Preparation and measurement setup
I first measure the current flow on the 3.3-volt and 5-volt rails on the 24-pin connector of the mainboard. To keep these values as low as possible, I at least left all external USB ports open, switched off RGB (except for the RAM) and disconnected both the mouse and the keyboard. I later subtracted the sum of these currents, which also flow back via the ground, from the return flow of the 24-pin. This is not entirely correct or particularly accurate, but I can’t disconnect all the mainboard’s ground lines. That’s about right and is less than one ampere.
I use my current clamps to measure the EPS, the 24-pin of the mainboard and of course the PCIe connection of the graphics card. I also determined the differences between the rails for a plausibility check, including the voltages between the ground points. The system consists of a Core i5-13600K on an MSI MAG B660M Mortar WiFi, 16 GB DDR4 RAM, a GeForce RTX 4070 Ti Super and a low-power PCIe 3.0 SSD. The graphics card runs under full load with Furmark, the CPU has an almost constant power consumption of just under 55 watts in a light CPU load scenario. The entire PC generates an average current flow of 31.4 amps on the secondary side over 15 minutes of measurement, which corresponds to around 380 watts at an average of 12.1 volts. So far, so unexciting. But how is this distributed in reality?
When the mass is askew
First, I measured and compared in the diagram how the three rails behave in terms of inflow and return flow. We see almost 20 amps on the 12V2x6 connection of the graphics card, while the EPS and the 24-pin are almost on a par. The approximately 3.4 amps of the graphics card, which then flow internally in the mainboard from the 24-pin to the PCIe graphics card slot, I can only log in the inflow, the return flow then takes place as in the watering can principle via the various ground lines of the mainboard. But that’s not the issue, but it also causes some detours.
I have now compared these items in percentage terms. Let’s start with the supply from the 12-volt source and look at the shares of the three rails. You can see very clearly that the 12V2x6 is the largest item here in terms of current. Also in terms of power.
And now comes the return journey, which is more than exciting! Because we can now see very clearly why the ground pins of the 12V2x6 connection are never damaged directly, but only as a result of damage in the immediate vicinity. In the return flow, the graphics card connection is suddenly only the smallest item!
Of course, you can draw a lot of conclusions from this and also learn a lot for yourself. As a graphics card manufacturer, I would also consider how the PCIe slot can be relieved of ground connections. This already starts with the voltage converter design. Please turn the page again!
184 Antworten
Kommentar
Lade neue Kommentare
Neuling
Veteran
Urgestein
1
Urgestein
1
Mitglied
Urgestein
Veteran
Urgestein
Urgestein
1
Urgestein
Urgestein
Veteran
Urgestein
Mitglied
Urgestein
Urgestein
Alle Kommentare lesen unter igor´sLAB Community →