![\"\"](\"https:\/\/www.hwcooling.net\/wp-content\/uploads\/2023\/04\/sapphire-rx-7900-xt-coils-test-01.jpg\")
<\/a><\/p>\nIn the tests, there is only a focus on noise in the exclusive coil sound band (4\u201320 kHz). Traditionally, spectrographs are used for the analysis, which are obtained using the procedure described in the corresponding chapter of the standard graphics card testing methodology<\/a>.<\/p>\n One of the test scenarios represents gaming (CS:GO@2160p), the other 3D rendering in Blender accelerated by OpenCL.<\/p>\n Significantly louder coil whine is experienced during gaming at significantly higher power draw \u2013 328.4 W (Sample 1) and 329.3 W (Sample 2). The reported power draw values are very accurate, as we measure them with current clamps and directly on the PCI Express slot<\/a>. So the power draw in the tested situations scales really perfectly. However, the difference in coil whine is minimal. At 5747 Hz, one sample does have a higher peak than the other, but only by 1.5 dBu, at which you cannot “feel” for sure which of the cards is noisier. Otherwise, the shape of the graph formed by the noise intensity at each frequency is very similar. There are small differences, depending on the differences in the manufacturing of the coils, but to such a small extent that when you measure with a noise meter, which outputs a single value in dBA, you end up with the same number.<\/p>\n A similar situation is repeated at lower power draw (~216 W), where again “Sample 1” is slightly noisier, but again to such a small extent that you have to have extremely good hearing to pick the quieter of the pair of cards. At this load, moreover, the noise level is significantly lower (you can see it in the ratio of blue with noise to white background), but on average you would already measure a difference of one or two tenths of a dBA. One of the samples is a bit noisier especially in the 6.5\u20137.5 kHz range.<\/p>\n The reason for “whining” (although especially for graphics cards we consider “crackling” to be a more accurate term) of the coils is that due to the switching frequency of the MOSFETs of the voltage regulators, the coils oscillate. The impact between their individual threads or between the threads and the ferrite core thus results in various vibrations. These can be different for each coil depending on the slightly different design. There are always more coils in a graphics card than one, and since each one is slightly different, their noise operates over a fairly wide band of sound.<\/p>\n The noise level of the coils can be reduced quite easily, but it would be at the cost of more expensive manufacturing. When someone talks about quieter coils, they must be coils in which the vibration between the threads is more thoroughly suppressed. This is done with an adhesive, the application of which, however, is usually not very thorough. While on one coil with the same electrical parameters the adhesive may indeed be beneficial, on another it performs almost no function and such a coil is significantly noisier. These differences are well seen in devices that do not have the coils shrouded (by a small “box”).<\/p>\n Apparently, in common practice, not much care is taken to minimise coil whine. A handful of demanding users have complaints about noisier coils, but they are clearly a marginal group. Most users are not bothered by noisy coils, so there is room for investigation in this direction. But then it’s a bit sad to look at graphics cards like the Asus RTX 4080 Noctua Edition<\/a>, whose main advantages kind of lose their meaning when combined with traditional coils.<\/p>\n This article is not meant to discourage anyone from looking for graphics cards with quieter coils, they certainly exist. And it’s not even out of the question that you’ll find such in the case of the Sapphire RX 7900 XT Pulse, although the results between the two samples tested end up too close together to make it likely that you’ll come across any audibly better piece.<\/p>\n But of course, after testing ten samples (instead of just two), the variance of the measured values would certainly be wider. However, if one is in opposition with someone and enjoys the quieter coils of their card, then in the end, perhaps the higher threshold of perceived noisiness is what will bias the different interpretations in a subjective evaluation.<\/p>\n English translation and edit by Jozef Dud\u00e1\u0161<\/em><\/p>\n<\/a><\/p>\n<\/a><\/p>\nWhy do coils emit sound?<\/h2>\n
<\/a><\/p>\n