MSI teases new power supplies with 'instant protection' against melting RTX 5090 cables — promises to solve 12V-2x6 connector woes by monitoring individual wires

News
By published

USB C could see real-time monitoring and control via your motherboard

MSI MAG A850GLS PCIE5 PSU
(Image credit: 3DTested)

It seems like every month we see a new incendiary GPU story with reports of a 12V-2x6 connector being the culprit. The issue is not limited to just Nvidia's RTX 5090 anymore either, as at least five AMD RX 9070 XT variants with this plug have faced the same fiery wrath. Companies have tried to come up with solutions to monitor the 16-pin power connector, and MSI is the latest to join this line of cautionary warriors.

If even one of the pins goes out or stops making proper contact, the power supply will instantly trip, shutting down the system and protecting your expensive hardware from getting scorched. We make this assumption because the best power supplies already regulate their voltage rails that control power going to your PC, with various protection features such as OCP, SCP, and more. MSI's addition, therefore, is a specific safeguard meant for power hungry graphics cards.

In the teaser, we can also see a USB-C port at the back of the power supply, which means you can connect it to your motherboard to potentially control it via software. It should still stop the "invisible problem" of melting GPUs without that, but it'd be interesting to see the expanded feature set. Perhaps, you'll be able to see real-time stats similar to third-party tools like WireView Pro, allowing you to set hard limits in Windows.

As the name suggests, the MSI MPG Ai1300TS & Ai1600TS are flagship 1300W and 1600W units, respectively, so they're very clearly targeting the RTX 5090. Just a few days ago, we saw another one of those catch on fire in possibly the worst incident so far. And the victim was even using a native 12V-2x6 cable that came with their 1000W ATX 3.1 power supply. Let's see if MSI's offerings can finally curtail this fiasco.

Google Preferred Source

Follow 3DTested on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

TOPICS
Hassam Nasir
Hassam Nasir
Contributing Writer
14 Comments Comment from the forums
  • ezst036
    This is exactly what Nvidia wanted all along.

    Nvidia gets to reduce its own costs. Who cares if it increases costs on the other side at the power supply side right?
    Reply
  • Notton
    While this is a nice feature to have, it's also not very helpful if the PSU doesn't have a debug LED or something to tell you why it tripped.
    And that's assuming you can even see the PSU, as the current trend is to cover and hide it.

    A breaker switch is easy to figure out. You'll hear an audible "click" and the switch will be flipped if you overload a circuit.
    Reply
  • hotaru251
    & more wasted $ on an issue that isnt theirs and doesn't "fix" the issue it just retroactively powers off. (Really hope you werent in middle of something important you wanted to save)

    All the $ various companies have wasted on this flawed connector they could have ACTUALLY redesigned it by now to actually fix the issue. It is 100% ok to blame nvidia and pci-sig....they arent your god/deity.
    Reply
  • KE=MV2
    It can be easy to not have the connector fully inserted when you believe it is, as this can require some force. The blue color used on the cable pins are useful in ensuring it's fully inserted. You should see ZERO color when properly inserted into the GPU and PSU.
    Reply
  • hotaru251
    KE=MV2 said:
    The blue color used on the cable pins are useful in ensuring it's fully inserted.
    Derbaur has shown that the actual plastic being in doesn't mean its properly connected.

    Just by pulling/pushing on the actual wires can change the contact inside the pin.

    So even if the cable "looks" perfect if you or the tension pull on the actual wire enough it could still not be correct & there is no visible sign for it happening.
    Reply
  • Gururu
    hotaru251 said:
    derbaur has shown that the actual plastic being in doesn't mean its properly connected.

    Just by pulling/pushing on the actual wires can change the contact inside the pin.

    So even if the cable "looks" perfect if you or the tension pull on the actual wire enough it could still not be correct & there is no visible sign for it happening.
    Yes that is pretty insane. They product they make WireView Pro II GPU looks awesome but so expensive to add on to a card after forking over $1000+.
    Reply
  • George³
    Is the question how to sell you solutions to problems? It's not complicated, first we will sell you the problems, and then wll sell the solutions to them.
    Reply
  • thestryker
    Nothing is ever going to fix the inherent problem of there not being enough margin of error. With the 8-pin connectors paying much attention wasn't particularly required because each one is wired separately and it is has a giant margin of error. With the 12VHPWR/2x6 connectors one really needs to be mindful of plugging them in entirely, the degree of cable bending and making sure no cables were pulled on.

    I don't think anyone involved with the design really took into account all of the things that minimized issues in the past. There's nothing particularly alarming from a technical standpoint, but fewer safeguards are a bad thing. Had the connector been limited to 300W I'm not really sure we'd really be talking about it even with the melting that has happened on lower than halo cards.
    Reply
  • Jabberwocky79
    hotaru251 said:
    & more wasted $ on an issue that isnt theirs and doesn't "fix" the issue it just retroactively powers off. (Really hope you werent in middle of something important you wanted to save)

    All the $ various companies have wasted on this flawed connector they could have ACTUALLY redesigned it by now to actually fix the issue. It is 100% ok to blame nvidia and pci-sig....they arent your god/deity.
    1000% this.
    Reply
  • derekullo
    What about any raids the computer is running?

    "Your raid 1's should be fine"
    Reply