Expert review and test of the Soundblaster E3 DAC/Headphone Amplifier by Australian Hi-Fi Magazine. Free pdf download.

Is it too cheeky to include in these august pages a review of a DAC that costs a mere $350? Indeed, a DAC that comes from what is essentially a computer peripheral company? A DAC that’s also an ADC, a microphone pack, a powered headphone amplifier and an audio processor?

After running it through its paces, I don’t think it’s too cheeky, because it turns out that you can get some very respectable sound out of this device.

The Equipment

The company is Creative Technology. It didn’t invent computer sound for IBM compatible PCs (this was way before Windows), but brought together the synthesis of sound and digital to analogue conversion in computers with its Sound Blaster cards for computers. That was 1989 and the DAC capabilities were by today’s standards somewhat primitive. Forget about high resolution audio. It could handle only 8-bit mono audio at 12kHz sampling. For some years Creative dominated the field, and although those glory days are long gone, it managed to survive the general incorporation of audio facilities into computer motherboards by aiming at the higher end of the computer user market, primarily gamers.

The device under review is the Sound Blaster E5, a small unit which is far from anything that would be placed inside a computer. It has lots of capabilities, but let’s start with the Micro-B USB port. Plug this into a computer and it becomes a USB DAC. A USB DAC with a headphone amplifier. In that role, it kind of performs like the Sound Blasters of old (except it doesn’t don’t have an FM synth any more… all that stuff is done in software these days), but it performs much better since it supports digital audio with bit depths to 24-bits and sampling rates to 192kHz. Indeed, using the Cirrus Logic CS4398 DAC chip it features a dynamic range of 120dB, while noise and THD is specified at –107dB. The chip supports DSD, but this has not been implemented in the Sound Blaster E5.

There’s a line output (all analogue connections use 3.5mm stereo sockets), two headphone outputs and an optical digital audio output, just in case you prefer to use your own DAC. Both headphone and line output levels are controlled by a rotary volume control on the unit.

There is no setting for a fixed line-level output, which is something I would have preferred. That way one could dedicate the level control to headphone output and let one’s audio amplifier look after level control from the line output.

Driving the headphone outputs is a Texas Instruments TI6120A2 headphone amplifier chip. I know we tend to be keen on discrete components and all that, but before rushing to judgement, consider that this amplifier has a maximum output of 1.5 watts, an A-weighted signal-to-noise ratio of 128dB, and THD of 0.00024%. It uses current feedback architecture, facilitating high slew rates (it’s rated at 1,300 volts per microsecond). The implementation in the E5 results in an output impedance of just 2.2Ω, leaving headphones of varying impedances across their frequency range largely unaffected by the voltage divider effect.

There’s also a USB Type-A socket into which one can plug an iPad or iPhone (it didn’t seem to work with an iPod Nano) or some Android devices for music playback. There is also a combo 3.5mm analogue audio input and optical digital audio input, so you could use it as a CD DAC.

But there are more inputs than that. Specifically, Bluetooth. Impressively, in addition to the standard SBC codec the device supports both the higher-quality AAC codec used on Apple devices, and the higher-quality aptX codec supported by many premium Android devices.

In this review I am restricting myself to those functions and features. But there’s another side to the device. It has microphones and a combination microphone, line and optical digital audio input, along with an analogue-to-digital converter. You can use it with an iOS device or a computer for recording… or use it as a hands-free system with a smart phone. There’s a button that works to answer calls and hang up, the rest of the time doing duty as a switch for the audio processing features.

There’s also a built in rechargeable battery with a 3200mAh capacity. That allows the unit to operate as a DAC for an Android phone or iOS device, or run with Bluetooth sources away from any external power.

 The battery is rated for up to eight hours of operation. The E5 measures 70×24×111mm (WHD) and weighs 164 grams.

It also has a DSP built-in, which can do things such as EQ the audio or create a virtual multichannel effect for headphone listening. Those things are controlled using the driver control software which you install when using the E5 with Windows or on a Mac, and similar features can be controlled by app on iOS and Android devices.

I did not use the DSP but kept things simple with a straight-through audio path.

Subjective Performance

The Sound Blaster E5 delivered quite a few surprises, most of them good, but not all.

There was one thing I just did not like at all, and that was what I assume to be an auto input selection. I would have far rather had manual control over this. I started to try to work out which inputs had priority, or if there was just some kind of mixing going on, but soon managed, just by using first the USB Type B input, then the micro-B USB input, to hopelessly upset the unit. Music was played with a loud clatter, like the sound of a playing card against the spokes of a bicycle wheel. Fixing that took both several switches on and off and a reboot of the computer, so I abandoned the effort.

I can report that the unit worked well at playing back music from an iPad Mini 4 and an Android phone with On-The-Go support. (Most premium Android phones and not a few middle models provide this support, which allows peripherals to be plugged in, including DACs.) Likewise the optical and analogue inputs, and the Bluetooth connection. You can have two devices paired to the E5 by Bluetooth at once. To add a third you have to disconnect one of the existing ones.

I did most of my listening using the E5 as a DAC for a Windows computer. I mostly used Foobar2000 and JRiver Media Center as players because with them I could choose a more direct connection to the DAC, using the WASAPI interface. The driver package also includes ASIO support which tends to be what professional software prefers. Ordinary Windows processes can send music to the device using the ordinary Windows interface. That means you can play Spotify music and YouTube clips and such through it. But it is best to go WASAPI or ASIO if player software supports them because Windows insists on converting everything to a fixed sampling frequency (not to mention that it mixes in all other system sounds).

