Who you gonna call? The Blur Busters! For Everything Better Than 60Hz™
https://forums.blurbusters.com/
please can I ask you how bad was your issue? I'm not trying to diminish your problem but from what I saw until now, there are many different levels of this problem, some people have way worse symptoms than others so I do believe that your problem might be fully fixed because your interference was lower. so thats why I want to ask, was you able to play at high or decent level before the psu change? at what level could you play before? do you have any comparison or measurement thats possible to see the difference in your performance?Roxtar wrote: ↑29 Mar 2025, 13:43I'm just replying here because today I got a 1000w psu which is totally an overkill for my rig, and my floaty aim and others issues disappeared instantly, I must say as the OP did, that this might not be the solution for every case of floaty aim, but at least for me it was.
internetexplorer4 wrote: ↑30 Mar 2025, 16:29please can I ask you how bad was your issue? I'm not trying to diminish your problem but from what I saw until now, there are many different levels of this problem, some people have way worse symptoms than others so I do believe that your problem might be fully fixed because your interference was lower. so thats why I want to ask, was you able to play at high or decent level before the psu change? at what level could you play before? do you have any comparison or measurement thats possible to see the difference in your performance?Roxtar wrote: ↑29 Mar 2025, 13:43I'm just replying here because today I got a 1000w psu which is totally an overkill for my rig, and my floaty aim and others issues disappeared instantly, I must say as the OP did, that this might not be the solution for every case of floaty aim, but at least for me it was.
So in summary, you fixed this problem with new PSU with more Voltage?ChristophSmaul1337 wrote: ↑16 Apr 2025, 03:09TL;DR at the bottom!
Good day everybody!
To start this post off, I want to clarify that everything is still working. The "bigger" (higher wattage) power supply still is providing me with clean, no-BS gameplay throughout. However, I have done more tests and noticed that the actual wattage of the power supply isn't the real reason my setup starts to work with this new unit. I can now use any power supply with enough wattage, as I should be able to, and with my workaround, all of them work fine. If you want to get straight to the workaround, go to the TL;DR at the bottom.
One thing I've always kept in the back of my head is the possibility of a ground loop somewhere. In my case, at first glance it could make sense that there would be a ground loop, because of the monitor I'm using. It has a 3-prong power plug which goes straight into the monitor. It doesn't feature an external power brick like some monitors do, so the PSU for the monitor is baked right into it. When taking continuity measurements with a multimeter, it clearly shows that there's continuity between any of the monitor's exposed metalwork and the earth pin within the power outlet.
And here is the possible ground loop happening: From the power strip, to the computer's power supply, to the graphics card via the PCIe slot and power connectors as well as the computer case itself, over the DisplayPort cable to the monitor, and back to the power strip. This might cause interference, as many audio technicians will attest to.
The attentative reader might already have spotted a problem with this logic though, and it's exactly what I've thought to be the case as well. What you're hearing when a ground loop is happening is the flow of current on the signal wire when there should not be current flowing there. In order for current to flow between two points, there needs to be a difference in potential. However, when you're plugging in the monitor and the computer into the same power strip, their grounds are on the exact same potential. Sure enough, the ground loop is actually there and the electrical connection is made, but there should not be current flowing as the device's ground potentials are equal. At least that's what I thought.
I've gotten the good old clamp meter out recently and tried to measure currents on wires where current should definitely not be found on. Sure enough, I could measure a current on the DisplayPort cable running between the GPU and the monitor. This gave me a hint that there is some sort of ground loop current flowing here. To sanity check, I've connected another monitor which has an external power brick and shows no continuity to the power strip's earth pin. Sure enough, when using this monitor, the current on the DisplayPort cable is gone. Easy enough fix, ground loop eliminated.
In order to test if this supposed ground loop is to blame for all of the BS I've been experiencing, I've connected one of the "problematic" power supplies and testet the gameplay. As you might have guessed, the problems were still there even after eliminating the ground loop. Back to square one.
Going back to the grounded monitor, I've looked at the amount of current in more detail. When idling on the desktop, the clamp meter hardly picked up any current and most of the time displayed 0A. However, when starting a game, the value would rise with more GPU loading. There is a clear correlation between GPU load and the amount of current that's on the DisplayPort cable.
At first, I was disappointed that I didn't find anything of relevance. After all, this current is also present when using the SuperFlower PSU, which shows no problems. I concluded that the presence of this current, while probably still in the wrong place, can't be responsible for the problems we're having.
