Saturday, 19 March 2022

MEASUREMENTS: Topping HS01 - USB 2.0 Isolator & Ground Loop Eliminator (and a listen to Santana's "Blessings and Miracles")

Hey everyone, as you might be aware, I'm not a fan of USB "doohickeys" in general use. You know, all those USB boxes that are supposed to clean up your signal, or maybe reduce jitter, or "regenerate" this and that. Especially if you're not sure if you actually need it; some of these devices cost hundreds of dollars. On the lower price end, I had a look at the AudioQuest JitterBug FMJ last year which was a rather disappointing product.

Having said this, there are devices that can provide benefits like USB isolation for things like DACs. For example, the USB1 inexpensive ADuM4160 devices like this Nobsound ADuM4160. The problem is that this is only operating up to USB1 12Mbps "full speed". Furthermore, one would be limited to 24/96 performance even if the device is able to fall back and negotiate as USB Audio Class 1.

Enter the Topping HS01 (~US$70), a little black metal USB2.0 box advertised as providing galvanic isolation (to 1kVrms), for data and power lines with USB-B (to computer), and USB-A (to audio device) ends. Furthermore, there's a USB-C plug for devices that need extra power.

[Note: This device is not compatible with USB1.0/1.1. So don't be plugging in keyboards and mice to this. Non-audio devices like USB2 memory sticks and even my smart phone were fine. I have not tried plugging a USB2 hub up to this point so there might be issues there.]

There's not much else in the generic-looking off-white Topping box other than the short USB-B to USB-A cable, a brief user manual and Topping pamphlet with advertising info for their other devices like DACs, and warranty terms.

The little metal box measures a small 2.3 x 5.3 x 0.9cm. Here's a peek at the bottom and we can see the USB-A and USB-C connectors better:

I see that Topping also advertises an insignificant 0.29μs latency added by this device.

The enclosure appears to be robust. No rattling. And I generally like this kind of a clean design with detachable cables than something like the JitterBug with its USB connector sticking out on one end. The unit does warm up in use but doesn't get hot.

To be honest guys and gals, I don't worry about "noise" in the USB system itself when we add a good modern DACs in typical sound systems. Today's good DACs already output signals with low noise floor, great dynamic range, very low distortion and some already implement galvanic isolation. The only times I've run into issues are when doing measurements with both my DAC and ADC plugged in to the same computer which can result in ground loops (as discussed with the NobSound USB1 isolator article). In principle, this Topping box should be able to remove such issues while operating at USB2.0 / UAC2 speed.

So then, let's set this up with the Topping D10s DAC and E1DA Cosmos ADC with the HS01 isolator/ground loop breaker in the chain for testing (this is similar to the NobSound USB1 isolator article although there I used the RME ADI-2 Pro FS ADC):

We'll use the Topping D10s DAC's single-ended output (running linear phase sharp filter firmware) to demonstrate the noise issue since typically balanced devices (like the Topping D10 Balanced) will not have problems.

Here's a picture of the testbench with the Topping D10s and Cosmos ADC plugged in to the NUC:

Raspberry Pi 4 streamer with touchscreen right edge. Used in test as well.

Notice that I'm using the 2 front USB outputs of the NUC; one going to the Topping D10s DAC, the other to the E1DA Cosmos ADC. Both devices are USB powered. A standard shielded RCA cable connects the DAC to ADC. Dangling off the USB connector is the Topping HS01 isolator which the DAC is plugged into.

Also notice that I'm not even connecting a separate power input to the HS01 for the D10s. The Topping D10s DAC is very efficient. As per the manual, a simple 2-prong 5V SMPS is preferred if needed - "Linear power supplies and other earthed power supplies are not recommended".

Using the RightMark Audio Analyzer with its usual suite of results - running at USB2 speed of 24/192:

The first 2 results on the left are with the Topping D10s connected to the computer USB output without and with the HS01 isolator. The result on the far right is with the DAC plugged into the Raspberry Pi 4 "Touch" player running Volumio. Looking at the numbers, we can see that the Topping HS01 Isolator indeed worked to reduce the distortions when both the DAC and ADC are connected to the computer. In fact, notice that with the isolator in place, results are basically identical to using a separate Pi 4.

Some graphs demonstrating in greater detail:

Nice. As expected, the frequency response will not be affected. The take home message is that with the isolator in place, elevated noise and distortion resulting from the multi-USB device loopback to the computer has been eliminated. In fact, if we look at the noise floor graph, the HS01 was even relatively free of 60Hz mains hum compared to the Raspberry Pi 4 + Topping D10s combination.

Here's a look at the 1kHz THD+N from the Topping D10s, connected to NUC +/- Topping HS01:

Left channel only. E1DA Cosmos ADC running in stereo mode.

Beautiful reduction in the noise floor and distortions. Nice to see that the 8kHz USB PHY noise which often shows up on measurements like this has been controlled with the isolator. For completeness, I plugged in a Samsung SMPS 5V phone charger power supply to double check that external power will not cause problems and in fact the THD+N was even slightly better.

Note that we're running at 192kHz samplerate and there are no noise issues out to almost 100kHz; even with the switching power supply (for those who still worry that SMPS can cause noise issues of some sort anywhere close to the audible range).

Finally, a look at the J-Test:

We can see the high noise level and sideband issue without isolator. Once the Topping HS01 is added on, no problem and the graphs are equivalent between the isolated DAC on the NUC and when playing through the Raspberry Pi 4.

Subjective Sound & Conclusions

Among other albums, I've been listening to some Dayglow this past week - Fuzzybrain (2019), and Harmony House (2021); fun alternative/pop recordings from front man Sloan Struble. Have a listen to "Can I Call You Tonight?" for a taste of the kind of music I'm talking about. While the recordings are a bit compressed (average DR10 and DR7 respectively), it's not too terrible.

I've also been having fun listening to the new Michael Giacchino soundtrack for The Batman (2022, DR9). Love the foreboding but in ways sweet "Sonata in Darkness". The album isn't particularly high resolution with higher noise floor so there's no need for 24-bits.

As expected, whether I have the Topping HS01 in line with the USB DAC or not really doesn't make a difference soundwise in the sound room where there are no ground loop issues.

While it might be nice to add galvanic isolation between USB devices (some DACs already implement isolation), in practice, unless you know that there's an issue, obviously there's no reason to get worried or excited! Remember that "Bits Are Bits" with modern digital audio gear. No need to lose touch with reality in claiming any different unless one can show evidence of concern with specific gear.

Bottom line: The Topping HS01 is a no-nonsense USB 2.0 high-speed (480Mbps) isolator / ground loop breaker that works as expected. It's a great little tool for my setup when testing things like a USB dongle plugged directly into my computer and connected to the ADC for measurements. It's built well, has simple plug-and-play functionality (transparent - does not show up in Device Manager), "sounds" great as it will improve noise when encountering ground loops or noise on the data lines; otherwise, does not impede sound quality or affect jitter (compared to say a Raspberry Pi streamer).

There are other devices out there worth considering. For example the iFi iDefender+ ($US60-70) is advertised to reduce/remove ground loops [some concerns about this product - see comments], works for USB 3.0 (5Gbps) devices, but not galvanic isolated on the data lines as far as I can tell. For galvanic isolation, the iFi nano iGalvanic3.0 (~US$350) looks like it will do the job also with USB3 speed. Intona's USB2 isolator (~US$280) has been available for years as well.

The Topping HS01 therefore is one USB2.0 doohickey I can recommend for those of you that need USB isolation without much expense.

--------------------

Guys and gals, these are the kinds of results I would have loved to see with something like the AudioQuest JitterBug FMJ when I had a look at it last year. Alas, it could not deliver despite claims of "filtering" and using "jitter" in the name is inappropriate since it hints at having an effect (it does not). It's a reminder that companies like AudioQuest and their advertising can say all kinds of things but unless there is independent measurements and verification to actually clarify what the device actually does and in what circumstances, it's hard to have full "faith" in the claims.

