Latest Casualty ...

[GorfTheChosen]

Went into the studio yesterday and the G4 800 Dual was off, when I had left it running the night before. Push the power button on the front panel and it lights but the unit does not power on. No spinning drives, no fans, no chime ... nada.

I recently went through this machine and gave it a really thorough cleanout, which included pulling the PSU and the case and processor fans, vacuuming/blowing them off and then cleaning the latter two items with rubbing alcohol. In the process it appears as though I swapped the case fan around, with it blowing air in, rather than exhausting air out ...    Vacuumed and blew out the rest if the case, etc. Replaced the battery.

After doing the above the machine ran fine for better than week. In that time frame, I did a couple of upgrades: added 4th internal (PATA) hard drive, added an Adaptec 4-Port USB 2.0 Model USB2connect 4000 PCI card.  That card does have an issue with sleep IIRC ... and it's possible that I may have forgotten that and put the machine to sleep before I left the studio.

At this point the machine will not power on.

What I have tried so far (in roughly the order tried):

Disconnected from power (30+ minutes), removed battery, pressed front panel switch, and later re-installed battery (which is reading 3.6 - 3.7 volts) Reset the CUDA switch. No joy.

Pulled and then reseated the two PSU connectors on the logic board. No. joy.

Pulled all ram and reseated the DIMMs in their original slots. No joy.

Pulled all ram and reseated the DIMMs in different slots. No joy.

Pulled all ram and tried one DIMM only, each one at a time. No joy.

Removed and reseated the cable from the mobo to the front panel power board on both ends. No joy.

Connected the old Microconnectors USB keyboard (with the Power button) from the Mac Mini and tried to power it up that way, bypassing the front panel board. (Normally this would work) No joy.

Disconnected power and signal lines on all four hard drives (two on the mobo ATA, two on an ACARD ATA 133 RAID card) and tried to see if it would boot with only the DVD-R hooked up. No joy.

Pulled the two PCI cards (ACARD RAID and Adaptec USB) and tried to boot it. No joy.

Pulled the AGP card (GeForce2 MX) and replaced it with the one from the G4 Dual 450 (think that one is an Apple OEM 32MB Radeon 7500) No joy.

The only thing I get when I press either power button (keyboard or front panel) I get a brief flash from the red LED on the logic board ... but no joy.

Last night I went ahead and pulled the heatsink and the dual processor daughtercard, cleaned those up and redid the thermal paste, and then reinstalled ... thinking that at this point what could it possibly hurt ?

I have not tried to power it up since then.

The only additional thing I have not done at this point is check the PSU to see if the appropriate voltages are present ... so I guess that's next.

Any insights or thought appreciated.

[FdB]

I have not tried to power it up since then.

The only additional thing I have not done at this point is check the PSU to see if the appropriate voltages are present ... so I guess that's next.

Very damned thorough approach. Kudos Gorf.

Try to power up after re-paste. The PSU is highly suspect.

Before PSU removal, check to see if PSU fans attempt to even “bump” on momentarily.
Most likely, PSU has expired. But do check voltages. (Is it an MDD or a Quicksilver?)

[IIO]

exactly, if something else than the PSU is wrong, the machine should at least attempt to power on for some 0.5 seconds or so (makes a short noise for the fans to switch on)

[ssp3]

This sounds like a trickle power problem to me. Most probably dried out capacitor(s) in that circuitry. Exchange them for the fresh ones and you'll be good. I had similar problem with my G4 Digital Audio a couple of years ago.

[GorfTheChosen]

Very damned thorough approach. Kudos Gorf.

Why, thank you ... (bows ... humbly ...  )

Try to power up after re-paste. The PSU is highly suspect.

Good.

Wasn't sure whether that red led flashing on the mobo was indicating that the PSU might be good or not.

I wouldn't be at all surprised if it were the PSU ... this machine has seen a lot of "power on" hours, a good deal of which where it was working very hard. Plus all the dust and crud ... which I should have kept a better handle on.

Before PSU removal, check to see if PSU fans attempt to even “bump” on momentarily.

Will do.

Will probably let it set awhile longer before I put the power to it though ... have seen some stories in my travels of "I let it set with the power disconnected for xx hours and when I tried to start it, it fired right up !"

Most likely, PSU has expired. But do check voltages.

If the PSU is all it is, that would be very welcome news ... particularly since that nice mail lady just delivered a new 500w ATX power supply about 15 minutes ago ... along with the diodes and an external 24v power supply to do an ATX--->G4 PSU conversion.

If it's not, then perhaps the new PSU will go in the G4 450 Dual ... which, sadly, is in a similar condition ... 

One thing I have noticed - now that I'm paying attention - is that when plugging in the AC power into the PSU there is a definite arc between the plug PSU power receptacle and the power cord ... can't recall whether that is normal or not.

(Is it an MDD or a Quicksilver?)

Just an original Quicksilver (2001 ?) ... (replete with the Rev. A logic board which doesn't support drives over 127 GB ?)

Thanks.

[GorfTheChosen]

exactly, if something else than the PSU is wrong, the machine should at least attempt to power on for some 0.5 seconds or so (makes a short noise for the fans to switch on)

Yeah ... I don't think that is happening ... but I'll verify on the next attempt to power up.

Thanks.

[GorfTheChosen]

This sounds like a trickle power problem to me. Most probably dried out capacitor(s) in that circuitry. Exchange them for the fresh ones and you'll be good. I had similar problem with my G4 Digital Audio a couple of years ago.

Caps in the PSU ?

[ssp3]

Yes, in the PSU. Look for the string of several resistors in series that runs from main large capacitors (high voltage) to a small-ish cap after that string (or something similar, PSUs do differ in construction). That cap is the most likely culprit.
No need to make a franken-mac with non-original PSU

[FdB]

Here's two different PSU's from two different Quicksilvers.
(They do vary considerably & can be more difficult than MDDs.)
But yes, do a visual check for swollen or leaking caps before
attempting replacement. Maybe you'll find one easy to replace?

[GorfTheChosen]

Well here's what I've discovered thus far:

Pin 22 - which is the white wire - is supposed to be +25v (Standby) ... and has around +25v present with the machine plugged into AC power.

Pin 15 - which is the green wire - is supposed to be +5v (Power On) ... and has +4.89v present with the machine plugged into AC power.

Pins 4, 6, 20, and 21 - which are all the red wires - are supposed to be +5v ... but thus far none of those appear to have any voltage present. (I'll be rechecking these when I resume)

Of course, the probes on my meter are really too large to really fit down in the plug to get to the metal terminals ... so it's hit and miss. So I tried to cobble something together ... first using small paper clips (too small), and then using small finishing nails (almost too large) Maybe large paper clips will be the sweet spot of "just right".

Part of the difficulty of doing this - and having some real certainty of the result - is a lack documentation of a/the procedure. What little I've been able to figure out has come from around 3 or 4 different sources: jcsenterprises.com MDD PSU testing writeup, ATXG4.com's "How to adapt an ATX PSU to replace your G4's PSU" write up, and outofspec.com's similar writeup on adapting an ATX PSU to power a FrankenMac he built from a Sawtooth I think.

Two of the three are focused on the "how-to" of an adaptation to replace a PSU, and the third - which does focus on testing and diagnosis - is for a different model Power Mac model than I have, with a different PSU.

Not really knowing exactly which wire should be "live" when powered off kinda hampers knowing exactly what I'm seeing.

On another note, the replacement PSU looks and feels pretty impressive for a cheap ATX power supply ... it's hefty, has a snotload of wires ... including the proper power connectors for SATA drives ... and the metal case actually appears to be ... chrome plated ... 

[GorfTheChosen]

Yes, in the PSU. Look for the string of several resistors in series that runs from main large capacitors (high voltage) to a small-ish cap after that string (or something similar, PSUs do differ in construction). That cap is the most likely culprit.

Good, thanks.

No need to make a franken-mac with non-original PSU

Well ... I have no experience whatsoever doing component-level board repairs ... but I haven't let things like lack of experience stop me before ... and I did watch 25+ minute 'tube on "Soldering Tools and The Basics for Beginners" earlier today ... 

But point taken.

Here's two different PSU's from two different Quicksilvers.
(They do vary considerably & can be more difficult than MDDs.)

Yeah ... pretty sure both the ones I have (for the Dual 450 and the Dual 800) are similar to the model on the left ... although they are from different manufacturers (one is an AcBel and the other is a Delta)

But yes, do a visual check for swollen or leaking caps before
attempting replacement. Maybe you'll find one easy to replace?

Will do ... thanks.

