Post by thread:[OT:]PC Power supply dodgyness

Ignoring for the moment that there are better solutions. what pitfalls do you guys see in running a PC off 2 standard ATX power supplies, IE one running the drives and peripherals (fans etc) and the other running the mother board. obviously the 2 need to be connected at ground and it will have perhaps half the reliability of a single unit (though they will be at reduced load which may count for something) I cant see there being any issues with the 12V and 5V supplies being at slightly different voltages. However any advice wrt any bad things that may happen if one psu is down and the other isn't? devices connected across IDE and U2W SCSI buses. I'm also wondering if it may be possible to make some kind of nifty board that took in 3 ATX type outputs and gave some kind of hot swapping, fault tolerance, and big power output for somewhat less than the standard $700 that kind of PSU starts at obviously the board would have to be PIC controlled so it could chop between power supplies without ever connecting them in parallel ;-> -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Jake Anderson wrote: > I'm also wondering if it may be possible to make some kind of > nifty board that took in 3 ATX type outputs and gave some kind > of hot swapping, fault tolerance, and big power output for > somewhat less than the standard $700 that kind of PSU starts at > obviously the board would have to be PIC controlled so it could > chop between power supplies without ever connecting them in > parallel ;-> I had at one time thought about doing this. Here were some ideas I had: --- Use Allegro hall effect current sensors on the low voltage inputs to make sure they are sharing the load. --- Use .01 ohm resistors on the 5v line to prevent the supplies' regulating circuits from freaking out and wasting energy. --- Low Rds(on) logic level MOSFETs to turn individual inputs on/off [would require at least two floating 5v supplies at a few mA] --- Fine-tune the feedback or synchronize the switch clock of the PSUs --- Add a dummy load for hot spares if you don't actually want them to share the load - they need a load to regulate the output. --- So I determined that the risk was too great and I don't have enough money to experiment with everything. But good luck if you do it! -- -- Martin Klingensmith http://infoarchive.net/ http://nnytech.net/ -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Assuming you don't actually connect the power busses together (which would be bad without at least some additional load sharing circuitry) then the voltage drops inside each device should be greater than the difference in voltage output. In other words, the HD may have 5.05V in, while the motherboard may only have 4.95V in. But the HD is only going to output a logic level as high as 5.05V - [voltage drop of circuitry] which the motherboard should be fine with. In fact, you can get external IDE drive cases with their own powersupply while a ribbon cable connects to the motherboard IDE header. If a powersupply goes down then the devices it is connected to may get power from the i/o from the attached powered device. This won't supply 12v, and may damage the output drivers of the source, but it probably won't affect the unpowered device other than to, perhaps, write bad data to the hard drive or something terrible like that. 'Probably' being the operative word here... Of course you'll want to be careful in your choice of how to power each device. A video card with a seperate power input, for instance, may connect that input directly to the AGP or PCI power bus without any protection, so the MB and video card must have the same power supply. -Adam Jake Anderson wrote: {Quote hidden}>Ignoring for the moment that there are better solutions. >what pitfalls do you guys see in running a PC off 2 >standard ATX power supplies, IE one running the drives >and peripherals (fans etc) and the other running the mother >board. >obviously the 2 need to be connected at ground and it will have >perhaps half the reliability of a single unit (though they will >be at reduced load which may count for something) > >I cant see there being any issues with the 12V and 5V supplies >being at slightly different voltages. However any advice wrt >any bad things that may happen if one psu is down and the other >isn't? devices connected across IDE and U2W SCSI buses. > >I'm also wondering if it may be possible to make some kind of >nifty board that took in 3 ATX type outputs and gave some kind >of hot swapping, fault tolerance, and big power output for >somewhat less than the standard $700 that kind of PSU starts at >obviously the board would have to be PIC controlled so it could >chop between power supplies without ever connecting them in >parallel ;-> > >-- >http://www.piclist.com hint: The PICList is archived three different >ways. See http://www.piclist.com/#archives for details. > > > > > -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

>> I'm also wondering if it may be possible to make some kind of >> nifty board that took in 3 ATX type outputs and gave some kind >> of hot swapping, fault tolerance, and big power output for >> somewhat less than the standard $700 that kind of PSU starts at >> obviously the board would have to be PIC controlled so it could >> chop between power supplies without ever connecting them in >> parallel ;-> ... >Low Rds(on) logic level MOSFETs to turn individual inputs on/off >[would require at least two floating 5v supplies at a few mA] I used to service a disk array that had this sort of arrangement, except I believe it used AT type supplies instead of ATX ones. It appeared to have some sort of FET switching arrangement on the outputs so if a supply went dead it did not load the rest of the system. Unfortunately I did not get the opportunity to work out just how they did it. We just did module swaps if anything went wrong. I suspect they used the "power good" signal to turn on the FETs, on the basis that if anything went wrong with the supply then most likely the switch mode would drop out, and hence the PG signal disappear. -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.