Nonetheless, I've looked further into the reaon for this current. GPU's come with a I/O shield (I'm gonna call it a "bracket", as I lack the correct vocabulary) that you screw into the case. When using a case made out of metal, it should create an electrical connection between the GPU bracket and the power supply, as the PSU should also be screwed into the case, ideally with metal screws. This should create equal potential between the PSU screws, the case and the GPU bracket. In reality, this doesn't happen. Depending on the load on the GPU, there is a voltage difference between the GPU bracket and the PSU screws, and the differential in voltage increases with higher GPU loading. I've measured up to 10 millivolts.
This is exactly where the current is coming from: The monitor and the PSU are on the same potential, as they're plugged into the same power strip. DMM measurements corroborate this. The GPU, its bracket in particular where the DP cable is making contact, is at a slightly higher potential. This enables current to flow, and with increased loading, more current can flow as the difference in potential gets larger.
Another thing the attentative reader might have spotted here. A power supply should have 2 different types of "ground". There should be the "ground" used for the DC side of the PSU, where the return current flows. As many know, electrical current isn't "consumed", because it can't be. Current going into a device needs to return to the source in full, it can't be created or destroyed. In must equal out. When sending current to the GPU on the 12v rail, it shall return on the DC "ground" wires.
Another use of a "ground" is the safety ground. A computer with it's metal casing is a class 1 appliance. This means that it's not double insulated, so it needs a safety ground, a 3-prong power cord. All metal parts of the computer must be connected to this safety ground.
In an ideal world, the safety ground, which the metal case is connected to for example, should not be used as the DC ground. Ideally, these two types of ground should be entirely separated from one another. In reality, this doesn't happen. Most power supplies, in fact all of which I'm currently owning and probably the vast majority of consumer grade PSUs, bond these two grounds together. This is the perfect explanation of the current that I'm observing on the DP cable:
When sending, let's say for example 300W of power (25A) over the 12v rail to the GPU, the same 25A need to return to the source. In an ideal world, all of the 25A arriving through the two 8-pin PCIe power connectors would return on the ground wires of the 8-pins. In practice, this doesn't happen, as there are multiple paths back to the source:
1. The PCIe power connector
2. The PCIe slot through the CPU EPS power pins
3. The PCIe slot through the 24-pin ATX plug
4. The case, if made of metal, provides a ground path
5. The monitor via the DP cable, power socket and PE wire
The current takes all paths back to ground in proportion to their resistance. If the PSU for example is worn out and has weak modular cable connections, the resistance might be substantially higher than for example through the motherboard & CPU EPS or 24-pin cable. This way, the majority of the return current might not return on the 8-pins, but take a different path entirely, for example through the PCIe slot and the CPU EPS power connectors.
Current flowing through the motherboard, instead of the PCIe power connectors, will create interference on the motherboard. I don't have the right measurement tools at hand to test for interference, so surely all of this is conjecture and baseless rambling without actual scientific proof or real measurements, right? Well, not so fast. Turns out this exact problem has been discussed in realtion to the melting 600W connectors on recent NVIDIA cards. There's an article on igor'sLAB about this exact problem. Here it is: https://www.igorslab.de/en/the-imbalanc ... stigative/
This article asks the question: Why are the 12v pins melting, but the ground pins are almost always fully intact? It concludes that the current going into the GPU doesn't take the same path back out and it also points out the possible problems this can come with. You can see an overview of the current flowing on page 3 of the article. This is a real thing and Igor has clear measurements of this happening. We can be 100% sure this exists and is a problem.
I wanted to test how all of my PSUs would behave if I relieve the above mentioned problem a little bit. My workaround focused on creating additional low-resistance paths back from the graphics card to the PSU, so the current doesn't have to take the route through the mainboard. To achieve this, I took a heavy-duty copper wire (probably 8 AWG if I had to take a guess) and placed one side of the wire underneath the top part of the bracket of the GPU, where the screw goes in. Then, I screwed the GPU down to the case as normal, which secures the wire in place and makes excellent electrical contact. For the PSU side, I simply took an old random piece of metal with a hole in it, jammed the other end of the wire between it and the case and screwed the PSU in as usual. It's a crude solution, but it sure as heck works.
