Alternative low power mode support
Aris has also published some interesting information about ALPM support, which is essential for any power supply that has to meet this specification. In the table below, Intel recommends that the T3 timing for ALPM support should be below 150 ms. T1 (power-on time) is usually under 100 ms for most power supplies, so this is not really a problem, but T3 (power_OK delay) is much harder to achieve. Intel does not specify in the public ATX 3.0 specification exactly which rail defines the criteria for passing this test. The T3 time must be measured on all rails, but the pass criterion is based on 3.3V only!
But why does the 3.3V rail now decide whether the T3 parameter passes or fails? According to Intel, this is because the rise time for the 3.3 V, 5 V and 12 V rails includes a sequencing requirement that requires the 5 V and 12 V rails to be within 20 ms of the 3.3 V rail.
The diagram above shows how to measure the T3 time with an oscilloscope. Section 4.2.8 of the ATX 3.0 specification states the following:
This means that the 3.3 V rail is the last to reach 95% of its rated voltage in the worst case. So by having the passing criteria based on that, it allows for more than 150 ms T3 time at 12 V and 5 V.
Summary
Let’s summarize once again: Power supplies with more than 450 W power can definitely be ATX 3.0 compatible, even if they don’t have a native 12VHPWR connector! That’s where a simple, included 12VHPWR adapter with at least two 4×2 cable ends as accessories is all that’s needed to get around the tough standards. So here, in case of doubt, the buyer must inform himself and look twice at the box. For ALPM support, although the T3 time is measured on all rails, the pass criterion is based only on the T3 time of the 3.3 V rail. As a normal buyer, you still can’t see any clearer after all the wrangling and the unclear formulations, but unfortunately there’s a method to it, too. Those who inform themselves are, of course, better off. As always.
My personal thanks go to Aris of hwbusters.com(read his article in the original here), Jon Gerow (Jonnyguru), two board partners, and the unnamed donor of all my papers.
43 Antworten
Kommentar
Lade neue Kommentare
Mitglied
Urgestein
Neuling
Veteran
Veteran
Urgestein
Urgestein
Veteran
Mitglied
Veteran
Mitglied
Mitglied
Mitglied
Urgestein
1
Urgestein
Mitglied
Veteran
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →