• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Review and Measurements of Chord Hugo 2 DAC and Headphone Amplifier

stalepie

Active Member
Joined
Jan 3, 2018
Messages
142
Likes
21
#41
JH Audio Angie dips below 4Ω at points. It's a niche case where <0.5 Ω is needed for output impedance much like the Hifiman HE-6 is for power required.
"Niche case" means "any" is inaccurate in the sentence "At 1.7 ohms, it is very low making it suitable to driving any headphone without changing its frequency response."
 

Candlesticks

Active Member
Joined
Jan 19, 2018
Messages
108
Likes
124
#42
"Niche case" means "any" is inaccurate in the sentence "At 1.7 ohms, it is very low making it suitable to driving any headphone without changing its frequency response."
There is no headphone that low in impedance.

There are always going to be niche exceptions to a rule. People who buy the niche exceptions know about it beforehand.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,469
Likes
4,301
Location
Seattle Area
#43
Also @amirm how does the Utopia sound subjectively (It is $4000 :eek:).
Unfortunately I had no time to listen to this specific sample. I did listen to them at the last show (RMAF 2017). It did not do too much for me but this was with someone else's music.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,469
Likes
4,301
Location
Seattle Area
#45
JH Audio Angie dips below 4Ω at points. It's a niche case where <0.5 Ω is needed for output impedance much like the Hifiman HE-6 is for power required.
As usual you guys know more about breadth of headphone products than I do. :)
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,469
Likes
4,301
Location
Seattle Area
#47
I meant to note that. :) There are some accuracy errors here with respect to frequency used, length of wires, temperature and exact value of dummy load, etc. So the measurements should be used as ballpark and not exact.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,469
Likes
4,301
Location
Seattle Area
#48
I just added this to my initial post:

Edit: Speaking of corrections, my friend who loaned the Chord Hugo 2 to me tried it on a Windows machine and it worked with in-box class drivers. So something went wrong on my machine but support is there. So I take my reservation away regarding driver support.

 

drconopoima

Member
Patreon Donor
Joined
Jan 14, 2018
Messages
49
Likes
27
#49
Candlesticks said:
JH Audio Angie dips below 4Ω at points. It's a niche case where <0.5 Ω is needed for output impedance much like the Hifiman HE-6 is for power required.
"Niche case" means "any" is inaccurate in the sentence "At 1.7 ohms, it is very low making it suitable to driving any headphone without changing its frequency response."
There is no headphone that low in impedance.

There are always going to be niche exceptions to a rule. People who buy the niche exceptions know about it beforehand.
I would be surprised in a bad way if such an exception to the rule wasn't designed to sound their best with some variation in their target frequency response compared to the one obtained with a negligible low output impedance source. Since there almost don't exist low enough output impedance sources for a sub-5 ohm headphones, they should have taken that into account. That said, I haven't tested it nor either seen reports on how they sound their best.
 
Last edited:
Joined
Feb 7, 2018
Messages
44
Likes
12
#50
The claim to fame of Chord DACs is a super high-tap FIR filter for reconstruction. This one claims over 490,000 taps. This means each (upsampled) audio sample goes through this many computations before a value spits out.
Can someone explain what this is all about? I kind of computations are they doing and how does that effect the output? I've never heard anything about a tap filter before.
 

RayDunzl

Major Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
5,116
Likes
1,158
Location
Riverview, Fl
#51
I think:

More taps give more control at low frequencies, and more taps for higher sample rates.

The taps store a number of consecutive samples of the music (mine has 6144 taps, or .128 seconds of the input file at 48kHz ) and a mathematical manipulation (the filter, which in my case has 6144 coefficients) is performed upon all those samples to come up with one new sample (filtered) to output.

Then, shift the music signal stored in the taps by one sample (bring in the next sample, lose the oldest), and repeat.

200 FIR filter values around the center of a filter (room correction for my JBL), plotted:

upload_2018-2-14_18-53-1.png
 
Last edited:
Joined
Feb 7, 2018
Messages
44
Likes
12
#52
I think:

More taps give more control at low frequencies, and more taps for higher sample rates.

The taps store a number of consecutive samples of the music (mine has 6144 taps, or .128 seconds of the input file at 48kHz ) and a mathematical manipulation (the filter, which in my case has 6144 coefficients) is performed upon all those samples to come up with one new sample (filtered) to output.

Then, shift the music signal stored in the taps by one sample (bring in the next sample, lose the oldest), and repeat.

200 FIR filter values around the center of a filter (room correction for my JBL), plotted:

View attachment 10642

Thanks for the reply, but I'm still not sure what their all about. I will do some more research online, maybe checkout chord's site to try and get a handle of this. I have no idea how to interpret that graph, not sure what's going on and what it's suppose to be showing. :confused:
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,469
Likes
4,301
Location
Seattle Area
#53
Can someone explain what this is all about? I kind of computations are they doing and how does that effect the output? I've never heard anything about a tap filter before.
Briefly, the output of every DAC needs a "reconstruction" filter to remove the high frequencies above half the sampling rate. When done in digital domain using a "FIR" filter a figure of merit is the number of "taps." Each tap requires an audio sample and a multiply+add arithmetic function. Here is a quick animation:



Usually by the time to you get to a few hundred taps or so, you have a near ideal response. Rob Watts on the other hand thinks you need to go to 100,000 taps or even more! When computed, the difference this increased number of taps makes is infinitesimally small yet he says it is audible.
 
