Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[LPED]

Hi all,

could need some help as a newbie on PC sound, plz. (I have just hanged from old styled CDs to PC...)

I do know that my newly bought "used" DAC2 from EMM Lab has great potential. However I am not quite pleased with the sound quality.

Especially the Sī and Tī sound horrible. And this goes both if I am using my Stax headphones or my Martin Logan speakers. My Pre amp and Power amp are Conrad Johnson. So all in all the potential for great sound is present.

I wonder if there is something in my JRiver settings which is completely wrong?

I might add currently I am ripping to an external Hard Disc, using a cheap Acer Laptop. Does this have any influence? 

I am using USB to USB from PC to DAC. BTW I discovered great sonic influence from the USB cable itself - possibly strange but without any doubt a big tweak.

All feedback will be highly appreciated :-)

[mwillems]

Welcome to the forums 

What do you mean by the "S' and T' " sounding horrible? 

There are a few settings that are worth checking on.  What are your audio device settings in JRiver?  Are you using WASAPI or something else?  What are your buffer settings?  How does the DAC sound when you play from another media player on your PC (like windows media player)?

[LPED]

Thank you.

īSī and īTī sound harsh/hard and are overdisplayed.

The JRiver settings are:
Output Mode: Direct Sound
Buffer Settings: 0,25 s
Prebuffering: 6 s

JRiver is by far superior to Windows Media Player using the DAC from EMM Lab.

All tips/tricks are welcome 

[6233638]

Nothing in Media Center should be causing sibilance.
It is often caused by the source material. It's also possible that your previous DAC had a rolled-off high frequency response which reduced it. (or the new DAC has a treble emphasis - though that seems less likely)
 
I believe the Martin Logan speakers have a reputation for being somewhat sibilant, and I find sibilance to be more problematic with lossy sources compared to lossless ones.
You might want to try the parametric EQ with a low-pass filter around 4KHz to see if it helps.

[felix2]

My opinion is that the EMM Lab DAC2 has serious design faults. The result is significant distortion of output such as the annoying Ss and Ts.

The EMM Lab DAC2 is designed for DSD signal and does not handle PCM correctly. Already a few years old the USB receiver of the DAC2 lacks the refined design of today's receiver chip. It claims to support up to 24/192 but it is not implemented correctly. But the most serious design fault is it upsamples everything to 2XDSD frequency. Such a huge amount of upsampling create a great deal of high frequency noise, which must be blocked using the most drastic 'brick wall' filter design. But such a filter will mess up DSD analog output big time. So DAC2 soften up the brick wall to make DSD work, but this allow a high amount of high frequency distortion to pass. The result is the large amount of Ss and Ts you hear.

The design of the DAC2 is fundamentally faulty which is why this product was discontinued quickly. Media Center produces the very best bit-perfect audio streams of any such app in the market so don't blame it. I have use MC to feed many of today well-designed DAC products and they all yield some of the best sound.

[6233638]

If the problem is with its handling of PCM signals, you could try Media Center's realtime DSD encoding? (in the DSP Studio > Output Format settings)

[Arindelle]

The JRiver settings are:
Output Mode: Direct Sound
Buffer Settings: 0,25 s
Prebuffering: 6 s

Did you miss this guys?

Have no experience with this DAC at all but your basic config is not, IMO, bit perfect using Direct sound. Try with WASAPI (or ASIO if it comes with your hardware) and get back to us. Using direct sound there are so many reasons why it could sound like crap.

Also your buffer setting seems odd if you are just starting out. Is there are reason for this? I'd try the default of 100 milliseconds first

The Staxx and Martin Logans are very precise components -- they are also unforgiving.  Before splitting hairs about the DAC's problems in PCM, might want to give it a chance with a bit perfect source 

[mwillems]

You might want to try the parametric EQ with a low-pass filter around 4KHz to see if it helps.

I think some kind of lowpass is definitely a good idea for troubleshooting here, but I would recommend setting it higher than 4KHz: on all but the 6dB setting you would be losing almost all of the audio content above 8Khz.  On the 6dB setting you'd have 8 dB of attenuation around 10 KHz, and 10 dB of attenuation at around 12 or 13 KHz, which is quite a lot of attenuation in the audible band.  It only takes a dB or two of over-emphasis to cause sibilance. 

