schiit yggdrasil LIM technical measurements

schiit yggdrasil LIM technical measurements

Standard Prologue
If you are unfamiliar with audio measurements please use a search engine with the query:
"audio measurements" or "audio measurement handbook"
Look for publications by Richard C. Cabot and also by Bob Metzler, both from Audio Precision. There are other useful publications as well. These will provide basic knowledge.
Interpretation of the following measurements is beyond the scope of technical measurements posts.

The data presented were collected as follows:

1. PrismSound dScope III, picoscope 5243B, Cal Test CT2593-2 balanced probe
2. Lifatec glass fiber toslink cable
3. Balanced XLR cables Belden 1800F with Neutrik 110R AES connectors
4. Single Ended cables Mogami 2964 1 meter with Amphenol RCA connectors
5. 100 Kohm load used for measurements
6. dScope analyzer sample rate 48 KHz unless otherwise noted
7. DAC 44.1 KHz sample rate, 24 bit depth unless otherwise noted
9. Audioquest Forest and Schiit Pyst USB cables
9. Vaunix Lab Brick USB hub
10. Shielded 14AWG and 16AWG power cables
11. Reaper Digital Audio Workstation (DAW) for WASAPI input
12. jRiver media player for WASAPI input
13. ASIO4ALL for ASIO to Windows 10 Class Compliant driver

Measurements are made in accordance with AES17:2015

schiit yggdrasil LIM
Measurements for record commenced after warmup of at least 168 hours.
Measurements and listening evaluation were performed over a period of several months.
Select balanced measurements were performed twice, one month apart for consistency verification.

What follows is an extensive examination of yggdrasil LIM performance.

Index
Post 1 - measurement setup description, highlights
Post 2 - AES input - Bal XLR outputs
Post 3 - AES input - SE RCA outputs
Post 4 - Opt input - Bal XLR outputs
Post 5 - Opt input - SE RCA outputs
Post 6 - USB ASIO / WDM input - Bal XLR outputs part 1
Post 7 - USB ASIO / WDM input - Bal XLR outputs part 2
Post 8 - USB ASIO / WDM input - SE RCA outputs part 1
Post 9 - USB ASIO / WDM input - SE RCA outputs part 2
Post 9 - Dante ASIO 88K input - Bal XLR outputs
Post 10 - -90 dBFS ASIO / WDM / WASAPI comparision part 1
Post 11 - -90 dBFS ASIO / WDM / WASAPI comparision part 2
Post 12 - square wave response part 1
Post 13 - square wave response part 2
Post 14 - reserved for additional data and corrections

Notable highlights:
yggdrasil LIM demonstrates excellent performance, as has become expected of schiit products
Low distortion
Power supply noise spectrum are at or below -150 dBFS in balanced output !!!
Balanced output Dynamic Range of 105 dB, very good for a 16 bit DAC
Balanced output Cross-talk is very good with > 125 dB isolation
Balanced output Gain Linearity is nearly perfect to -110 dBFS, less than ±1 dB to -130 dBFS
Exceptionally low jitter
Single Ended performance is slightly less than Balanced output, as is typical
Single Ended performance is still exceptional compared to numerous other DAC SE outputs

Well done schiit!


Commentary:
Power supply is at least 50% of any audio component's design. The higher the level of performance the more the power supply matters. schiit recognizes this, enhancing yggdrasil LIM's power supply, keeping noise to 100s of nanovolt levels while supplying 10s of volts throughout the system. Exceptional performance by design.


Measurement setup example:


The yggdrasil LIM is an engaging DAC.
Here are three particularly synergistic listening setups (of many) I used to audition the LIM:

EC Studio B

A rich, luxurious presentation. After acclimating to the LIM, Spring2 KTE sounded dark in comparison. However after acclimating to Spring2 KTE, LIM sounded bright in comparison. Either provides a satisfying experience with Studio B.

SW51+

This amp continues to amaze me with so many similarities in sound quality to higher end tube amps. Many a satisfying hour was spent listening with this combination.

ECP DSHA3F

Magic happens with the LIM, ECP audio DSHA3F and HD800S for me. Many a good intention of listening only to one or two songs became entire album sessions.

Recap: all three setups shown provided immersive, gratifying listening sessions.

WASAPI special considerations
dScope application software was created over a decade ago. It has two Sound Card interface modes: ASIO and WDM, where WDM means Kernel Streaming (KS). It does not support a direct method of accessing WASAPI. dScope WDM access is subject to Windows mixer (and limiter) as discussed here:
https://www.superbestaudiofriends.o...-dac-technical-measurements.5770/#post-190596

To use WASAPI as a digital audio transport method to yggdrasil LIM it was necessary to create test signal wavefiles from the dScope:

Generating test signal wavefiles


The test signals were played through both a DAW (Reaper) and a media player (jRiver) while results were measured through analog inputs on the dScope.

playing wavefiles via WASAPI using jRiver / Reaper


Due to this multistage process only a select few measurements were performed with WASAPI as the digital transport method.

yggdrasil LIM power supply special note

yggdrasil LIM A04 THD+N THD nth-HD FFT AES input Bal output

Power supply noise spectrum are at or below -150 dBFS in balanced output !!!
0 dBFS = +14.3 dBu
-150 dBFS = -135.7 dBu
There is NO power supply noise. None. Nada. Very well done schiit!

yggdrasil LIM A04 THD+N THD nth-HD FFT AES input SE output

Single Ended RCA output has some power supply noise in the spectrum.
But the largest component at 120 Hz is -122 dBFS!
0 dBFS = +8.3 dBu
-122 dBFS = -113.7 dBu or approximately -114 dBu

Here are sound levels that would be produced by a variety of headphones at -114 dBu


Only the insanely sensitive Shure SE215 might allow one to observe the tiniest hint of AC mains noise but only if in an extraordinarily quiet space with less than 25 dBC SPL ambient. So there is no reason to develop excessive nervosa about yggdrasil LIM SE outputs.

 

AES input Bal output measurements

yggdrasil LIM Dynamic Range AES input Bal output

Excellent performance for a 16 bit DAC system

Dynamic range, in simplified terms
1. find maximum output voltage at 0 dBFS
2. find residual noise with a -60 dBFS 1 KHz stimulus, removed from analysis by window notch
3. Dynamic Range is the difference between maximum output and residual noise
Specifics are in AES17 section 9.3 (measurebators do your homework please)
Example here:
https://www.maximintegrated.com/en/design/blog/spec-dynamic-range.html

Why we have a -60 dBFS stimulus
Some clever codec designers include an output gate that shuts off when there is no signal present. This has the dual benefit of lowering output idle noise while also gaming the measurements. AES became wise requiring a -60 dBFS stimulus so any such gates are open during measurement and true Dynamic Range performance measured.

yggdrasil LIM A04 THD+N THD nth-HD FFT AES input Bal output


yggdrasil LIM Gain Linearity AES input Bal output


yggdrasil LIM Gain Linearity AES input Bal output -140 dBFS range


yggdrasil LIM Distortion vs Frequency AES input Bal output - right channel


yggdrasil LIM Imaging AES input Bal output


yggdrasil LIM Residual Noise AES input Bal output


yggdrasil LIM Inferred Jitter AES input Bal output


Complete AES input Bal output analysis report pdf attached for download.

 

AES input SE output measurements

yggdrasil LIM Dynamic Range AES input SE output


yggdrasil LIM A04 THD+N THD nth-HD FFT AES input SE output


yggdrasil LIM Gain Linearity AES input SE output


yggdrasil LIM Distortion vs Frequency AES input SE output - Left Channel


yggdrasil LIM Imaging AES input SE output


yggdrasil LIM Residual Noise AES input SE output


yggdrasil LIM Inferred Jitter AES input SE output


Complete AES input SE output analysis report pdf attached

 

Opt input Bal output measurements

yggdrasil LIM Dynamic Range Opt input Bal output


yggdrasil LIM A04 THD+N THD nth-HD FFT Opt input Bal output


yggdrasil LIM Gain Linearity Opt input Bal output


yggdrasil LIM Distortion vs Frequency Opt input Bal output - Left Channel


yggdrasil LIM Imaging Opt input Bal output


yggdrasil LIM Residual Noise Opt input Bal output


yggdrasil LIM Inferred Jitter Opt input Bal output


Complete Opt input Bal output analysis report pdf attached

 

Opt input SE output measurements

yggdrasil LIM Dynamic Range Opt input SE output


yggdrasil LIM A04 THD+N THD nth-HD FFT Opt input SE output


yggdrasil LIM Gain Linearity Opt input SE output


yggdrasil LIM Distortion vs Frequency Opt input SE output - Left Channel


yggdrasil LIM Imaging Opt input SE output


yggdrasil LIM Residual Noise Opt input SE output


yggdrasil LIM Inferred Jitter Opt input SE output



Complete Opt input SE output analysis report pdf attached

 

USB ASIO / WDM input - Bal XLR outputs part 1

This following sections will demonstrate why I prefer using Dante ASIO to DAC AES or spdif input.

