This is a detailed measurement and review of the Chord Hugo 2 DAC. It retails for $2,700 in US. The unit I had in for measurements was from an ex-Microsoft colleague and recent forum member. I only had an hour or so with it and it is now returned.
This is an odd looking box to be sure. As you see above the oddity continues in its ports with S/PDIF input being a stereo 3.5 mm jack (it has dual S/PDIF inputs as a result). I had my friend operate the unit as I did not even want to try to decode the button lights. The unit has batteries and that is how I tested it, likely giving it an edge in mains related noise/distortion.
The claim to fame of Chord DACs is a super high-tap FIR filter for reconstruction. This one claims over 49,000 taps. This means each (upsampled) audio sample goes through this many computations before a value spits out. FPGAs are custom ICs that are very well suited for this type of application which is what this DAC uses. The tap number has been positioned as a figure of excellence with each newer model sporting more and more taps. Marketing people rejoice!
Rob Watts the designer says that there are audible improvements with increased taps. Unfortunately when I asked him whether he had evaluated such merit in blind testing, he said no.
On the software front this DAC unfortunately gets a failing grade by requiring a custom driver or it does nothing. No USB class is supported and plugging in the device without drivers did absolutely nothing in Windows 10. This is a bummer because if the OS driver interface changes and this company is no longer in business, you have a $2,700 door stop.
For my testing I installed ASIO drivers and used bit-exact output to drive it from the PC. Other tests use S/PDIF digital output from my Audio Precision analyzer.
Measurements
For this test I had to pull out the only DAC I have on my bench that rivals it in price, namely the Exasound E32 ($3,400). For a test or two I also threw in the much cheaper Topping DX7 at $400 although I purchased mine for $300.
First, let's run our usual J-Test for jitter and noise. A test tone of 12 Khz should be the only thing visible here with the rest being noise and distortion. I ran the test two ways: using USB output from the PC and S/PDIF generated by Audio Precision Analyzer. Comparison is made to Exasound E32 over S/PDIF:
We see a very competent execution here with USb and S/PDIF matching each other. The Exasound is just as good but mains related hum and noise is there at very low level which can be remedied with a better power supply. Even as is, that noise is at very low level so not an audible concern either way. So little to complain about in either device. BTW "better" in the graph means better than Exasound E32.
Next let's get another boring graph out of the way, namely frequency response with different filters:
The default filter in red has the best frequency domain response with nearly perfect response. The others have small amount of roll off (about 1 dB at 20 Khz). I let subjectivists play with those filters and report back.
Let's get into my favorite test, linearity. And ideal DAC would have a flat line meaning it outputs what it is told to output:
Both DACs eek out about 18 bits of output before deviation exceeds 0.1 dB (my personal limit). In the case of Exasound it seems to just get noisier. On Chord Hugo 2 it gets funky jumping up and down. Both of them are very respectable though.
Both of them unfortunately get schooled by the Topping DX7 which costs a fraction of them:
OK, let's not spoil our dinner and continue.
Let's look at how a very low level sine wave at -90 db looks like in these two DACs:
Please excuse the dim output for Exasound. It was at a different size window and I had to resize it to match. Regardless, hopefully you see that the Exasound E32 is a closer match to a sine wave and freer of noise/glitches. Its DC offset changes a bit but that is due to power supply hum. Overall Exasound E32 wins.
Let's look at Harmonic Distortion+Noise:
Here, the Chord Hugo 2 DAC has higher noise floor but better control of distortion products. I give it the nod since noise is more benign than distortion products.
Next let's look at SMPTE intermodulation test:
The Exasound in green is the clear winner here and by good bit. The Chord Hugo 2 matches the Topping DX7.
Let's compare the distortion and noise products with a 1 Khz signal that itself has been filtered out:
Chord Hugo 2 pulls way ahead here. Yes it has higher noise floor but otherwise has less distortion spikes. The Exasound produces lot of glitches in 2 to 3 Khz which I suspect is due to mains distortion products.