Although I'm singling out AudioQuest here due to the JitterBug, I am reminded of issues with this company over the years: ridiculous "audiophile" ethernet cables, HDMI cable scamquestionable utility of their Niagara power conditioners (that AQ promo video with Garth Powell is hilarious!).

Consumers IMO need to demand that companies like this engage in the objective/measurement process and be accountable with demonstrations of efficacy. Over the decades, I think a lot of money has been spent on things that practically make no difference to sound quality. 

--------------------

On the music front this week, thanks Giraffe recently for pointing out that Tears For Fears has a new album out called The Tipping Point. I've had a listen to the stream and enjoy it enough to want to add to my music collection. I have an order in for the Blu-Ray with multichannel 5.1 and Dolby Atmos mix done by Steve Wilson; hoping the sound quality will be a significant step up since Wilson has been known to maintain good dynamic range and sensible use of surround channels in his work. Will have a good listen when it arrives!

Also had a listen to Santana's new one Blessings and Miracles (2021, DR6). It was nice hearing Steve Winwood on "Whiter Shade of Pale" and Chris Stapleton did a great job on "Joy". Unfortunately, dynamic range compression on this album is quite heavy handed.  The other issue is that it's trying too hard to get onto the pop charts (especially "Move" and "She's Fire") to the point where it sounds like a compilation album with Santana as a session musician contributing some guitar work much of the time, interspersed with headliner tracks like "Santana Celebration" or "Song for Cindy".

Tracks like "Santana Celebration" and "Rumbalero" could have been so much more engaging if there were some subtlety allowed to infuse into the layers of instrumentation. Instead, the levels were pushed to 11/10 as you can see - absolutely no need to do this in the 2020's:

It sounds chaotic, somewhat annoying, and ultimately dull because listeners will shift their attention elsewhere as a result. IMO, this is not how enduring albums are created. I wish the studio audio engineers could at some point change this way of doing things, who knows, maybe it's the record companies asking for quality like this? I certainly hope Santana himself doesn't have a desire for his album to sound like this!

One more thing you can see from the image above - don't bother with the hi-res 24/96 version; at best this is an upsampled 16/48 recording! To still see upsampled stuff like this in the "hi-res" audio catalogue is a bit of a scam in the third decade now of the 21st Century.

Classic Santana albums like Santana, Abraxas and Caravanserai were sonic gems. Sound quality with Santana albums have been forgettable for the last two decades. And sadly, neither great blessings nor remotely miraculous content was found on this album (certainly not from the standpoint of sound quality!).

Alright, back to Spring Break with the family. ;-) Hope you're all doing well and enjoying the music!

--------------------

Addendum: March 22, 2022

As per the discussion in comments to Vladimir, here are the results of a balanced DAC, the Topping D10 Balanced. The set-up is identical to the picture of the work bench above except for the balanced DAC and cable connecting to the ADC of course:

As expected, balanced connections are free of the noise from ground loops. The USB isolator made no difference in this set-up.

27 comments:

  1. I purchased a device similar to this (iFi iDefender+ powered by an Apple USB wall wart) to solve a noisy PC USB port powering my Sennheiser GSX 1000. Apparently Sennheiser didn't have adequate filtering built in, so this was the only solution. About the same price, but I'd much rather had bought a Topping branded one. Thanks for the review!

    ReplyDelete
    Replies
    1. Excellent Rabbit,
      Glad to hear that it worked out as expected. Too bad the Sennheiser GSX 1000 is noisy though!

      Wondering how good the 7.1 --> Binaural processing sounds over your headphones?

      Delete
    2. Not my experience at all with the iFi device, though I was not using it to provide better power so I can't comment on that aspect.

      I occasionally run in to strange ground loop problems (made worse by a big FM transmitter nearby) on my bench between my desktop PC and test gear. I usually switch over to a laptop running on battery as that can sometimes help but there are times where I want the desktop.

      The iFi is advertised as eliminating ground loops. Not sure what planet their engineers are from, but you can not break a ground loop with a direct ground to ground connection from input to output. Total false advertising.