I have to say that the sound using the Sound Blaster E5 as a DAC for my computer was simply excellent.

Not just excellent for a lowish-cost device like this, but just all-round excellent. The simple matters were of course fully covered. The tonal balance was perfect. There was no audible noise. Nothing untoward appeared in the signal. But there was a very classy tangibility in the stereo imaging, and surprising subtlety in the delivery. As I write this, for example, I have the Giles Martin-prepared album ‘Love’ playing, consisting of substantial remixes and cut-and-pastes of Beatles tracks.

The tiny birds tweeting (in the traditional meaning of the word) around the song Because appear here, there, up above the speakers, down between them, and at various distances behind them.

That was repeated with all manner of music, little of it from audiophile sources. The imaging surprised me with its depth and precision. I was simply not expecting this.

Using the headphone outputs the performance was of a similar high quality, but there are two important things to note. First, there was no apparent bias of frequency response with headphones which vary in impedance according to frequency, which tends to corroborate the company’s claim of an output impedance of just 2.2Ω. Secondly, there was no practical limitation on output volume. Even using my elderly Sennheiser HD535 headphones, with their rather low sensitivity and highish impedance (around 160Ω), the Sound Blaster E5 could push them to remarkably high levels cleanly, with fine control. And that was without switching the E5’s gain switch from low to high, which would have added ten decibels. It managed those headphones effortlessly, with an open sound, as spacious as can be managed with headphones.

With lower impedance (26Ω), higher-sensitivity headphones, again things were clean and controlled. There was no noise introduced into the sound by the E5.

Objective Performance

I ran some tests on the E5 using RightMark Audio Analyzer software. The results were mixed, although mostly very, very good. Let’s start with what most of us spend most of our time listening to: 16-bit/44.1kHz audio. Using the line output the E5 produced a flat frequency response, down by 0.1 decibels at 20.7kHz, hitting a brick wall shortly beyond that. And after accounting for a 0.2dB droop at 20Hz from my measurement rig, down by around 0.25dB at 20Hz.

Initially the noise performance was so-so: –88.4dB A-weighted. Inaudible, to be sure, but not at the theoretical limits of 16-bit sound as you’d hope. But then I repeated the test with the computer disconnected from power, running on its own battery, and the noise levels dropped considerably to give a result of –98dBA.

And that figure is, I think, pretty close to the best I’ve ever measured with 16-bit audio. THD was 0.002% and IMD+Noise down at 0.0047%.

All that said, noise from a connected computer really ought to be blocked by one’s DAC, not be permitted to affect the analogue signal… and especially not allow it to be overlaid with spurious spikes (the worst at –93dB) at simple fractions of the sampling frequency.

With 24-bit sound the noise level fell further to –109dBA. Also very impressive. Of course, this was with the computer running on battery. I’d learned my lesson. The frequency response at 96kHz sampling was extended at the top end, to be down by less than 0.5dB at 30kHz and only –0.9dB at 40kHz. THD: 0.0018%, IMD+noise: 0.003%.

The DAC supports 192kHz signals, but the wheels seemed to fall off a little. Noise and both THD and IMD were just about the same as for 96kHz, but the frequency response was weird. It took up a narrow wobble above 2kHz, down by nearly 0.2dB at 6.5kHz, back up to nearly full level at 12.5kHz, down at 19kHz, up at 24kHz and with another brief local peak (albeit half a decibel down) at 34kHz before plummeting.

Audible significance? I defy anyone to notice in blind tests but these kinds of things make me worry that something has gone wrong with my measurements. So I replicated it but used a different brand of portable DAC/headphone amp as the DAC. It exhibited the expected smooth roll-off to –3dB at greater than 62kHz. Then I re-measured the E5 with identical results to the earlier measurements.

But what about the headphone output? Given the 192kHz wobbles, I measured that for 96kHz/24-bit signals. The frequency response graph precisely aligned with that for the line level output. Noise was, A-weighted, four decibels higher than for the line level. The measured level at –105.8dB was roughly a mile below audibility.

At maximum setting, the line-level output was 2.01V RMS for a full modulation sine wave. Into a 295Ω load the headphone output managed 1.84 volts RMS, which is 11.6 milliwatts, which in turn means some 10.6 decibels more output than the sensitivity specification of headphones. If your high-impedance headphones have a rating of 95dB for 1mW input, you can expect better than 105dB SPL using the E5. (More using the high gain setting).

With a 15.9Ω test load maximum output reduced. Or, rather, I reduced it because at full output it was clipping. For the 1kHz and 10kHz test frequencies dragging it down to 1.25-ish volts output eliminated clipping, while it had to go all the way down to 0.81 volts output into 100Ω to avoid clipping.

But even at 100Hz that was over 40 milliwatts output, yielding 16dB above the headphone sensitivity rating.

Conclusion

I guess if you pay a lot more than $350 you can get a DAC with comparable performance, plus improved noise rejection from the USB input. But that aside, I’d have to say the quality of the performance of the Sound Blaster E5 makes it a true high-fidelity DAC. So it’s quite the bargain as just a DAC, and if you can make use of all the additional functionality available it becomes even more of a bargain. # Stephen Dawson

Manufacturer's Site: www.soundblaster.com