ASIO = ASIO4ALL interface to Windows 10 Class Compliant driver
WDM = Kernel Streaming

yggdrasil LIM Dynamic Range ASIO input Bal output


Dynamic Range is approximately 101 dB which is a bit lower than AES input. Why this result occurs will be examined in the -90 dBFS performance in post 10 below.

yggdrasil LIM Dynamic Range WDM input Bal output


Dynamic Range is now on par with AES input at 105 dB, again discussed in post 10 below.

yggdrasil LIM Dynamic Range WASAPI (jRiver) input Bal output


Also on par with AES input for Dynamic Range.

yggdrasil LIM Distortion ASIO 0 dBFS input Bal output


yggdrasil LIM Distortion WDM 0 dBFS input Bal output


Note how distortion is significantly higher and spectrum looks considerably worse than ASIO at 0 dBFS. This is due to Kernel Streaming with a shared application path through Windows mixer and limiter as discussed here:
https://www.superbestaudiofriends.o...-dac-technical-measurements.5770/#post-190596


yggdrasil LIM Distortion WDM -0.3 dBFS input Bal output


When sine stimulus level is decreased to -0.3 dBFS the Windows limiter no longer engages and distortion returns to normal.

yggdrasil LIM Distortion WASAPI (DAW) 0 dBFS input Bal output


WASAPI in Exclusive Mode will bypass the Windows Mixer / Limiter. But only in Exclusive mode, which must be enabled in the driver properties for yggdrasil LIM AND requested by the DAW or Media Player. Both must occur to avoid having increased distortion for signals above -0.131 dBFS.

yggdrasil LIM Distortion WASAPI (jRiver) 0 dBFS input Bal output


Similar performance to the DAW WASAPI exclusive mode digital data transport method in the previous measurement.

yggdrasil LIM Gain Linearity ASIO input Bal output


While ASIO4ALL avoids the dreaded Windows Mixer / Limiter problem, it has an unusual interaction with yggdrasil LIM. It limits data to 16 bits which has considerable negative impact on Gain Linearity below -90 dBFS. This is an ASIO4ALL exclusive issue, and not a problem with yggdrasil LIM.

yggdrasil LIM Gain Linearity WDM input Bal output


Kernel Streaming (and all other modes) pass greater than 16 bits and has much better Gain Linearity at low levels.

 

USB ASIO / WDM input - Bal XLR outputs part 2

yggdrasil LIM Distortion vs Frequency ASIO input Bal output


yggdrasil LIM Distortion vs Frequency WDM input Bal output


Performance is substantially similar in either USB transport method.

yggdrasil LIM Imaging ASIO input Bal output


yggdrasil LIM Imaging WDM input Bal output


Performance is substantially similar in either USB transport method.


yggdrasil LIM Residual Noise ASIO input Bal output


Residual Noise floor is higher than either WDM or WASAPI shown below.

yggdrasil LIM Residual Noise WDM input Bal output


yggdrasil LIM Residual Noise WASAPI input Bal output


yggdrasil LIM Inferred Jitter ASIO input Bal output


Complete ASIO input Bal output analysis report pdf attached

 

USB ASIO / WDM input - SE RCA outputs part 1

See post 7 for detailed discussion between various USB Digital Data Transport methods used for SE measurements.

yggdrasil LIM Dynamic Range ASIO input SE output


yggdrasil LIM Distortion ASIO 0 dBFS input SE output


yggdrasil LIM Distortion WDM 0 dBFS input SE output


Similar to balanced measurements above 0 dBFS engages the Windows Mixer / Limiter which increases distortion.

yggdrasil LIM Distortion WDM -0.3 dBFS input SE output


Below the Windows Limiter threshold of -0.131 dBFS distortion returns to normal.

yggdrasil LIM Gain Linearity ASIO input SE output


As discussed above ASIO4ALL limits data to 16 bits when interacting with yggdrasil LIM, and is not a LIM issue.

yggdrasil LIM Gain Linearity WDM input SE output


yggdrasil LIM Distortion vs Frequency ASIO input SE output


yggdrasil LIM Imaging ASIO input SE output

 

USB ASIO / WDM input - SE RCA outputs part 2

yggdrasil LIM Residual Noise ASIO input SE output


yggdrasil LIM Residual Noise WDM input SE output


yggdrasil LIM Inferred Jitter ASIO input SE output



Dante ASIO 88K input - Bal XLR outputs

These four examples demonstrate Dante to AES input as having the best performance of all previous USB digital data transport methods and avoids Windows Mixer / Limiter issues while maintaining 24 bit depth communication to DACs.

