Schiit Loki Mini+ EQ Review and Measurements

Discussion in 'Headphone Amplifiers and Combo (DAC/Amp) Units' started by Marvey, Jul 7, 2021.

  1. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Wuut? When did the Loki Mini+, a revised Loki suddenly appear? That's the question no one seems to have an answer for - except for @rhythmdevils who I believe may have snagged one not too long after he was defrosted - after being trapped in the ice for decades by fascists. Oh wait, that's a movie.

    The original Loki I never really gave much thought to. As an audio purist somewhat, the OG Loki was just another unnecessary piece of gear in the signal path, mucking things up. Yes, I could hear the OG Loki when it was switched in the signal path. However, the truth is that I didn't know what to do with the wide EQ bands of the Loki. I knew how to use PEQ and/or digital EQ to get the results I wanted - in a surgical way. This is where @rhythmdevils came in. I've always loved his approach because it's experimental as he uses his senses without the aid of measurements, numbers, etc. It's a human approach. Probably too much to the chagrin of today's objectivists. But who cares? When was the last time we went to a headphone to meet up with real people with shared interests and bumped into an objectivist. Never. Anyway, I heard that @rhythmdevils that the Loki EQ is supposed to be fun, a right brained activity. Do it Marv, you have good ears. Use the Force. I know you don't need measurements. Have fun. And you know what, he's absolutely right!

    I finally figured it out: the Loki ain't an EQ, it's four way tone control. Go at it.

    Before we move on, first I will present my "RealTHD" Tic-Tac-THD panel.
    upload_2021-7-7_16-10-34.png

    We do not see any concerning issues. The measurements we see here are quite good! The higher THD with the low level sounds 2mVrms is at a voltage level 1000 times lower than max. This tends to be normal because we are starting to bump into dynamic range / noise floor limits. The high frequency THD at the max voltage is worse relative to the others, but on an absolute level, nothing to be concerned about. These are test signals, so it's more or less impossible that we will experience a 2Vrms 9.3kHz signal in real music. Finally, these numbers are academic. If you believe in the temple of SINAD, or that THD of -75 is really bad, then go away. SBAF is a forum of reason and science, not dogma and religion.
     
    • Like Like x 14
    • Epic Epic x 1
    • List
    Last edited: Jul 7, 2021
  2. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    One the first questions from the purists, myself included, is how transparent is the Loki Mini+? And is it more transparent than the OG Loki. To the second question, yes, it's more transparent than the OG Loki - and shockingly so. While using it with Magni 3+, L30, Jotunheim 2 even, I don't I even noticed when the EQ circuit was switched in and EQ knobs.

    The only time when I noticed the Loki was in way was when I brought my big-boy mostly open loop (only 1.5db of global feedback) 45 tube amp. (The 45 tube has a very linear output tube). And even then I had to strain to hear the difference. It takes about 5% of the engagement factor away, not enough for me to forgo using it for transducers with severe linear distortion problems such as this one. There is a difference, but it's close enough where I wouldn't bet my life on it or even bother with a double blind test. I think I could train myself to pass such a test, but too much effort to prove something so inconsequential.

    Random Graph and Panel (for your amusement)
    SINAD 1kHz, 100 kohm load, 2Vrms output, 20kHz bandwidth
    upload_2021-7-7_16-33-12.png
    upload_2021-7-7_16-34-6.png

    Probably not sufficient by social-media objectivist standards. But sure... listen to your headphones with massive linear distortion (frequency response irregularities) while you nitpick small amounts of inconsequential amounts of non-linear distortion - keeping in mind that Loki Mini+ can be used as a linear distortion corrector! Only a dummy would take massive linear distortion just to have lower non-linear distortion of no consequence. When and how did people turn so stupid in the modern world?
     
  3. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    So here's the really funky part, and a huge reason as to why the Loki Mini+ sounds so transparent: discrete parts and open loop. You guys will keep hearing this from me open loop (or minimize as much feedback as possible) will always sound better, more immediate, more engaging - at long as distortion is held in check reasonably. This is why opamp designs with 100db of gain sounds like shit.

    The easy way to have designed this EQ would have been to use opamps. Opamps for each band of the EQ circuit and finally another one for the buffer. Use caps in the feedback loops of the opamps to determine the bands. Voila. Done. This sounds meh. If you don't believe me, trying breadboarding one. Schematics for such designs are everywhere, including in many opamp datasheets. Opamps, caps, done.