[ssp3]

Here's some starting point for you:
https://audioxpress.com/article/repairing-switching-mode-power-supplies

Apple PSUs are a bit more complicated than those mentioned in the article, but the operating principles are the same.

[GorfTheChosen]

The larger paperclip seems to be a better fit as a test post/probe.

Updated results:

Still nothing showing up on all the red wires (Pins 4, 6, 20, and 21) and are supposed to be +5v.

All orange wires (Pins 1, 2, 8, and 12) which are supposed to be +3.3 read 0.00v

(Actually Pin 1 has two wires crimped into the terminal in the plug - one orange and one brown)

The purple wire (Pin 9) has +25v on it.

The yellow wire (Pin 10) reads 0.00v.

The blue wire (Pin 12) reads 0.00v

All remaining wires are black (Pins 3, 5, 7, 11, 14, 16, 17, 18, and 19) and should be GND.

That pretty much completes all the readings for the P1 plug.

The only wires with any juice are Pin 9 (purple, +25v), Pin 15 (green, +4.9v), and Pin 22 (white, +25v)

Oddly enough, my Quicksilver wiring harness from the PSU looks (notice how these images are labelled) like this:

http://atxg4.com/images/Gigabit.jpg

Not this:

http://atxg4.com/images/Quicksilver.jpg

... except that I have the extra P2 lead with the two yellow 12v lines and two GNDs show in the second image.

The pinout image for Quicksilvers @ http://www.outofspec.com/frankenmac/wire.shtml jives better with what I have.

On my P2 plug:

The two 12v lines on P2 lead read 0.00v when checked to the two black ground wires adjacent to them.

Tried powering the unit up on the bench with all the drives disconnected, still no dice ... just a quick flash of the red LED on the mobo and the fan inside the PSU does absolutely nada ... no hint of even trying to rotate.

[GorfTheChosen]

Here's some starting point for you:
https://audioxpress.com/article/repairing-switching-mode-power-supplies

Apple PSUs are a bit more complicated than those mentioned in the article, but the operating principles are the same.

Thanks - appreciate you linking to that.

Just glancing over it briefly, it looks to be a good primer on the basic design theory behind how PSU's work.

It will be some work for me to get through it and truly understand it, as I'm no EE ... but I'm sure it will be helpful.

[FdB]

To possibly further confuse things… the Gigabit Pinout example looks ok when compared to the (right) PSU pictured above / with a similar, single connector.



The Quicksilver Pinout is a bit confusing and the example shown is from an Acbel PSU (the left one, pictured previously above). If you rotate the Gigabit Pinout image 180˚ to the left, does it then not match the Quicksilver Pinout as the connectors actually are? (Same goes for the QS Pinout image graphic / top P1 connector.) AND what you refer to as the P2 lead (with the yellow and black wires) is labeled P4? The QS Pinout image graphic is perhaps just a little whacko?



Gorf, have you tested trickle pins with the “power on” (green wire pin 15) jumpered? Trickle voltage should be present without this (but I always enjoy the possibility of seeing sparks fly). 
(Disconnect all power connectors from mobo completely.) Might be a total shame if your type A board was somehow at fault and you’d then have to upgrade to a B board?

AND… no evidence yet of swollen or leaking capacitors inside? And - no evidence of burnt traces, etc.? The “sniff test”.

Andy Cuffe ([email protected]) down in Austin Texas for PSU refurb service. Andy even offered members here a discount (previously) so you might check with him. Old eBay link for his services: https://www.ebay.com/itm/REPAIR-SERVICE-Apple-PowerMac-G4-Mirror-Drive-Doors-Power-Supply-MDD/141030901102?hash=item20d618c56e:g:~RsAAMXQysxSAA61

Have never attempted the ATX route myself and there are those here that view that approach as sheer and utter sacrilege… so there is that risk of being ostracized. 
It is however difficult at best to resist the wiles of a nice mail lady und of course,  zee chrome plating.

“Klaatu barada nikto”?

And… a very good reference link there, ssp3. Thanks.

[GorfTheChosen]

To possibly further confuse things… the Gigabit Pinout example looks ok when compared to the (right) PSU pictured above / with a similar, single connector.

Right.

The Quicksilver Pinout is a bit confusing and the example shown is from an Acbel PSU (the left one, pictured previously above).

A bit.

If you rotate the Gigabit Pinout image 180˚ to the left, does it then not match the Quicksilver Pinout as the connectors actually are?

Yes, it does.

(Same goes for the QS Pinout image graphic / top P1 connector.) AND what you refer to as the P2 lead (with the yellow and black wires) is labeled P4? The QS Pinout image graphic is perhaps just a little whacko?

Just a little whacko.

There's one specific error in the Quicksilver image that I can see: Pin 9 (+25v) is shown as being yellow in the Quicksilver pinout picture, when in reality (apparently on both your end and mine) it is actually purple.

Error on the part of the person who created the image … or an accurate reflection of the variation of the PSU’s that are out there, in terms what has has actually been produced and deployed in the field ?

I really don’t know … but I do know that when I run into a contradiction like that I find it confusing at best and it makes my head hurt.

Gorf, have you tested trickle pins with the “power on” (green wire pin 15) jumpered?

No.

I had done some digging around on the 'net to try and find a procedure for metering a power supply after "tricking it" into thinking it was on by jumpering it. That ended up being sort of a mixed bag, although I saw at least one article where the author said that a PSU should not be “turned on” without something attached to it.

How would I do that … what pin would I jumper Pin 15 to ?

And what are these “trickle pins” of which you speak ?

Trickle voltage should be present without this (but I always enjoy the possibility of seeing sparks fly). 

(Disconnect all power connectors from mobo completely.) Might be a total shame if your type A board was somehow at fault and you’d then have to upgrade to a B board?

Assuming I could find a working one at a reasonable price, I would not be heartbroken.

What “trickle voltage” value am I looking for ?

And where should it be present ?

AND… no evidence yet of swollen or leaking capacitors inside?

I haven't taken the PSU back out of the machine yet and opened it up again. I’ll do that here in a little bit.

When I had it out previously, I did take the cover off in order to clean it out ... and nothing really jumped out at me at that time as far swollen or or leaking caps ... but then it was working at that point, and I was wasn't really trying to diagnose an immediate problem.

And - no evidence of burnt traces, etc.? The “sniff test”.

Yeah ... no, not that I've noticed anyways. Pretty sure I would.

When I had the processor daughtercard off to redo the thermal paste I used a paint brush to clean the dust off ... I got a pretty good look at it then and nothing really stood out. And I was kind of looking for it too, as I had read of one tale where a fellow had claimed his father's (or mother's brother's uncle’s ... or something) daughtercard had a hole burned right through it when it failed.

Not really seeing anything obvious on the mobo either.

My assumption has always largely been that it IS the PSU that is at fault.

Have seen some saying that these machines should not be left connected to AC power if they are going to be shutoff for long periods of time, due to heat generated in the PSU and its fan not being on.
 

Andy Cuffe ([email protected]) down in Austin Texas for PSU refurb service. Andy even offered members here a discount (previously) so you might check with him. Old eBay link for his services: https://www.ebay.com/itm/REPAIR-SERVICE-Apple-PowerMac-G4-Mirror-Drive-Doors-Power-Supply-MDD/141030901102?hash=item20d618c56e:g:~RsAAMXQysxSAA61

Yeah thanks ... saw the thread (started by GaryN here ?) last night after I replied to you and ssp3.

Andy’s repair/refurb service appears to be pretty impressive - very fast turnaround and very reasonable pricing.

Have never attempted the ATX route myself and there are those here that view that approach as sheer and utter sacrilege… so there is that risk of being ostracized. 

I can only imagine ... 

It is however difficult at best to resist the wiles of a nice mail lady und of course,  zee chrome plating.

Of course.

“Klaatu barada nikto”?

For now anyways … if I have to spend another 12 hours trying to get frickin’ W2K to reinstall on the PC that could change however.

 

[GorfTheChosen]

Some images.

G4 800 PSU:

[GorfTheChosen]

G4 450 PSU:

[GorfTheChosen]

Except for one small cap (behind the orange wires) in the first image, I don't see any signs of anything leaking or being swollen.

The top of that one small cap does seem to be slightly pushed out/up.

No obvious signs of anything being fried ... to me anyways.

[IIO]

it is actually not a bad idea to learn how to fix the PSU yourself when it dies, but otoh a good soldering kit is like 50 dollars, it is more than one hour work, you can f*ck it up if you dont know what you are doing, you have to find the right parts before...