Anyone with Sun knowledge willing to share?

It seems like detailed technical information on Sun equipment is a closely
guarded, expensive commodity. Certainly too expensive for a hobbyist, if
not outright impossible to obtain without the right friends.

I was given a fairly stripped down e450 and figured I'd just use the case
for building an AMD box. I never played with ultrasparc stuff before, and
I would be interested in doing that; I'm just afraid that after rounding
up the few needed parts from ebay/etc, I would be sorely disappointed in
the machine's performance:electrical-bill ratio.

I thought it would be kinda neat to keep the old hot-swappable power
supplies but beyond locating what may be equivalent to an ATX's PWR_ON, as
well as the main power busses (the latter were actually found on Sun's
site) I haven't gotten anywhere.

Shorting one pin to ground (looks like pin 11) on one of the 2 small
24-pin connectors to the mainboard while both connectors are disconnected,
kicks on the power supplies for a fraction of a second. I imagine it's
wanting feedback from somewhere to determine that the power is OK in order
to maintain.

I think the answer lies, at least in part, in determining the pinout for
J4503 (one of the 2 24-pin connectors on the mainboard, and the only one
that is required to be connected in order for the power supplies to stay
operational).

Has anybody tried such a thing before, or is anyone willing to share the
needed information?

Thanks.

On Thu, 29 Apr 2004 21:39:38 -0500, fred wrote:

It seems like detailed technical information on Sun equipment is a closely
guarded, expensive commodity. Certainly too expensive for a hobbyist, if
not outright impossible to obtain without the right friends.

I was given a fairly stripped down e450 and figured I'd just use the case
for building an AMD box. I never played with ultrasparc stuff before, and
I would be interested in doing that; I'm just afraid that after rounding
up the few needed parts from ebay/etc, I would be sorely disappointed in
the machine's performance:electrical-bill ratio.

I thought it would be kinda neat to keep the old hot-swappable power
supplies but beyond locating what may be equivalent to an ATX's PWR_ON, as
well as the main power busses (the latter were actually found on Sun's
site) I haven't gotten anywhere.

Are they of suitable amperage for the intended system? Primarily you'll
be needing roughly 20A @3.3V, 30A @ 5V, & 3A on 12V. This is for the
traditional power distribution, most common on Athlon systems. Then
there's the motherboard with 12V for CPU power (as evidenced by the 4-pin
connector) that need far less 5V amperage but over 10A @ 12V.


Shorting one pin to ground (looks like pin 11) on one of the 2 small
24-pin connectors to the mainboard while both connectors are disconnected,
kicks on the power supplies for a fraction of a second. I imagine it's
wanting feedback from somewhere to determine that the power is OK in order
to maintain.

It would be just as likely, possibly even moreso, that you need to supply
a load on the 5V rail. A hard drive should do it, or a automotive tail
light bulb, or a ~ 6 Ohm 5W resistor, etc.

I think the answer lies, at least in part, in determining the pinout for
J4503 (one of the 2 24-pin connectors on the mainboard, and the only one
that is required to be connected in order for the power supplies to stay
operational).

Has anybody tried such a thing before, or is anyone willing to share the
needed information?

Have you narrowed down the possibilities at all? For example,
determination of all voltage levels listed on the power supplies, and the
number of leads of each color? I'd try the load, having a multimeter
handy so you can take voltage readings on the other pins. It is not
necessary to have or hook up -12V, -5V, power-good, to run an Athlon
system "unless" the motherboard's hardware monitor causes the system to
halt, in which case you'd be seeking a setting in the bios to disable the
feature or the individual power rails omitted.

The other major issue is 5VSB (standby power).... you need it for a normal
configuration. The system might run if you don't have it, by keeping the
Power-on signal pulled low to ground then turning the AC on/off via a
switched outlet (like a surge strip) or since you're digging into the
power supply anyway, using the old AT method of having an AC switch on the
case's front bezel.