    What a lot of people do not appreciate is what a brilliant (actually more like crazy) design Jason is at Schiit. That mofo decided to do an LC (that is inductor and cap) EQ. A quad of open-loop JFETs drives the LC network. The output stage looks like a miniaturized Magni 3+ type.

    PXL_20210707_204842928.jpg

    The only exception is the bass, the lowest EQ band, where it appears a gyrator circuit (note the two green electrolytic caps) is used. BTW, I learned that those inductors (black/gray cylinders across near the knobs) are totally 100% custom. We cannot find them in Mouser, Digikey, or whatever. The value of the lower-mid inductor is actually 150mH. Holy cow. I told you he's one crazy dude. The only reason the lowest band doesn't have an LC network is such a inductor would have been the size of the board (and probably also cost more than the EQ too). Inductors sort of go exponentially bigger with larger values. (BTW, the OPA there is used to drive the gyrator circuit for the bass band. If anywhere we need to compromise, it would in the lows. Besides, it was the only practical way.)

    I can't wait for Danny Ritchie to offer an upgrade kit that replaces the two green electrolytic caps ("signal path") on the board. :)

    BTW, I just noticed something in the THD panel - see below. This THD behavior looks more like it's caused by the reactivity of the passive LC components than the buffer circuits. It's all academic really and part of us learning what makes things tick.

    upload_2021-7-7_16-59-28.png

    Of course an opamp design with caps in their feedback loop will measure better. However, trust me, there is a reason why talented designers, or at least those who are capable of designing discrete (a discrete LC EQ isn't horribly complex), go with such designs despite their headaches (in this case, asking folks to make parts that did not exist). LC EQs sound better. They are more transparent, more engaging. They don't lose the essence of music. Anyone who tells you otherwise is lying do you, has insufficiently developed senses, looks at SINAD all day, or may have had Bill Gates tear him a new asshole each day.
     
    • Like Like x 17
    • Epic Epic x 2
    • List
    Last edited: Jul 7, 2021
  4. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Effect of knobs from leftmost to right. Let's call them bass, lower mids, upper mids, treble. Note that the reverse counterclock is the exact inverse of what is see here. Note that there is quite a bit of increase from the 3 o'clock position to the max at 5 o'clock. I can speak to this with real life use: we need to crank it a bit to get more than a subtle effect. It looks sensitive in this region, but it's really not during actual use.

    Bass knob maxxed out and at 3 o'clock
    upload_2021-7-7_17-45-0.png

    Lower mid knob maxxed out and at 3 o'clock
    upload_2021-7-7_17-43-42.png


    Lower mid knob maxxed out and at 3 o'clock
    upload_2021-7-7_17-42-10.png

    Note that that EQ controls at the ends (bass and treble) are about +/- 12db and the ones in the middle (lower mids and upper mids) are +/- 6db.

    Treble knob maxxed out and at 3 o'clock
    upload_2021-7-7_17-41-12.png
     
    Last edited: Jul 8, 2021
  5. gixxerwimp

    gixxerwimp Professional tricycle rider

    Pyrate
    Joined:
    Sep 30, 2015
    Likes Received:
    5,773
    Trophy Points:
    113
    Location:
    small island claimed by China
    Is this with all the knobs @ 12 o'clock and tone control active? How is the THD affected when the knobs are twiddled?
     
  6. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Good question. I'll crank all the knobs up, show the EQ profile, and take the same set of measurements tonight.
     
  7. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    upload_2021-7-11_10-12-2.png

    Middle Row +8db
    With both lower and upper mids cranked up (they overlap and will combine for more gain than +6db at 1kHz when used alone), we can get a bit more than +8db at 1kHz, but I turned the dial back slightly to get a good +8db so we could get the same with the 37Hz and 9.3kHz signals. Basically when the EQ knobs are set so that the gains at each of the frequencies are 8db, those are the THD figures. Note that all knobs are set to about the 4 o'clock position.
    upload_2021-7-8_23-22-59.png

    Top Row +12db
    The top row is when the bass and treble are cranked up (remember, these go all the way up to +12db for massive abuse). The lower mid and upper mid controls won't go up to +12db, so I didn't bother hence the greyed out middle column.
    upload_2021-7-8_23-20-29.png

    Note that analog EQs work differently than digital EQs. With digital EQs when we want to apply gain, WE BETTER set a negative pre-gain because digital clipping is nasty. Digital will not go higher than 0dbFS! The result is that if we want to add +12db to the bass, we end up losing 12db (moving up the noise floor), about 2-bits. Analog EQ lets us avoid this. However, the penalty is higher distortion. I thought this was worthy of mention because the SINAD nerds are probably freaking out over this kind of stuff. Sometimes analog has advantages.

