Sunday, 21 February 2016

MEASUREMENTS: Google Chromecast Audio - Part I: Analogue Output


So you've no doubt seen this device at the local electronics store like Best Buy. It's part of the second generation line of Chromecast devices, the Chromecast Audio released late 2015 (the other one being the HDMI video-enabled Chromecast itself). At a list price of US$35 (CAD$45), this is clearly the least expensive ways to stream audio off one's phone or other compatible network enabled server. Remember, this light, plastic puck-like device doesn't have much in the way of "intelligence" built-in which is why it needs to be connected to an actual streaming device that will basically "push" the audio data to it, either from a local server or link it with a cloud-based music source. Effectively, this is a basic "audio renderer".




In the box, we have a standard stereo phono jack to patch into the analogue input to your (pre)amplifier. There's a little button on the back to reset the unit to factory settings (hold it down 25 seconds). The standard micro-USB power input goes in the back and in the front (if you look closely in the picture above), the single 3.5mm phono plug output which doubles as an optical TosLink like on Mac computers. As you can see, I have a little TosLink adaptor on the table (not included in the package) to use.

When you initially point your WiFi enabled phone/tablet/laptop at "chromecast.com/setup", it will find the Chromecast Audio at which point you will enter the settings which are saved into the Chromecast Audio so it logs into your home network - worked for me without a hitch... There's a little white LED that lights up when it's ready.Very intuitive install procedure!

If you use a Windows computer, the app will look something like this:
Selection of network settings...
Scroll down a bit... A few Chromecast Audio settings like HDR as well as firmware ID.
The unit gets just slightly warm after an hour or so. As you may have heard, as of December 2015, the firmware has been upgraded to allow streaming up to 24/96 (I'm using firmware 49061 as above in these tests). That "High Dynamic Range" setting doesn't seem to change anything in testing when I'm streaming a high quality signal from Plex. I suspect it only affects the sound from lossy streaming with the default being that some kind of dynamic range compression is activated and setting this will disengage the dynamic compression (this seems to be the suspicion in discussions here). In fact, the documentation indicates that the HDR setting is automatically turned "on" when using a TosLink connection.

Thereafter, stream away with Chromecast enabled apps/programs. Note that for best quality "bit perfect" streaming, I agree with this Computer Audiophile post to use Plex from my server computer connected to my ethernet. I then would use the Plex control app from my Android cell phone. For these tests, I'll just focus on casting through a music library on Plex running on my Windows Server 2012 R2 machine, using the Samsung Galaxy Note 5 for Plex controls. The WiFi router in my home is the Netgear R7000 which is running DD-WRT firmware - very stable with a current uptime of >100 days.

Another server option that's worth looking into is CastBridge for all you Logitech Media Server / Squeezebox users. As the Squeezebox devices age, this could be a very inexpensive replacement! So far, I haven't had time to investigate how well CastBridge works (and with a Squeezebox Touch replacement stowed away for backup, probably will not worry about it in quite awhile!).

Tip: For those running into issues with the "Can't play a sideloaded song remotely" error if you're trying to just cast from a local music folder on your Android phone with the standard Play Music app, what you have to do is go to the Chromecast app, tap the top left triple-line icon --> Cast screen / audio --> select the device to cast to --> CAST SCREEN / AUDIO button. From here just go back to the music player app and start playing. Note that casting from a phone in this way would not result in best sound quality as the audio is run through lossy compression to stream based on some quick testing. I hear that the app AllCast works as well. I'm sure there are other solutions out there...

Check out the teardown of the Chromecast Audio here. We see that the DAC inside is the AKM 4430 chip, a reasonable budget chip with rated dynamic range of 104dB. It is capable of 192kHz but realistically speaking, 24/192 over WiFi could be hit-or-miss depending on signal strength; probably best to not support this to prevent too many complaints. There's 2Gbit of RAM (250MB) in there, so probably a good amount of buffer space available to avoid buffer under-runs. Presumably the embedded OS running on the Marvell ARMADA 1500 dual-core 1.3GHz Cortex A7 processor will be very memory efficient. The interesting thing is that this Chromecast Audio is using the same processor as its video-enabled sibling. Considering the lower processing demands of audio, that's quite a bit of speed under the hood...

Now of course, for high-fidelity purposes, best practice would be streaming through the digital output to an outboard DAC to maximize sound quality. But before looking at the digital output in Part II, let's see what this AKM 4430 DAC-based analogue output can do...


I. Output impedance, 1kHz Square Wave, Impulse Response and Digital Filter Composite

The analogue output is a stereo 3.5mm standard headphone jack. Indeed, you can just plug in some headphones and stream music from your device using the volume control. Using a 1kHz sine wave and a 20-ohm load, I measured an output impedance of 34.5-ohms. Clearly this is high and not meant to be used with the typical low-impedance headphones these days. Not a problem when plugged into a preamp of course.

Let us now have a peek at the 1kHz 0dBFS square wave:

That looks pretty good. I double checked with a sine wave to make sure there was no clipping. Peak of 2.8Vmax (or a good 2Vrms). Very nice channel balance.

Impulse response:

A typical symmetrical linear phase filter with equal amounts of pre- and post-ringing. Notice that the phase is inverted.

Digital Filter Composite:
Okay, with 16/44 test signals, we're seeing a digital filter that's quite good with mild interpolator overload with the complex and loud white noise signal. Small amount of intermodulation and aliasing distortion with the 19 and 20kHz signal.


II. RightMark Tests & Comparisons

Alright, let's see how this thing performs with the audio test battery... As usual, I'll be using my E-MU 0404USB ADC for these measurements described many times over the years.

A. 16/44
As per convention, let's start with the 16/44 test signal - important to make sure this is done "well":

Confirmation that the Chromecast Audio is capable of 16/44 with ease. As you can see, I've compared it with a number of other portable devices, the Light Harmonic Geek Out V2 DAC, as well as a couple Squeezebox streamers (Touch, and Transporter).

Frequency Response

Noise Floor

IMD+N Sweep
B. 24/48
I've been measuring the 48kHz samplerate more these days with my portable devices. Also, as I have seen quite a number of 96kHz digital files being nothing more than upsampled 48kHz (or what looks like 48kHz material run through an analogue console), 48kHz has become more important to me since I routinely with downsample.

Summary:




As you can see, the Chromecast Audio measures very well! It's achieving about 17.5-bits of resolution; clearly capable of benefiting from the extra dynamic range of 24-bit audio data.

C. 24/96
As of the December 2015 firmware upgrade, the Chromecast can now handle 24/96 streaming.

Summary:

What really separates itself from the pack here is of course the Logitech/Squeezebox Transporter; hardly fair given the price differential of course. But it does give the opportunity to compare objectively the range of capabilities between the different DACs and players as it applies to high-resolution audio.

Frequency Response - note that the iPhone 6 resampled 96kHz signal. Of note, the Chromecast Audio is -1dB at 20kHz.

Noise Floor

IMD+N Sweep - again, the iPhone 6 is incapable of true 96kHz playback. Impressive result from the Transporter.

D. 24/192
The Chromecast Audio failed to stream at 192kHz through Plex - as expected. There was no audio output at all.

III. Jitter


As you can see, jitter is a non-issue even with a <$50 streaming device receiving the audio data off a WiFi network. For the sake of completeness, there are a pair of obvious sidebands evident in the 24-bit J-Test of very low level corresponding to +/-250Hz. In terms of amplitude, these are 115dB below the primary 12kHz signal! Absolutely irrelevant in ability to impact sound quality.

IV. Subjective

Look, I can write all kinds of things about the music I listened to, etc... But I won't bore you and encourage you to have a listen yourself if interested in this device. Seriously, the analogue output sounds good; just don't plug some low impedance headphones into this thing and expect the best potential audio from it. Also, remember that it's best to use a dedicated streaming source like Plex Media Server for best quality. If you simply cast from your phone or tablet, you're likely sending a lossy and dynamically compressed signal (this is where the "High Dynamic Range" setting on the Chromecast could improve the sound quality).

The fact is I don't have a need for this device to act as a particularly high-fidelity source at the moment (I like having it for convenience) but I did listen to the analogue output for about 2 hours in my main stereo system one night - Emotiva XSP-1 preamp (RCA input, 47kohm input impedance), dual Emotiva XPA-1L amps, and Paradigm Signature S8v3 speakers with SUB 1 subwoofer. Sounds great, no harshness nor any obviously objectionable characteristic. Recently, I've been on an Asian music kick. Late last year, going through China and Hong Kong, I picked up a copy of Dadawa's Sister Drum (阿姐鼓, 1995). A Mandarin album consisting of New Age and Tibetan influences. Plenty of dynamics, excellent rendition of nuances, great tonality. On of the most important qualities I enjoy in a good DAC is the ability to render the "depth" of the recording and indeed the Chromecast sounded very good in portraying the "3D" characteristic of this album.

V. Summary

This concludes Part I of a 2-part look at the Chromecast Audio. I had no issues at all with the setup. Remember that for best quality, I did set up Plex server for streaming to the device. As usual when streaming wirelessly through WiFi, make sure the signal strength is adequate, especially if you're going to be doing 24/96. I did not notice any issues with just 2.4GHz Wireless N on the main floor of my house with the router in the basement. Even upstairs where the signal strength dropped below 40%, I did not run into any issues. I suspect the Chromecast has a very good buffering system to prevent under-runs.

