Member
Joined 2009
Paid Member
I notice a lot of excitement in the audio press over this unit.
Products: Hugo mobile DAC/headphone amp
Anybody actually had their hands on one ?
Do any of the DIY projects on this forum offer comparable sound quality ? 😉
Products: Hugo mobile DAC/headphone amp
Anybody actually had their hands on one ?
Do any of the DIY projects on this forum offer comparable sound quality ? 😉
Attachments
The design is interesting for sure, but if one takes what Rob Watts (the designer) at face value then a passive DAC output filter should beat his digital filter for SQ. He's claiming 'the more taps in the filter, the better the sound' - an analog filter has effectively infinitely many taps which trumps his few thousand digital taps.
Very long thread on it over at Head-Fi where Rob Watts contributes useful technical background - http://www.head-fi.org/t/702787/chord-hugo
Very long thread on it over at Head-Fi where Rob Watts contributes useful technical background - http://www.head-fi.org/t/702787/chord-hugo
Last edited:
No there is a big difference between an analogue filter and a FIR digital filter. An FIR filter, due to the filters long group delay. looks into the future and the past. An analogue filter, like an IIR filter, only uses the past as it's stimulus.
Sampling theory requires to know the future and the past, and an infinite tap length FIR filter will perfectly reconstruct the bandwidth limited signal - as if sampling had not actually happened. This would categorically not happen with an IIR or analogue type filter.
Rob
There's nothing inherent in an FIR filter that it must 'look into the future' - its entirely a matter of how its coefficients and delay elements are structured. Asymmetric FIR filters (with short group delay) are just as possible to realize as symmetric (long group delay) ones.
How does sampling theory require knowing the future?
How does sampling theory require knowing the future?
Member
Joined 2009
Paid Member
Well I've read enough to get an idea that this is a very well regarded DAC. I can't make myself pay that much though so my interest is to know what alternatives are available that might offer 90% of the SQ for very much less conga!
And it's clear from the response of the market that Rob knows his stuff.
And it's clear from the response of the market that Rob knows his stuff.
Reading the spec, it looks like (same dac chip?) a Ifi Nano Digi...which is a HuGo further with is 10 hours of cells. The Chord seems to have more inputs possibilities: spidf, RCAs, spidf...
@Bigun - are you looking for all the bells and whistles or would playback of 44k1/16bits suit you?
@Eldam - no the Hugo doesn't use a 'DAC chip' - its done in an FPGA with 'discrete' style output stage.
@Eldam - no the Hugo doesn't use a 'DAC chip' - its done in an FPGA with 'discrete' style output stage.
Member
Joined 2009
Paid Member
I don't have any hi-res material. I have a CD collection that I want to get rid of - by putting it onto my computer. Actually I have been quite interested in the new Sony HAP line as they offer internal storage so it can operate independently from a computer once the files are transferred. I'm new to this whole DAC thing, just asking questions and seeing what knowledge I can pick up.
Seems then that a DAC optimized to deliver RBCD would be most suitable for you. Certainly one can be built which'll cost well under 10% of the current Hugo price (in components, not counting DIY time). I've put some of the necessary sub-modules for such a DAC up on my blog.
Read a review or two - the same old story, extremely high quality sound is achievable from digital if enough measures are taken to troubleshoot SQ issues, but commercial units are then sold at pretty inflated prices, and the ideas tend not to trickle down too well to normal, value for money products.
Take a look at:
HTML:
http://en.wikipedia.org/wiki/Whittaker%E2%80%93Shannon_interpolation_formulaTo perfectly reconstruct the original bandwidth limited signal (that would then mean you have effectively not sampled the signal as the recovered signal is identical to the un-sampled signal) the interpolation filter needs a sinc function impulse response. A sinc function converges on zero in the infinite past, and the infinite future.
Clearly we can't process an infinite data set. But if you took the view that 16 bit sinc function coefficients were OK, thus ensuring time domain errors were below 16 bit, then you would be looking at getting on for 1,000,000 taps for an 8 times OS filter! In the case of Hugo I use 26,368 taps, this filter has a group delay of 40mS, so it uses 80mS of 44.1kHz data.
Rob
it seems this dac functions similar to the new PS Audio NuWave DAC that is getting great reviews....
It seems to me that the math is taking as read that the sequence of real numbers is perfectly band-limited. However no engineered filter is perfect, perfection only exists in math. So how do real world anti-aliasing filter considerations impinge on the beauty of the mathematical formulation? Are any AAFs close to giving 16bit time domain performance?
I'd misunderstood what you meant by sampling theory needing to know the future, it seems now that you were meaning that ideal reconstruction needed to consider the figurative 'future' samples in the dataset. I was sitting here wondering how an ADC was going to be able to see into the future in order to sample ideally....😕 But perfect bandlimiting (anti-alias filtering) seems to me to require there to be an infinite delay before the ADC gets to see anything at all so engineering considerations (rather than pure math) need to prevail do they not?
Yes it's the ideal reconstruction filter where you need very long symmetrical tap lengths.
I have been working on ADC bandwidth limiting filters for an ADC project. Its not that difficult to get 24 bit performance - 144dB rejection at FS/2, although you do need to use more taps than conventional decimating filters. Sampling theory only requires no signal above FS/2, and this is no where near as difficult as the reconstruction or interpolation filter at the DAC, when to get ideal performance you then need silly numbers of taps.
Hugo is at 26,368 taps, I have a prototype filter with 6 figure taps, and you can still hear improvements even at this level with increasing tap length, albeit the changes are getting smaller.
Rob
Member
Joined 2009
Paid Member
I took a look at some other forums and see that the NuWave is not attracting the same level of praise as the Hugo. Perhaps Rob's filters are the key here...
Perhaps somebody will work out how to implement those filters into a DIY DAC project
I had a Hugo that I sold because it could not fit my interconnect(solved since then) but I have nothing but high praise for its sound and functionality - it was great to be able to just feed it whatever type of file I had and it sounded simply great. Of course it could be just a bit better to kill my very expensive CD player and I would be the happiest person in the world, I guess I will have to wait for a new QBD... Make no mistake though - Hugo needed just 5% improvement to reach the level of the player that cost 4 times as much and the functionality of the Hugo was 500% better. Those damn 5%...
Hugo is a indeed a technical marvel, a friend of mine was also very impressed that you could get such a bass from a battery on 3.3V...
Hugo is a indeed a technical marvel, a friend of mine was also very impressed that you could get such a bass from a battery on 3.3V...
Chord were expensive from the beginning ! I have an amp 1200 buyed on second hand ten years ago...still happy with it ! Certainly not the best amp of the world but for the price I ruled to have and my 85 DB speakers it is just good.
Very fast bass, treble is not harsch : happy of this product ! But heard Chords was not so good before with digitals units...
just two cents of course
Digging up the old page, any measurements on chord hugo? I would like to see some measurements on this gem.
ps. Especially the IMD CCIF.
ps. Especially the IMD CCIF.
- Status
- Not open for further replies.