However you want to do it, the goal is to create a wired connection between the GPU bracket and the PSU's screws. The thicker the wire the better, as it can carry more current and is of lower resistance. After doing this, the difference is stunning. No PSU is showing any problems any more and I can use my computer with any PSU now. All of them feel excellent, the desync is completely gone and all of the weirdness disappeared.
In my opinion, this basically confirms what the Chief said in the EMI's introductory post: These return currents, going through the motherboard where they should ideally not be going through, create interference. In the past, this hasn't been much of an issue, mainly because if you told someone as early as 10 years ago that one day a GPU will get a 600W power connector and can draw way more than that in practice, he would've laughed you out the building. Also, back in the day, error correction mechanisms haven't been as advanced as they are today, and if there was too much interference, the computer would just malfunction and turn itself off. Today, there's much more error correction going on and everytime errors have to be corrected, it takes time. As the article from Igor states, this messes with USB signal integrity, it can likely also screw with the Ethernet connection and also with sound when using an internal sound card.
Because this is so complex and so many variables are in play, the most simple thing can seemingly "fix" the problem. One USB port might be influenced more than anther, so plugging the mouse into another USB port might change the feeling entirely. It depends on so many factors, it would take days to write all of them down.
Some users have reported that undervolting their components helps with the problem. It makes sense, because lower voltage means lower current, less of which needs to be returned to the source in the first place, alleviating the problem. Taking the PC to another place to test it there? Maybe the other place actually has a less noisy ground and that makes all the difference, maybe you just so happened to plug the mouse into a different USB port that's not affected as much. A new PSU might make the feeling better, because the connectors are still making excellent contact, especially the PCIe power connector. As the connectors wear out and the PSU gets older, this can easily change and the resistance on the PCIe power connector increases, sending more current through the motherboard. Motherboard topology changes the amount of current, a high-end motherboard might have a well-built and thick ground plane, sending more current through it, while low-end or midrange boards might not have that beefy ground plane, increasing resistance and sending current elsewhere. Has the manufacturer of that top-of-the-line motherboard thought about it and designed the board with the interference in mind? How is the motherboard heating up? More heat means more resistance...
As you can probably tell, there are SO MANY variabes in play to the point where it's basically a gamble if you'll get a combination of parts that will work together. However, from the tests I've done so far, using my workaround will always alleviate the problem and gameplay becomes clean every time.
Can this explain every scenario? To my knowledge, no, but there are many questions I can't answer as I lack the understanding of it. I don't yet understand how this can be of relevancy in cases where the feeling changes with time of day. It could be related, maybe the safety ground is noisy or something, but if and how that influences the return current over the computer's case I really don't know. Electrical engineers can answer this way better than I ever could.
I think it's worth giving it a try if you suffer from problems. If you have a copper wire around, it's even free. Good luck to everyone who tries. It's certainly not a good solution, but as Igor points out in his article, PC manufacturers have to rethink how PCs are made and how the electricity is flowing within them.
For me, this will stay in place indefinitely, even though the SuperFlower PSU works great even without it. As the PSU is still relatively new, the PCIe 8-pins are still making excellent contact, but as soon as they wear out a bit, the problem will likely come right back.
TLDR: The majority of the return current from your GPU doesn't go through the PCIe power connectors but takes different paths back to ground depending on each path's resistance. More current through the motherboard creates interference, which will screw with data integrity. There's an article from igorsLAB linked above which goes into a lot of detail, including measurements. A workaround that significantly improves the problems on any of my systems is to create a low-resistance path from the GPU to the PSU directly. I've done it with a beefy copper cable from the GPU's I/O bracket to the PSU's screws inside a fully metal case.
Yes and no. A new PSU "fixed" the problem, but it's not about the Voltage or Wattage. The reason I think my problem was "fixed" by a new PSU is because new units have pristine connectors on both sides (PSU and GPU). This creates a low resistance path back to the PSU which reduces the current flowing through the motherboard. As a PSU ages, the connectors wear out ever so slightly, increasing resistance and therefore sending more current through the mainboard, thus creating more interference. In my opinion it's this interference that's causing problems for me.
Because I don't think you can do much about it. You definitely can't do anything about this problem if you wanted to actually fix it. It's a design flaw as far as I understand. It's up to the manufacturers of PC components to design their PCBs in such a way that this doesn't happen in the first place. What I'm doing with a wire, and what a better connection would do is to be a workaround. Fixing is only possible by attacking the root cause, and that's nothing an end-user can do on their own.
Sure. Here you go! https://imgur.com/a/ruXGZWr