[jwhat]

Hi Megaimg & Weblacky,

I took the power supply out of my Fuel and opened it up as I was curious to see if it had seperate controller board like the Sparkle...

Mine is: PSU 430W: - SGI PN 060-0140-004 Rev B, Model GM430WTXW01SSV (from NMB Technologies - Made in Thailand)

As with all things SGI, I did not find what I was hoping for:
- seperate control circuit, rather I found that in this case all the circuit electronics are on the main boards (there are two layers as per Sparkle) and
- the fan is two wire only .

I took pictures and put it back into machine.

Pictures are of:
- Top Layer:


- Bottom Layer:



So based on this I think I am now on hunt for SGI Sparkle which can be recapped.

Cheers from Oz,

jwhat/John.

[jwhat]

Hi Megaimg & Weblacky,

seems that Dallas DS1780s are everywhere in SGI boxes.

In reading the spec sheet, it says that each DS1780 has 2 x fan speed monitor.

So for every 2 fans need another DS1780, reason I raise this in this thread is that I wonder if the SGI Sparke addition circuit is based on DS1780 and so could be easily replicated without even sighting the original.

Once you have the monitoring circuit you could just cut off existing power terminals and rewire/recrimp new plugs on.

IRIX Network members kubatyszko and cz7asm both know the details of the wiring spec , as per this sticky thread: https://forums.irixnet.org/thread-3376.html

I raise this as then we could convert PC Sparkles (lots of these about) into SGI Sparkles, which would be nice near original solution to Fuel PSU replacement. Could possibly even apply some approach using “better” more modern ATX PSU.

NOTE #1: the 1 DS1780 per 2 fans is confirmed by the O350 env thread: https://forums.irixnet.org/thread-3424.html where there are 5 fans and 3 x DS1780s.

NOTE #2: Maybe I am barking up the wrong tree, here (I have before ;-) ), and need to read more of the historical threads... does the Kuba / wiring ATX adaptors just simulate a "pulse" which is then passed to an existing internal DS1780 on Fuel system board or is it sending a result back down the ATX "wiring" to some other circuitry on the Fuel ?

NOTE #3: found the original Kuba Nekochan thread on this: https://web.archive.org/web/201605100013...0#p7348338

Cheers from Oz,

jwhat/John

[weblacky]

My crude understanding for FUEL (Tero, I do not know) is that Pin 12 and 24 are fan signals (but not directly from the PSU fan) on the the WTX connector for Fuel (no additional plug). The Kuba adapter already "fakes" this "fake, not real, RPM TACH" fan signal for you. Either way...it's NOT a native fan signal, it's already been altered in the PSU (I thought). I don't recommend reverse engineering Kuba's adapter for the reasonable price it is but since the Fuel doesn't command the PSU fan speed (is only informed). It doesn't really matter and plugging in a "fake fan rpm signal" is okay in my book. That's what Kuba's adapter does. It has a tiny SMD IC (didn't actually read what it was, likely a timer or something) that sends the right pluses for a "correct" PSU fan speed to the Fuel MB. It's a fake out.

I do not know if the "receiving end" on the MB is just a DS1780 either. You could check by seeing if pins 12 or 24 connect directly to a DS1780 on your MB but otherwise with an adapter readily available I just don't see the want/need to convert PSUs. It's hard enough to find correct units that meet physical and electrical standards the adapter already mods the MB so you can swap future PSUs. And remember you deal with BOTH a Fan signal AND a WTX to ATX pinout different. Even if you solve the FAN signal on the main board...you'll need to solve the other on the PSU plug or something...the current adapter does BOTH so you just slamming ATX PSUs into it is really golden. Takes the motive for me right out for Fuel (at least).

[jwhat]

Hi Weblacky,

your understanding: "FUEL (Tero, I do not know) is that Pin 12 and 24 are fan signals (but not directly from the PSU fan)"

I had not considered that....

This goes back to the variation of the SGI Sparkle vs the PC Sparkle.

The SGI Sparkle has an additional small circuit board and a 3 Wire fan, while PC just has a 2 Wire fan.

So you are saying the small circuit board in the Sparkle PSU is just a signal simulator ...

I can test if the Fan Signal goes to a DS1780 pin with volt meter, the extent of my electronic measurement equipment.

I am not wanting to reverse engineer the ATX adaptor, rather I am trying to see if you can convert a PC Sparkle to be same as an SGI Sparkle.

However if the SGI Sparkle is just using a "fake" signal as well (as per ATX adaptor) then that has no value.

Cheers from Oz,

jwhat/John

[cz7asm]

I also spent some time studying PSU's used in Fuel and I came to the same conclusion - the PSU fakes the fan speed. Not sure how intelligent this faking is though. Whether it's based on at least some feedback from the fan, like consumption or it's purely made up based on temperature with no data from the fan at all. Actually, there must be some feedback because I remember fixing once a broken PSU fan which didn't spin and the speed reported was correctly 0.

The fan signal is then fed to one of the DS1780. Already forgot which one. But some time ago I was tracing the signal when developing my version of the PSU adapter (here)

As stated in this thread - the adapter is faking already faked signal, that's very true

[jwhat]

Hi Weblacky & cz7asm,

thanks for info, I also read the nekochan thread in more detail and looked more closely at my photo of the GM430WTXW01SSV PSU from my fuel above and saw that the PSU has a 74HCT chip (with what looks like a Philips logo on it) just next to fan plug and the Nekochan Kuba thread has 74HCT as one of options to generate signal.

So my summary is that yes even original SGI had “fake” signal, though to cz7asm’s point the circuit in SGI had enough smarts to at least detect if fan was spinning.

So my new speculation is that the extra board in SGI Sparkle has a 74HCT chip or similar in it and not a DS1780, as initially guessed.

