ATX PSU in PEB mod

[+helocast]

6 hours ago, GDMike said:

Any tips before I start to replace my peb PS?

I'll be using this PS.

I found this was mentioned by Mike:

Instead of bypassing the regulators, simply wire the ATX +12v to *both* the TI +16v and +8v lines. The card regulators will take the 12v down the 5v quite well, and the drop on the 12v lines is still good enough to run the card. (Also wire the -12v to the PEB's -16v line). (He tested on RS232 and HDFC, I'm testing on the TI 32k and Disk, and Corcomp RS232). This does in fact work and is a much! better solution.

ok, which are the TI +12, -12, -16, +8 lines

You'll have to fool your ATX supply on the power good signal. Here's the project for that https://www.overclockersclub.com/guides/atx_psu_startup/ with a couple variations.

Also, once you disassemble your PEB, you should have your power/ground planes already marked and color coded for you:

 

Doug

[HOME AUTOMATION]

1 hour ago, helocast said:

You'll have to fool your ATX supply on the power good signal.

This is an output! The need for loading has led to this misconception, somewhat.

 

[HOME AUTOMATION]

6 hours ago, GDMike said:

Will start disassembling Peb soon.

Remember the one about...

 

...All the kings horses and all the kings men...

 

Oh, and,

 

If it ain't broke...

 

...Bye, bye... Supernotes!

 

 

But if you must!

 

...

 

[GDMike]

Oh yes auntie M!

 

Edited July 24, 2020 by GDMike

[GDMike]

Step 1. Remove NEW PS from case. And extend wiring for PEB power connector and power switch. I'll also be using the"closed" jumper pin 14 to ground so I can just use the PEB power switch for on/off. I'll also be including a switch for my rasberry pi onto the back of the PEB case. 

Edited July 24, 2020 by GDMike

[GDMike]

12 hours ago, helocast said:

You'll have to fool your ATX supply on the power good signal. Here's the project for that https://www.overclockersclub.com/guides/atx_psu_startup/ with a couple variations.

Also, once you disassemble your PEB, you should have your power/ground planes already marked and color coded for you:

 

Doug

So basically, I'll put +12 from new PS to my green and brown ti wiring, and the -12 to the Yellow and black leads on the TI peb board.

I'll take a vcheck of the board first before I start just to confirm 12,16,8 etc..

Edited July 24, 2020 by GDMike

[GDMike]

PEB apart. I find things are usually easy to come apart, this one had stripped screws and the too cover is not screwed on. I'll have to fix that upon reassembly.

I've already tested my new power supply for power up/off per switch. But the switch will be swapped with the TI front panel switch.

Edited July 24, 2020 by GDMike

[GDMike]

I mounted it like this, using the original mount and a screw. and I'll throw a plastic zip tie around it once I'm done. I'm looking to see if I can place another fan. I'm thinking about using the area where the studds are located that held the original transformer.

BTW, this powers up and the single fan runs. Next I've got to wire it to the PEB board.

Edited July 25, 2020 by GDMike

[+helocast]

50 minutes ago, GDMike said:

I mounted it like this, using the original mount and a screw and I'll throw a plastic zip tie around it once I'm done. I'm looking to see if I can place another fan. I'm thinking about using the area where the stuffs are located that held the original transformer.

BTW, this powers up and the single fan runs. Next I've got to wire it to the PEB board.

?Looks like you're progressing ... always good.

I just wanted to address terminology in my last "help". Thierry does theory WELL (as do probably most of the seasoned contributors here).

I know enough to be dangerous, but I always try to consolidate links, pics, sources since projects and references also seem to be scattered to the four winds over the years.

With that said, http://www.harmlesslion.com/text/atx_peb.htm seems to be the project you are following. Might I also suggest Chris (Shift838 here) as a source since I know he did a great write-up on solving all of this AND putting PEB cage in a modern PC case.

 

What I observe:

It looks like your PSU has six rails supplying max peak of 750W. If you "math" what each rail could provide max, you get 823.6W, yet you are still limited to any combination of rails not totaling PSU rating of 750W.