As for objective sound quality, honestly, I was impressed by what I'm seeing here. If much of this is the result of AKM's new generation of 44xx chips and the budget 4430 is of such high quality already, I'd imagine the AKM "Verita" AK4497EQ "flagship" should be objectively impressive! Of course the question is whether human ears would be able to appreciate the quality...

From a subjective perspective, what else does one really need? I would certainly have no qualms about using the analogue output from this remarkably economical device to feed a good stereo system.

One might ask the age-old question of whether the inverted phase as demonstrated by the impulse response makes any difference; I certainly do not believe I have ever heard a difference flipping polarity back and forth in the comfort of my listening chair with the Transporter (other than perhaps a slight "tick" at the point of the polarity switch). It's truly amazing when reading stories of audiophiles having customized settings for "correct" or even "absolute" polarity for each album they listen to (here's a strange and extremely long "Abstract")! I agree with J. Gordon Holt's article (Stereophile back in 1980) and suspect that it's only a factor if one's speakers have significant excursion nonlinearity; basically if one hears such a problem, the speakers are probably at fault and should be checked out for damage or maybe needs to be upgraded if there is an inherent issue. Of course, as usual, I am happy to be proven wrong if someone can provide an example to try...

Stay tuned for Part II as we have a look at the TosLink/optical digital output.

-------------------------
Have a great week ahead everyone!

23 comments:

  1. Thanks for the posting Archimago, I don't expect I'll see anything in the Absolute Sound about Chromecast Audio.

    Also, I just posted my first Medium article on my take of the audiophile world and mention you and your blog - here it is.

    https://medium.com/@skikirkwood/truth-lies-and-fraud-in-the-audiophile-world-a365e56c97c4#.odzhhbh1u

    ReplyDelete
    Replies
    1. Hi Jim,
      Yes, I suspect you are right... Don't expect the Chromecast Audio to get many words published on just how well it works. If it does come up on a page somewhere in the "mainstream" audiophile press, I'm sure it'll be along the lines of comparisons with something more expensive and be provided as an example of a typical "utility" device far below what the audiophile target market would want (and I doubt I'll go bankrupt betting that it doesn't sound as good according to them)... Nonetheless, it is IMO these kinds of easily accessible devices that can provide great joy and promote the love of music by making it accessible - and sounding great doing it!

      Wow! What a blog entry. Love your clarity of thought and sharing of your life journey, man. And that's not just because you mention this blog in a positive light :-). Seriously, welcome to the world of audiophile blogging and I look forward to more of your findings and ideas! Just be prepared with some flame-retardant outerwear when wandering in some corners of the Internet.

      I think your comment about the migration and expectations of the shift from print to Web media is spot on. At this juncture in history, for a relatively small hobby like high-fidelity audio, after years of a certain (subjectivist) point of view, democratization of the tools to create and disseminate content I believe will be a big factor in shaping the hobby ahead. I think history will show in time the true allegiance of the "silent majority" as it pertains to audio fidelity and the hardware.

      Cheers!

      PS: Nice Raspberry kit. Tempted to put one together myself after seeing that!

      Delete
  2. Good to hear they measure well. I Just ordered two of these as replacement for the horrible wireless connections of my the Marantz Consolette (MS7000). While these are great sounding all-in-one units I have never seen anything as bad when it comes to software support and conectivity. I plan to use the usb connectors of the MS7000 as power supply for the chromecast, so there hopefully will not be be to much clutter.

    I am planning to use jrivers gizmo (or jremote, should this be available later this month) and a jriver server to stream to the chromecast units.

    ReplyDelete
    Replies
    1. Cool. Let us know how this works out! That does sound like a great way to stream to the Consolette on USB power.

      From looking at the JRiver board, I assume you will need to run BubbleUPnP to make the CC Audio appear as a DLNA streaming device...

      Delete
  3. I think you're mistaken regarding the FDR/HDR option in the Chromecast settings. It seems to be nothing more than a simple output level switch, there is no compression either way.

    With FDR on, the output is 2Vrms, but this can be too much for some devices. Case in point, my onboard sound card clips if I try to record the Chromecast's output at full scale, so I had to use a spare USB ADC to record it.

    When two recordings with and without FDR are amplified to the same peak level and compared, their Replaygain and DR values are identical.

    The sheer amount of misinformation regarding this option is astounding, and probably fueled by Google's vague language on their help page.

    ReplyDelete
    Replies
    1. Thanks Kozmo. Yeah, I admit much of it seems like speculation... Interesting point. I'll see if I can try the same the next time I give this a look. At the moment I lent the device out to a friend.

