O2 Measurements & Review

Discussion in 'Headphone Amplifier Measurements' started by tomchr, Jan 26, 2017.

  1. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    513
    Trophy Points:
    93
    Location:
    Calgary, Canada
    Home Page:
    The O2 is widely hailed as state of the art headphone amp. It was designed by ‘NwAvGuy’, who has now disappeared into the ether. The design is distributed under a Creative Commons type license. Prodded by many here and curious to see what the hub-hub was about, I decided to pick one up and run it through its paces on my Audio Precision APx525 audio analyzer.

    You can find my full review and measurements here: http://www.neurochrome.com/o2-headphone-amp/

    A couple of the more telling plots are below.

    [​IMG]
    [​IMG]
    [​IMG]

    As the O2 is battery powered, I also measured its performance as the batteries die. The graph below shows the O2 delivering 200 mW into 32 Ω for about 2.5 hours at which point the batteries run out of charge and the amp starts misbehaving.
    [​IMG]

    Once the batteries die, this is what the output voltage of the amp looks like vs time.
    [​IMG]

    The O2 contains a comparator-based low-battery detection circuit. Unfortunately, the circuit does not appear to have enough hysteresis. This means the low-battery detector will detect low battery voltage and turn the amplifier off. This in turn causes the battery voltage to rise a bit as the amp is no longer loading down the battery. So the low-battery detector will now think the battery isn't fully discharged after all and turn the amp back on. The result is the pulsing behaviour shown above.
    The instruction sheet from JDS Labs does include a warning that this behaviour is known to happen. They also warn that users should not operate the amplifier in this state with the headphones connected as damage to the headphones may result. Now, I don't know whether the pulses shown above can destroy a pair of headphones, but the warning in the instruction sheet does make me nervous, in particular as I tend to listen to music late at night and often forget to turn the amp off before going to bed.
    This problem could have been solved by using an actual power management IC or by adding an SR latch (two logic gates) with a few passive components to ensure that the power was turned off and remained off once a low-battery condition had been detected.

    Once the batteries are dead, it takes about 14-16 hours to recharge them fully.
    [​IMG]

    I personally wasn't terribly impressed by the O2 in listening tests, but I'm very likely biased and/or spoiled as I'm used to amps that provide considerably better technical/measured performance but also cost ten times as much. Tradeoffs, tradeoffs.

    On the positive, the O2 does have a nice and low noise floor and, as you'd expect from a battery powered amp, its mains hum is very low. DIY'ers or tweakers will also enjoy the leaded build with socketed ICs. One could entertain the option of ditching battery operation and plugging in a couple of better performing opamps, thereby potentially improving the sound quality of the O2. I will leave that as an exercise for the reader.

    Tom
     
    Last edited: Feb 1, 2017
  2. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,462
    Trophy Points:
    113
    Location:
    Winnipeg
    Wow, those battery-dying pulses are hitting nearly 7V peaks...
     
  3. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    513
    Trophy Points:
    93
    Location:
    Calgary, Canada
    Home Page:
    Yep. It's anything but graceful.
     
  4. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    Same. It was an okay amp, but nothing more. Would recommend as a first time build.
     
  5. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,462
    Trophy Points:
    113
    Location:
    Winnipeg
  6. Kattefjaes

    Kattefjaes Mostly Harmless

    Pyrate
    Joined:
    Sep 5, 2016
    Likes Received:
    4,521
    Trophy Points:
    113
    Location:
    London, UK
    Thanks for satisfying our morbid curiosity, Tom- it was an interesting read!
     
  7. MattRG

    MattRG Facebook Friend

    Contributor
    Joined:
    Jan 11, 2017
    Likes Received:
    235
    Trophy Points:
    43
    Location:
    Taylorsville, Georgia - USA
    Very cool. Thanks for putting the time and energy in to this and posting it for us. As a low cost, common license amp I believe the O2 accomplishes what NwAvGuy designed it to do and I wish he was still around so we could ask him how he feels about how it all turned out. At any rate, ignoring the funky low battery behavior the measurements look "fine" for what the O2 is.
     
  8. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,462
    Trophy Points:
    113
    Location:
    Winnipeg
    I'd love to see a regular tin can cmoy measure better than the O2 somehow :p
     
  9. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Nwavguy has CMoy measurements on his own site and was admittedly impressed by it.
     
