Hardware Insights Forum

Big T1 integrators like HP, Dell, Lenovo already started pissing about Intel's obsolete ATX connector standards and using 12V PSUs already (with +5/+3.3V VRMs on the MoBo only for the HDDs and ODDs), which creates some problems when these units die. They are often TFX or 1U units, but when you don't know the pinout, you cannot replace them with standard units utilising 20(+4)pin Main ATX. So lets take a look at those you come around.

First one is the Lenovo ThinkCentre M93p, using FSP240-40SBV 80 PLUS Bronze rated units with only +12V and -12V outputs (and SB rail). This has 14pin main connector with only 12 pins used. It is the ordinary Molex Mini-Fit Jr. 5557-14R:

- pins I, II and III are +12 V
- pin 4 is -12 V
- pins 5-6 are not conencted
- pins 7-10 are ground
- pin 11 is +5 V SB
- pin 12 is PWR Good
- pin 13 is PWR On
- pin 14 is a white wire, not sure what for, but likely +12V feedback

It just came to my attention that Intel has answered these with a new standard in the meantime; using a 10‑pin Mini‑Fit Jr. variant with three +12V lines (pins 8‑10), three ground (pins 2‑4), a +12VSB (pin 7), the existing on/off (pin 1) and power good (pin 6), and pin 5 unused. The 6‑pin PCIe connector has also been reused for additional main power (here it can be loaded up to the actual connector rating, rather than the PCIe specified 75W). SATA drives are again powered through the board (though without +3.3V since it was removed from SATA specification as of V3.2) using Micro‑Fit style connectors (with 4 pins for 2 drives or 6 pins for 4 drives), while the peripheral connectors are still provided from the PSU but with only +12V and ground (pins 1 and 2).

It may well end up being a mostly-OEM standard (a bit like BTX, though hopefully less of a failure than that was), though adapter cables (with a converter for the standby) are provided to support transitioning.

In many ways this mess seems to be a matter of bad timing in the industry, with the move to +12V distribution coming too soon after basic ATX to start afresh.
This also goes for SATA; rather than wasting three pins on +3.3V, they could have made +12V only HDDs quite easily.
(By the early 2000s only the head preamplifier used +5V – and a locally-converted −5V – directly; even a linear regulator would do fine for that, surely less wasteful than the linear regulators used until ≈2005 to drop +5V to the +1.5V or thereabouts for the MCU core .)
Although, this would conflict with the established +5V only for laptop drives (but they could still drive the same motors using a lower PWM duty cycle; and perhaps even implement adaptive support for either voltage as provided)…

Good to know, thanks. Wonder when (if at all) T1 producers will adopt it. From their point of view, having non-standard PSUs is an advantage rather than otherwise, they often have customised motherboards and even though providing voltage at some specified current is general requirement without any need for a specific power supply, they will use that MoBo customisation as an excuse to only provide their own branded PSUs for systems in question.

Good that there are often chinese producing cheap aftermarket cable adapters to use more ordinary ATX/SFX/TFX power supplies with different OEM systems, you can get those from fleebay, ali and other typical sources. I surelly know about such for Dell and HP workstations.