PDA

View Full Version : AT vs XT PSU (difference )?



bobwatts
May 13th, 2010, 04:39 AM
Hi Gang !

I'm getting ready to pitch a massive pile of PSU's, but want to keep a few for future "projects".

As I was going through these, it occurs to me that there might be a difference between AT and XT PSU's, but I can't remember what it is, nor can I find it searching.

Anyone know the difference, if any ? Or will an AT PSU work on an XT board ?

Thanks !

bobwatts

mikey99
May 13th, 2010, 05:11 AM
The main differences are :

1) the physical size of the case, the AT PSU is taller.
SO they're not interchangeable between the XT/AT cases.

2) the wattage of the AT PSU is probably higher. I think the XT PSU
is 130 Watts, not sure about the AT. If its a genuine IBM AT PSU
its probably not much more than 130. But if its a clone AT PSU
it could be 200 or higher Watts.

mikey99
May 13th, 2010, 06:21 AM
The PSU cases also come in handy for projects..... I used one from a dead
AT PSU to build a homebrew Variac :-)

hargle
May 13th, 2010, 07:43 AM
There also seems to be some difference between the handoff of Power Good or some other signal to tell the cmos battery that it is or is not needed depending on the state of system power on/off.

I ran into this not too long ago where I was trying to use an older power supply on a newer machine, and kept losing my cmos settings on power up. I ended up moving an AT supply's guts into the older supply's case and my problem was solved.

bobwatts
May 13th, 2010, 02:17 PM
There also seems to be some difference between the handoff of Power Good or some other signal to tell the cmos battery that it is or is not needed depending on the state of system power on/off.

I ran into this not too long ago where I was trying to use an older power supply on a newer machine, and kept losing my cmos settings on power up. I ended up moving an AT supply's guts into the older supply's case and my problem was solved.

Hello !

Thanks for the information. I was thinking there was an electrical issue concerning the difference between a
PC/XT-AT PSU. Now I will try to track down which of these PSU's are what.

By the way Hargle, how did the XT-IDE controller turn out ?

bobwatts

Unknown_K
May 13th, 2010, 02:50 PM
The XT motherboards have square pins that connect to the power cables while AT have flat pins that might be wider.

Also:

http://www.pcguide.com/ref/power/sup/partsMotherboard-c.html

"The PC/XT, AT, Baby AT and LPX form factors all use the same pair of 6-wire connectors, usually called "AT Style" connectors. They are typically labeled either "P8" and "P9" (what IBM originally labeled them) or "P1" and "P2". (Actually, the PC/XT form factor omits the +5 V signal on pin #2 of P8, but otherwise is the same.)"

Chuck(G)
May 13th, 2010, 03:30 PM
Any power supply capable of supplying +12, -12 and +5 (and -5, if it's a 16-64K 5150) can be used to run a 5150/5160/5170. If you don't need any RS-232 I/O, you can probably get away with just +5 and +12. I used to buy XT clone motherboards for cheap, mount them to a sheet of plywood and use whatever power supply I had around I've used power supplies scavenged from old terminals and even a linear open-frame supply to run these. The important thing is that it's able to supply the required current.

"Power Good" is easy to fudge with a simple RC combination tied to the +5 to give a few hundred millisecond delay after the +5 comes up.

mikey99
May 13th, 2010, 06:24 PM
I'm getting ready to pitch a massive pile of PSU's, but want to keep a few for future "projects".
.....

Are the XT power supplies from IBM XT's ? or clones ?

kishy
May 13th, 2010, 07:52 PM
Alright, so I pose a question...

When I power an XT clone (or real 5160) mobo, I'm using a "baby AT" PSU from a substantially newer system. I do not trust or like particularly old power supplies and never, ever intend to use an actual XT power supply again...in an XT or otherwise.

What, if any, harm can come from this combination? I've yet to witness a failure to POST or anything on 3 different boards, and without a means of storing CMOS settings to begin with (or any CMOS settings to store) it seems that particular issue is irrelevant.

Chuck(G)
May 13th, 2010, 08:59 PM
What, if any, harm can come from this combination? I've yet to witness a failure to POST or anything on 3 different boards, and without a means of storing CMOS settings to begin with (or any CMOS settings to store) it seems that particular issue is irrelevant.

None at all--just be sure that if your motherboard is using 16K DRAMs (4116), that there's -5 on the power connector for substrate bias. You'll usually only see this on 64K 5150s and similar very early clones.

modem7
May 14th, 2010, 12:19 AM
There also seems to be some difference between the handoff of Power Good or some other signal to tell the cmos battery that it is or is not needed depending on the state of system power on/off.
I ran into this not too long ago where I was trying to use an older power supply on a newer machine, and kept losing my cmos settings on power up. I ended up moving an AT supply's guts into the older supply's case and my problem was solved.
One of the causes of that symptom (occasional losing of CMOS settings) is due to a difference between the XT and AT power supplies. I'm not confining this discussion to IBM made machines. I ran across the symptom in the 80's when people upgraded the motherboard in their computer from an XT (or PC) type to an AT type. The power supply was adequately rated. The cure was always to upgrade the power supply from an XT type to an AT type.

What I was told at the time (by 'experts' we had access to): The CMOS chip on an AT motherboard (146818) is intolerant of it's voltage supply rising too fast on power up of the computer and falling too fast on shut down of the computer. If one looks at the datasheet for the 146818 chip, one sees parameters of "minimum rise time of Vcc" and "minimum fall time of Vcc". In the move to the AT, IBM had to design the AT power supply so that the rise and fall times of Vcc weren't too fast for the 146818. Apparently, many PC and XT power supplies do not meet the minimum Vcc rise/fall times required by the 146818 chip.

bobwatts
May 14th, 2010, 02:54 AM
Hi Gang !

Thanks again for the information, it's appreciated.
To answer a question, I have no original IBM "big ol" PSU's, I just have a pile of clone stuff.
I was mainly concerned if I decide to test out or hook up an PC/XT board someday, and didn't have a PSU
to do this with.
I *thought* I remembered that an AT PSU would work, but wasn't sure.

Thanks again !

bobwatts

Chuck(G)
May 14th, 2010, 08:51 AM
One of the causes of that symptom (occasional losing of CMOS settings) is due to a difference between the XT and AT power supplies. I'm not confining this discussion to IBM made machines. I ran across the symptom in the 80's when people upgraded the motherboard in their computer from an XT (or PC) type to an AT type. The power supply was adequately rated. The cure was always to upgrade the power supply from an XT type to an AT type.

One of the problems is that the original XT POWER GOOD signal was never completely specified. Even among clone AT PSU makers, the implementation could vary wildly. The situation got to be so bad that some board makers left PG as a no-connect and implemented their own power-sensing circuit (I have a couple of those).

If the time constant of the PG signal is a problem, then it's easy to remedy with a change to the PSU or even re-implement with a resistor and a capacitor.

N.B. Very few "Power Good" signals as implemented actually serve that function. Usually, all they indicate is that the voltage on the +5 line has risen to somewhere close to +5V. Overvoltages and lack of voltage on other lines are not usually diagnosed. In other words, "Power Good" doesn't really mean the power's good. :)

michell22
May 14th, 2010, 08:13 PM
Hi
The PS2 PSU is made for motherboards with the older AT form factor.
The ATX PSU is made for ATX form factor motherboards introduced with the Pentium 4s.

These links should help you discover more:

amtrade.com/pwrspply/pwrspply…
wikipedia.org/wiki/Power_suppl…



-------------------
mac book case (http://www.mymacbookcase.com/)