Beefier Power Supply for SE/30

[tt]

JAMECO might not be the best place to look, I found more options at Digikey, but either too expensive or too big. I don't mind harvesting a PSU board from a CPU PSU, I am just unsure if it makes sense given my earlier questions; at least cost-wise it is attractive. Even the bare PSU's from JAMECO will have forced air cooled/convective cooled ratings. Here is a PSU I thought looked interesting, but it only has one 12V rail even though the nameplate has an A & B listed for 12V:

SeaSonic SS-300TFX

http://www.hardwaresecrets.com/article/Seasonic-SS-300TFX-Power-Supply-Review/1021/2

and on power distribution (rails): http://www.hardwaresecrets.com/article/Seasonic-SS-300TFX-Power-Supply-Review/1021/6

 

[JDW]

While it did merely serve to drive my internal hard drive, I nevertheless am speaking from experience on JAMECO parts:

http://www.flickr.com/search/?q=se%2F30+psu&w=66071596%40N00&m=tags

That's not to say nothing better is out there, nor is it to say that you would "never, ever need a fan." But consider well that the stock SONY PSU has no fan in it. And it needs no fan (technically) as the SE/30 itself has one. Furthermore, if you look at the photos of my dual-PSU rig, you will see I added a secondary fan to move more air around inside the SE/30 chassis.

But again, none of the JAMECO-sold PSU's have a turnkey solution. Even if you go with the one I suggested previously which has a separate +24v output, you'd still need to DC-DC that down to +12v. That requires extra time, extra expensive and extra bother (since you would need to not only solder it in place but mount it as well). Even so, most of us are "hobbyists" in many ways (as you can tell I am from my SE/30 rig photos). So the "bother" it entails may not be significant for some.

 

[techknight]

And why is the split 12v rails a requirement? Why does the sweep supply have to be separate? They dont have to be separate if you put separate EMI Pi-filters on the output for each circuit path direction, and use diode clamps for voltage transients, so noise dont get back into the mobo from sweep, and vice versa.

Steal ideas from the automotive world, There is only 1 supply, So look at the different methods used to eliminate noise in that regards.

 

[JDW]

TechKnight, that is my question too. But after all these years, no one has answered me on it!

By logical deduction one can only "assume" that two isolated +12v lines are needed. Otherwise, why didn't Apple/SONY use a single +12v in their original design?

 

[techknight]

because back in those days, it was easier to distribute current via multiple 12V rails instead of a single one. SMPS power supplies were a relatively new thing back in those days. Components were not NEARLY asefficient then as they are now. Most SMPS based systems were current-mode based, becuase they used bipolar transistors as chopper transistors. By the time the SE came out though i think MOSFETS that had enough power hit the scene and they just started using them, but the on-resistance was not good. Still inefficient and generated heat, however though still the efficiency in the saturation region on bipolar was horrible. Nowadays high power MOSFET came along.

Today, we have MOSFETS which are voltage-controlled systems, and have saturation on-resistances in the thousandths of milliohms. orders of magnitude more efficient than we had back then.

Thats the only thing I can figure, as back then it was much easier to make multiple rails of smaller amperage, than one large single rail thats high current.

 

[JDW]

TechKnight, your explanation is logical and plausible. And if one analyzes the Sweep circuit and cannot conclude that it would in any way interfere (e.g., through feedback, noise, etc.) the Disk Drive circuit, or visa versa, then a single +12v PSU output with a 3.35A rating (or higher) should work fine. All that's left is an actual Se/30 test, to determine if our theoretical matches the experimental. All 3 of the "MEAN WELL" PSU's I mentioned before satisfy the output requirements.

 

[tt]

While it did merely serve to drive my internal hard drive, I nevertheless am speaking from experience on JAMECO parts:

JDW, I'm not knocking Jameco or their parts, I'm just saying they might not have the most extensive selection in this area. There might be a shop that has what we are looking for off-the-shelf we are not aware of.

techknight, when you mention borrow techniques from the auto industry do you mean using something like this (12V DC noise filter):



Are there off-the-shelf solutions with Pi-filters, etc?

Maybe someone would be able to determine if the CRT sweep can influence the noise level significantly of a common 12V rail (and other rails?) for a "modern" PSU... What tools would be needed, an oscilloscope?