yggdrasil LIM Dynamic Range Dante 88K input SE output


yggdrasil LIM Distortion Dante 88K input SE output


yggdrasil LIM Gain Linearity Dante 88K input SE output


yggdrasil LIM Inferred Jitter Dante 88K input SE output


Complete ASIO input SE output analysis report pdf attached
Complete Dante 88K AES input Bal output analysis report pdf attached

 

-90 dBFS ASIO / WDM / WASAPI comparision part 1

yggdrasil LIM does not have custom ASIO drivers available and relies on Windows 10 Class Compliant USB audio drivers. ASIO4ALL can provide an ASIO interface for this situation. While ASIO4ALL solves one problem, avoiding Windows Mixer / Limiter and is an easy method to avoid installing many custom DAC device drivers, it has potential to cause other performance issues.

One important control for ASIO4ALL is the Force WDM Driver To 16 bit:


However ASIO4ALL and Windows Class Compliant USB audio driver appear to not behave correctly with yggdrasil LIM. It made no difference how the control was set. Output to DAC was always limited to 16 bit as can be seen in the next three measurements:

yggdrasil LIM -90 dBFS ASIO input Bal output – Force WDM to 16 bit unchecked


yggdrasil LIM -90 dBFS ASIO input Bal output – Force WDM to 16 bit checked


yggdrasil LIM -90 dBFS ASIO input Bal output – Force WDM to 16 bit unchecked
no dither generated by dScope

Dither is normally on when sending dScope digital signals


Why I have standardized on Dante digital audio data transport in my lab.
Here are three measurements where dScope sends signal to Dante via ASIO which is output to AES on the Dante receiver and analyzed with dScope digital input.

dScope -90 dBFS to Dante ASIO to Focusrite D16R AES out to dScope digital in


dScope -120 dBFS to Dante ASIO to Focusrite D16R AES out to dScope digital in


dScope -130 dBFS to Dante ASIO to Focusrite D16R AES out to dScope digital in

The two previous measurements indicate Dante is sending accurate 24 bit data, no funny business. Most DACs have really good performance with AES or spdif inputs. Further, Dante word clock may be disciplined by an external clock such as my Rubidium clock + LiveClock combination for lower digital audio phase noise and long term frequency stability.

 

-90 dBFS ASIO / WDM / WASAPI comparision part 2

Now for a comparison of WDM (KS), WASAPI and Dante.
dScope has the ability to send 16, 24 and 32 bit data out the digital audio interface.

yggdrasil LIM -90 dBFS WDM input Bal output – 16 bit no dither

Results are as expected with DAC output toggling between least significant bits.
Note the FFT distortion present.

yggdrasil LIM -90 dBFS WDM input Bal output – 16 bit with dither

With dither the noise floor is raised slightly but with the benefit of eliminating distortion components. What is left is a fairly pure -90 dBFS sine.

yggdrasil LIM -90 dBFS WDM input Bal output – 24 bit with dither

Noise floor drops significantly while the -90 dBFS sine appears cleaner.

yggdrasil LIM -90 dBFS WDM input Bal output – 32 bit with dither

Sending 32 bit data does not make any further improvements.


The following measurements were performed as described in the first post using Reaper for the DAW and jRiver for the Digital Media Player.
All signals were sent 32 bit with dither to yggdrasil LIM.

yggdrasil LIM -90 dBFS WASAPI input Bal output - Reaper

Performance is substantially similar to WDM 24 bit with dither above.

yggdrasil LIM -90 dBFS WASAPI input Bal output - jRiver

There was a 6 dB gain somewhere in jRiver that I didn’t track down, but performance again was substantially similar to WDM 24 bit with dither above except for the 6 dB gain.

yggdrasil LIM -90 dBFS Dante ASIO input Bal output – dScope direct

Dante ASIO is also substantially similar to the others above with no funny business.

yggdrasil LIM -120 dBFS Dante ASIO input Bal output – dScope direct

At -120 dBFS Yggdrasil LIM is still producing a credible sine. This would be true for either Dante ASIO or WASAPI.

 

square wave response part 1

Balanced XLR output 0 dBFS 20 Hz sine stimulus measured 11.5 Vpp
The following measurements are from the balanced outputs

Yggdrasil LIM 20 Hz 0 dBFS sine Opt input

Sine 0 dBFS calibration measurement. Vertical rulers are adjusted to peak excursions.