None of this is audible concern though due to very low levels. Highest peak is at -120 dbFS.
Let's skin this cat differently and look at THD+N vs frequency:
Here the Chord Hugo 2 is the winner (in red). It has similar distortion+noise to Exasound E32 at low frequencies. At frequencies above 2 Khz the noise+distortion jumps up in Exasound. The Topping DX7 and Hugo 2 do the same but at much lower amplitude.
Headphone Measurements
I measured the output impedance of Hugo 2 and compared it to others on record:
At 1.7 ohms, it is very low making it suitable to driving any headphone without changing its frequency response. I need to measure and add the Exasound to this graph.
EDIT: there have been complains about this measurement saying that Chord advertises a much lower number (0.02 ohm?). The only way my measurement is off is if the unit clips under load of 33 ohm. If it does, then that obviously shows a larger drop resulting in impedance to be higher than if it were not clipping. I will try to get my hands on the loaned unit again in the future to confirm this.
Next I am going to show a new test which is the THD+N not against a dummy resistive load, but real headphones. This is based on work done by Benchmark where they demonstrate this problem with rising low level distortion when real loads are used instead of artificial:
Sadly we see that noise and distortion rise especially in low frequencies. Worst performance is with the Focal Utopia. Best was with Hifiman HE1000 V2. I will need to run this test on my other headphone amplifiers and see how they behave.
Summary
The Chord Hugo 2 is competently designed with measurements that show no obvious design/execution problems. Its high number of filter taps though provide it with no advantage against traditional implementations such as in Exasound E32 or even Topping DX7.
Despite these good measurements, I cannot recommend it due to requirement for proprietary driver. If that is not a concern for you, then I don't see any grave reason to not buy it other than high cost.
As always comments, corrections, feedback, bad jokes, etc. are welcome.
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.
If you like this review, please consider donating funds for these types of hardware purchase using Patreon (https://www.patreon.com/audiosciencereview), or upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
This is an odd looking box to be sure. As you see above the oddity continues in its ports with S/PDIF input being a stereo 3.5 mm jack (it has dual S/PDIF inputs as a result). I had my friend operate the unit as I did not even want to try to decode the button lights. The unit has batteries and that is how I tested it, likely giving it an edge in mains related noise/distortion.
The claim to fame of Chord DACs is a super high-tap FIR filter for reconstruction. This one claims over 49,000 taps. This means each (upsampled) audio sample goes through this many computations before a value spits out. FPGAs are custom ICs that are very well suited for this type of application which is what this DAC uses. The tap number has been positioned as a figure of excellence with each newer model sporting more and more taps. Marketing people rejoice!
Rob Watts the designer says that there are audible improvements with increased taps. Unfortunately when I asked him whether he had evaluated such merit in blind testing, he said no.
On the software front this DAC unfortunately gets a failing grade by requiring a custom driver or it does nothing. No USB class is supported and plugging in the device without drivers did absolutely nothing in Windows 10. This is a bummer because if the OS driver interface changes and this company is no longer in business, you have a $2,700 door stop.
For my testing I installed ASIO drivers and used bit-exact output to drive it from the PC. Other tests use S/PDIF digital output from my Audio Precision analyzer.
Measurements
For this test I had to pull out the only DAC I have on my bench that rivals it in price, namely the Exasound E32 ($3,400). For a test or two I also threw in the much cheaper Topping DX7 at $400 although I purchased mine for $300.
First, let's run our usual J-Test for jitter and noise. A test tone of 12 Khz should be the only thing visible here with the rest being noise and distortion. I ran the test two ways: using USB output from the PC and S/PDIF generated by Audio Precision Analyzer. Comparison is made to Exasound E32 over S/PDIF:
We see a very competent execution here with USb and S/PDIF matching each other. The Exasound is just as good but mains related hum and noise is there at very low level which can be remedied with a better power supply. Even as is, that noise is at very low level so not an audible concern either way. So little to complain about in either device. BTW "better" in the graph means better than Exasound E32.