Another page on PSU/ATX hacking (Japanese): http://take103.blog.fc2.com/blog-entry-22.html

 

[James1095]

Try it and find out. Any salvaged AT or ATX PSU should have all the necessary voltages. If you hook that up and it works well, you can be reasonably confident that a similar arrangement with a power supply that fits in the available space will work too.

It can be notoriously difficult to eliminate noise and interference though. You have to watch out for ground loops in particular.

 

[tt]

There is concern here about phantom issues that changing the PSU may bring up. Combine those potential issues with risking rare PDS cards and other possibly flaky parts of the system, I would like try to make as an informed decision as possible before hacking hardware. That being said, overhauling the PSU should probably be done after recapping the analog board.

 

[JDW]

Please define "phantom issues" more specifically.

(Sorry, but some folks reading this may confuse your comment with "phantom power of audio equipment.")

 

[wally]

Guide to the Macintosh Family Hardware, Second Edition has more extensive power supply specs for the SE/30 on pages 259-260 that relate to this discussion. In particular on page 259 there is a peak load specification for disk seek servo current on the +12 V disk (4A, with certain duration and duty cycle limits and an allowed sag to 11V) and on page 260 there is what I consider a very tight requirement on +12 V sweep: The way I read it is that the + 12 V sweep supply volts will not vary by more than +_10 mV in response to any one load step on +5 V, +12 V disk, or -12 V, while supply rails other than the one being step load tested are at max load, and line input AC volts are anywhere within the wide AC spec range 85/135 or 170/270 VAC. The individual load steps are not full size between each rail's min/max current, but each designated load current step is considerable.

I speculate that the designers did not want the raster to shrink or move to any visible degree when disk seek activity or future expansion cards demanded peak current. Load step tests should be done with special attention to raster visual stability of position. It may be that commercial supplies cannot meet the strict 0.010 V sweep rail requirement, but that the raster visual appearance is satisfactory for your purposes.

Some cautions:

1. As detailed as pages 259/260 are concerning output limits, they nontheless fall far short of a design spec and/or requirements document. In particular they do not address in what order the rails decay when AC power is removed. It may not matter, but sometimes the magic smoke gets out when most rails rapidly go to 0 volts but one rail hangs in there extra long.

2. It's always safer to do some bench load testing with switched resistor loads or an active load instrument by applying a series of increasing size step load steps while monitoring response with a scope, before connecting your treasure and finding out that at minimum load the supply goes out of regulation or oscillates and threatens to smoke stuff.

 

[techknight]

Yes, but if the pwoer supply has 10 to 20 amps or so, flickering/shaking under load isnt going to be a problem, especially if the power going to the CRT sweep circuitry is filtered through an external filter as i mentioned before. Ill give it a shot once I find a suitably sized power supply. the tiny ones I have at work only have 0.1A on the -12v which is too low. So i cant use them, I would say if the filter is good enough, any EMI spikes that occur on the digital side isnt going to get into the sweep circuitry, and vice-versa.

And since I am partnered with a Car audio shop, getting filters for testing isnt going to be an issue.

Yes, those are the types of filters I mean from the automotive world. They are pi-filters. at least should be.

The biggest problem for me right now is time. I am strapped of it. I have been doing a major R&D project for a display corporation on new scoreboard technology, and thats been taking up nearly all of my time, even on weekends to get the hardware design and programming done.

 

[tt]

Please define "phantom issues" more specifically.

I mean subtle instability issues that may be difficult to attribute a cause to. One example is I tried a loaded SE/30 config with multiple PDS cards and I started getting random freezing and system bombs. It seemed like after several restarts I was able to run benchmarks flawlessly for maybe 20 mins, so I am thinking the PSU's got warmed up. I checked the voltages with a DMM and they were close to nominal, but I don't really know what was going on. Another example could be your floppy drive errors discussed earlier.

Guide to the Macintosh Family Hardware, Second Edition has more extensive power supply specs for the SE/30 on pages 259-260 that relate to this discussion...

Good idea to check there and some useful info here thanks. It does seem like the sweep specs. are tight.

Ill give it a shot once I find a suitably sized power supply. ...The biggest problem for me right now is time. I am strapped of it. I have been doing a major R&D project for a display corporation on new scoreboard technology, and thats been taking up nearly all of my time...

Sounds like a cool project. Thanks for your input so far! It would be great if we could eventually find a reference PSU to use that is commonly available and low cost that can fit into the old chassis for a clean restore. :b&w: