ATX Power Supply for PEB - seeking more detailed information on replacing the original PS

[XLERB]

Hi all: I have a PEB with a fried power supply board (Diodes at least are shot, judging by the fireworks). I’ve been seeing and hearing that it’s possible to replace the big transformer and the PSU board with an ATX power supply, with the Mean Well PT-45 power supply mentioned several times. I’ve been scouring the threads (including Shift838’s massive tower project, which is far beyond my capabilities), looking for more information. I’ve also seen references in a couple threads to Tim Tesch’s way of connecting an ATX power supply, but can’t locate anything on that (apologies that I don’t know Tim’s handle on here).
 

It seems (and I’ve gotten indications from Arcadeshopper) that the ATX voltages (+5v, +12v, and -5v, rather than TI’s 18v and 8v) can be used with TI peripheral cards and most others (except the HRD) without modifying the cards, since the cards will still handle the lower voltage OK. There was some controversy in this thread, but the prevailing opinion seemed to be that it can be done.

 

But I haven’t been able to find any details about how this might be done, especially by someone who has done it.
 

I’m pretty handy with a soldering iron and making physical modifications to devices, but my circuit knowledge is pretty rudimentary. Can anyone steer me to some more detailed directions, or even a schematic (beyond the PEB schematic)?

 

I’d like to have a backup (that won’t damage my precious cards) so I won’t have to go back on eBay (and add another broken PEB to the growing pile in my basement) when something goes wrong - and I figure this may be of use to the community as working PEBs dwindle in the wild… Thanks!

[RXB]

I have the Shift838 P-box replacement board that is now in a PC case with all my old P-box cards.

[XLERB]

36 minutes ago, RXB said:

I have the Shift838 P-box replacement board that is now in a PC case with all my old P-box cards.

Good to know, thanks! I'm trying to stick with the existing firehose, but this is another possibility.\!

[+jedimatt42]

The rumors I have heard are that you run the PEB bus power rails for +8v at the ATX +12v, and the +18v at the ATX +12v. 

The PEB drive bay, should get power from the standard +12v & +5v molex drive connectors.

 

The 7805 +5v regulators on most of the cards will handle the +12v input. 

TIPI has another +3v regulator on board, that would then be getting +12v... which should be ok, but totally depends on the part used - my BOM lists an LM1117-3.3 which has a maximum voltage input of +20v. But I haven't seen a single board that has exactly that 3.3v regulator, substitutions are plentiful.

 

To trust that plan, I'd want to read the datasheets for the operating voltage inputs and expected output of each of the regulators on each of the cards in this mode of operation. 

 

https://www.ti.com/lit/ds/symlink/lm1117.pdf?HQS=dis-mous-null-mousermode-dsf-pf-null-wwe&DCM=yes&ref_url=https%3A%2F%2Fwww.mouser.com%2F&distId=26 -- says the +3.3v max input voltage is +20v

https://www.mouser.com/datasheet/2/389/l78-1849632.pdf -- says the +5v max input voltage is +35v, this sheet covers common +12v regulators too, same max input, but we'll be operating them at only +12v input. If I'm reading it correctly, in this mode, a 7812 will output ~+12v for an input of just +12v with minimal regulation. Maybe that is ok, given the ATX supplied +12v is also regulated. 

 

Why don't we just run the PEB card +5v regulators on +5v input? The datasheet doesn't describe that mode of operation. It in fact says you should have a minimum input voltage at least +2v above the output target. But then for the 7812, it indicates that the regulation operates down to just 12V input voltage. 

 

Each board may have different regulators with different tolerances. But most are going to fit this scenario. I probably want to read more about the +12v regulator at just +12v input. 

 

Then there are the negative voltages... 

 

I have my favorite push button PEB with a dead power supply in it just hanging around, waiting for me to do something about it. I'd certainly rather fix it than switch to a PC case. 

 

[Tursi]

I run my PEB using that approach (+12V on both lines). Most of the old regulators have voltage drop, that's why you can't put +5V on the 8v line and have it work. You'd need LDOs instead. A lot of people did the approach of using +5v there, and modifying the cards (You have to modify the flex cable foot too - I ran into that when I did it), but then you have cards that will probably be destroyed if they are put into a normal PEB.

 

My PEB works fine with the TI disk controller, Corcomp RS232/PIO, and TI 32K. I /think/ I also ran a HRD+ and Thierry's IDE in there, but it has been a long time and I'm not 100% sure when I did the mod.

 

You DO want to make sure there's a heat sink on the regulators, a 7805 taking 12V down to 5 will throw a lot of heat. But it will do it. (I particularly remember the IDE card not having one and shutting down till I added it, that's why I think that I had already modded it by then.)

 

[XLERB]

1 hour ago, jedimatt42 said:

 

TIPI has another +3v regulator on board, that would then be getting +12v... which should be ok, but totally depends on the part used - my BOM lists an LM1117-3.3 which has a maximum voltage input of +20v. But I haven't seen a single board that has exactly that 3.3v regulator, substitutions are plentiful.

 

To trust that plan, I'd want to read the datasheets for the operating voltage inputs and expected output of each of the regulators on each of the cards in this mode of operation. 

 

I hadn’t mentioned the TIPI but that would be a huge concern for me. Thanks so much for all that information, it’s really helpful.

[+InsaneMultitasker]

I ran my primary system in the manner described (12v to the 5v/12v lines and -12v to -12v line) for 20 or more years, pretty much 24x7x365. I took it offline to replace the ancient power supply with something from this century, though I haven't finished the project

 

It is important that the regulators can handle the higher input voltage, that you have heat sinks installed with thermal compound where needed, and that you keep watch and/or periodically replace electrolytic capacitors that are subjected to the higher heat. This last item can depend on how the card is installed in your box of choice; heat radiates and rises, so if you use a PC tower and mount the cards sideways, it could change the heat transfer direction for better or for worse.

 

There are a few threads where this topic has been discussed that may have further how-to instructions.  There have also been a few conversations about replacing the 5v regulators with DC-to-DC converters and buck converters.   These devices run "cool to the touch" at 12v input.  I opted for the CUI DC-to-DC converters for a few of my high heat producing cards 4-5 years ago with success. 

 

Whatever path you decide, be careful to review specs and expectations. Don't mistakenly install a 12v regulator in place of a 5v regulator, and if you decide to remove regulators and jumper the power, double and triple check your work and be sure not to place that card in a standard box.  Like Tursi said, you'll almost certainly destroy the card if you make that mistake.

 

[RXB]

I have discussed this with many people and the Shift838 P-box card with the Shift838 replacement for the Firehose card with a PC case works great.

Very quiet with 4 120mm fans running from the Shift838 P-box card fan jumpers.

And you can hardly hear it running.

 

The Regulator are warm to touch but not so hot you can not hold them, much improved over normal P-Box.

[GDMike]

13 hours ago, InsaneMultitasker said:

I ran my primary system in the manner described (12v to the 5v/12v lines and -12v to -12v line) for 20 or more years, pretty much 24x7x365. I took it offline to replace the ancient power supply with something from this century, though I haven't finished the project

 

It is important that the regulators can handle the higher input voltage, that you have heat sinks installed with thermal compound where needed, and that you keep watch and/or periodically replace electrolytic capacitors that are subjected to the higher heat. This last item can depend on how the card is installed in your box of choice; heat radiates and rises, so if you use a PC tower and mount the cards sideways, it could change the heat transfer direction for better or for worse.

 

There are a few threads where this topic has been discussed that may have further how-to instructions.  There have also been a few conversations about replacing the 5v regulators with DC-to-DC converters and buck converters.   These devices run "cool to the touch" at 12v input.  I opted for the CUI DC-to-DC converters for a few of my high heat producing cards 4-5 years ago with success. 

 

Whatever path you decide, be careful to review specs and expectations. Don't mistakenly install a 12v regulator in place of a 5v regulator, and if you decide to remove regulators and jumper the power, double and triple check your work and be sure not to place that card in a standard box.  Like Tursi said, you'll almost certainly destroy the card if you make that mistake.

 

Exactly. I've got a test board running this way and it's been just fine. Use your existing firehose too.

[XLERB]

Thanks to everyone for all the information. I have a pretty fair idea of what needs to be done, but being a slow learner in matters circuit-wise, I want to be sure I’m being careful. Jedimatt42, your links to the datasheets should be very helpful. My fantasy that someone could make a step-by-step tutorial or video (similar to the 12v cooling fan instructions) is part of my insecurity but I may just need to trust myself and jump in. I have a spare 32K card and disk controller card that I can try first before I risk anything from my main setup with the TIPI and SAMS card.

[XLERB]

The other approach would be to start replacing components on the existing power supply boards, starting with the capacitors on the one that didn’t display fireworks, and see if it still blows fuses. I believe it was a p-code card that caused these failures, but I don’t know how to test any of the components on that. There are no obvious shorts.

 

UPDATE:

 

Ordered the capacitors and voltage regulators to replace the ones on the original PEB power supply board. Can't hurt to try. If that doesn't work then the AT power supply is plan B. I appreciate all the information and suggestions.

Edited March 15, 2023 by XLERB
Update!

[BradB2]

 I did this about 12 years ago but forgot how I wired it. I gutted a peb and put it all in a newer pc case. I'm about to do another one so I'll be using my first as a reference. I remember I did remove and jump all regulators on my cards for the first one. It's running a disk controller card, ams super card, and flex cable card. It also has 2, 1/2 height 5.25" drives as drives #2 & #3 and a 3.5" drive as #1. It's been running fine with no issues for years now. It sat for quite a few years but is back in use again 😊

[XLERB]

16 hours ago, BradB2 said:

 I did this about 12 years ago but forgot how I wired it. I gutted a peb and put it all in a newer pc case. I'm about to do another one so I'll be using my first as a reference. I remember I did remove and jump all regulators on my cards for the first one. It's running a disk controller card, ams super card, and flex cable card. It also has 2, 1/2 height 5.25" drives as drives #2 & #3 and a 3.5" drive as #1. It's been running fine with no issues for years now. It sat for quite a few years but is back in use again 😊

Wow! That looks exactly like a PC tower that my son left in our basement when he moved to Kansas. My wife made me get rid of all that stuff he left behind (including a bunch of power supplies). Oh well…

[+InsaneMultitasker]

On 3/16/2023 at 12:00 AM, BradB2 said:

 I did this about 12 years ago but forgot how I wired it. I gutted a peb and put it all in a newer pc case. I'm about to do another one so I'll be using my first as a reference. I remember I did remove and jump all regulators on my cards for the first one. It's running a disk controller card, ams super card, and flex cable card. It also has 2, 1/2 height 5.25" drives as drives #2 & #3 and a 3.5" drive as #1. It's been running fine with no issues for years now. It sat for quite a few years but is back in use again 😊

 

Looks like a nice install!      My original full-size tower case was set up like this, with the cards inserted horizontally and no guide/edge on the front (right side) to keep the card level.  I mention this because over time, my heavier cards warped a bit due to gravity.  I had to install a different PEB cage to give the cards some support, and eventually replaced the backplane because a few of the connectors were no longer making proper contact.  My current case has the cards installed vertically.  I don't know if Shift838's rails would be of use here, just mentioning in case your cards start to show similar signs of fatigue. 

 

 

[RXB]

In my PC case my TI cards are standing up with the LEDs at bottom of case.

So you can not see the LEDs on unless you remove side panel as they are against the card support at bottom of PC case

Thus very unlikely to warp the cards or come loose from card slots.

Also as I have 3 super silent 120mm fans all running full speed you can barely hear the case running, my AMD ROG tower on other side of room is louder.

[XLERB]

Well, I thought I might also try repairing the power supply board as Plan B. But when I went to desolder the components, I discovered that TI used that military-grade solder that doesn't melt until it reaches 1000 degrees. They used the same thing in the RF modulators, as I discovered when I went to disassemble and use one of them to hold a composite cable I was making. It's more like welding than soldering. I may have to just cut the leads of the old capacitors and solder the new leads to the old ones instead of to the circuit board. And I don't know what I'll so with the two surface-mount voltage regulators. What a pain.

[+jedimatt42]

Flux paste will lower the melting point. If you don't have that applying fresh solder will too. 

 

The traces or ground plane on a power supply board will have higher heat soaking characteristics too, making it harder to keep heat in the joint. 

 

Making sure your iron is free of oxidation is important for heat transfer. If you turn the heat up too high it will oxidize faster. Clean and tin the iron if that happens. 

 

I used to feel like I was battling these old boards.

 

 

[XLERB]

3 hours ago, jedimatt42 said:

Flux paste will lower the melting point. If you don't have that applying fresh solder will too. 

 

The traces or ground plane on a power supply board will have higher heat soaking characteristics too, making it harder to keep heat in the joint. 

 

Making sure your iron is free of oxidation is important for heat transfer. If you turn the heat up too high it will oxidize faster. Clean and tin the iron if that happens. 

 

I used to feel like I was battling these old boards.

 

 

Thanks, Matt, this is all really great advice. I will try all of these.

[+jedimatt42]

I took the plunge and replaced my broken PEB PSU with an ATX  PSU. 

 

Here is a gallery: https://photos.jedimatt42.com/s/13k5pdurj6/ti-99-4-peb-psu

 

I didn't want to grind or drill out the welded in bolts that the TI transformer and fan mount to. Instead I printed some parts to adapt mounting an ATX PSU onto. 

 

 

I bought an ATX PSU breakout board, https://www.amazon.com/dp/B08LPRHZL9, seen in front.

 

I borrowed the ATX PSU's AC power jack, as it had a lower profile. Connecting 'earth' to the PEB chassis just as the TI part had done. The ATX PSU has it's own fuse, so I removed the TI one from the circuit. I also removed the ATX AC power switch and wired the 'N' and 'L' AC lines directly to the ATX PSU pcboard. The stock TI switch is double pole single throw and normally fully switches both of the AC lines and the PSU board. I chose to have the PEB power switch connect the ATX PS_ON and COMMON to enable or disable the PSU.

 

 

The ATX PSU I used, also had a 115/230 switch on it. The 'closed' position matched '115' when I checked with the multimeter. So I removed that switch and hard-wired it to the 115 setting. 

There were points on the ATX PSU pcboard that are labelled as ground, and connected to the original metal housing, so I ran a line from them to one of the mounting posts for the original transformer. 

 

The PEB backplane then got the GREEN = +12v, BLACK = COMMON, YELLOW = -12v, BROWN = +12v wiring to the breakout board terminals. 

 

I extended and ran one of the normal drive power lines from the ATX PSU out to the PEB drive bay. 

 

It is all back together, and working. 

 

As for the PEB cards... I am running only 3 at the moment. Shift838's Flex card replacement, A 2016 1Meg SAMS, and my TIPIPeb board. 

 

The SAMS card has a CUI V7805-500 DC-DC converter ( I think I replaced a bunch of the 'LS chips with 'HCT as well ) and that part is only about 3 degrees F over room temperature. 

 

The Shift838 Flex card heat sink is 23 degrees F over room temperature. 

The TIPI regulator region (there are 2) are at 16 degrees F over room temperature.. The TIPIPeb doesn't have proper heatsinks, I cheated and used several vias to connect a copper patch of the front side of the pcb with a copper patch on the backside. 

Room temperature is 77 F.

 

These are not precise measurements. But they are taken after it has been running for about 4 hours continuously using all 3 boards. 

 

I have said on the Zoom call that I wouldn't go near this 12v/12v solution cause it just didn't sound right to me. But after this thread kicked me to actually read the datasheets for the various voltage regulators in play, it seems sound... along with Tim and Tursi's testimony. 

 

I don't have any boards using the YELLOW and BROWN voltage rails on the backplane. I know a Rave99 peb-speech adapter will use the ~-12v rail with a -5v regulator.

 

I imagine the RS232 board uses it as well. It always appeared to me that the cast metal housing for original TI PEB boards is able to connect thermally and serve as the heatsink.  

 

Anyway, that was a fun adventure. 

 

[+arcadeshopper]

Nice work!

[+InsaneMultitasker]

Nice work. I really like that breakout board.  I hard-wired mine by splicing into a short ATX extender; your approach is much cleaner and I'm certain a lot easier to work with.  And... you can power a full height mechanical hard drive just to hear it spin and whine, not that you'd want to.... or maybe you do...?   

 

1 hour ago, jedimatt42 said:

The SAMS card has a CUI V7805-500 DC-DC converter ( I think I replaced a bunch of the 'LS chips with 'HCT as well ) and that part is only about 3 degrees F over room temperature

I installed three capacitors per converter when I modified my Geneve and Myarc FDC.  These came at the recommendation of the person who first recommended the part and after some conversation with @Swim, based on our interpretation of the data sheet.  I no longer remember if this was specific to the Geneve or all cards in general, so you may wish to check the app notes for completeness.   It seems to me that the older horizon ramdisks are poor candidates to drop in a dc-to-dc converter but I don't remember for certain if that is true... or why. 

 

1 hour ago, jedimatt42 said:

I don't have any boards using the YELLOW and BROWN voltage rails on the backplane. I know a Rave99 peb-speech adapter will use the ~-12v rail with a -5v regulator.

 

I imagine the RS232 board uses it as well.

The HFDC and Geneve use a 12v regulator for the hard drive controller and composite ckt/modulator 12v output, respectively. 

 

[XLERB]

15 hours ago, jedimatt42 said:

I took the plunge and replaced my broken PEB PSU with an ATX  PSU. 

 

Here is a gallery: https://photos.jedimatt42.com/s/13k5pdurj6/ti-99-4-peb-psu

 

I didn't want to grind or drill out the welded in bolts that the TI transformer and fan mount to. Instead I printed some parts to adapt mounting an ATX PSU onto. 

 

 

I bought an ATX PSU breakout board, https://www.amazon.com/dp/B08LPRHZL9, seen in front.

 

I borrowed the ATX PSU's AC power jack, as it had a lower profile. Connecting 'earth' to the PEB chassis just as the TI part had done. The ATX PSU has it's own fuse, so I removed the TI one from the circuit. I also removed the ATX AC power switch and wired the 'N' and 'L' AC lines directly to the ATX PSU pcboard. The stock TI switch is double pole single throw and normally fully switches both of the AC lines and the PSU board. I chose to have the PEB power switch connect the ATX PS_ON and COMMON to enable or disable the PSU.

 

 

The ATX PSU I used, also had a 115/230 switch on it. The 'closed' position matched '115' when I checked with the multimeter. So I removed that switch and hard-wired it to the 115 setting. 

There were points on the ATX PSU pcboard that are labelled as ground, and connected to the original metal housing, so I ran a line from them to one of the mounting posts for the original transformer. 

 

The PEB backplane then got the GREEN = +12v, BLACK = COMMON, YELLOW = -12v, BROWN = +12v wiring to the breakout board terminals. 

 

I extended and ran one of the normal drive power lines from the ATX PSU out to the PEB drive bay. 

 

It is all back together, and working. 

 

As for the PEB cards... I am running only 3 at the moment. Shift838's Flex card replacement, A 2016 1Meg SAMS, and my TIPIPeb board. 

 

The SAMS card has a CUI V7805-500 DC-DC converter ( I think I replaced a bunch of the 'LS chips with 'HCT as well ) and that part is only about 3 degrees F over room temperature. 

 

The Shift838 Flex card heat sink is 23 degrees F over room temperature. 

The TIPI regulator region (there are 2) are at 16 degrees F over room temperature.. The TIPIPeb doesn't have proper heatsinks, I cheated and used several vias to connect a copper patch of the front side of the pcb with a copper patch on the backside. 

Room temperature is 77 F.

 

These are not precise measurements. But they are taken after it has been running for about 4 hours continuously using all 3 boards. 

 

I have said on the Zoom call that I wouldn't go near this 12v/12v solution cause it just didn't sound right to me. But after this thread kicked me to actually read the datasheets for the various voltage regulators in play, it seems sound... along with Tim and Tursi's testimony. 

 

I don't have any boards using the YELLOW and BROWN voltage rails on the backplane. I know a Rave99 peb-speech adapter will use the ~-12v rail with a -5v regulator.

 

I imagine the RS232 board uses it as well. It always appeared to me that the cast metal housing for original TI PEB boards is able to connect thermally and serve as the heatsink.  

 

Anyway, that was a fun adventure. 

 

Wow, that’s exactly what I was trying to do only better! I have the ATX power supply and I’ve removed the fan, transformer, and circuit board from my old-style PEB. I’ve also puzzled out the penciled-in color notations on the old schematic (yellow for -16VDC was especially hard to read).

 

I’m at the point of thinking about grinding off those TI bolts, and I may end up doing that since I don’t have a 3D printer. I was hoping to avoid deconstructing the ATX power supply but the box it’s in is just too big to comfortably fit in the PEB.

 

The breakout board I’ve ordered is very similar, and I was fixing to wire the push button switch to pins 15 and 16 of the 24-pin ATX connector as well.

 

I’m going to study your method of fitting everything in. Absolutely impressive! Glad my questions sparked this project!

 

One more comment - you could start a cottage industry with those 3D printed mounting parts!!

 

EDIT: The ATX PS comes right out of its case, so I won’t worry about that. I might even be able to use the plastic bracket from the TI board with a few new holes drilled to hold in in just the right position. The breakout board can be mounted somehow, I’ll figure something out.

Edited March 26, 2023 by XLERB
Added comment

[XLERB]

One other update: I’m still working on trying to replace the caps and voltage regulators on the other PEB circuit board, but even with generous applications of flux, it’s hard to get all of that solder off (and I’m sure the heat is frying the bigger voltage regulator since I heard it “singing”). But I’ll keep after that one (that’s my original rocker switch PEB, purchased during the great TI clearance in 1984 when my kids were all toddlers/preschoolers, so it kind of has sentimental value).

 

All this fiddling around is keeping me busy during and after my radiation treatments for prostate cancer, which is a welcome distraction. (They’re finished, and I’m told I have an 85% chance it’s gone for good). But my wife wishes that I would get the tree trimmer out and get a new front door before I order any more electronic parts from Amazon and eBay.

[+jedimatt42]

Cottage, like when you leave a glass of milk near a window and come back to it a week later... ? LOL...

 

Pushbutton replacement - I didn't need this, but someone else asked me for the measurements, so I modelled the original : https://cad.onshape.com/documents/72e403f5d507511b828e4226/w/86e436bcd8f5c2ef164c777d/e/ad8b60fcda21230e675f7b1f

 

Breakout board mount : https://cad.onshape.com/documents/81683f3688039da627ca59d3/w/8f7da5f36b52acfeb0a76249/e/4e0f33657350c17f8fddef24

 

PSU pcboard vertical mount and guard : https://cad.onshape.com/documents/89ba6baa51fce2b6c0c1a9c8/w/6585cb204e0cd64c64442fe1/e/365aff479824751fa2060b3a

 

With the free access to OnShape, this work is freely available for all to consume. IDK if 9/10 ATX PSUs will have the same mounting dimensions... so anyone hoping to use these should measure the mounting holes in modelling software before you print. 

[GDMike]

18 hours ago, jedimatt42 said:

I took the plunge and replaced my broken PEB PSU with an ATX  PSU. 

 

Here is a gallery: https://photos.jedimatt42.com/s/13k5pdurj6/ti-99-4-peb-psu

 

I didn't want to grind or drill out the welded in bolts that the TI transformer and fan mount to. Instead I printed some parts to adapt mounting an ATX PSU onto. 

 

 

I bought an ATX PSU breakout board, https://www.amazon.com/dp/B08LPRHZL9, seen in front.

 

I borrowed the ATX PSU's AC power jack, as it had a lower profile. Connecting 'earth' to the PEB chassis just as the TI part had done. The ATX PSU has it's own fuse, so I removed the TI one from the circuit. I also removed the ATX AC power switch and wired the 'N' and 'L' AC lines directly to the ATX PSU pcboard. The stock TI switch is double pole single throw and normally fully switches both of the AC lines and the PSU board. I chose to have the PEB power switch connect the ATX PS_ON and COMMON to enable or disable the PSU.

 

 

The ATX PSU I used, also had a 115/230 switch on it. The 'closed' position matched '115' when I checked with the multimeter. So I removed that switch and hard-wired it to the 115 setting. 

There were points on the ATX PSU pcboard that are labelled as ground, and connected to the original metal housing, so I ran a line from them to one of the mounting posts for the original transformer. 

 

The PEB backplane then got the GREEN = +12v, BLACK = COMMON, YELLOW = -12v, BROWN = +12v wiring to the breakout board terminals. 

 

I extended and ran one of the normal drive power lines from the ATX PSU out to the PEB drive bay. 

 

It is all back together, and working. 

 

As for the PEB cards... I am running only 3 at the moment. Shift838's Flex card replacement, A 2016 1Meg SAMS, and my TIPIPeb board. 

 

The SAMS card has a CUI V7805-500 DC-DC converter ( I think I replaced a bunch of the 'LS chips with 'HCT as well ) and that part is only about 3 degrees F over room temperature. 

 

The Shift838 Flex card heat sink is 23 degrees F over room temperature. 

The TIPI regulator region (there are 2) are at 16 degrees F over room temperature.. The TIPIPeb doesn't have proper heatsinks, I cheated and used several vias to connect a copper patch of the front side of the pcb with a copper patch on the backside. 

Room temperature is 77 F.

 

These are not precise measurements. But they are taken after it has been running for about 4 hours continuously using all 3 boards. 

 

I have said on the Zoom call that I wouldn't go near this 12v/12v solution cause it just didn't sound right to me. But after this thread kicked me to actually read the datasheets for the various voltage regulators in play, it seems sound... along with Tim and Tursi's testimony. 

 

I don't have any boards using the YELLOW and BROWN voltage rails on the backplane. I know a Rave99 peb-speech adapter will use the ~-12v rail with a -5v regulator.

 

I imagine the RS232 board uses it as well. It always appeared to me that the cast metal housing for original TI PEB boards is able to connect thermally and serve as the heatsink.  

 

Anyway, that was a fun adventure. 

 

Did you add any pwr LED to your case. I didn't. But was wondering