On Fri, 30 Apr 2004 07:23:48 +0000, kony wrote:

Are they of suitable amperage for the intended system? Primarily you'll
be needing roughly 20A @3.3V, 30A @ 5V, & 3A on 12V. This is for the
traditional power distribution, most common on Athlon systems. Then
there's the motherboard with 12V for CPU power (as evidenced by the
4-pin connector) that need far less 5V amperage but over 10A @ 12V.

Yes, they should be plenty suitable. I planned on running just one supply,
keeping the others for replacement parts. It can hold up to 3, using up to
2 simultaneously with the 3rd for backup. Each supply is rated for 605W,
and is spec'd to gives the following maximum output:

+3.3V @ 47A
+5V @ 64A
+12V @ 20A
-12V @ 1A
+5VSB @ 2.4A

Shorting one pin to ground (looks like pin 11) on one of the 2 small
24-pin connectors to the mainboard while both connectors are
disconnected, kicks on the power supplies for a fraction of a second. I
imagine it's wanting feedback from somewhere to determine that the power
is OK in order to maintain.

It would be just as likely, possibly even moreso, that you need to
supply a load on the 5V rail. A hard drive should do it, or a
automotive tail light bulb, or a ~ 6 Ohm 5W resistor, etc.

You'd probably have to see how a 450 is arranged to understand the
following, but the power supplies all attach to a distribution board. The
board then supplies connectors for the main 3.3/5/12v busses (rather
large conductors), drive power, fans, and safety interlocks for the side
panels. In addition, there are (2) 24-pin connectors that go to the
motherboard. They are fine (2 mm connector spacing I think), flat ribbon
connectors similar to what is often used in laptops and are probably (?)
just for signaling/logic rather than power distribution. They are nowhere
near the size of your typical 20-pin ATX connector. With just one of the 2
24-pin connectors (J4503) attached, the power supplies will switch on
and stay on as expected. If I disconnect J4503, the supplies will not turn
on. Grounding pin 11 on the connector will get the supplies to start for a
fraction of a second, then kick off. This is what I thought was either the
equivalent of PWR_ON, or maybe some sort of Wake-On-Lan/etc.

I don't think it is a load issue, as I don't believe the connector in
question carries any significant load. I could be wrong, but compared
to the way the power buses are built, I think it is pretty safe
to assume. It does have a couple of old drives in it, which spin up
normally when the power supplies kick on, but don't make any difference
regarding keeping the power supplies going.


I think the answer lies, at least in part, in determining the pinout for
J4503 (one of the 2 24-pin connectors on the mainboard, and the only one
that is required to be connected in order for the power supplies to stay
operational).

Has anybody tried such a thing before, or is anyone willing to share the
needed information?

Have you narrowed down the possibilities at all? For example,
determination of all voltage levels listed on the power supplies, and
the number of leads of each color? I'd try the load, having a
multimeter handy so you can take voltage readings on the other pins. It
is not necessary to have or hook up -12V, -5V, power-good, to run an
Athlon system "unless" the motherboard's hardware monitor causes the
system to halt, in which case you'd be seeking a setting in the bios to
disable the feature or the individual power rails omitted.

I have only been looking at the one 24-pin connector so far. While
disconnected, I have been able to identify 5 pins that appear to be GND, 3
pins that appear to be 5V supplies, the pin that kicks the power supplies
on when grounded is at 5V when open, but is not a supply. 2 pins that are
at 0V, but apparently not GND, one other pin at 5V that is not a supply,
10 other pins that vary from ~.1 to .5V, and two pins that are 4-4.5V.
I have it all mapped out, the sheet is not handy at this time though.
This is with the connector unplugged. Next step is I plan to check
them with the connector plugged in, both with the power supplies on and
off. Maybe I will notice a change that can be used to trick the supplies
to staying on. If the supplies are kept on by some active signal I'm
probably just out of luck.

Colored leads? Nope. Not a one. Maybe inside the power supplies
themselves, but I haven't opened them yet. Everything else is black or
white except for the power buses going from the distribution board to the
mainboard. They are rather thick copper strips in blue teflon-like
sleeves. The latter are clearly labeled on the distribution board though.

The other major issue is 5VSB (standby power).... you need it for a
normal configuration. The system might run if you don't have it, by
keeping the Power-on signal pulled low to ground then turning the AC
on/off via a switched outlet (like a surge strip) or since you're
digging into the power supply anyway, using the old AT method of having
an AC switch on the case's front bezel.

Not a problem. The supplies do have a +5V standby output.

Thanks for the reply. I'd be willing to hear any other comments or
suggestions.

On Fri, 30 Apr 2004 13:40:22 -0500, fred wrote:


You'd probably have to see how a 450 is arranged to understand the
following, but the power supplies all attach to a distribution board. The
board then supplies connectors for the main 3.3/5/12v busses (rather
large conductors), drive power, fans, and safety interlocks for the side
panels. In addition, there are (2) 24-pin connectors that go to the
motherboard. They are fine (2 mm connector spacing I think), flat ribbon
connectors similar to what is often used in laptops and are probably (?)
just for signaling/logic rather than power distribution. They are nowhere
near the size of your typical 20-pin ATX connector. With just one of the 2
24-pin connectors (J4503) attached, the power supplies will switch on
and stay on as expected. If I disconnect J4503, the supplies will not turn
on. Grounding pin 11 on the connector will get the supplies to start for a
fraction of a second, then kick off. This is what I thought was either the
equivalent of PWR_ON, or maybe some sort of Wake-On-Lan/etc.

I don't think it is a load issue, as I don't believe the connector in
question carries any significant load. I could be wrong, but compared
to the way the power buses are built, I think it is pretty safe
to assume. It does have a couple of old drives in it, which spin up
normally when the power supplies kick on, but don't make any difference
regarding keeping the power supplies going.

It sounds like you need to somewhat reverse engineer that 24 pin
connector's attached circuit(s). Then again you might be more likely to
find specs for Sun somewhere other than this (general-purpose, mostly "PC"
oriented) newsgroup.



I have only been looking at the one 24-pin connector so far. While
disconnected, I have been able to identify 5 pins that appear to be GND, 3
pins that appear to be 5V supplies, the pin that kicks the power supplies
on when grounded is at 5V when open, but is not a supply. 2 pins that are
at 0V, but apparently not GND, one other pin at 5V that is not a supply,
10 other pins that vary from ~.1 to .5V, and two pins that are 4-4.5V.
I have it all mapped out, the sheet is not handy at this time though.
This is with the connector unplugged. Next step is I plan to check
them with the connector plugged in, both with the power supplies on and
off. Maybe I will notice a change that can be used to trick the supplies
to staying on. If the supplies are kept on by some active signal I'm
probably just out of luck.

It would seem extremely unlikely that there's any "data" on those leads,
beyond a simple hi/low signal. In that case it would be a matter of just
determining that pin and using a relay to energize when unit turns on for
that brief period. That is, if a mechanical relay is fast enough, else
going with one solid-state.

Possibly the power supply circuit board has some of those functions
silkscreened onto the PCB?

Sun docs can be found here;
http://www.sun.com/products-n-soluti...450/index.html
Hope they are useful.

"kony" wrote in message
...
On Fri, 30 Apr 2004 13:40:22 -0500, fred wrote:


You'd probably have to see how a 450 is arranged to understand the
following, but the power supplies all attach to a distribution board. The
board then supplies connectors for the main 3.3/5/12v busses (rather
large conductors), drive power, fans, and safety interlocks for the side
panels. In addition, there are (2) 24-pin connectors that go to the
motherboard. They are fine (2 mm connector spacing I think), flat ribbon
connectors similar to what is often used in laptops and are probably (?)
just for signaling/logic rather than power distribution. They are nowhere
near the size of your typical 20-pin ATX connector. With just one of the
2
24-pin connectors (J4503) attached, the power supplies will switch on
and stay on as expected. If I disconnect J4503, the supplies will not
turn
on. Grounding pin 11 on the connector will get the supplies to start for
a
fraction of a second, then kick off. This is what I thought was either
the
equivalent of PWR_ON, or maybe some sort of Wake-On-Lan/etc.

I don't think it is a load issue, as I don't believe the connector in
question carries any significant load. I could be wrong, but compared
to the way the power buses are built, I think it is pretty safe
to assume. It does have a couple of old drives in it, which spin up
normally when the power supplies kick on, but don't make any difference
regarding keeping the power supplies going.

It sounds like you need to somewhat reverse engineer that 24 pin
connector's attached circuit(s). Then again you might be more likely to
find specs for Sun somewhere other than this (general-purpose, mostly "PC"
oriented) newsgroup.



I have only been looking at the one 24-pin connector so far. While
disconnected, I have been able to identify 5 pins that appear to be GND,
3
pins that appear to be 5V supplies, the pin that kicks the power supplies
on when grounded is at 5V when open, but is not a supply. 2 pins that are
at 0V, but apparently not GND, one other pin at 5V that is not a supply,
10 other pins that vary from ~.1 to .5V, and two pins that are 4-4.5V.
I have it all mapped out, the sheet is not handy at this time though.
This is with the connector unplugged. Next step is I plan to check
them with the connector plugged in, both with the power supplies on and
off. Maybe I will notice a change that can be used to trick the supplies
to staying on. If the supplies are kept on by some active signal I'm
probably just out of luck.

It would seem extremely unlikely that there's any "data" on those leads,
beyond a simple hi/low signal. In that case it would be a matter of just
determining that pin and using a relay to energize when unit turns on for
that brief period. That is, if a mechanical relay is fast enough, else
going with one solid-state.

Possibly the power supply circuit board has some of those functions
silkscreened onto the PCB?

On Fri, 30 Apr 2004 21:08:12 -0400, Pen wrote:

Sun docs can be found here;
http://www.sun.com/products-n-soluti...450/index.html
Hope they are useful.


Thanks. Unfortunately the freely available documentation from Sun is very
light on technical details.

I think you need to be a certified Sun guru to get your hands on the
worthwhile stuff.

On Fri, 30 Apr 2004 21:47:00 +0000, kony wrote:

It sounds like you need to somewhat reverse engineer that 24 pin
connector's attached circuit(s). Then again you might be more likely to
find specs for Sun somewhere other than this (general-purpose, mostly
"PC" oriented) newsgroup.

Yeah, but I looked on the sun groups, and they seem to be all business. I
am not sure they would even reply to some hobbyist who wants to hack a Sun
to death. I'd probably be run out on a rail if they knew I needed info to
gut an ultrasparc system and put lowly x86 stuff in it

The documents I have been able to find on the net (Sun's site is the only
place I've really found anything) do not contain any specific, applicable
information.


It would seem extremely unlikely that there's any "data" on those leads,
beyond a simple hi/low signal. In that case it would be a matter of
just determining that pin and using a relay to energize when unit turns
on for that brief period. That is, if a mechanical relay is fast
enough, else going with one solid-state.

Pin Off On
7 .8 5
10 0 11
11 5 5 - grounding this pin fires it up for an instant
14 0 11.3
18 0 11.8
19 .15 -12
21 .15 -12

I tried connecting the pins that change to their respective "On" voltages
without any effect. Pins 19 and 21 are actually outputs, but maybe need a
load. Maybe some need to be grounded or attached to a specific "return."
I'm basically SOL if I don't find the needed information.

Possibly the power supply circuit board has some of those functions
silkscreened onto the PCB?

Nothing but component numbers silk-screened. When I opened one of the
supplies, I thought it was interesting that there is a bundle of six wires
connecting the top and bottom halves. They each have a color that could
correspond to the standard ATX colors (or it could be complete
coincidence), but no green wire to even try as a PS_ON.

Numbering the pins on the socket like so on an E450 mainboard,

Socket J4503
Pin
1 2
3 4
....
23 24

grounding the corresponding pins on the disconnected plug works for me
like so:

Grounding pin 11 momentarily will start the power supplies.
Holding pin 3 at ground will keep the power supplies running.