  10. Psalmanazar

    Psalmanazar Most improved member; A+

    Pyrate Slaytanic Cliff Clavin
    Joined:
    Sep 27, 2015
    Likes Received:
    5,345
    Trophy Points:
    113
    Were you able to measure the clipping @tomchr ? I found the O2 audibly clipped quieter masters and the RCA jacks on some versions letting you use it as a pre-amp clipped all the time.
     
  11. shipsupt

    shipsupt Admin

    Staff Member Pyrate
    Joined:
    Sep 25, 2015
    Likes Received:
    1,119
    Trophy Points:
    93
    Location:
    Bay Area CA
    Gents, there is some intersting data and discussion here, but I'm going to shut this one down.

    Tom, I am sure your intentions are good, but here is not the place for a MOT to review other manufacturers gear. Clearly you can see how it's the kind of thing that could get out of hand. We love having direct contact with manufacturers, but there have to be some common sense guidelines to keep things on the rails.

    Thanks...
     
  12. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    So guys, I started playing around w the SR1 and measured my Massdrop O2 Headphone Amplifier, which is black.

    The gadget comes with two gains: 2.5 ("low gain") and 6.5 ("high gain"). It comes brickwall power thingy rated at a 15VAC/500mA output (7.5W). It can charge my NiMH 9V batteries that are currently mere parasites living of the O2 brick.

    The package deal can take consumer line level (about 0.316 Vrms) all the way up to about 1 Vrms w/o clipping on both low and high gain.

    Here is THD 1 Vrms in / 1 Vrms out in low gain (volume pot half way): -106 dB or 0.0005 % (which is awesome)

    THD_1V_1V_106dB_lowgain.JPG

    Here is THD 1 Vrms in / 2.5 Vrms out in high gain (volume pot half way): -102 dB or 0.0008% (which is also awesome)

    THD_1V_2p5V_102dB_highgain.JPG

    The increase in distortion is likely due to the increase in gain (less feedback maybe). So don't use high gain with this magic box, unless you need to.

    However, pro line level of about +4dBu (1.228 Vrms) will clip in high gain (output level was 1 Vrms, but it doesn't mater, clipping occurs before the pot which acts as an attenuator):

    THD_1p228V_1V_30dB_highgain.JPG

    Some DACs output 2.0 Vrms and obviously those will clip in high gain also. My 2i2 can give about +10dBu (2.45 Vrms) in balanced mode, but I guess it's more like 1.25 Vrms or so unbalanced, cuz half yo. That will clip too anyway, and I've experienced that. I avoid high gain anyway.

    All that said, low gain does not clip for me even with 2.45 Vrms unbalanced level, output set to 1 Vrms (-104 dB or 0.0006%):

    THD_2p45V_1V_104dB_lowgain.JPG

    I set the volume on max with 2.45 Vrms input voltage and got an output of about 6 Vrms or so (as expected for 2.5x gain). THD was about -100 dB or 0.001%:

    THD_2p45V_6V_100p4dB_lowgain.JPG

    The above is in some ways worst case scenario. I don't have a DAC putting out 2.45 Vrms unbalanced, and 6 Vrms output is stupid high for my cans. So "low" gain in the O2 has me covered.

    I also did some sweeps of THD vs. Vin and Vout vs Vin.

    Here are the "low gain" ones:

    THD vs. Vin
    Sweep_THDvsVin_lowgain.JPG

    Vout vs Vin
    Sweep_VoutvsVin_lowgain.JPG

    The above tells me this thing will clip with around 2.6 or 2.7 Vrms input which corresponds to about 6.7 Vrms output. Otherwise life is good.

    Here are the "high gain" ones:

    THD vs Vin
    Sweep_THDvsVin_highgain.JPG

    Vout vs Vin
    Sweep_VoutvsVin_highgain.JPG

    Which mean, don't use high gain unless your DAC output is less than 1.25 Vrms.

    And for now that's all. Maybe I'll play around with IMD and multitone. But not today.

    BTW, the sweeps are linear instead of log. Maybe later I'll do log. Dunno. Also impedance is "high" right now, as in connected directly to the analyzer in HiZ, so maybe later I may do sweeps with other more common impedances or actual headphone loads later. I also need to figure out how to do THD vs Vout and other stuffs (don't want to do post processing on raw data). One gets the picture though.
     
    Last edited: Dec 2, 2018
  13. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Shit! Fogot! How does it sound? Ok.

    It is a little brighter than the 2i2 POS headphone out, and more resolving. The 2i2 headphone out is an afterthought, but it is convenient.

    I'll measure the 2i2 next.

    BTW, the SR1 is a f'ing beast. I only wished I could remote desktop to it. Currently passing captures through a shared folder inside the "Users" directory in the SR1.
     
