arestavo

kurtley

jankerson
 
Your T2 1000 will be fine.

Oh good. Thanks, jankerson. So, the issues are PSU related then. Good to know.

Some are - and some are cabling (PCIE power or sometimes even other cables, due to increased total system power drawn), some are botched MCU firmware updates (likely other programs interfering, like AV), some are driver related (mostly resolved), some are video cable related (mostly HDMI 2.1 cables that can't handle the bandwidth, but some are DP cables that don't work right), and some are cards that really do poop the bed seemingly randomly as with any product in history (and that's why I get EVGA cards for their warranty - based in the US no less (even if it is the dictatorship of california)).

jacoffey85

jankerson
Right.
 
The transient power spikes only last for a few microseconds, the real issue is they can be twice what the measured power would be. So for simple example a 400W draw could be 800W transient spike for a few milliseconds.  
 
So as it was explained to me by someone who designs PSUs the issue is on the secondary side of the PSU. It's how big the caps are (rated at) and or how they are used. If they can't handle that load the PSU will simply shut down to protect itself.
 

The transient power spikes you’re talking about in the other thread the user posted with the power graph aren’t coming from the PSU. In that aspect transient power spikes aren’t really a thing, at least from the hardware side. What that person was probably talking about is transient voltage spikes which isn’t the same thing in this context. Transient voltages spikes come from collapsing magnetic fields around capacitors and even more so in inductors.

The reason the power spikes isn’t a thing on the hardware side is because an electromagnet field collapses when current is removed. The collapse causes a voltage spike proportional to the strength of the field, but it’s very little power when this happens because there’s little to no current. Transformers work off of this principle. Even though it’s low power, the voltage can cause silicon in solid state devices to degrade quickly. I have to load balance frequency drives and electrical motors sometimes to eliminate these voltage spikes or they’ll damage the insulation in motors.

I read in depth the article you talked about by Aris and the part you’re talking about with small voltage spikes on the voltage ripple after the rectifying phase. He mentions transient voltage spikes during the first stage of rectifying which is right. Voltage spike doesn’t mean power spike, and also that’s just during the first stage of the the rectifier. It goes through a couple more filters and had more suppression afterwards. The only part that’ll truly have an effect on the load in his breakdown is the LLC Resonant Filter right before the last transformer at the output. It will run capacitive, neutral, or inductive depending on the demand of the load. Even then, it’s still electrical isolated through an isolation transformer. His guide is very good, but still is only as good as the person interpreting it.

ATX specification require the voltage to stay within a very tight range and current draw is going to determined by the load. The erratic power draw is coming from the card and either how it is controlling its power limit with the bios, or its coming from the applications you’re running.

jacoffey85

jankerson

jacoffey85

jankerson
Right.

The transient power spikes only last for a few microseconds, the real issue is they can be twice what the measured power would be. So for simple example a 400W draw could be 800W transient spike for a few milliseconds.  

So as it was explained to me by someone who designs PSUs the issue is on the secondary side of the PSU. It's how big the caps are (rated at) and or how they are used. If they can't handle that load the PSU will simply shut down to protect itself.

The transient power spikes you’re talking about in the other thread the user posted with the power graph aren’t coming from the PSU. In that aspect transient power spikes aren’t really a thing, at least from the hardware side. What that person was probably talking about is transient voltage spikes which isn’t the same thing in this context. Transient voltages spikes come from collapsing magnetic fields around capacitors and even more so in inductors.

The reason the power spikes isn’t a thing on the hardware side is because an electromagnet field collapses when current is removed. The collapse causes a voltage spike proportional to the strength of the field, but it’s very little power when this happens because there’s little to no current. Transformers work off of this principle. Even though it’s low power, the voltage can cause silicon in solid state devices to degrade quickly. I have to load balance frequency drives and electrical motors sometimes to eliminate these voltage spikes or they’ll damage the insulation in motors.

I read in depth the article you talked about by Aris and the part you’re talking about with small voltage spikes on the voltage ripple after the rectifying phase. He mentions transient voltage spikes during the first stage of rectifying which is right. Voltage spike doesn’t mean power spike, and also that’s just during the first stage of the the rectifier. It goes through a couple more filters and had more suppression afterwards. The only part that’ll truly have an effect on the load in his breakdown is the LLC Resonant Filter right before the last transformer at the output. It will run capacitive, neutral, or inductive depending on the demand of the load. Even then, it’s still electrical isolated through an isolation transformer. His guide is very good, but still is only as good as the person interpreting it.

ATX specification require the voltage to stay within a very tight range and current draw is going to determined by the load. The erratic power draw is coming from the card and either how it is controlling its power limit with the bios, or its coming from the applications you’re running.

 
Or for those of us who aren't engineers it comes down to PSU design in the end really.
 
Just to keep it fairly simple.
 
 

I can agree with that, and you’re completely right, PSU quality matters. Cheap PSU’s don’t have the LLC resonant filter at the end of the circuit and puts more strain on the voltage regulation side of the GPU.