Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[ssp3]

indibil, take a look here: http://macos9lives.com/smforum/index.php?topic=5341.msg39045#msg39045

[aBc]

Yes, for more in-depth, related info read through that entire thread beginning here:
http://macos9lives.com/smforum/index.php?topic=5341.0

[Jacques]

You may have a dead mainboard / motherboard...it's a possibility.

Have a look at this YT vid, similar symptoms and Nick found the issue was the mainboard.

https://youtu.be/vAi-L6KWwZs?si=xhVaLR-qMScBJQU2

[aBc]

You may have a dead mainboard / motherboard...it's a possibility.

Have a look at this YT vid, similar symptoms and Nick found the issue was the mainboard.

https://youtu.be/vAi-L6KWwZs?si=xhVaLR-qMScBJQU2

Interesting video.

And after watching through it a few times I’m uncertain as to whether or not he actually replaced the original mobo with a different one in the end - OR if he actually put the original mobo back in, making certain to properly seat all of the mounting screws? (Especially that one screw that was not Apple original.) And that one in particular is important screw for the CPU daughtercard circuit. No stable, solid connection there? Then no power up. So yes, definitely DO check ALL mounting screws and Molex connectors. No complete circuit power through the CPU = no boot and shut down.

Also… not a CUDA button. But instead… a PMU reset button. They are different and PMU needs only a fraction of a second to be depressed for a reset. I often confuse this terminology too, coming from older Macs with a CUDA button.

AND no new heatsink paste applied, nor any anti-static procedures followed?

[indibil]

I'm testing the power supply outside the QS, bridging green wire with black.

I have replaced several capacitors, checked several suspicious tracks, checked the 4A thermofuse... and it still doesn't work. It doesn't seem like a capacitor problem, I think there is a short circuit in some component but I can't detect it with the naked eye.

[aBc]

While I seem to be on a posting binge today…

After indibil’s comment that his PSU was like the one that Jacques shows, I noticed that Jacques’ sub board (“Fan control board”?) looked amazingly like mine. And while the bottom of Jacques’ main board seemed a bit different, his noted locations of the capacitors on his main board were very similar.

D’oh! Same API number and same Apple P/N number.



So, compared some other AcBel’s noted in this thread and came up with the following.

AcBel,  API1PC12. APPLE P/N:  614-0157     aBc         QS xxx

AcBel,  AP10PC24. APPLE P/N:  614-0146    gert79     QS 933

AcBel,  API1PC12.  APPLE P/N:  614-0157    ovalking  QS 800

AcBel,  API0PC24.  APPLE P/N:  xxx-xxxx     ovalking   QS 933

AcBel,  API1PC12.  APPLE P/N:  614-0157    Jacques   QS 867

AcBel,  API1xxxxx. APPLE P/N:  614-0157     indibil      QS 800

Would like internal images of gert79 and ovalking’s QS 933’s.
[AP10PC24 or API0PC24 PSUs.]

[Jacques]

For your reference table, mine is a QS 867Mhz.

In my PSU I noticed the resistor with the sleeve next to the one capacitor had actually caused that capacitor to discolour from heat as it had been touching. This is also the one that had bulged the most, surely this is not great design?

[aBc]

Thanks Jacques.
Have updated the reference table and changed your API number too.

And if referring to the same resistor, may note greater signs of heat on mine.
(Red dots on the resistors in the last image above.) R44

Likely result of sporadic power surges due to poor solder joint on mine, over time.
(As opposed instead to close proximity to failed (or failing) nearby capacitor.)
But that’s just my theory. And no… not a great design, all considered.

Having difficulty remembering and locating that exact PSU I recapped, now
over four years ago to determine just what all I did replace to correct it.

Do please keep us posted of your progress. Thanks.

[robespierre]

The large resistor is a 3 W power rated load resistor for the 12 V output. They are part of the design because when there are no hard disks installed, there might not be any consumer of 12 VDC power. The supply needs a minimum load (minimum power consumption) in order to keep the output regulated. So that resistor burns power all the time just to supply the minimum load.
They are typically 50 Ω in value (since P = V^2/R, or (12*12)/50 ≈ 3) and are scorching hot. If you always have 12 VDC consumers attached you can just cut the resistor out. If the PSU is run with no 12 VDC consumer, regulation may fail on the 12 V rail which could damage the fan operating at that voltage.