      Delete
  4. I am glad to see my subjective impressions confirmed. I have been using one in my pretty revealing main rig (QUAD 2805 electrostats plus sub equalized by Antimode 8033) and it sounded just fine streaming from the web. It is so convenient....

    ReplyDelete
    Replies
    1. Yup! Based on sound quality alone, this baby works well :-).

      Delete
  5. Dear Archimago,

    some time ago we spoke here about 24 bit support on Lumia 735/Windows Mobile 10. Have you obtained any information about this ? Thank you

    Honza

    ReplyDelete
    Replies
    1. Sorry Honza. Been busy with other stuff and work :-). Plus haven't met up with the Lumia-using friend. I'm curious is there anything special about this phone? Are they claiming any special sound quality?

      Delete
    2. No, it is a standard Windows Phone with Snapdragon 400 CPU and no special DAC. But I have some 24 bit FLACs and would like to know if i have to convert them to 16/48 format to play without truncating in Groove Windows Mobile.

      Delete
    3. You should probably just convert them anyway, to save space. There's no audible difference.

      Delete
    4. According to this link newer lumias support 24 bit audio

      http://www.qobuz.com/be-fr/info/hi-fi-guide/smartphones-compatibles-audio-hd/ecoutes-de-la-section-audio-du177091

      They also claim higher sampling rates support but I do not use them much because of nyquist, especially on mobile I use almost always 44.1.

      Delete
  6. Doers the standard (HDMI) Chromecast work for casting lossless 5.1 audio?

    ReplyDelete
    Replies
    1. Good question... Don't know. Alas, don't have one at the moment.

      Delete
  7. For the digital test I'd be very interested to know how tightly the multi-room sync feature works, though this assumes you've a) two of them and b) and a recording interface with a pair of synchronised digital inputs.

    If the're very tightly synced then it allows quite a neat solution for active speakers incorporating digital crossovers (e.g. minidsp or hypex plate amps). E.g. two chromecasts for L/R each which eliminates the requirement for any wires between your music source and the speakers, or between them (i.e. to duplex the digital source signal).

    BTW to eliminate any wireless issues you can connect these with ethernet, either using the official Google Ethernet adaptor or a variety of usb ethernet adapters + 'micro usb y-cable'. I've a Linksys USB3GIG which is rock-solid for both audio and video.


    [Wall socket] --2-core extension power-flex--> [ Powerline ethernet] -power and ethernet-> [Google Chromecast Ethernet Adapter] -power and toslink-> [digital plate-amp w/crossover]

    A further note/ thought is that you should be able get an electrical SPDIF signal out the 3.5mm mini jack with some judicious soldering.

    ReplyDelete
    Replies
    1. Thomas, brilliant! However that would be very much beyond the capability of the typical user. Alas, I don't have 2 of these so would not be able to test at this time.

      Cool. Didn't know about the ability to connect wired with the Y-cable. Even with the powerline ethernet, I would not be surprised if latency would be too much for tightly paired L/R stereo though! Even with a typical DAC when running through DiffMaker, it's easy so see the amount of samplerate drift over time. I would imagine with 2 independent devices, there'd be significant drift and tolerance would likely not be something ideal like synchronized within a handful of samples at most.

      Hey, if anyone gives this a try as a DIY solution, I'd love to hear about it and how it goes!

      Delete
  8. I'm using three CCAs in a multiroom setup, and I certainly haven't been able to detect any significant latency in that setup, while playing to all three devices at once.

    ReplyDelete
  9. Same here: no audible latency when in multiroom use. As for 5.1 from the video Chromecast: from all reports this works withour problems, but of course you need an AV receiver or similar with hdmi input, or a 5.1 audio extractor.

    ReplyDelete
  10. I have 4 CCA within earshot and have noticed giant and unacceptable delay between devices as one is connected via 3 party wired Ethernet (ugreen USB hub). Setting that single device to 100ms delay has solved that problem. But I wonder if L/R usage of CCA as discussed above is possible if no one has measured latency between CCA devices (assuming all wired or all wireless).

    ReplyDelete
    Replies
    1. The time/sample drift between the two devices would have to be extremely low, otherwise you'll run into weird phasing and soundstage issues.

      It would be a really fun experiment, though.

      Delete
  11. Any thoughts on what impedance would be presented at the 3.5mm output when a CCA is unpowered, or not streaming? The issue relates to using a simple resistive mixer to take a number of remotely controllable inputs, and pass them to active powered speakers. Whichever input device is enabled and generating audio will pass through to the speakers, but the loading presented by other mixer inputs would change the attenuation of the mixer circuit.

    ReplyDelete
  12. We know that Chromecast is the best source to watch all the desktop or smartphone content on a large screen(like TV, Projector screen)
    how to rest chromecast

    ReplyDelete