Is the 12VHPWR amperage distribution suitable?
Is the 12VHPWR amperage distribution suitable?
If the readings suggest it, this indicates the final three pins exhibit marginally lower contact resistance, resulting in a bit more current flow. In my view, 12V-6X6 (12VHPWR) connectors are inexpensive, of poor quality, only moderately sufficient, and likely to fail. However, given my experience as a design engineer on MilSpec and Aerospace projects, I would prefer connectors costing hundreds of dollars. This approach is more practical for consumer GPUs, which is why I opt for 12VHPWR.
Isn't it more than that, though? Although increased resistance will cause the affected wires to become hotter than usual, I believe it's the increased current that leads to other pins exceeding their limits and melting. I'm uncertain whether higher current actually passes through those pins.
If increased resistance in a mating pair of connector pins leads to reduced current in that specific lead, the amount of power dissipated in the copper wire will decrease slightly. The relationship is defined by Power = I squared R, where I represents the current and R denotes the wire resistance. On the other hand, when resistance is lower, more current flows, resulting in a bit higher power loss in the copper wire.
In the first case, pairs with greater contact resistance tend to heat up more, while the wire cools slightly because less current passes through. Conversely, when resistance drops, pairs may feel cooler, but the wire experiences a small increase in temperature due to higher current flow.
Precise measurement is challenging, as varying contact resistances influence currents and consequently the heating levels. In a 12VHPWR/12V-2X6 setup, most heat tends to be absorbed by the connectors themselves, especially when current is kept near its maximum without significant derating.
During my time as a design engineer for MilSpec and AeroSpace systems, we implemented a "healthy" derating rule for connectors and cables. We reduced derating when contacts are more numerous or currents are higher, aiming to prevent overheating. A maximum of 30% derating for bundled wires in connectors was standard to avoid excessive heating.
This contrasts sharply with the common "design practice" for 12VHPWR / 12V-6X6, where pushing current through components is accepted without concern. It's crucial to understand that ignoring proper derating can lead to severe consequences.
If a pair of pins shows unusually high contact resistance—potentially only a few hundred milliohms—the entire pair could overheat. This might cause permanent damage, possibly resulting in the contacts becoming open circuits.
The remaining five pairs would then carry the full current, but due to insufficient derating, especially with high-power components like a RTX 5090, the current through each of these five pairs could exceed their rated capacity. This would trigger further overheating, eventually causing the system to fail entirely—potentially leading to a complete shutdown or even visible smoke and flames.
RMA is the only viable solution. Without the ability to monitor and adjust for abnormal current distribution across all six wires, a 12VHPWR / 12V-6X6 setup poses a significant risk. Picture the consequences if you connected a 12V-6X6 connector to a satellite and pushed 600W through it. The premium price of these connectors justifies their use in professional environments, but they would be impractical for commercial GPUs.
It’s likely that nearly all 12V-2X6 users won’t face issues, but some may experience problems—such as card failures or hardware damage. Reviewing thermal images typically shows the highest temperatures in the 12VHPWR / 12V-2X6 connector pair, not in the cable itself.
https://www.hwcooling.net/en/12vhpwr-cab...t-temps/2/
Copper wire efficiently conducts heat, so some of the generated warmth from the connector pins travels along the cable, raising its temperature beyond what self-heating would cause. This behavior aligns with basic thermodynamics principles, acting as a heat sink for the hot pins.
Hi all, understood on why this is a problem, but that doesn't really answer the question of this is okay.
Update, ran furmark bench and can see that load is skewed towards the last 3 pins.
Is this still in spec?
Imbalance problems likely originate from the cable wires themselves. Determining what might be causing them is challenging due to various influencing factors such as time, contact quality, copper condition, and current differences. There’s no consistent pattern, so the only approach is to keep monitoring.
If every one of the six wires forming the +12V connections in a 12VHPWR cable is cut from the same wire reel during production, it's likely each wire would have resistance almost the same, just a few milliOhms off.
When all six +12V wires are joined into a single spade terminal at each end and linked to a +12V DC supply with a matching load, the current distribution should remain nearly uniform, differing only slightly by tens of milliAmps.
I think the issue on the original post stems from the fact that the +12V feeds power two separate circuits on the graphics card, rather than a single large VRM with multiple phases.
If there are two distinct VRM circuits, each running on three separate +12V wires, this could explain the current being split into standard 6.9A and 8.2A feeds.
Any minor variations in current flow are probably caused by differences in contact resistance at the connector pins. This remains just a hypothesis, and a closer look at the card would be needed to confirm the presence of multiple VRMs.
The amperage allocation you receive is what you receive. The RTX 3090 used active current-balancing through shunt resistors to balance the load across all wires, but this safety mechanism was removed on the RTX 5090 FE due to its low-end/budget position. It's likely the savings were spent on purchasing more leather jackets for the CEO, who earned ten times more from Nvidia in high-end and AI markets compared to average gamers. The ASUS ROG Astral lacks active load balancing, relying instead on the "Power Detector+" feature in its GPU Tweak software to track uneven current distribution and advise precautions. Alternatively, you might try disconnecting and reconnecting to assess if the balance improves.