Yggdrasil LIM 20 Hz 0 dBFS square Opt input

Vertical rulers calibrated in previous measurement allow an observations DAC output design headroom. Observe very slight amount of clipping on the filter ringing. A good compromise in a design tradeoff against Dynamic Range achieved. Also observe DC coupling as seen in the inset.

Yggdrasil LIM 20 Hz 0 dBFS square Opt input

17.1 uS 10 to 90% transition time infers 20.5 KHz bandwidth.

Yggdrasil LIM 20 Hz 0 dBFS square AES input

Switching to AES input the output is observed as identical to Opt input measurement above.

Yggdrasil LIM 20 Hz 0 dBFS square WASAPI USB input via DAW Reaper

Again the output performance is basically identical to Opt and AES above.

Yggdrasil LIM 20 Hz 0 dBFS square Dante ASIO to AES input 44 KHz sample rate


Compare the filter ring time above at 44 KHz with 88 KHz below.

Yggdrasil LIM 20 Hz 0 dBFS square Dante ASIO to AES input 88 KHz sample rate

Filter ring duration is shorter at 88 KHz than at 44 KHz, as expected due to doubling the sample rate.
Observe how the filter ring is no longer slightly clipped. Another bonus for 88K.

Yggdrasil LIM 20 Hz 0 dBFS square Dante ASIO to AES input 88 KHz sample rate

8.65 uS 10 to 90% transition time infers 40.5 KHz bandwidth, as expected due to doubling the sample rate.

Yggdrasil LIM 20 Hz 0 dBFS square Dante ASIO to AES input 44 and 88 KHz

Overlay of the 44 KHz with 88 KHz transition for visual comparison of both filter ring duration and transition Rise Time.

 

square wave response part 2

Single Ended RCA output 0 dBFS stimulus measured 5.7 Vpp
The following measurements are all from SE RCA outputs.
Performance and is substantially similar to the Balanced output XLR in the previous post.

Yggdrasil LIM 20 Hz 0 dBFS sine AES input

Vertical ruler calibration

Yggdrasil LIM 20 Hz 0 dBFS square AES input

Same very slight clipping of filter ringing.

Yggdrasil LIM 20 Hz 0 dBFS square AES input

17.04 uS 10 to 90% transition time infers 20.5 KHz bandwidth.

Yggdrasil LIM 20 Hz 0 dBFS square Opt input

Essentially identical to AES above.

Yggdrasil LIM 20 Hz 0 dBFS sine WASAPI USB input via DAW Reaper

WASAPI vertical ruler sine calibration

Yggdrasil LIM 20 Hz 0 dBFS square WASAPI USB input via DAW Reaper

Same very slight clipping of filter ringing.

Yggdrasil LIM 20 Hz 0 dBFS square WASAPI USB input via DAW Reaper

17.01 uS 10 to 90% transition time infers 20.6 KHz bandwidth.

 

Reserved for additional data and corrections.

 

Amazing dedication. We don’t deserve you Bob.

 

Great stuff Bob, thanks.

Re: the Windows limiter, am I understanding correctly that, for practical purposes when not playing with WASAPI/ASIO output and instead playing through the mixer, the way to prevent it from kicking in is to set Windows volume for the device (e.g. if using Unison USB w/ LIM in this case) to "99%" or lower?

 

The windows limiter threshold is -0.131 dBFS which corresponds to 98.5%. That is the level below which will be safe when using a USB audio path that goes through the mixer. This level adjustment must be prior to input to the mixer.

 

Bob, thanks for all that you do. Here is a newbie question for you. Naturally, this being a 16-bit dac and with all of the "higher" bit dacs around these days, how would you describe the subjective difference between the two?

You mention "Balanced output Dynamic Range of 105 dB, very good for a 16 bit DAC", other DACs like the Holo May for example exhibit a much higher spec in this area. Does that mean there will be a perceivable difference between the most quiet areas of a track and the more dynamic portions? How does that spec translate to what we hear? (if at all).

Very happy LIM owner FWIW.

 

Don't let technical nervosa detract from your auditory enjoyment.
LIM is behaving as if it delivers 17.15 to 17.65 honest bits.
There are numerous DACs on the market quoting the data stream of 24 or 32 bits and using that as marketing bits. In reality many only deliver 15 to 16 honest bits.
Holo May is on another level but for many reasons beyond Dynamic Range. However, if you aren't a fan of NOS R2R then you may not find Holo May a good fit with your personal preferences.

 

Thanks Bob. Just to clarify though, I’m far beyond the stage of measurements nervosa. Just trying to understand your comment about this measurement being good for a 16-bit dac.