    In any event, analog EQ with physical dials is for more right brain oriented people: creatives, musicians, or people who actually like to listen to music any way they like it. Digital EQ is much better for surgical EQ and obsessive perfectionists who love to look at headphone frequency response plots to get the perfect curve to match Sean Olive's consumer target (even though they will fail because everybody's ear is slightly different - unless they cut off their ears and attached a GRAS in their place).
     
    • Like Like x 14
    • Epic Epic x 2
    • List
    Last edited: Jul 11, 2021
  8. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    And yes, I actually wrote a computer program to draw Loki knobs in the way that I turned them. :D
     
    • Epic Epic x 11
    • Like Like x 9
    • List
  9. Metro

    Metro Friend

    Pyrate
    Joined:
    Dec 27, 2016
    Likes Received:
    1,590
    Trophy Points:
    93
    Location:
    San Francisco
    Last edited: Jul 9, 2021
  10. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,239
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    I love jerking my Loki knobs!

    What @purr1n says is true. It’s a very fun device. I’m too much of a purist to use it with full size headphones but also because all my headphones are orthos modded by myself so EQ feels like cheating. If there’s a problem I want to fix it with mods.

    But I use it all the time with iems. There’s already so much EQ going in on with the multiple drivers and crossovers in iems and only the Gaudio Nair has perfect FR.

    I just adjust the knobs until it sounds right. Using my speakers as a reference. I don’t own measurement gear or use any test tones.

    You can do it to! Use the force! What sounds good is good. As long as you have a neutral reference to keep you calibrated.

    This post makes me want to upgrade to a Loki+ But I honestly don’t hear a transparency issue with iems. Im skeptical that even the mighty Solaris is not transparent enough for it to make a difference. But I’ll probably get one anyways. How can I resist?
     
    • Like Like x 7
    • Epic Epic x 1
    • List
  11. nishan99

    nishan99 Friend

    Pyrate
    Joined:
    Dec 1, 2019
    Likes Received:
    1,617
    Trophy Points:
    93
    Location:
    Layla
    So if my upstream components have more gain than what I need and very low noise floor that I will never hear then the digital EQ is the better fit for me?.

    I use digital EQ moderately (+-3dB max) for all my systems and the only concern I have is possible loss of fidelity of negative pre-gain so it's why I am interested in trying transparent analog EQ if it provide better fidelity. I haven't used one so idk.
     
  12. dBel84

    dBel84 Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    3,229
    Trophy Points:
    113
    I might need to play with one of these @rhythmdevils has been advocating for the fun it brings to listening
     
  13. Taverius

    Taverius Smells like sausages

    Pyrate
    Joined:
    Dec 27, 2017
    Likes Received:
    3,026
    Trophy Points:
    113
    Location:
    Rapallo, Italy
    Yah I'm already using ReplayGain at basically all times, so I'd rather not add more digital preamp stacking.

    I'll pick up a plus when the maxi comes out I guess and move the small one to my bedroom 2ch.
     
  14. rhythmdevils

    rhythmdevils MOT: rhythmdevils audio

    Pyrate
    Joined:
    Apr 15, 2020
    Likes Received:
    12,239
    Trophy Points:
    113
    Location:
    Bay Area, CA
    Home Page:
    I have an extra non + Loki you (or anyone else) can borrow if you want.
     
  15. dBel84

    dBel84 Friend

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    3,229
    Trophy Points:
    113
    let me see what the non + is - I like the JFET open loop design idea of this one, I always thought the LOKI was just passive ..dB
     
  16. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Last edited: Jul 9, 2021
  17. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Oh, I forgot the distortion profiles. IMO, this is just as important if not slightly more important than overall THD. The below are for 2Vrms input.