Next let's get another boring graph out of the way, namely frequency response with different filters:
The default filter in red has the best frequency domain response with nearly perfect response. The others have small amount of roll off (about 1 dB at 20 Khz). I let subjectivists play with those filters and report back.
Let's get into my favorite test, linearity. And ideal DAC would have a flat line meaning it outputs what it is told to output:
Both DACs eek out about 18 bits of output before deviation exceeds 0.1 dB (my personal limit). In the case of Exasound it seems to just get noisier. On Chord Hugo 2 it gets funky jumping up and down. Both of them are very respectable though.
Both of them unfortunately get schooled by the Topping DX7 which costs a fraction of them:
OK, let's not spoil our dinner and continue.
Let's look at how a very low level sine wave at -90 db looks like in these two DACs:
Please excuse the dim output for Exasound. It was at a different size window and I had to resize it to match. Regardless, hopefully you see that the Exasound E32 is a closer match to a sine wave and freer of noise/glitches. Its DC offset changes a bit but that is due to power supply hum. Overall Exasound E32 wins.
Let's look at Harmonic Distortion+Noise:
Here, the Chord Hugo 2 DAC has higher noise floor but better control of distortion products. I give it the nod since noise is more benign than distortion products.
Next let's look at SMPTE intermodulation test:
The Exasound in green is the clear winner here and by good bit. The Chord Hugo 2 matches the Topping DX7.
Let's compare the distortion and noise products with a 1 Khz signal that itself has been filtered out:
Chord Hugo 2 pulls way ahead here. Yes it has higher noise floor but otherwise has less distortion spikes. The Exasound produces lot of glitches in 2 to 3 Khz which I suspect is due to mains distortion products.
None of this is audible concern though due to very low levels. Highest peak is at -120 dbFS.
Let's skin this cat differently and look at THD+N vs frequency:
Here the Chord Hugo 2 is the winner (in red). It has similar distortion+noise to Exasound E32 at low frequencies. At frequencies above 2 Khz the noise+distortion jumps up in Exasound. The Topping DX7 and Hugo 2 do the same but at much lower amplitude.
Headphone Measurements
I measured the output impedance of Hugo 2 and compared it to others on record:
At 1.7 ohms, it is very low making it suitable to driving any headphone without changing its frequency response. I need to measure and add the Exasound to this graph.
EDIT: there have been complains about this measurement saying that Chord advertises a much lower number (0.02 ohm?). The only way my measurement is off is if the unit clips under load of 33 ohm. If it does, then that obviously shows a larger drop resulting in impedance to be higher than if it were not clipping. I will try to get my hands on the loaned unit again in the future to confirm this.
Next I am going to show a new test which is the THD+N not against a dummy resistive load, but real headphones. This is based on work done by Benchmark where they demonstrate this problem with rising low level distortion when real loads are used instead of artificial:
Sadly we see that noise and distortion rise especially in low frequencies. Worst performance is with the Focal Utopia. Best was with Hifiman HE1000 V2. I will need to run this test on my other headphone amplifiers and see how they behave.
Summary
The Chord Hugo 2 is competently designed with measurements that show no obvious design/execution problems. Its high number of filter taps though provide it with no advantage against traditional implementations such as in Exasound E32 or even Topping DX7.
Despite these good measurements, I cannot recommend it due to requirement for proprietary driver. If that is not a concern for you, then I don't see any grave reason to not buy it other than high cost.
As always comments, corrections, feedback, bad jokes, etc. are welcome.
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.
If you like this review, please consider donating funds for these types of hardware purchase using Patreon (https://www.patreon.com/audiosciencereview), or upgrading your membership here though Paypal (https://audiosciencereview.com/foru...eview-and-measurements.2164/page-3#post-59054).
Last edited:
