Saturday, 1 April 2023

MEASUREMENTS: Topping HS02 USB 2.0 isolator. (This works with the E1DA Cosmos ADC; and the Topping + Intona "double isolation" of DAC and ADC.)

Hey guys and gals, I must be a sucker for relatively inexpensive USB galvanic isolators. To be honest, this is mostly because I look for ways to lower noise on my measurement testbench, not that I'm concerned with the actual audiophile sound system these days! ;-)

Today, let's have a look at the recently released Topping HS02 USB2.0 isolator (currently ~US$100). This is an update to the previously reviewed HS01 last year. As you can see from the picture and specs, the box now has selectable USB-A/B connectors or USB-C. Isolation is not just applied to the in/out USB data connectors, but also the auxiliary power input with filtering which means the device could stay free of noise even if we plug in a power source with some noise in it. As usual, be mindful of power-hungry USB-powered devices which will require auxiliary power input.

Furthermore, Topping has improved compatibility now with the ability to negotiate USB2.0 low (1.5Mbps)/full (12Mbps)/high speeds (480Mbps). I can confirm that my low speed wireless keyboard dongle works fine.

The diagram showing the path of the data and power connections is relatively clear printed on top. Inside the box are a couple of pamphlets - the black one is the warranty card and Topping catalogue, and the white user manual. There's a short USB-A to USB-B cable in the plastic bag.

Above, we see the view from the input end with the USB-B connector along with the USB-C. The small metal box weighs approximately 45g and measures 6 x 4 x 2 cm.

Connecting the isolator up to DACs and other USB devices is straight forward. Topping in their literature claims that the isolation voltage is >1500kVrms and latency is basically nothing at 0.073┬Ás - impressive.

Let's do a quick test of isolation ability using this device; here's a diagram of the connections I'll use to demonstrate improvements objectively:

You can compare this diagram with the one I used when recently testing the Intona 7055-C. Notice this time, I put the HS02 isolator between the Beelink SER4 (discussed previously) mini computer and the E1DA Cosmos ADC measurement system. Yup, unlike the other isolators I have tried to date, this is the first time I've been able to get the E1DA Cosmos ADC to play well with a USB isolator. The only caveat on my system was to make sure I plugged the isolator directly into one of the computer ports rather than into a hub and on the Beelink SER4, I noticed that the USB-C port did not seem to provide enough power to the ADC so I stuck with one of the computer's USB-A connectors.

With this kind of set-up, I can measure the audio output from a direct "computer-to-DAC" connection without needing to put an isolator between the two.

Apologies for the mess of wires and dummy loads in the background! Here's a picture of the testbench using the Topping HS02 measuring the single-ended Topping D10s as per the flow diagram above with the E1DA Cosmos Scaler prototype, APU, and ADC "stack":

Note the lit white LED on the front port connected to the Cosmos ADC to indicate power going into the isolator.

To keep the measurement system as clean as possible, I used lithium batteries to power the APU and Scaler. Auxiliary 5V was not needed for the Cosmos ADC.

Results comparing with and without isolator using the single-ended DAC in a loop-back configuration prone to ground loops...

Apologies using different X-axis... Both measurements done with DAC and ADC at 32/192. Too lazy to recapture. ;-)

Excellent. The isolated THD+N at almost -113dB is basically what I got for the Topping D10s as well measured a few months ago with the Intona isolator. As discussed before, let's not bother with a balanced DAC this time since it's these single-ended devices that are much more prone to ground loops and other noise.

For fun, let's do a "double isolation" test measurement:

In the above set-up I've now isolated both the DAC and ADC connections to and from the measurement computer. Let's see how much difference this made:

Very slight improvement in the THD+N result; just an extra -0.1dB due to slightly better N. It's a nice demonstration I think that so long as we break the ground loop at some point in a complex system like this, it'll be fine.

We can peer deeper down in the noise floor as well using a -60dBFS signal and the APU's 34dB gain to see the difference; this gives us an accurate measurement of the dynamic range of the DAC as well:

Dynamic range obtained with isolation showing an excellent 121dB (60dB for -60dBFS signal + 61dB from measured THD+N) with the single-ended DAC. Notice the noise floor anomalies in the un-isolated situation. "Double isolation" with both Topping HS02 and the Intona 7055-C again only showed an insignificant difference.

A quick obligatory look at the jitter test with the Topping D10s + HS02 just in case:

No jitter sidebands. All noise below -140dB.