TI spec'd/budgeted original supply to not more than ~4W on unregulated +8V rail to include a floppy drive, ~3.75W on unregulated +15/16V rail to include a floppy drive, and finally ~4.5W on unregulated -15/16V rail?

Divide each of those rails up among 8 PEB slots plus a drive respectively and you can see there wasn't a lot of excess power and why only one drive was officially supported - actually quite puny!

I'd suggest keeping your two yellow +12V rails separated, one powering backplane and one powering noisy accessories (fans, drives), common grounds. I don't know what your full complement of cards and accessories for your system are.

I'd also suggest using the +5VSB rail (see attached picture as an always-on source for a TIPI if you have one and perhaps a standby panel light.

 

Thinking aloud:

I personally can't see how a modern PSU provides enough overhead for the relatively power hungry 7805s or 7812s of the day. +8V was chosen to provide 2.5-3V overhead just as the +16V was chosen to provide 3.5-4V overhead to respective regulators.

Also of course over time, some users/techs may have bypassed regulators on individual cards to control heat and chose to supply regulated +5/+12/-12V.

If they did that they should have annotated on the card/shell that was done to prevent plugging in a modified card into an unmodified PEB.

 

I guess it has been successfully done, so I'm not trying to impune anybody's successes previously.

As I said, I just try to consolidate and combine info into a one-stop reference ... and those aren't all that pretty (tough conveying all the information and I suck at that).

My 2 cents. Interested to see how this turns out and as always, good luck and low magic smoke release!

 

Doug

 

[GDMike]

Great write up Doug. Thx for that information and doing the math for us, at which I suck at.

 

 

 

The standard computer power supply unit (PSU) turns the incoming 110V or 220VAC (alternating current) into various DC (direct current) output voltages suitable for powering the computer’s internal components and with a little bit of imagination it is possible to convert ATX PSU to a TI PEB power supply.

 

ADVERTISING

 

This isn't mine, but I'll bet for the most part that the wiring labels listed are the same as what I'm using.

Safety. Yeah, whatever.. I've been bitten a couple times and to me it's not an issue, but to others if you haven't encountered it, its s

"shocking".  I'm sure you'll get over it - just keep your socks dry, yeah I kind do things in my socks, unless it's a sensitive component, then NO socks allowed. But anyway, because some have died doing this here's the safety presentation.

 

Firstly and more importantly before you start to convert ATX PSU, make sure that the PSU is unplugged from the mains supply and discharged by letting it sit unconnected for several minutes before you start. This is important! as it could result in a potentially dangerous or even lethal situation due to the high voltages inside the PSU if you decide to dismantle it. Also make sure that the metal box of the PSU is correctly earthed or grounded. You are responsible for your own safety!.

We can not just simply plug the PSU into the mains supply and expect to get the required 5 or 12 volts output. The standard PC power supply unit has two safety mechanisms that prevent it from being switched “ON” without the motherboard attached.

Unfortunately you can not just have the wires left open, luckily both of these issues are easily fixed.

There are several different coloured (hmmm.. is that spelled weird).. wires attached to the 20-pin ATX connector providing several different voltage outputs such as +3.3V, +5V, +12V, -12V, -5V as well as a number of black ground wires and a couple signal wires as shown in the following image along with their colour-code and description.

20-pin Molex ATX Connector

 

Pin outs of the 20-pin connector with the colours of the wires used in a standard ATX PSU connector.

Pin	Name	Colour		Description
1	3.3V		Orange	+3.3 VDC
2	3.3V		Orange	+3.3 VDC
3	COMMON		Black	Ground
4	5V		Red	+5 VDC
5	COMMON		Black	Ground
6	5V		Red	+5 VDC
7	COMMON		Black	Ground
8	Pwr_Ok		Grey	Power Ok (+5 VDC when power is Ok)
9	+5VSB		Purple	+5 VDC Standby Voltage
10	12V		Yellow	+12 VDC
11	3.3V		Orange	+3.3 VDC
12	-12V		Blue	-12 VDC
13	COMMON		Black	Ground
14	Pwr_ON		Green	Power Supply On (active low)
15	COMMON		Black	Ground
16	COMMON		Black	Ground
17	COMMON		Black	Ground
18	-5V		White	-5 VDC
19	5V		Red	+5 VDC
20	5V		Red	+5 VDC

Ok. I'll be working on this tonight. Since the yard is freshly mowed and I'm ready to do this.

I'm taking the, well, advice of most including Doug, keep that "yeller" wire as my 12+(-) feed. Oh, Doug I forgot to say what cards I have. 

The TI interface card, probably the most crucial and I'm not sure how it handles this change, but we'll see

Also have the SAMs memory 1 MB card. I'm thinking without looking, but I'll look later, I think it has sm voltage regs. 

Also in the mix is an RS232 TI card. But I'm probably not going to install it 

I do have the TI disk controller card.

And lastly, I have the PEB version of TIPI card. It too looking at it, has its own an voltage regs.

Im using a small case fan for venting out from behind the case and a larger fan to blow onto the power supply board.

 

Edited July 25, 2020 by GDMike

[GDMike]

Already had a problem last night. After verifying power all the way up to the point where I put the PS into the PEB them all of a sudden, just like that, no power. I mean as soon as I put it in the case the power supply quit working. Yeah I was teed off for 5 minutes then decided I had to pull it all apart again to find out what was going on. Found out that this part had come off on one side. not ever trying to touch that on purpose, hello! but maybe that's where that shock came from that I felt earlier I don't know.. but apparently that leg came off and it  took me about 20 minutes to find it as I was doing a pretty darn good job at not touching anything on the board hello. Anyway I fixed it and we're back up again so I put the board back in the PEB and yay it tested fine as a standalone.

just remember when you do these as I have been trying to do don't touch parts as unnecessarily these things are probably made in China but regardless I think they're still made cheap.  parts are probably ready to pop out of the board at any moment. And I want to do what with this? And I want it to last how many years duh this is probably a stupid idea.

Anyway I won't talk myself out of it no I won't. I make lots of dumb decisions with no help.

Edited July 25, 2020 by GDMike

[GDMike]

Well, uh.  failure! That's how to put it. 

I had a device, it's a 12v temp sensor/card reader. It worked for me last night before I ended my night as I thought surely it can safely power this Along with a fan and a couple of hard drives for a load. And it did. For hmmm.30 mins I guess I let it run. I shut everything down for the evening, turned it on like I left it yesterday and wholla! A burst of on then straight to off. I can't find anything causing the supply to shut down. No frayed wires, no loose components, I pulled the load I pulled everything but the fan,as my original power up test was only using a fan. But nothing keeps it alive. Pin 5 on the logic chip when dropped low pulls the power supply back up to operation, but it's a quick up down as something is telling it to Change state. I'm at my ends.

I'll have to put PEB back together, not a hard thing, but I'll blame this on a cheap supply, meaning build. Because there was no reason for it to have changed from last night. The original choke filter I had trouble with earlier yesterday was fixed with no more issues afterwards. But the symptoms are the same as if that choke wasn't connected. So I'm guessing that's the culprit. But weird that it worked the rest of the night. And I Tried bypassing it, same results. So, yeah. I assume the crappy, brand new supply is the failure and nothing regarding procedure was at fault. That choke that I found this connected by accident I'm pretty sure I never touched that choke I don't like getting shocked on purpose they do hold a charge for a while sometimes. I figure if components start popping off the board just because you're holding and or moving the board and if you're not being rough it should maintain its integrity. Anyway  it is it is what it is. 

Well put Mister PEB back together.

Edited July 25, 2020 by GDMike

[GDMike]

This is how I left it last night.

But that is not how it acts today. 

And this just plugs into the molex connection, so nothing weird about that. As it works like it's supposed to. But something doesn't add up.

 

Edited July 25, 2020 by GDMike

[GDMike]

I changed fans, I have resoldered, I've pulled power switch and bypassed what I could for testing, and I get same results, on then off. Just like when that choke had a loose connection, so I resoldered it I even pulled it outta the mix and I still get the same results. So I'm left to abandon.

[GDMike]

So I started looking at reasons why an Atx supply would act like this, and a column I read said, make sure the voltage on the supply is set to 115. Hmm..I thought, surely I didn't mess with that. But let's look. Well wth. Why is it working on 225? This gets weirder by the day!!!

 

[+helocast]

19 minutes ago, GDMike said:

So I started looking at reasons why an Atx supply would act like this, and a column I read said, make sure the voltage on the supply is set to 115. Hmm..I thought, surely I didn't mess with that. But let's look. Well wth. Why is it working on 225? This gets weirder by the day!!!

It's Quora, which I give about equal footing on the level of Wiki for delivering dependable answers from "experts", but:

https://www.quora.com/Some-powersupplies-can-be-switched-between-110V-and-220V-Will-leaving-the-switch-in-the-wrong-position-damage-only-the-powersupply-or-also-anything-connected-to-it#:~:text=If the switch is left,power supply or device results.

 

Doug

[GDMike]

I found out that it works because it's pulling 115 from the wall and the maximum power. It can't put out any more voltage than what's coming in as it's not a transformer, and that's why it works. Now what's wrong with the divider or the splitter, just crappy power supply I guess. I'm assuming it's not detecting the right flow coming from that coil that came apart earlier that's at the beginning of the input phase. And apparently it's not going to hurt anything to run it because it's putting out 115 like it supposed to be. And the power supply  is giving my 12 and 5 volts so I don't see nothing wrong with it now.

Edited July 26, 2020 by GDMike

[GDMike]

I'm going to let it sit with a hard drive (an old useless one) for a load, a fan and my temp sensor sitting in the ps. I'll let it run all night.

[GDMike]

Burn test...no, not the house!!

Edited July 26, 2020 by GDMike

[GDMike]

Voltages test and Temp, real PSU temp.

 

[GDMike]

So I'm guessing, but so is the PSU. But apparently the PSU isn't receiving the splitter/divider signal correctly. And that's leading up to the one-off pulse. But telling it via 225/115 switch, to look at the voltage again and pass anything above 110. I guess.

My voltages all look good and my dummy hd is spinning.

 

Edited July 26, 2020 by GDMike

[GDMike]

PSU running, well pick this back up again tomorrow.

[HOME AUTOMATION]

A well designed P/S will run over a wide range of input voltage... so, a good thing, that it can run somewhat, even at half the rated input voltage. It will similarly, likely run on 60v, at the 120v setting.

 

This P/S has likely been under usage in an upright position... Old electrolytic capacitors, that run hot, from high current, charge/discharge cycling... can obtain a thermal performance pattern, relative to the vaporization/condensation of the remaining electrolytic fluid. When operated in a "sideways" orientation, performance can differ greatly. This can at times become acute, esp. when cycled for the wrong amount of thermal/time initially.

 

The uneven heating of the can can cause the remaining electrolyte to bead up on the lower inner wall of the can. Without enough heating to the underside of the can, the fluid does not easily re-vaporize. This allows the paper separator to dry out, resulting in high internal resistance! In short, capacitors that run under hot/cold cycling conditions, need ample fluid to run satisfactorily. ...Perhaps this is not always taken into account by designers/manufacturers.

[GDMike]

When my PC died I ordered this PS. But after installing it, the PC still didn't work. So I assumed that the PC was indeed dead. But with all that's happened, I'll reorder a Corsair ps. Maybe that ole pc will live again.

Edited July 26, 2020 by GDMike

[HOME AUTOMATION]

Oh, that's different...

 

...nevermind.

 

I almost sprang for one of these...

 

     ATX P/S TESTERS...

 

...when my 64-bit Gigabyte Desktop, died last year.