[aBc]

Thanks @robespierre for the detailed explanation of the R44’s intended role and its scorching hot operational nature.

With the latter in mind perhaps as Jacques intimated, its location and close proximity to that sub-board and especially to C29, with less than adequate heat shielding (if any at all) it could possibly have contributed greatly to the baking, swelling and possible failure of that single capacitor over time. And with that in mind… somewhat a poor design referring to the PSU’s internal layout and lack of heat shielding?




The black rubber / composite(?) shroud noted here above on Jacques’ R44 was nearly 75% degraded (melted away) on mine and perhaps more interesting… is that the PSU continued to function for six months before total failure, after I had supplied supplemental power directly to the PSU fan from an independent outside power source. The fan had stopped working beforehand. Perhaps the damage had already been done.

Also, given the scorching hot nature… also perhaps a contributor to why my R44 seemed to have a loose, poor solder joint? Weakened perhaps over countless working hours and at those scorching temps?

If you always have 12 VDC consumers attached you can just cut the resistor out.

You’re not suggesting that by not having the 12 VDC demand of a conventional hard drive alone present, that this could lead to the eventual failure of R44 and the PSU? Especially when considering all other 12V demands on the PSU (like mobo, other internal 12V components and the case fan, etc.)? Or, just a warning about powering on a PSU outside of the machine, unloaded?

If the former - then perhaps that does not bode well for SSDs, mSATA and/or other likewise “power-adapted” storage media actually requiring less than 12 VDC?

[ssp3]

If you always have 12 VDC consumers attached you can just cut the resistor out.

Don't mess around with original design, unless you know 100% why any particular component is there!!!

First Google hit gives you the correct answer:  https://tinyurl.com/2czp2yff

Typical minimum loads for switching supplies are 10% of the maximum load current and usually one ampere, or less. Some supplies have internal resistors to provide the minimum loading; however, they will only work when the loading is slightly above this internal load.

EDIT. Almost every component inside computer is a variable load. Their current consumption at any given time is unpredictable. Constant loads, such as resistors, are put in there for a reason.

@aBc - iment what I ment.

[robespierre]

The colored stripes indicate a 69 Ω resistor with 5% tolerance. (12*12)/69 ≈ 2.1 W.
Hard drives may not present a load on 12V all of the time; when they sleep by spinning down, they stop drawing power. But case fans operate consistently whenever power is on. It doesn't take too many case fans to add up to 2.1 W.
The stability of regulation can be tested using an instrument called a DC Electronic Load, which would be required to "check out" any modification.
It's also worth pointing out that some computers use the 12V rail to run the CPU, through a "point of load" (POL) buck regulator next to it.

[gert79]

Hello everybody

Sorry I couldn't follow the forum for a while due to real life priorities.

I still have all the photos of my QS PSU disassembly but I am a fan of decent resolution images and don't want to go thru the annoying work of resizing them for upload and furthermore have a loss of detail.

Sequence Of Events photos are the only way for me to be able to reassemble the PSU, so I will not post any single image. It is not my way of working.

Factually I was even a bit annoyed that my photos were scales down for "conformity". Imho we live in the age were it does not cost you much to shoot proper photos and also storage is available. You can always delete redundant photos later but you can never generate a missing photo when the work is done. And most of the time disassembly photos are assembly photos if you swap the order ...

I come from control cabinet wiring. Usually I shot thousands of photos each year in the highest resolution that was available on mobile device at the given time.

Currently 4000x3000 is good enough for me, and my intent was to have a photo that clearly shows the Apple Partnumber while being so high res that you could reproduce the PCB from that photo or even reverse engineer the schematics if you have good nerves. I like to think about the future, solve issues before s*it hits the fan.

Furthermore I am a bit stuck since I have to find the caps somewhere in Europe and I don't know with which substance I can replace the white goo, since it is needed to protect against mechanical stress, to prevent unwanted movement of caps closer to heat sources, but also to dampen any kind of coil noise. I had to break the hardened goo in some places and a full clean and re-goo later on has to be done once I have the caps sourced and mounted.

Going a bit OT as usual, I can share my photos via Google photo gallery but I absolutely don't take any responsibility if someone wants to do it like I did.

I don't want to come across as rude or arrogant but too many times in my life people thought I am an eejit and I am tired of it, being treated like that. So everything I explained above is just to add some more depth to my avatar gert79, which is actually a male name in Europe.

@aBc - please send personal message here in the forum and I will prepare the photos for you. Please handle with care like you usually do, I guess you just want to compare types of layouts anyways. If you have any special ideas which components I could replace, as the PSU is still disassembled, I would be very grateful

[aBc]

PM sent.  



*Example images. Clickable for larger views.

[gert79]

Hello everybody!

I am in a bit of a crisis since I obtained a mid 2012 MBP with FireWire 800 port, to hook up my MOTU Ultra lite Mk3 dual using a multilingual FW 400/800 cable... I only tested the midi clock and it outperformed the Opcode Studio 4 by far (Deviation was 0.04% in logic pro, measured with the midisizer.com midiGal utilising Midiclock V1.01 measuring firmware)...

But nonetheless, I will try to finish what I have begun, the workflow and functionality in SVP is still a thing for me, although I plan to also send control voltage on the MOTU...

That was the off topic part, I hope it will be accepted.

I decided to take my work back on the ground ;-) so I will attach here is a hand painted capacitor parameter page.

You can hardly beat pen and paper :-)

I still don't know what type I should replace, low ESR? Higher voltage rating? Unipolar?

I will approach the refurb slowly, also I will downgrade my GPU, maybe also CPU speed, install more SSDs and add proper noctua fans to PSU and CPU, also I have docs from a German manufacturer that describe proper aluminium cooling and all its parameters.

My GF hates the G4 because of the noise and the smell, so replacing the fans is inevitable.

Best option for me will be to run SVP and logic pro together, best of both worlds.

✌️ 🕊️ from the heart of Europe :-)

[gert79]

So well not much happened here?

By now I found a decent fan replacement and caps, but of course the caps have long lifetime and low ESR - I am unsure if they are good for the QS supply since original caps are rated quite low voltage, I need to research more if it is smart to put slightly larger voltage tolerance/Low ESR.

I did not go for a Noctua fan as it is very expensive and does not really move a lot of air, I will probably use a MagLev fan which can run low speed at very low voltage.

I sourced from a Polish retailer but the types of parts I will go for should be available from mouser and the usual sources aswell.

I spent some time to select these so I cannot guarantee that this will work but I hope so, especially the caps.

New caps will be from Panasonic, I will select fitting types from the EEUFR0J222B series. They are green/gold or green/silver, not super expensive.

The fan will be a Sunon MF80251V3-1000U-G99 and is not super expensive but ticks all the right boxes for me.

I looked at more expensive EBM Papst and other fans but they don't come with magnetic levitation bearing and go only at high speed. As said the Sunon runs on voltages as low as 5VDC and you can really balance things well to have less noise but still proper airflow.

Sure the Noctua fans are an elite design but this is something for gamer PCs imho. Sunon is a well known brand and their MagLev fans seem to be something that can last for a long time without any kind of bearing that will wear out.

I spent about two hours of comparing parameters and datasheets for the fan so my choice can certainly not be changed much anymore... If this helps someone I am ok with it too, but please don't complain if this fan doesn't work for you, I only know for me it might be the best one.

The caps are a preselection, I will post detailed partnumbers later and also report if they worked in the end, once I refurbish the whole thing.

As of now I have not ordered anything yet, I also need some tactiles for my MPC1000 and what not, so I can order all these things in one big order.

There is still a pending thing which would be the probable release of my disassembly photos. To be honest I would prefer to see them wrapped up in a pdf file. Posting them on this forum will make them unreadable imho, a tech would want a pdf file that he can work thru page by page, sequence of events like.

Of course the pdf will include proper warnings and lightning strike icons... But: Ultimately I still don't want to be associated with it. I cannot bear the responsibility of *someone* following these photos and eventually hurting or killing himself. So the best would be to keep this in an "inner" circle of the community, where the techs have access to it.

As soon as this would be out in the wild everyone and his grandmother will approach QS PSU repair... There is no info anywhere how to do it and in Europe "Golytronic" does not repair them anymore... He had Schematics for all the PSUs and knew everything about it, also how to measure and load test them.

As far as I know his business was shut down due to death... May he rest in piece.

I don't know how capable the repair Techs in USA are, but shipping a PSU overseas is not a cheap option and anyways a PSU has to be supplied, even the broken ones are not cheap anymore... I searched several weeks to buy my spare for a high price from UK...

No I don't want to fix European PSUs for a living, I just want my supply and the spare to work

[ssp3]

Caps.
Low ESR is always good.
Higher voltage is also good. FYI -higher voltage caps are more expensive to manufacture (thicker films).

Did you know that Silicon Graphics used very expensive ulra-low ESR caps by Sanyo called "OsCon" in their machines?

Fans.
Sunon is good industrial stuff. I've seen Sunon fans in Apple's IIci, IIcx, Q700 and PM7100 PSUs.
Papst is good industrial stuff too, although, I have a feeling that quality went a bit down when they moved production out of Germany. I've used 8412 L(?) model in the past for my non-Apple projects.
Noctua looks like something from a "boutique department" to me. Lots of TMs, lots of smoke and mirror. Designed in Austria, but made in ..?

[robespierre]

Also Panaflo, Comair Rotron, and Delta for fans.

Oscon are solid polymer aluminum capacitors, now more or less standard across the industry for computer applications. They are limited to low voltages, however. One advantage is not having any liquid to potentially leak out.

Sanyo also developed the Poscap, a solid polymer tantalum capacitor. Unlike standard tantalums with a manganese dioxide cathode, the Poscap won't combust if damaged.

Both types are now available from multiple manufacturers.

[ssp3]

Lets not confuse gert79 regarding capacitors that he might need, ok. Those Panasonics are just fine.

* I mentioned OsCon only as a curious fact in the context of low ESR.

[gert79]

Thanks for all the inputs. These other caps are not available in the needed values and also they are not exactly cheap.

Using these Panasonic types is probably already five times better than the original caps, as long as the machine outlives me the mission was a success.

I understand there is always new bleeding edge stuff but this is a more than 20 year old computer and somehow it doesn't feel right to upgrade it to the latest components. It seems that some caps have a plain unisolated aluminium housing which is maybe not good with the coils, I prefer more traditional caps with plastic sleeve cover.

I also heard from some people that they avoid Tantalums because of the horrible conditions that people in 3rd world countries have to suffer to mine this substance...

So going forward only a bit, the concrete partnumbers are:
EEUFR1C222B C10 2200uF 16V
EEUFR0J222B C13,C16,C19 2200uF 6.3V
EEUFR1A332B C21 3300uF 10V

Off-topic: I will do some tests with a low performance GPU as mentioned and also check which fans or which airflow is actually needed.

Just replacing all fans will maybe not solve the noise issue. I am thinking of disabling the case fan and have a fan drawing air over the CPU aluminium brick and blow it out on the holes on the rear, SSDs don't need so much cooling.

I somehow have the feeling that the cooling on the G4 QS is a bit botched, or they ran out of time. But yea PCI cards can get hot and there were no SSDs back then.

But for today's modern/modded machines there are sure better approaches. I saw some posts by people solving it differently, I will also try that, just for myself, how I imagine ideal cooling.

The basic principle is always that hot air goes upwards by itself, so a cold air inlet should be on the bottom as much as possible. But then for fan directions and locations it gets interesting. The G4 enclosure is quite airtight by design so grill holes can be used or modified/closed, new ones can be drilled.

The PSU will always need a fan and since it is on top the fan can only blow hot air to outside, but at the same time there could be inlet holes for cold air on the supply case itself, ideally at the position of the aluminium cooler assemblies. Cold (outside) air supply could be coming from the case fan or the grill holes under the disabled case fan. I will spend some time with that later, trying not to experiment too much and kill the supply again...