Yeah, no problem. Jitter? What jitter? Even with just a $100 USB DAC connected to a standard AMD Ryzen 7 mini computer these days. Remember this when you're confronted by companies and reviewers who want to make a sale claiming noise and jitter performance superiority - like say this Innous PhoenixUSB for example - a mere US$3,749. Where are their graphs? ;-)


It works. The Topping HS02 USB isolator (Canada link) is compatible with a broader range of devices including "low" and "standard" speed USB2.0 than the first generation Topping HS01 although a bit more expensive.

As a demonstration of that wider compatibility, it's good to see that the E1DA Cosmos ADC used for measurements worked with this isolator. If you need isolation for USB 3.0 5Gbps and don't need E1DA Cosmos ADC compatibility, go have a look at the Intona 7055-C discussed recently.

Apologies to those looking for a subjective "How does it sounds?" section to this review. There's simply no need to write paragraphs describing whether sound stage expanded, transients changed, bass got deeper or treble extended beyond the frequency range audible to dogs. When a system is well isolated and you're using a DAC with excellent temporal performance (ie. low jitter), it just sounds like "Bits Are Bits" because the noise and interference have been neutralized, which is what we're aiming for in high fidelity reproduction in the first place. To claim that one hears significant differences or that one isolator is better than another due to subjective beliefs would be folly. Yes, I listened with the HS02 connected between a fanless MeLE Quieter2Q computer and the Sabaj A20d 2022 DAC the other night and the sound remained excellent (no, an isolator is not needed in my audio system).

As I have suggested before in other USB isolator discussions, you likely will NOT need a USB isolator most of the time with most systems. I do not recommend just buying these things because some audiophile reviewer guy recommended it. These are useful however if you have ground loops or notice noise leaking into your DAC from the computer or streamer; these anomalies will sound like a hum (eg. 50/60Hz mains frequency), or intermittent electrical noise when the computer/streamer CPU is under load or accessing storage.

And of course, do not buy useless stuff like the AudioQuest JitterBug and their suggestion to hang one of these "filters" off every USB port to lower the noise! Be wary of brands like this making unsubstantiated claims. Even if inexpensive, think twice before handing out the credit card numbers.

Happy April everyone. Great to see the sun out more and cherry blossoms in bloom. Hope you're enjoying the music!

[Reminder: The 16-bit vs. 24-bit Blind Listening Test and Audiophile Survey are still ongoing - 1 more month to get your results in!]


  1. It's a pity you don't own an LRC meter. The efficiency of isolation can be easily measured as coupling capacity between input and output. No need to turn the unit on. Short all input pins of USB-B as one side, all output pins of one socket as the other side, measure capacitance between them at audio frequencies. The old Intona (no USB 3) has a few picofarad only.

    1. Thanks for the tip Techland,
      I do have an LRC meter so I could give this a go later...

    2. Ordered one HS02 as well. Read about it in ASR. It has one small disadvantage - uses the Analog Devices chip which registers as USB hub, so no full transparency as the Intona. But I doubt that matters much.

    3. And just read that HiFiMe also has a new isolator, chip undisclosed, for half the price of the HS02, available in May.

    4. Just read in ASR that one guy there has spread misinformation - the chip used on the Topping seems to be fully transparent, there is no hub at all stealing USB resources. That's good news!

  2. Hi Archi

    Great, thanks. I have ordered it right away.


    1. Have fun Juergen!
      Could be a nice addition to a mobile measurement rig...

  3. How much power can the HS02 provide to usb powered dacs without adding an external supply? Can you measure it with different sources without groundloops, ie. rpi, etc to see if you can find any measurement improvements? Also would be interesting to measure the power supply filtering claims. Thanks

    1. Hi Audioidiot,
      The manual mentions 800mA out. Not sure if that's with the auxiliary input. Might have to ask Topping on this one.

      I would be surprising I think to see much benefit from the RPi but I'll check this for you.

      Yeah, maybe I can try plugging some crappy 5V USB source (like say the USB outlet from a power bar) compared to Li battery to see if the filtering prevents any increase in noise level or hum...

  4. Can't wait for your take on MQA bankruptcy

  5. I wish someone would put out an HDMI equivalent at reasonable cost. I tried the Fibbr HDMI optical cable you posted about a while back but it actually increased the noise level: it was mechanically grounded and poorly shielded.

    1. Thanks for the note Phil. Good to hear the feedback, at least it wasn't expensive to try...

      Yeah, very little in the HDMI-for-audiophiles space. Will keep an eye out and see if something might fit the role...

  6. Can this work to clean power and data signal from a noisy tv/pc? They say it reclocks the audio data:

  7. Archimago, apart from hearing buzz noise, how do I know my dac is effectively neutralising noise and interference? I mean, if it does well with your Eida stack, why whouldn't aswell in my system? I heard dac chips are extremely sensitive to noise in data bus.