[weblacky]

However, please note I’m not calling what you said wrong. It’s just not a correct assumption to make with many switching power supplies.

Someone could make a design where they want a doubler all the time (more components), someone could be constrained in a way where they do odd things. I’m still not sure how to tell what’s going on in random, denser, PSUs because there are a large number of variations in each step of the PSU workflow.

I’ve seen large poly caps between the FET and the drain to the main chopper. I’ve seen weird doubling of rectifiers, put back to back, instead of a larger/better rectifier in PC power supplies!

My point is that you need to remain flexible to realize that realistically for any random PSU there will always be a measure of mystery for much of the details.

Repair is different from design and doesn’t really take the same level of understanding as the designer.

You have the knowledge of basic theory and the basic steps. How they are accomplished can happen several different ways, each.

I’ve learned over my studying to not take any one guide as gospel, because I’ll find another design that looks nothing like what you’d expect.

For example Apple AC adapters now use a type of very high frequency zero-cross switch technology that allows an incredibly small main transformer. Looks nothing like the intro guides...but you still have a high voltage and low voltage sides. It’s just hard to figure out the exact separation due to many of the components being placed in one package instead of several discrete ones.

I have high hopes for SGIs because there are not complex when compared to the latest tech. They are dense, but not hiding anything...they were made by third parties and don’t have much anti-repair technics in them.

You should keep reading and you did get good info. It’s just that it doesn’t apply in a lot cases, so not a great generalization. But yes, it is something you’ll see given enough repair work.

Oddly by the way there are only four things I know about primary filter caps.

1. Fast and slight voltage fluctuations across the cap often indicates a secondary short.

2. The cap voltage is expected to be at least 130v or higher in all situations. Lower isn’t an option.

3. All dc measurements on the high voltage side are to be done with your meter’s negative lead on the negative post of the filter cap. All secondary side dc measurements use PCB ground as negative reference.

4. The primary filter caps responsible for smoothing rectified DC to a kind of wiggle instead of humps. They also help prevent noise in both directions (from PSU to outlet, from outlet to PSU). Degradation of these means noise travels in both directions and stress every component. Which degraded them faster as they are experiencing more ripple than designed and cannot keep up (being beaten to death). Normally your secondary caps will degrade first, but failing health in primary caps will tip nearly everything over the edge as it’s supposed to act as the largest and earliest filter in the entire setup.

Now you know about what I do:-).

[Irinikus]

Thanks very much for the valuable info! I appreciate it!

It's good to be corrected from time to time, as thats how you learn!

[Dylanear]

Interesting thread. I'll leave the PSU issues without comment and just say, for me, I love the old slow machines as much as the faster, newer ones. ALL SGIs are blown away by more modern and a lot more boring computers. To me they are classics, a reminder of times past and of the changes in the history computers over the years. That they are different and yes, slower, is half the point. I still have my R3000 Indigo expressly because it's slow and of a similar speed to the early 4D machines I first learned Irix on. Sure, if you are doing a lot of rendering or gaming on IRIX, the newer higher spec machines have huge advantages, but depends on why you happen to be spending time and money on these ancient artifacts.

[Raion]

Old systems are fun, I enjoy my Onyx2 more than the Tez tbh.

[jwhat]

Hi Irinikus,

I am with you on performance front when it comes to "useful".

Useful for me is: has a software application that is "useful" for some purpose (games == entertainment, Maya == valuable for 3-d modelling, flame == valuable for video processing) and it it able to run the application with sufficient performance that you can use it for its targeted purpose.

So applications is the key for me.

For Irix I put the: Irix Synthesizer, media utilities including DATMAN tape player, Maya, Flame and few other things in this category.

By way of example I put SoftWindows in the not useful bucket as I can do this with and a whole lot of other emulation/virtualisation stuff with QEMU / KVM much much more easily and with better performance.

As the IRIX stuff I am using requires a well performing machine I am with you: start with Octane and go from there, anything else is to slow ;-) .

When I want to have a slow nostalgia play I can use "Dark Castle" and "Lemmings" on my Mac Colour Classic ;-)

But I guess old retro computers are a bit like old vintage cars, in that you spend more time fixing and keeping them running than you do actually driving them.

Cheers from Oz,


John.