3.00    2.46
XIO 3.3V aux    Enabled  10%   2.97/  3.63  20%   2.64/  3.96    3.29

Description    State       Warning RPM  Current RPM
-------------- ----------  -----------  -----------
FAN 0  EXHAUST    Enabled          920         1256
FAN 1       HD    Enabled         1560         2321
FAN 2      PCI    Enabled         1120         1573
FAN 3    XIO 1    Enabled         1600         2475
FAN 4    XIO 2    Enabled         1600         2475
FAN 5       PS    Enabled         1600         2065

Advisory   Critical   Fault      Current
Description    State       Temp       Temp       Temp       Temp
-------------- ----------  ---------  ---------  ---------  ---------
NODE 0            Enabled   60C/140F   65C/149F   70C/158F   41C/105F
NODE 1            Enabled   60C/140F   65C/149F   70C/158F   44C/111F
NODE 2            Enabled   60C/140F   65C/149F   70C/158F   34C/ 93F
PIMM              Enabled   60C/140F   65C/149F   70C/158F   52C/125F
ODYSSEY           Enabled   60C/140F   65C/149F   70C/158F   43C/109F
BEDROCK           Enabled   70C/158F   75C/167F   80C/176F   45C/113F


Thanks for your feedback, I'm glad it worked.

For the general public - at first it didn't want to work, then I sent toasty a revised version of the board (this is when I got the idea for the kickstart button). After that we were able to conclude what was wrong with his Fuel, it turned out to be the motherboard and monitoring chips (voltages were off, some were at 0v, some fan speeds not showing up) . After replacing the M/B it works as you can see.

Donations are open again - I've done my part.

Oh bummer, I just missed this post - I guess the trip already happened?

Anyway, after 10 years in Tokyo I moved out of Japan to SoCal (LA) so I can't be of any local help anymore, just throwing this for the future reference...


There are very few shops in Akiba left with used stuff, one is just near the station (roughly here - 35.698252, 139.772083) but all I saw there were used MAC's and maybe one X68000, nothing "rare".

There is an interesting shop called "super junk" - mostly electronics but sometimes other random stuff pops up - 35.698559, 139.770187

Another shop is right on the corner here 35.700389, 139.769372 - they have some gaming and retro stuff.

Sometimes shop chain "hard off" has retro hardware - there's one just near the one mentioned above.

Yahoo auctions might be the best bet - but you need to be a paying member to buy items over 5000 JPY directly. Sellers would never deal with shipping abroad (I was always the exception), locals are very protective and one time I even got a neutral comment saying that I was probably a foreigner judging by how I treated the buyer..

I *hate* buyee - I do know that they provide useful service and enable buyers from abroad, but since I did offer international shipping I didn't want to go through middle-men... Buyee is so weird - once *they* bought 2 or 3 items from me at the same time and asked precisely to ship them in separate packages and write some long ID strings on the box (which means they brainlessly forward packages abroad).


Cheers

Depending on terminal, ALT+key might also work, it's basically META+key and ALT+ or ESC+ are the same thing. IRIX accepts regular .Xdefaults for Xterm, might be doable.

You might have better luck finding a GFX 1600 SW PCI card and simply wiring up power (5v, 3.3v and ground) to it - it won't look pretty but it should be cheap. The card was meant for ordinary PC's and is basically multilink in a PCI form factor.

As far as multilink - I've repaired a couple of them simply by replacing capacitors, I vaguely recall chopped image on the screen and hissing noise coming out. I don't remember ever seeing plain black screen though.

hamei wrote: Jesus Christ. Can someone who has a clue please step in here ?

It was COMMON to swap out the pre-300 mhz processors. Worked in both single and dual forms. But at 360 mhz SGI went to soldered processor chips, which are a bear to replace. Even with the O2 boards and soldering on new outta the box chips, doing small lots of bga chips is expensive and has a lower-than-100% yield.

In practice, no one is going to make dual 600 boards from lower-spec boards because it would cost twice as much and only have about an 80% chance of success. Plus there's no source for new 600 mhz cpu's. What is the plan here, buy two old single-p boards, unsolder the chips to resolder them onto a dual-p setup ? Totally not sensible.

Not only <100% yield, those modifications even require X-raying the modules to ensure the soldering is done correctly, changes in voltages I think and most importantly modifications of PROM to correctly recognize the new CPU (and since there are no PROM sources it's all reverse-engineering and guesswork).

It's most certainly not worth the cost.

FYI - Those adapters are made by another nekochan member - michaelpastras

Yep, back in Tokyo I've deliberately chosen my provider because they had native IPV6, IRIX worked flawlessly. Call me lazy but it's so much easier when I can set up all my hosts all over the world without crazy routing necessary.