| General > General Technical Chat |
| What is the motivation of intel to want PSUs that are 12V only? |
| << < (5/14) > >> |
| mariush:
My guess is that this is requested a lot by OEM manufacturers like HP and Dell who want to reduce the number of wires and save a few pennies this way and get less complaints about proprietary / custom connectors on their power supplies. HP is known for using custom connectors in their workstation systems for example, Dell also has some power supplies with custom connectors. It's an OK move but it's sad that Intel didn't have the guts to go further and work with power supply and motherboard manufacturers to innovate further. Seems like they just stuck to 12v standby and 12v to power everything. I would like to see a new revision of the standard to introduce 20v (-10% .. 5%) (to allow possibility of powering a system from DC In barrel jack using 18.5v adapters common in laptops), while keeping 12v for backwards compatibility. 20v is standardized in USB, so having 20v and 12v would make it possible to use USB connectors to power a monitor for example, or fancier external storage. It shouldn't be very hard for the psu to include some kind of microcontroller or something with a usb 1.0 or 2.0 connection to the system, allowing the system to communicate with the psu and enumerate capabilities ... For example i'm 12v only, i'm 12v and 20v, i'm 20v only (motherboard could have this tiny chip by the DC In connector and detect 12v or 18.5v..20v and pretend it's a psu with 12v only or 20v only ... useful for small systems with an APU that consume 40-60w max... power them from a 90w laptop adapter or a 200w atx psu)... psu could also say current on each voltage, total wattage, indicate if the 12v is derived from 20v using dc-dc converters.. motherboard can query and get number of fans, fan rpm, may even be able to tell psu to adjust fan rpm (but psu would have the final say as in motherboard can't override psu's safeties ... if psu controller needs to spin fan at 50% to keep cool, user would not be able to force fan to shut down) maybe support rgb if the power supply has leds inside... With just two output voltages it could be possible to have a current shunt on each 12v or each 20v pair of wires ... or worst case per 2x4 connector (eps, pci-e 6/8 pin etc)... Some would argue 5v should also stay as an option , as that could allow future very low power laptops and tablets to be powered completely from a 10-15w 5v laptop adapter but nah, lots of usb chargers have option to go up to 12v or higher and dc-dc converters are tinier and tinier and more efficient. |
| Circlotron:
--- Quote from: Circlotron on March 05, 2020, 12:19:23 pm --- --- Quote from: epigramx on March 05, 2020, 12:08:36 pm ---That raises the question though why not go 15V or 20V or 24V since they are relatively human safe. --- End quote --- My guess is because you would use a synchronous buck regulator for local conversion to the lower voltage, and when you start going to say 24V input to the reg you need a pair of mosfets with say 30V rating and consequently a higher RDSon figure than a lower voltage mosfet, hence higher conduction losses. Don't forget you have to switch the same current that goes out of the reg but at the voltage that goes in. Maybe 12V is the sweet spot? --- End quote --- Further to this, to maintain the same peak to peak ripple current in the buck reg inductor, you need proportionally greater inductance as the input voltage goes up. More copper means greater losses. |
| tszaboo:
--- Quote from: OwO on March 05, 2020, 04:19:06 pm ---Personally I would go with 70V. It's the highest I can still comfortably touch and at this level the higher the better. The DC-DC converter on the motherboard doesn't have to be transformerless. --- End quote --- Nah. There are only 28 buck converters from TI above 70V, with max current of 3A. There are over 240 with max voltage between 14 and 24V, with max current of 40A. And you dont end up with large ripple voltage and current, and big capacitors and inductors. 12V is a good compromise. |
| ajb:
--- Quote from: tunk on March 05, 2020, 12:29:03 pm ---The cynic in me says it's because the total cost is lower. A positive side effect may be higher efficiency. --- End quote --- The total cost is lower *because* it's more efficient, though, no cynicism required. At least within the context of a desktop PC, what's the disadvantage of moving to a single power rail and PoL converters wherever needed? |
| David Hess:
--- Quote from: SiliconWizard on March 05, 2020, 04:35:50 pm ---12V is part of the legacy I think, but there's another reason IMO. The idea of down-converting voltages more locally is generally a good one. But 12V is likely close to the sweet spot for the following reason: step-down converter ICs that operate with higher input voltages than this are just going to be much more expensive, as this gets into the "high-voltage" territory for current CMOS processes. 48V or over would make local step-down converting much more expensive. You could say that you can always chain them to overcome this, but then you have a chain of efficiencies which would make efficiency for the low voltages (which currently draw the most current on computer motherboards) pretty poor. Just a thought. --- End quote --- The ratio of input to output voltage for a transformer-less converter is a factor. 48 volts to 1 volt yields about a 2% duty cycle which makes things difficult, although not impossible, so it is usually better, and more efficient, to invert 48 volts to a lower voltage and then use a buck converter for the lowest voltage supplies. Vicor actually sells inverters and regulated converters to do this. |
| Navigation |
| Message Index |
| Next page |
| Previous page |