Sjostrom QRV08

Discussion in 'Headphone Amplifier Measurements' started by tomchr, Dec 2, 2016.

  1. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    499
    Dislikes Received:
    19
    Trophy Points:
    63
    Location:
    Calgary, Canada
    Home Page:
    Last edited by a moderator: Dec 2, 2016
    BMF Flyer, fishski13, bazelio and 6 others like this.
  2. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    @tomchr, moved this here to keep things well organized.

    This is very interesting.

    Is the DC servo installed? You think this may have contributed to the interesting looking THD+N hump at lower frequencies?

    Great review BTW. Not a bad discrete headphone amp at all.

    I may have overlooked, but what was the output impedance of the amp?

    Is it fixed gain?

    (BTW, screw it, if you can, do the O2... Pretty please? :) )
     
    fishski13 and bazelio like this.
  3. Armaegis

    Armaegis Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    3,700
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    Winnipeg
    The review does seem to jive with my impressions of the amp. I found it overall very competent, with some decent kick on the bottom, but overall a bit flat up top and lacking the "sparkle".

    I've been chatting with Tom for a couple weeks. The O2 is on the list o_O
     
    fishski13 and ultrabike like this.
  4. fishski13

    fishski13 Friend

    Friend
    Joined:
    Sep 25, 2015
    Likes Received:
    392
    Dislikes Received:
    0
    Trophy Points:
    63
    Location:
    Twin Cities, MN
    Thanks Tom. Super appreciate your work. At least it appears I didn't totally fuck up the build.

    No parts matching on my part with SMD sand.

    IIRC, the input coupling capacitor is a film 0.1uF SMD type.

    As far as off-board channel separation goes, the input runs are air wired with Belden 2 + shield with a Gold Point SA, minimizing run distance.

    While low powered, I've never been able get this amp to go clunk and clip, but I primarily use dynamics.

    Ultra,
    Yes the DC servo is installed. I forget what the gain I went with - IIRC it might be 2x? I just used the stock recommend component and values.

    https://solderworksaudio.wordpress.com/sjostrom-qrv-08/
     
    Last edited: Dec 2, 2016
    ultrabike likes this.
  5. Armaegis

    Armaegis Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    3,700
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    Winnipeg
    Modest words from a guy who assembled something with parts that look like they could literally fall out of a salt shaker...
     
  6. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    499
    Dislikes Received:
    19
    Trophy Points:
    63
    Location:
    Calgary, Canada
    Home Page:
    I did once have a customer who exclaimed, "those parts are smaller than what comes out of my pepper grinder!" at the sight of 0805 passives (huge by SMD standards). He was able to assemble the circuits he bought just fine, though. At the other end of the spectrum, one of the guys I used to work with could hand solder 01005 components (0.01 x 0.005 inch dimensions). Under a microscope, of course, but still... 01005!!

    The gain of the QRV08 is 12.7 dB (midband). Or 4.3x.

    It is possible that the DC servo impacts the THD at LF, but if that was the case, I'd expect the THD to decrease 20 dB/dec until the THD/noise floor. I'd also expect the same impact regardless of load impedance. As I wrote in my review, I think it's due to the biasing of the output stage. As far as I can tell, there's no adjustment of the bias current, so the design relies on part-to-part matching to get the correct bias point. Unless the parts are matched on assembly *and* operate at the same temperature, the amp will have considerable sample-to-sample variability in the bias current. The logical choice from a designer's perspective is to err on the side of caution (low bias) such that the end user doesn't end up with an amp that self-destructs.
    I'm not making too much of a stink about the THD profile. It's not really a problem per se. It's just one of those things that would have gotten me back to the drawing board (and circuit simulator) to figure out what the heck is going on. It could be an artefact of the current feedback design. Maybe the designer did go back to the drawing board and ended up concluding that the THD profile was as good as the design constraints would allow.

    The only downright disappointment is the 120 mA current limit. P-A specifies it to 1.0 A.

    I haven't measured the output impedance. I don't have a good test rig for that. I'll figure something out and at least provide measurements at a couple of frequencies. I can do that by hand. The output impedance is pretty low. The output voltage doesn't move much when I switch from 20 to 300 Ω load.

    Yes, the O2 is in the mail and should be here in a few days. I'll add it to the stack of gear that needs measurement. If anyone has a burning desire to own the best characterized and torture tested O2 in the World in a few weeks, toss me a PM as I'll be selling it when I'm done.

    Tom
     
    sphinxvc, Cspirou and ultrabike like this.
  7. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    Thank you Tom.