It's as simple as that...or is it? The fans on the drive-cage side are
not powered using this method. Hmmm...

So now the myriad of us (read: "me") wanting to use a discarded,
and rather dainty, Sun E450 case for x86 purposes may be able to use the
original power supplies and power distribution board.

I still wish I could find a friendly Sun technician who would be willing
to share a little information, but I realize that just isn't going to
happen.

With my luck, there is probably some voltage monitoring or regulating
feedback on the mainboard that, when omitted, greatly shortens the life of
the power supplies or dysregulates voltage regulation. I am hoping that
all of those functions are internal to the power supplies, but...

Time will tell. I have a sacrificial mobo/processor that I'll use first.

Now to figure out the fan issue and modify the keyswitch to get it to do
what I want...

Thanks to kony and Pen for their suggestions.

On Sun, 02 May 2004 02:23:14 -0500, fred wrote:

Numbering the pins on the socket like so on an E450 mainboard,

Socket J4503
Pin
1 2
3 4
...
23 24

grounding the corresponding pins on the disconnected plug works for me
like so:

Grounding pin 11 momentarily will start the power supplies.
Holding pin 3 at ground will keep the power supplies running.

It's as simple as that...or is it? The fans on the drive-cage side are
not powered using this method. Hmmm...

Good to hear it's working... if nothing else works just rewire those fans
to the 12V rail.


So now the myriad of us (read: "me") wanting to use a discarded,
and rather dainty, Sun E450 case for x86 purposes may be able to use the
original power supplies and power distribution board.

I still wish I could find a friendly Sun technician who would be willing
to share a little information, but I realize that just isn't going to
happen.

With my luck, there is probably some voltage monitoring or regulating
feedback on the mainboard that, when omitted, greatly shortens the life of
the power supplies or dysregulates voltage regulation. I am hoping that
all of those functions are internal to the power supplies, but...

Well you'll easily enough be able to check voltages with the meter and
mainboard health monitor once it's going... probably as typical the
regulation is based on 5V load, voltage, and that is certainly controlled
internal to the power supply. Perhaps the others are feedback for the
current-sharing? (I don't know.)


Time will tell. I have a sacrificial mobo/processor that I'll use first.

Now to figure out the fan issue and modify the keyswitch to get it to do
what I want...

Thanks to kony and Pen for their suggestions.

Now that other issue, are these even new enough power supplie to be a
viable alternative to a new replacement with standard ATX format (but
perhaps in a casing format that agrees with the chassis)?

On Sun, 02 May 2004 07:59:51 +0000, kony wrote:

Now that other issue, are these even new enough power supplie to be a
viable alternative to a new replacement with standard ATX format (but
perhaps in a casing format that agrees with the chassis)?

New enough? In what way? Physical age, or some specific, recent ATX
feature?

As far as age, I have found a significant number of parts inside the
system with manufacture dates of 2000. While I haven't found a date in or
on the power supplies, the original 450s came with 560W PSUs, while these
are the more recent 605W units with Artesyn labels, and Artesyn was
created in 98 following a merger. So they could be up to roughly 6 years
old. I have yet to hook up the old motherboard and check voltage levels
maintained under load.

If you are referring to the latest ATX spec feature set, the only problems
I see, which could be a serious ones, is that I haven't located a 3.3V
remote sense line or a PWR_OK type of signal. The supplies do have an LED
that lights after a self-test of DC voltage levels, and the front panel
has a PS error light, so I imagine one of the 48 pins from the PS does
what I need, but haven't located it yet. Certainly it would be possible to
rig the PWR_OK on the motherboard end, but that would not be the desired
way.

If my hack job doesn't work for whatever reason, it would be easy enough
to mount a new ATX supply inside one of the current PS cases. Heck, you
could nearly fit 2 in each PS case. It would require removing the power
distribution board, etc., which wouldn't be as clean of a modification,
but certainly possible if all else fails, and definitely simpler than what
I am currently attempting thanks to Sun's tight control on old technical
info.