    [​IMG]

    37 Hz
    0dBr is normalized to about 8Vrms.
    upload_2021-7-9_23-0-45.png

    9.3kHz
    0dBr is normalized to about 8Vrms.
    upload_2021-7-9_23-1-56.png

    Now the mids.

    [​IMG]

    1kHz
    0dBr is normalized to about 5Vrms.
    upload_2021-7-9_23-6-19.png

    Overall, the distortion profiles look very good. Mostly H2 and H3. The 37Hz signal when pushed all the way up has a nice triode type pattern with descending harmonics. The 9.3kHz maintains that shape somewhat, but with odd order just slightly higher then even order. H4 and above are a bit on the high side, but levels do descend. The 1kHz has the lowest overall distortion. Most of is H2 and H3 fairly even, followed up with odd order harmonics H5 and H7 at low levels.

    I really cannot comment on the subjective quality of these settings because they are pretty insane and what I would consider abuse of the EQ*. However, considering the level of boost: the bass distortion looks very good; the treble distortion is a bit on the higher side, still good, but harmonic pattern suggests a rich sound; the 1kHz doesn't have my favored distortion pattern of successive descending harmonics. However with H2 and H3 being fairly equal, no other even harmonics, and low H5 and H7, the latter into non-detectability by SOTA instrumentation - it's really hard to ask for better.

    *Actually I take that back. A few work buddies and I maxed out the bass and played some electronic music with deep bass on the HE400SE. No, we were not trying to kill the HE400SE (it's a very robust headphone), but trying to kill an amp. And yes, considering the SPL, the bass sounded surprisingly clean. (We listened at a distance of about 0.5 meters away). Unlike the guy at Audio Science Review, we couldn't hear the -77db THD.
     
    Last edited: Jul 9, 2021
  18. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    I will close this out with some a real world example before a conclusion. All the below settings were done by ear. The measurements are provided after the fact. Even when I opt to do surgical EQ, I always do an ear check.

    The reason for this is that frequency response measurements do not always align to perception of tonal response. There are often a variety of factors that come into play such as distortion, transient response, shape of ear, measurement irregularities, etc. A lot of people newer (or perhaps around a long time) actually think that frequency response is one of those an end-all be-all measurements. It's not.

    Grado RS-2 with Stock Bowl Pads
    GRY = Before WQ
    upload_2021-7-11_8-45-17.png
    upload_2021-7-11_8-46-45.png

    The bass dial set to cut is optional. I did a small cut because I wanted to back off midbass. The main issue with the lows is that it only needs more extension, not more midbass. A four band EQ won't be able to do what we want here. The highs I guess I didn't back off as much. Part of the reason is because they are uneven and I could only do broad strokes. This allows me to use the stock Grado pads (more open spacious sound) while obtaining a tonal characteristic closer to the TTVH pads (less dark actually).
     
  19. Marvey

    Marvey Super Friend

    Staff Member Pyrate BWC
    Joined:
    Jun 6, 2021
    Likes Received:
    1,712
    Trophy Points:
    93
    Conclusion: Skull Pyrate approves!
    DSC00260(2) (Large).JPG

    The Loki Mini+ is a broad-brush tool to correct linear distortion or set a headphones' tonal signature to whatever way that suits you. The transparency of the unit is astounding for its price with only the purists with the best gear being able to hear when it's in the component chain. (There's a switch to put the EQ circuit in or out of the way). The physical EQ knobs allows us to change stuff around in a matter of seconds without messing around with a mouse and keyboard. The analog aspect allows us to turn up the knobs without applying a precut as with digital (which will result is a dynamic range loss). The downside to its broad EQ bands is its inability to correct for narrow dips or peaks. Finally, there's nothing out there in the market like this.
     
    Last edited: Jul 11, 2021
  20. TheloniuSnoop

    TheloniuSnoop Friend

    Pyrate Contributor
    Joined:
    Oct 15, 2015
    Likes Received:
    588
    Trophy Points:
    93
    Location:
    Gold Canyon, AZ
    Be sure to have the unit powered on, even if the switch is set to bypass. Otherwise, nasty distortion. Don't ask me how I know that (we don't have to read any stinking manuals!)
     

Share This Page