    Last edited: Dec 2, 2018
  14. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    Also, as to why the 1 kHz fundamental is buried, that is because I'm using the analog generator and applying the THD notch filter in order to distill those precious THD numbers and plots.

    The BW in the PSD plots is 16 kHz because I want to zoom in on the action. Above 16 kHz there was not much interesting stuff going on, and I wanted all the resolution I could get.
     
  15. BenjaminBore

    BenjaminBore Friend

    Pyrate
    Joined:
    May 23, 2016
    Likes Received:
    2,842
    Trophy Points:
    93
    Location:
    London, UK
    O2 Five Point Review - The Lady’s Away & I Have A Cold Sunday Edition
    • Resolution and micro-dynamics downgrader
    • Clips from a standard 2vrms output
    • Has a lot of low end oomph for a battery powered amp
    • Sadly couldn’t match the high bar that was a 2014 mid-range Samsung TV headphone output
    • Recognising it was a POS was my first step on a path to true enlightenment
    5/5 Winner of the Golden Goldblum Award
    [​IMG]

    PS: @Kattefjaes @Thad E Ginathom Yes, the key takeaway here is the colour choice, a topic SBAF members are becoming the thought leaders in, clearly, and that some may have forgotten they’ve changed the default forum style on their account ;)

    (Just for you guys I’ll change it to something better...)
     
    Last edited: Dec 2, 2018
  16. Kattefjaes

    Kattefjaes Mostly Harmless

    Pyrate
    Joined:
    Sep 5, 2016
    Likes Received:
    4,521
    Trophy Points:
    113
    Location:
    London, UK
    Yep, manually colouring text is pretty special in readability terms.

    [​IMG]
     
  17. Thad E Ginathom

    Thad E Ginathom Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    14,130
    Trophy Points:
    113
    Location:
    India
    It would be ok to read, albeit dubious in terms of taste, except for the yellow.

    Yellow text should have been left behind in those old psychadelic days when it was supposed to dissolve and melt rather than be read. That was a brief interval in the centuries-old history of typography which knew better before and after.
     
  18. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    513
    Trophy Points:
    93
    Location:
    Calgary, Canada
    Home Page:
    Which load impedance did you use for the measurements?

    I suggest running at least the 300 Ω and 32 Ω cases. 50 Ω can be interesting as well.

    Tom
     
  19. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,960
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! I don't think this is poop. I think NwAvGuy was a bit narcissistic about it, and a bit of a dick to others. But then most folks tend to be biased to like what they design.

    I don't think it has bad resolution and microdynamics.

    It DOES NOT clip at 2 Vrms input @ 2.5x gain which is actually a relatively high gain. The bottom line is that this amp, using the 15Vac/500mA power brick, can do a maximum 7.6 Vrms output regardless of gain setting. I believe this is reasonable. It does clip @ 6.5x gain, which should not be used with 2 Vrms input levels.

    I do not use this as a battery powered amp. I plug it to the AC socket in my house.

    It destroys the mid-range Samsung TV headphone output (I have the LN46A750 "touch red" TV, no dead pixels, and awesomesauce since 2009 and going).

    It is a relatively bright amp (relative to other amps I've heard).

    Just the Hi-Z. I will get to testing it with other loads. I believe we will be getting very similar results. You hit about 0.0005% with 300 ohms, and that is as good as it gets with SR1 Hi-Z as well (-106 dB). Note the sweep is linear. A log sweep will likely look a lot more like what you have.
     
    Last edited: Dec 2, 2018
  20. BenjaminBore

    BenjaminBore Friend

    Pyrate
    Joined:
    May 23, 2016
    Likes Received:
    2,842
    Trophy Points:
    93
    Location:
    London, UK
    It’s possible the gain on the one I had was set differently. Epiphany Acoustics had changed it at least once during their version’s lifetime, but even the low setting would clip with regularity fed from a 2vrms output, and high gain would sometimes clip from less.

    Jokes aside that really was my experience of it. I found it to markedly degrade both resolution and micro-dynamics from the DACs I had at the time, and it sounded worse than direct from the TV, iPhone, D1, and GO450 I had back then. Of course it had more driving ability than any of those on mains or battery power, I believe the latter of which put out around 5vrms.

    One of my early posts here lead to a discusion about this with purr1n. If I recall correctly he attributed this to heavy use of negative feedback, and possibly the opamps used.
     
    Last edited: Dec 2, 2018

Share This Page