I'd suggest starting the lowpass a little higher (8 or 10 KHz if you're using the 6 dB setting, higher if you're using a steeper filter).  If it's not just treble overemphasis, but is instead intermod from ultrasonics as suggested by felix2, a  lowpass would definitely help with that as well.

If there's treble overemphasis between 2 and 10 KHz, I'd suggest using a lowshelf for troubleshooting, or even just playing around with attenuation in the graphic equalizer.

OP, I agree with Arindelle, you should definitely try WASAPI as your output setting and see if that helps, directsound is not bitperfect, and may (conceivably) be partly responsible for what you're hearing.

[TedSmith]

But the most serious design fault is it upsamples everything to 2XDSD frequency. Such a huge amount of upsampling create a great deal of high frequency noise, which must be blocked using the most drastic 'brick wall' filter design. But such a filter will mess up DSD analog output big time.

Probably I should let this pass, but this analysis is incorrect.  Upsampling to 2xDSD allows a much gentler filter to be used and it also allows a lower noise floor in the audio band.  See the first attached picture and note the vertical scale.  The three traces are for three different tradeoffs of noise floor against frequency response.  Note also that this is a simulation, no hardware actually has a noise floor that low.  The second attachment is an actual measurement of the noise in a 2xDSD upsampling system.  The measured noise floor is that of my 8 bit scope and the spikes are artifacts from my poor measurement grounding.  The filter required by 2xDSD can be quite shallow and the ultrasonic noise doesn't have to be significant.

I haven't personally listened to a EMM DAC2 for extended periods of time, but I have a EMM DAC6e and am sensitive to sibilance.  I wouldn't blame the DAC2 (or JRiver MC).

[felix2]

The issue of upsampling has received much debate.

Purist designers refuse any upsampling arguing that the biggest cause of digital noise is large jitter. They proceeded to build ultra-low jitter DACs and the result speaks loud and clear. Low jitter DACs has the finest sound, especially coupled with a ultra low noise power supply.

Some designers argue that a reasonable degree of upsampling is desirable because the DAC has to receive audio of various compressed and uncompressed quality in the real world. Upsampling improves lower quality audio input. Such DACs use up to 8X over-sampling, which yield some amount of ultra-sonic products, but which can be filtered out reasonably well. So this is a sound compromise approach.

But EMM Lab DAC2 uses not a reasonable 8X, but an astounding 64X, even 128X. There is absolutely no technical reason to do this, except to convert the input signal to fit the internal DSD core. Math analysis shows a 64X over-sampling yield insane amount of high frequency distortion, a good chunk of the harmonics are actually at within 20KHz band. An insane brick-wall filter is required, which make the DAC performance insanely bad.

Now you don't have to accept my analysis. Just check the market and find how many successful commercial DACs uses a 64X over-sampling design.

[TedSmith]

But EMM Lab DAC2 uses not a reasonable 8X, but an astounding 64X, even 128X. There is absolutely no technical reason to do this...

False, my graphs show the actual HF content, but this isn't the place for argument and it isn't on topic.

[Arindelle]

but this isn't the place for argument and it isn't on topic.

+1

C'mon guys - start a new post if you want. The OP is apparently just getting into digital, helping him set up JRiver to its defaults as a starting point with the equipment he has would be a good thing. ... but puttting down his DAC is not terribly motivational. (and I'd say jumping to conclusions, btw^^)

A slightly more positive take on this DAC (ok the 2X version) from Chris who put the 2x version on his top list >Computer emm-labs-dac2x-review. If I can get a Cambridge DAC Magic bought on eBay for 150 euro work well when paired with a fully active bi-amped Linn system, I think we can help the OP take "empirical" steps to make his SQ issues go away with this DAC ... just my 2 cents

PS - He might not be ready to understand implementing low pass filters and parametric eq in JRiver yet

[VM]

the problem could be that the output on the XLR is 7,2V (high position) and 4V (low position) and is very possible that the high voltage output is too high for Conrad Jonson preamp input circuits. I had the same problem and after modification input impedance of my preamp the problem was completly solved