As per cz7asm comment this just feeds straight through to an on board DS1780.

Maybe when you open up your SGI Sparkle Weblacky, you could confirm this.

Cheers from Oz,

jwhat/John

[kubatyszko]

Re faking the signal by OEM SGI PSU's, that's may not necessarily be true.

It's been a while since I looked at a NMB PSU, but I recall it had a PIC Microcontroller (mine had it fried), which may very well be controlling the fan voltage (if 2 or 3-wire), or even PWM (if 4-wire). Can't remember what the wiring of the fan was (but I might pull up old photos someday to verify).
I think that micro was the only thing different between the SGI NMB and standard ATX NMB (barring pinout differences).
These "lookup tables" are frequently seen in an embedded world.
The PIC may then independently report the Fan speed into Fuel based on the voltage it (in non-PWM setup) sends to the Fan, so even though the fan doesn't report directly, it's correlated.
This is a very common technique, for example "power/resistance curves" for bicycle trainers (one of my other hobbies) are well defined, and it's very easy to derive the power "requirement" given trainer's spinning speed.

As for Fuel-ATX adapters, this is the VERY reason why my adapter both "fakes" the signal but also has a jumper that allows you to feed the fan speed signal into it.
I've done it on one of my Fuel's and it was nice knowing that the L1 relies on actual fan speed data to protect my Fuel.

[jwhat]

Hi kubatyszko,

thanks very much for commenting and also thanks very much for helping to provide a second life to lots and lots of Fuels:-)

I can absolutely confirm that the SGI PN 060-0140-004 Rev B, Model GM430WTXW01SSV (from NMB Technologies - Made in Thailand), has 2 Pin Fan and the 74HCT chip.

As someone who obviously knows much more about this stuff than me can:
- you comment on how you might detect if 2 Pin Fan was spinning or not ?
- advise on the air flow rate of SGI Sparkle (has NIDEC Beta SL - 80mm, 12V, 0.22 AMP Fan) as fan is no longer made I cannot find any flow rate info (needed for alternate model) ?

Also thanks for the info about the Jumper on the ATX adaptor which allow for either "faked" or "real" fan signal, I was not aware of that.

I agree that having an SGI PSU is not really needed, its a bit like Weblacky's retro "Octane Light Bar" (which I currently have in my Octane2), it is for historical aesthetics not necessity.

Cheer from Oz,

jwhat/John

[weblacky]

I agree that having an SGI PSU is not really needed, its a bit like Weblacky's retro "Octane Light Bar" (which I currently have in my Octane2), it is for historical aesthetics not necessity.

To be far, it's a little more than that, while it might be nice to have a real Fuel PSU, it's still my theory they do a little ENV damage on their "way out", that's really not acceptable.  Even if I get to rebuilding Fuel PSUs...I doubt I will EVER use one in the long term as they've proven themselves suspect (compared to all other SGI PSUs).  If it fits into the case cleanly and screws in, ANY PSU that works is acceptable to me on an SGI..I don't consider the "original" PSU to be of great importance...only original functionality and outward appearance to be important.  If the ATX PSU hung out the rear, cables dangling off the Fuel...unable to auto-power on/off...I'd not accept that as a long-term solution. But the PSU is like the starter or alternator in a car, a car that has a new one doesn't have diminished value, in my eyes.

Likewise I demand to use incandescent bulbs in the Octane light bars because the LEDs do not replicate the original appearance, which is a big deal to me.  An outward mod simply reminds me that it's not original and has been tampered with.  If this were a functional device like a tractor or a swimming pool...I'd not care.  But these are on my shelves and walls every day and are both art and a hobby, I'd go insane staring at a grainy-looking 3D printed part or a huge crack or the wrong case color, etc... every day as a I work. I'll pay for good to mint skins because of that.

Part of my collection is to be as close to museum quality as I can get.  Most of my units can be placed behind glass (facing forward) and be acceptable in a showcase.  

It's a requirement for me.  Just like, if you're into pinball machines (I'm not) you don't do an LED conversion on a PURE bulb unit...it won't look the same.  Unless for you're going to monetize it by running it in a business to attract people, it's really going to be at home, turned on only occasionally just for you and on that occasion it better look right and transport you back to a yesteryear so you can enjoy playing, nostalgically.

Otherwise it's not a collection but just keeping an old device (like a lawn mower) running without caring about it's appearance.  I realize some people don't care but it's not like I use SGIs to surf the web or do anything.  I look at them way more than I use them.

Small damage or changes are just inside the fence of "uncanny valley" or disturbed apparence - like owning an O2 with no skin and just seeing the metal case...If I were unable to get a replacement skin...I might either have to sell the unit quickly and take it out back and "put it down" to prevent any more suffering on both our ends.  I cannot bare to see them like that.  Not for the value and history they represent.  It's not a Sega or Nintendo gaming system they made millions of that you can just go on eBay and get another one of...SGIs are too rare for that.  For me, they deserve our preservation like a painting or an art print.  Just as an art restorer tries to minimize or hide the damage, so too would I try to replace "like" with "like".

Sorry, but this core for me, I would not have this large of a collection over 20+ years if I didn't feel strongly on this issue. Likewise I want them to WORK, so I'm trying what I can on that...the two don't need to be at odds right now.  Functional/electrical issues and case appearance are often separate issues.

[jwhat]

Hi Weblacky,

np, shows the SGI collector community is not a single homogenous group and apologies if i caused offence.

My limited collection tends towards the "pragmatic", not sure why I find the Fuel PSU such an irritant ;-).

Ironically when ever I turn on my Octane now-a-days I always think something is wrong, as I am so used to the very bright blue LED lightbar and I have to remind myself, "ohh yes I put in the iridescent bulb lightbar".

Cheers from Oz,

jwhat/John