Isolating and filtering: Galvanically isolating USB and analog devices and filtering signals | Part 2 of 2

Preliminary remarks on the USB filters

USB interference filters are designed to improve signal quality over USB connections by reducing noise and electromagnetic interference (EMI). These filters may contain passive components such as ferrite cores or active circuits specifically designed to suppress interference that may affect data transmission. Improving data integrity is just one facet, but it is often audible.

So-called electrical noise can cause errors in the data transmitted via USB. Interference filters then reduce this noise level and thus ensure more reliable data transmission. There is also protection against electromagnetic interference, as PCs and other electronic devices can be exposed to EMI from various sources, including the graphics cards already mentioned several times. Such USB noise filters then help to minimize this interference, reducing the likelihood of malfunctions or data loss.

In addition to galvanic isolation, filters such as the PureClock from Oehlbach or the Jitterbug FMJ from Audioquest can help. I know that the notorious gold cable dealer with the oil in its name doesn’t have the best reputation with many people, but there are also useful products if you look harder. I also find it difficult, but the PureClock at least does almost what it’s supposed to, even if it’s not really cheap at 20 euros for a passive filter. The Jitterbug FMJ is even more expensive, but I can spoil it by saying that it does an audibly better job.

Maybe a piece of gold like this will help after all? That’s exactly what I had hoped for two years ago when I spontaneously ordered an Oehlbach PureClock and wasn’t really sure beforehand whether such a passive stick was pure snake oil or a passable solution. And yes, I have to give the Oehlbach company some credit, at least for this part, even if I am rather skeptical about other products. The stick hasn’t completely eliminated the problem (I don’t think that’s possible passively), but it has at least pushed it so far into the background that I can live with it. And that’s saying something.

The Audioquest Jitterbug FMJ for around 60 euros is much more efficient and better here, but I can no longer present any measurements, because you completely eliminate the remaining residue and really only measure the usual background noise. But whether you want to spend three times as much as a simple passive stick costs is something you have to decide for yourself. I use both, with the jitter bug hanging in front of the Nubert nuPro XS 8000 RC and my multimedia PC table without the galvanic isolation, in which a GeForce RTX 4090 is now sweating. The two Nubert speakers are already electrically isolated. However, three of my USB headphone DACs are suffering from the effects of the PC on the USB. Or rather, they suffered. Not anymore.

Jitter and high-frequency interference

We already know that high-resolution, digital audio signals in particular require a precise and stable clock as well as a clean and interference-free 5V power supply as the basis for high-quality D/A conversion in addition to good signal transmission. In addition, motherboards and their components (graphics card, SSD, hard disk, CPU) produce unpleasant electromagnetic interference and unwanted signal noise. Flippant sayings such as “You can hear what you see” are unfortunately the normal reality and by no means an exception. You can be lucky or unlucky with the strength, everyone is affected differently.

The dreaded clock jitter is also known as “jitter” and the high-frequency interference signals also quickly find their way to the external DAC via the USB cable. The relevance and audibility of jitter in audio and music production as well as in digital communication are important aspects that can affect both the quality and integrity of audio signals. Jitter refers to small, irregular variations in the temporal placement of signal events, particularly in digital signals. These variations can be caused by a variety of factors, including hardware imperfections, transmission errors and processing delays. The impact and perceptibility of jitter depends on several factors, including the nature of the audio material, the quality of the equipment used and the sensitivity of the listener.

External audio hardware can be as great and expensive as it wants to be, not every device has true galvanic ground isolation and really effective filters. This is exactly where the two sticks come into play to eliminate this interference and guarantee perfect synchronization between the USB interface and the converter. At least that’s what the manufacturers say. And since I didn’t want to connect another device with an additional power supply between the PC and the DAC, these sticks were just what I needed. Snake oil or a real solution? In any case, I was curious and simply bought both parts at different times.

If the completely passive PureClock is really supposed to do something, it is probably because components such as common mode chokes, capacitors, resistors and various other coils are installed inside. An empty stick would be pretty cheeky, so I’m sure I won’t come across an empty housing here. Nobody is likely to go that far. But how effective is the whole thing in practice?

Subjective impression and measurement

The starting position is clear. Since my Beyerdynamic A20 only has one analog input compared to the HIFIMAN, I have to connect a proper DAC between the PC and the A20. But then I can actually hear my desktop when I set the level so high that my headphones are properly driven up to full. It chirps like alcoholized cicadas behind a Mexican dice shack. If I now insert the stick in between, the chirping is minimized to an almost non-existent minimum, even with a fully loaded graphics card. At idle, it becomes virtually inaudible. Then I can move the mouse again without getting a nasty tingling sensation in my ears every time. So for now, the PureClock does what it’s supposed to do. I don’t measure the jitter bug again, because there’s nothing left. But how good is the PureClock really when you measure it once and leave out the USB isolator?

First, I set the level of the Beyerdynamic A20 so high that I reach full scale at 600 ohms without distortion becoming visible on the oscilloscope. A full 8 volts RMS or almost 108 mW RMS per channel is of course a revelation here and well suited to gallantly shooting many a headphone into orbit. The eardrums of the insanely daring, of course. But it’s always good to know what’s going on when you want it that way. The 20.3 dBu or 18.11 dBV is a pleasure to take with you. The electrical power level of -0.97 dB is also perfectly acceptable.

Let’s stop the 1 kHz hype train and measure the level of the noise that reaches the amplifier output. In other words, the whole bag of crickets and cicadas from the direction of the PC electronics! The peak voltage is 0.0062 volts (-41.9 dBu, -44.2 dBV), Vrms is 0.0043 (-45.1 dBu, -47.3 dBV) and the converted RMS power is 0.0003 mW, i.e. -75.2 dB as the power level. This is far above what can be accepted in the sum of the individual components of DAC and headphone amplifier and is, above all, something that is extremely audible and perceived as disturbing.

Let’s now put the PureClock in between, which does a visibly good job and can reduce the high-frequency interference components in particular. You can’t get rid of it completely, but it is no longer subjectively audible. The peak voltage shrinks to 0.0024 volts, i.e. -50.2 dBu or -52.4 dBV, while the more important Vrms drops to 0.0017 volts or -.53.2 dBu and -55.4 dBV. The power level at the 0.000005 mW is now -83 dB. You can really leave that alone, because the ideal values from the catalog can only be achieved on the PC with active components between the USB output and the DAC, if at all.

The PureClock from Oehlbach is not a solution to a possible grounding problem, as it does not provide galvanic isolation between the PC and DAC. But that is not what is promised. In the end, it is a very effective, completely passive filter in the USB line. No more, but also no less. The high-frequency noise, which is passed on to the chain via the power supply and ground, can be very effectively attenuated with this stick, but it is not possible to eliminate it completely. But it is completely sufficient for normal everyday use. And if you then use a USB isolator (see previous page), there is complete radio silence anyway. The Jitterbug FMJ is even better here and I can definitely hear the difference on the Nubert speakers when it comes to really good audio material.

And what about the jitter? The joke is that the effect can only be proven if there really is a genuinely jittered signal and the PC (graphics card) unfortunately also delivers what was not ordered. The filter can help to significantly improve the signal integrity so that clock interference can also be largely eliminated. However, the fact that it always “sounds better” per se is practically a legend. However, the lousier the signal source and the more electrical noise and EMI noise from the PC, the better it sounds.

On a home PC with a fat power supply and a powerful graphics card, however, something like this can really be the last resort in combination with the USB isolator. So you can look at it however you like, the 20 or 60 euros are quite appropriate for the product, even if they are not really cheap.

Summary

The use of galvanic isolation and USB noise filters in PC systems is based on the application of proven electrical and electromagnetic protection measures. These technologies help to improve the reliability and performance of USB connections by protecting the systems from harmful electrical and electromagnetic influences. Their importance is undeniable, especially in environments with a high risk of electrical interference or in applications where data integrity is critical. By applying sound scientific principles and engineering techniques, galvanic isolation and USB noise filters are essential tools in the arsenal of computer engineering that are far from any form of “voodoo”.

In the case of the so-called USB isolators, I can no longer hear the ground and EMI-induced noise at all; with the interference filters, the sound usually improves audibly if there was previously jitter and electrical noise. Of course, there are many more causes for a disturbed relationship with the acoustic environment, but this is where you can really start, because it is almost always worthwhile. Voodoo? Not at all. However, it’s all a question of trained hearing, your own standards and your monetary war chest in the fight against acoustic pollution.

And even mice and keyboards can benefit from a cleanly decoupled USB interface when it comes to clean signals. What use is an 8000 Hz polling rate if the data arrives later or even distorted due to necessary error corrections? Here too, of course, the sensitive perception of each individual is very different. But especially in e-sports, I wouldn’t compete without USB isolation and a jitter bug if current peaks like the ones we have unfortunately measured often enough occur in the PC.