      I cracked the case open just to make sure. There are some parts in the thing but I just haven't been motivated enough yet to reverse engineer it. I wouldn't be surprised to find the parts are just for show.

      Delete
    3. Binaural 7.1 is the only reason I use it. I've tried everything else and it's the best for me. Tried Dolby Atmos and DTS:X on Windows, CMSS-3D (still the best for EAX games), Out Of Your Head HRTF, and Dolby Headphone on Asus Xonar cards.

      Best compatibility as it always shows as a 7.1 channel device in Windows and the most convincing "behind the head" sensation for me.

      Delete
    4. Should also mention that the iFi device did nothing for noise unless connected to external power.

      Delete
    5. Thanks for discussions guys.

      In general, be careful about lifting the ground pin on USB. For example over the years, people have talked about doing this with USB cables - for example:
      https://www.head-fi.org/threads/ground-lifting-a-usb-cable.131544/

      You need ground for USB handshaking, but even after that process, not a good idea to just remove the ground connection.

      In the event of large voltage differentials between your components, if all ground connections are severed, the data pins then end up "seeing" this large differential and if severe enough, will result in the USB receiver chip getting fried. I have it on good authority that manufacturers have seen many devices damaged in this way over the years due to folks tweaking USB cables and doing stuff like putting tape over ground pins and such...

      That's disturbing about your findings Brewster. Looks like we need to be careful with the iFi Defender device if all it's doing is playing with the ground connection while not providing true isolation.

      Delete
  2. Hi Archi

    Great report. Will order this device, to have one, just in case I need it. Thanks.

    And yes, spring is coming. All the best,

    Juergen

    ReplyDelete
    Replies
    1. Thanks for the note Juergen... Hope you're doing well and looking forward to the spring!

      Just a note; I received some information from a friend that new generation high-speed USB isolator devices seem to be coming our way:
      TI ISOUSB211

      Analog Devices ADUM4165

      I'm guessing USB2.0 isolation will become more and more common in the days ahead at reasonable cost with support for "low" and "full" speed USB devices.

      Delete
  3. Thank you for the test. It would be great to test it with a balanced DAC. To be sure that there is no need to waste money.

    ReplyDelete
    Replies
    1. Sure Vladimir,
      Let me hook up the same set-up with the balanced Topping D10 Balanced and I suspect this would show no change to the noise level on a simple 1kHz signal for example even without the isolator... Will try this tonight.

      Delete
    2. There you go Vladimir.

      "Proof of concept"... No difference with the isolator.

      Delete
    3. Thank you Archimago, you have saved my money.

      Delete
  4. Does this work on usb interfaces too (signal in as well as out)? Also, if my interface or dac has balanced outputs, will this still be beneficial?

    ReplyDelete
    Replies
    1. Hi P.S.,
      As above, usually you should be OK with balanced devices.

      You can certainly use this isolator with your USB interfaces... Obviously when it comes to benefit, it'll depend on your set-up and whether you currently have issues!

      Delete
    2. Thanks for the reply! I have a few single-ended dacs and a Motu M2, so this article of yours really piqued my interest. Can't wait to test it on them :D

      Delete
  5. Intona is the product of choice... even now USB 3. Ground loop isolation is one thing, but deals/helps better on star grounding...

    ReplyDelete
    Replies
    1. Thanks. Will keep an eye on one of these as well!

      Delete
  6. Archimago, I discovered your blog a few seeks ago. I love it and have learned a lot. Thanks for writing it.

    Based on this article, I bought the HS01. I've had a noise issue with my computer speakers for years. The HS01 fixed it.

    I have two computers - my personal tower PC and my work laptop. I connect both to a single pair of AudioEngine A2 speakers. I use a KVM to share the keyboard/monitor/mouse.

    The tower PC connects via mobo USB to an AudioEngine D1 DAC, which then connects via L/R RCA to the speaker A input. The laptop connects via 3.5mm to 3.5mm directly to the speaker B input.

