Conclusion
Through a bit of happenstance and luck, today, we have the opportunity to look at what should simply be an academic question. However, it is one that has more variables than the “in theory” nature of the question often implies or addresses. That being; what happens when we change the cable length on our power supply in order to fit a case we are using or a cable management scheme we are interested in? Obviously, resistance of a wire is an easily calculable parameter. but much more goes into the overall picture than that. So, with a Corsair RM750x (2018), a pair of its normal cables, and a shorter set of Corsair’s Type 4 cables we set about to see what the real-world impact was of changing the cable length between what would be some realistic cable lengths. There are certain caveats to this, as we are using one power supply and just one set of each cable length, so we can’t reach true statistical power. However, we can get an idea if there appears to be something here that warrants further investigation. So, how did these cable lengths impact the voltage regulation? The efficiency? Does it really matter if it does impact them? Let’s see!
Load Testing
With our Corsair RM750x (2018) that we tested today we did, indeed, find measurable differences in our voltage regulation, as well as efficiency, depending on which cable set we used. However, there are a number of caveats to that beyond just the statistical issue mentioned above.
First, the differences were not very pronounced as seen above. On the 12v rail, the differences in peak changes were just 0.04v overall with the normal (longer) cable set doing worse. Then, depending on the connector examined, at times the longer cable set did better than that or even mixed. When we look at the minor rails we a smaller spread between our two cables sets but a larger fold change. However, that is one of those statistical lies as we are talking about differences on the order of 0.01v to 0.02v and those are practically rounding errors. The only reason they seem relevant is because of how small the magnitude of changes are that we are talking about on those rails.
Second, the variation we see on the 12v rail between connectors can largely be chalked up to the construction of the connectors/cables. Why? Well, when we look at our PCIe cables we see a difference between our normal cable length and our short cable length of only 0.01v and the overall voltage change was actually WORSE on the short cable set. Thus, indicating that it is something other than just cable length at play here. Then the connectors that we see the largest changes on (0.14v for the normal cable set and 0.10v for the short cable set) are the motherboard connectors which only have one 12v lead which makes sense as the PCIe example above has three 12v leads splitting the same load. Given the length of the cables involved, and what we see from our other examples, the differences seen here are unlikely due to the cable length alone (or mostly) and are more likely due to the construction of the connectors and crimping in those cables themselves. That being one of those reasons why the just academic drop in voltage over run-length due to resistance not being the be-all and end all of this discussion.
Third, these numbers don’t mean much. Why? The draw through the motherboard connector is going to be rather small. The draw through the PCIe connectors will be, generally, larger. In the end, the cables that have the higher draw have a different construction that has more leads to accommodate the higher current and, thus, experience less drop in voltage. The one “outlier” (if we can call it that given the terrible sample size) is the EPS connector. This means the changes seen by the most stressed parts of your system is going to be very small. This is coupled with the fact that the onboard VRMs of those components do not usually “notice” a difference between 12.11v and 12.09v as those values are well inside of the ATX12v/EPS specifications which these products are designed to work with.
Lastly, the efficiency. The efficiency is not the same kind of issue as voltage regulation, but it is intertwined with this whole process. In the end, we do indeed see differences between the normal length cables and the short length cables. Again, that difference is not great as it peaks at just over 1%. For users, that difference won’t mean much. Truly, it is merely an interesting data point. For vendors trying to make 80 Plus certification, however, it might. So, why not use shorter cables for your certification testing?
Final Points
Today, the bottom line is fairly simple. If you need to, or want to, change up the cable length you are using on your power supply then don’t worry about it affecting performance in the regards we have looked at today. Go with whatever reasonable length works for you. Nothing we see here today has shown a result that was so eye grabbing to make this a point of concern even with our limited sample size. Yes, there were measurable differences. No, those measurable differences won’t affect your system. That said, one last caveat. This is all pertaining to cables that are first-party factory cables. If you roll your own, or you buy third party cables for this kind of application, then I would say all bets are off. We have seen plenty of examples of cheap and/or knockoff cables being issues in all kinds of applications. So, be aware, play smart, and don’t worry about the reasonable cable length from the factory.