Joined
Feb 7, 2018
Messages
44
Likes
12
#54
Briefly, the output of every DAC needs a "reconstruction" filter to remove the high frequencies above half the sampling rate. When done in digital domain using a "FIR" filter a figure of merit is the number of "taps." Each tap requires an audio sample and a multiply+add arithmetic function.

Usually by the time to you get to a few hundred taps or so, you have a near ideal response. Rob Watts on the other hand thinks you need to go to 100,000 taps or even more! When computed, the difference this increased number of taps makes is infinitesimally small yet he says it is audible.
Thanks, amir! This explanation and the animation really helped out.

So on dacs that have selectable digital filters, these would be the reconstruction filters, correct? Do all digital filters use "taps," or is this just something implanted for FIR type filters? Does this question even make sense? ;) I've only been learning about digital audio and dacs from a more technical side for a few months now so this is all still very rough around the corners for me.
 

amirm

Founder/Admin
Staff Member
CFO (Chief Fun Officer)
Joined
Feb 13, 2016
Messages
12,469
Likes
4,301
Location
Seattle Area
#55
Correct. They are different filters with varying response in frequency and time domain. There are other digital filters like IIR which do NOT have taps.

This is a dense complicated field so don't worry about not being to keep up yet. :)
 

Soniclife

Senior Member
Joined
Apr 13, 2017
Messages
399
Likes
81
Location
UK
#57
Yeah, I've been learning a bit here and there on digital filters over the last few months and still feel like I know essentially nothing. Good Times! Truly love learning about this stuff.
Archimago has lots of good stuff about filters on his blog, I tried to find a really good start point and the following is the best I could do, best just to dive in, read a load of stuff till it starts to get clear in your head.

http://archimago.blogspot.co.uk/2017/02/musings-discussion-on-mqa-filter-and.html
 

Blumlein 88

Major Contributor
Joined
Feb 23, 2016
Messages
3,536
Likes
1,637
#59
Thanks, I will definitely take a look at this.
https://www.embedded.com/design/configurable-systems/4025591/Digital-filtering-without-the-pain

Though I am not filter guru a very simple approach like in the above link is what made it possible for me to get a handle on and start learning about digital filtering.

Digital filters are like moving averages. You can read the link above, but here is a taste of my own version. Yes it is simplified and leaves a little out, but is not conceptually crippled.

Suppose you have setup a 5 box moving average. You feed it samples as numerical values. When you have samples in 5 boxes you add those boxes together and divide by 5. This becomes the next output sample of your filter. You move in one more sample, and move out one sample, add the boxes and divide by 5 again and spit out another sample. That in essence is all a digital filter is doing. This one would be a 5 tap filter. Something of a low pass filter.

Now where they get complex is imagine the same filter except in box one I add that value to our sum of the 5 boxes. In box 2 I multiply the value by 2 and add it to our 5 box sum. In box 3 I multiply by 3 and add to our 5 box sum. In box 4 I multiply by 2 and add to our sum. In box 5 I simply add the box value. I take the sum and again divide by 5. This filter will produce a different output. I think this one would be something of a bandpass filter. The important thing is you have various versions of moving averages to accomplish different types of filters. The math for a few hundred tap filter with a coefficient for each tap and the function of the sum of the taps can become complex math in a hurry. In essence they are still varieties of moving average filters.

After this the next useful thing is to look at what happens in such a simple filter when upsampling. Suppose our little 5 tap (box) filter is running at 88.2 khz sample rates, but our input samples are running at 44.1 khz rates. A common technique is zero stuffing. You take your first 44.1 khz sample, and follow it with a zero value, then the next 44.1 khz sample and follow it with a zero sample value. So on and so forth. Our 88.2 khz digital filter is receiving its 88,200 samples per second. Half are from our 44.1 khz sample stream and half are zeroes we stuffed in there. It still will function as a digital filter. It will put out a useful stream that has the info in our original lower sample rates. It will allow us to do more complex multiplying and summing for more complex possible results. Things like noise shaping of the output to gain resolution at lower frequencies by placing noise at higher frequencies where we don't care.

I hope this helps, and I hope I haven't ruined some important concept to mislead you. If so our more knowledgeable members will hopefully correct me.

EDIT: thanks to wakibaki who explained this and more to me a couple years back. Not sure he is with us anymore as he had health problems. A loss as he was knowledgeable and helpful.
 
Last edited:
Joined
Feb 7, 2018
Messages
44
Likes
12
#60
Blumlein,

Thanks for the link and the quick summation. This did help me to get a better grasp of things here. Still plenty to learn but this does clarify a few things for me now.
 
Top Bottom