    Whenever both computers are connected to the speakers, I get moderate noise. I've dealt with it by unplugging the speaker connection for the computer I wasn't using at the moment. That pushed the noise level low enough that I didn't here it if I didn't crank up the speakers.

    The noise had been louder when I am on the laptop, so I focused my debug there over the years. New 3.5 mm cable. Nope. External USB sound card for the laptop. Nope. New USB cable. Nope. And recently, a brand new shielded 3.5 mm Blue Jean cable costing about $40. Nope.

    I read your article and thought ground loop. So I bought the HS01, put in inline between the tower PC and the D1 DAC. It worked. Now I can keep both computers connected all the time. I'm happy. I celebrated by playing some Bootsy Collins - loudly, without noise. :-)

    ReplyDelete
    Replies
    1. Well, I spoke too soon. Yes, the HS01 removes the ground loop, but my listening this afternoon was with my laptop connected to the A2 speakers via the 3.5mm cable.

      Tonight I tried listening to my tower PC, connected via USB to the D1 DAC. Unfortunately, when the HS01 is inline between my tower PC and the D1, Windows 10 does not see the D1. As soon as I remove the HS01, the D1 shows up in Device Manager.

      I then added power to the HS01 side port. The red LED "breathed", showing the connection was good through the HS01, but Windows still didn't see the D1.

      I'll have to contact both Topping and AudioEngine to see what's up.

      Delete
    2. Hey _cl.
      Could it be that the AudioEngine D1 is a UAC1 device (running at USB1.1 speed)? I believe it's limited to 24/96 over USB (although it might handle 24/192 over TosLink).

      If that's the case, you can just use the inexpensive ASuM4160 devices like this:
      http://archimago.blogspot.com/2021/09/measurements-usb-isolation-with.html

      I believe likewise the AudioQuest Dragonfly DACs are also UAC1 devices.

      Delete
    3. Thanks, Archimago. You solved the puzzle.

      My D1 is from 2012. Various reviews from that time say it uses the TI1020B USB controller. I can't find a data sheet on that, but others have equated this to the TAS1020B whose data manual does say it is USB 1.1 with UAC1.

      Topping's web site says the HS01 doesn't support USB 1.1. Well, there you go.

      What's weird is Topping's tech support never mentioned to me the problem might be the device is USB 1.1.

      Furthermore, I did connect the HS01 to my HP printer (2019). The printer also can't be seen with the HS01 inline. HP doesn't say what version the USB version is. Their connectivity specs just say "USB".

      Anyhow, I've spent enough time on this already, so I might just keep the HS01 on the DAC and use the TOSlink connection on the DAC.

      I only need 16/44.1 for my desktop speakers, so 24/96 is overkill.

      Thanks again.

      Delete
  7. Hey you’re welcome one both counts - happy to have shared something useful. Hope the surround sound mix is up to scratch.

    ReplyDelete
  8. Oh I forgot to ask, will ferrite rings/clamps be able to achieve results remotely close to these?

    ReplyDelete
    Replies
    1. Hey PS,
      No. Will not achieve the same thing. Ferrite rings can be helpful to reduce unwanted RF. It will not help the ground loop issues and will not provide galvanic isolation.

      Delete
  9. Hello Archimago, I have a ground loop issue on my (recent, i7) PC which I solved with a Phiree USB to toslink converter. It goes into a Sanskrit 10th DAC. No complaints about the sound but resolution is limited to 96. The DAC would display a default resolution of 96, but with a recent Windows update it went to 48. 96 no longer works, 44 does. No complaints about the sound either. Reading your review though, I'm wondering if getting the Topping unit would be worthwhile and if it is sure to defeat the ground loop issue I have - which I also had with a Motu M2 (balanced) interface by the way.

    ReplyDelete
  10. Did you ever try connecting the E1DA ADC to the HS01? I am unable to get that to work. When using the front panel USB socket on my desktop machine, it works intermittently (glitching at least once a second). When using the rear panel USB sockets, or any USB sockets on my laptop, it just glitches continuously.

    No such problems when connecting the HS01 to my Topping D10s.

    ReplyDelete