    Indeed, I the load seems to affect low frequency THD+N, and the servo noise/distortion should be somewhat independent of that (I think). Matching of transistors may indeed be the culprit. And it may be as good as it gets with the approach there. Peranders (Sjostromaudio dude) does seem to have a few current feedback designs there in his portfolio, which are not as usual as voltage feedback designs. Who knows. That said, IMO it's not shit and it's got it's kool factor there.

    It seems from the comments in Peranders website, that the QRV08 focused more on the power supply and in offering an all discrete approach.

    I also understand the current drive disappointment. 120mA IMO is not bad though, but probably on the low end for a discrete SS design, and it will likely not have what it takes to drive the likes of an HE6 (note @ohhgourami uses a nuclear power plant to power these cans). Dunno wazzup with not meeting the 1 A spec.

    The O2 IMO is not bad. The pot in the middle there may buy some noise performance, but it may clip with certain sources depending on gain. I dunno however if his 4mV "respectable" typical DC offset voltage is that respectable (I think that might be closer to 40mV worst case based on the datasheet of that part). I also dunno about his love for the NJM4556. It's not a bad part, but it's a bit on the slow side and may be a bit close if running full voltage swing. For most things it's proly not an issue though. His power management seems to me a bit Rube Goldberg-ish.
     
    Last edited: Dec 3, 2016
    bazelio likes this.
  8. Armaegis

    Armaegis Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    3,700
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    Winnipeg
  9. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    In Spanish it's Jesustrom.
     
    Last edited: Dec 3, 2016
    brencho likes this.
  10. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    BTW, I would like to point out that while I do not personally know or had any conversations with Peranders, I have a lot of respect for him given some of his fairly useful post about DC servos at the diyaudio website (IMO).

    I downloaded his LTSpice simulation and read some of his ramblings there, and I found them very useful in reviewing and understanding stuff. Here is a link to one of his contributions:

    http://www.diyaudio.com/forums/chip-amps/46891-dc-servo-2.html#post525874

    (I also find NwAvGuy's monologues very useful, one just has to filter the wheat from the chaff.)
     
  11. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    499
    Dislikes Received:
    19
    Trophy Points:
    63
    Location:
    Calgary, Canada
    Home Page:
    I have nothing against P-A (Per-Anders) either. My critique is of his circuit and not of him as a person.

    In a world where many want 5 W out of a headphone amp, 200 mW won't cut it. That's decidedly not my problem, though. :)

    I suspect the 1 A output current is either a rough estimate based on ideal conditions or is the instantaneous short-circuit current. Neither is relevant from a practical perspective. Had the amp provided, say, 800 mA I would have waived it off as a spec of typical performance and chalked it up to sample-to-sample variation. But when the actual implementation delivers one tenth of the specified output current, there's something wrong somewhere. At the very least, there should be a note in the spec about how the measurement was performed so it could be repeated.
    There is a mention on P-A's audio page about the ability to use different transistors with the board, so perhaps he measured with different transistors than the loaner sample was equipped with.

    I think I'll show some leadership and toss P-A an email. It's in his best interest to know what's going on. Perhaps he can explain how the output current was measured to arrive at the 1.0 A spec.

    As I mentioned in my review, I think the QRV08 is a fine amp. It has a few quirks but will work fine for most people. That said, my HP-1 beats it by >10x on nearly all parameters. The HP-1 is also nearly 10x the cost, so that seems to go hand in hand... :)

    Tom
     
    bazelio and ultrabike like this.
  12. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    Don't worry Tom. I feel we all need to take constructive criticism of our work hopefully to make things better and learn. And I honestly appreciate what you are doing.
     
  13. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    499
    Dislikes Received:
    19
    Trophy Points:
    63
    Location:
    Calgary, Canada
    Home Page:
    If I can help drive the field forward and push the boundaries of what's considered state-of-the-art, I'm totally okay with it. :)

    Tom
     
    Cspirou and ultrabike like this.
  14. Armaegis

    Armaegis Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    3,700
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    Winnipeg
    And when it comes to the O2, please don't hold back the punches. Nwavguy did his fair share of... well, slander is the wrong word, but he certainly didn't hide his disdain for many designs and used his measurements to his advantage. I think it would only be fair and fitting that his amp be placed under the same scrutiny.
     
  15. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    It's up to Tom. Given NwAvGuy's levels of public disdain I can understand. However, few times I have felt the urge to use the full power of the dark side. And to be fair, his write ups are not total bullshit. LOL! We have been a bit disdainful ourselves of this and that product. I can be a total asshole myself sometimes, and be wrong while at it.

    I'm not sure if NwAvGuy is an analog designer. But I have not meet very many analog designers that do not consider their designs "best in class" and "cures cancer" level. NwAvGuy is proly no exception.

    If fixing his little red wagon is in order, so be it though. For science!
     
    Last edited: Dec 3, 2016
  16. Armaegis

    Armaegis Friend

    Friend BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    3,700
    Dislikes Received:
    2
    Trophy Points:
    113
    Location:
    Winnipeg
    His writeups in general weren't bad, but he had a couple "favourite" people to pick on in reviews, sometimes making references back to how bad they were which was completely unnecessary in the context of the latest teardown. Plus you cannot deny that he pretty much built his name by trodding down well known entities in the field before deciding to bestow his graces upon the masses with the O2. He would not have garnered nearly the following that he has if he didn't pick his targets as he did. I view his entire story as a very elaborate setup leading up to the O2/ODAC. If he built up his name through solid and impartial analysis, I'd have no beef with him... but the disrespectful manner in which he built himself up by knocking others down (getting himself booted from several sites in the process) is where I take issue.

    Anyhow, back to the QRV08 vs HP-1... overall I did find the HP-1 more resolving than the QRV08. Copying some of my thoughts over from the other thread:

     
  17. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    499
    Dislikes Received:
    19
    Trophy Points:
    63
    Location:
    Calgary, Canada
    Home Page:
    Wow. That's quite a review. Thank you.

    Regarding the O2: I don't like how NwAvGuy presented the data. He measured some parameters, summarized them in a table and put the words "excellent", "very good", "good", etc. next to them with no definition of what he considered to be "very good", "good", etc. Of course his numbers were all better under the measurement conditions he chose so his were all "excellent" (whatever that means). That aside, I really have no beef with the guy. I do commend him for presenting the measurements, though, I can also see a few examples of how he was a bit creative on some of the measurements in the interest of bringing out lower numbers (THD+N measurement with 22 kHz measurement bandwidth, thus, only measuring 3rd harmonic up to 7 kHz, for example). I understand why he did so, but find that there are better and more honest ways of presenting the data. I plan to measure the O2 in the exact same fashion as I measured the HP-1 and QRV08.

    I'm not here to pick fights. I do my best to talk positively about my own products and avoid talking negatively about competitors' products. That said, to be as helpful to the community as I can, I do believe I should point out any concerns and oddities I find in my measurements. In case of the QRV08 (see, now we're back on topic!) I found a few oddities (THD profile), one potential concern (low output power), and one flaw (output current doesn't meet spec). Anybody can grab the loaner unit and repeat my measurements. I'll be happy to share my methodology.

    There's a certain sense of pride that comes with designing something that works well. I don't blame analog designers for having pride in their work. I do wish that some analog designers would measure their gear and compare it with the performance of other gear available, however. You know... Get some perspective. Marketing is a funny thing. It's easy to be best in class when you define the class narrowly enough that you're the only one in it. :)

    Tom
     
  18. fishski13

    fishski13 Friend

    Friend
    Joined:
    Sep 25, 2015
    Likes Received:
    392
    Dislikes Received:
    0
    Trophy Points:
    63
    Location:
    Twin Cities, MN
    The gorilla in the corner is that the builder possibly screwed up. I've never spent so much time populating and quadruple checking a headphone amplifier PCB, or any amplifier for that matter. I ordered tape/reel packaging for the resistors and even made sure I measured one resistor from each string of tape to double check correct values before soldering.
     
    Kattefjaes likes this.
  19. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Friend MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,602
    Dislikes Received:
    16
    Trophy Points:
    113
    Location:
    Irvine CA
    I haven't seen the amp. But Tom did mention this:

    "Overall, this particular build is quite nice. It appears to be built according to instructions with one modification: The PCB mounted E-I core transformers used in the original build have been replaced with toroidal transformers connecting to the PCB with short leads"
     
  20. tomchr

    tomchr MOT - Neurochrome

    Joined:
    Oct 28, 2016
    Likes Received:
    499
    Dislikes Received:
    19
    Trophy Points:
    63
    Location:
    Calgary, Canada
    Home Page:
    As far as I could tell (I didn't check the component values) the board is built to spec. I fully trust your ability to read a BOM, order, and populate parts. Nothing seemed out of order or like it'd been hacked. The only exception was the use of toroidal transformers rather than E-I core ones, which seems reasonable. I have every bit of confidence that the build was successful. It was nicely done in my opinion. I was a bit surprised that you didn't get one of the Neutrik phone jacks, though. In particular as you use the rather nice Neutrik RCA connectors. The NJ3FP6C-B would be perfect. It wouldn't make a lick of difference on the performance, but it'd look cool. :) Nice build, man. No worries.

    Tom
     
    fishski13 likes this.

Share This Page