Review and Measurements of Hifime UDA38Pro DAC

[έχω δίκιο]

If you want to see what someone with experience has to say about this topic, I'll point you here:
http://nwavguy.blogspot.com/2011/08/op-amps-myths-facts.html

perhaps it isnt.

Unless you are exceeding the drive capability of the OPA1612, confirmation bias is, by far, the most likely explanation for the audible differences you describe.

the human ears are capable of detecting the smallest changes in sound... sometimes changes that are difficult/impossible to measure with existing instruments.

But test equipment measured the OPA1612 and the OPA1622 typical THD+N @ 1 kHz figures of 0.000015% and 0.000018% respectively. If you can hear that 0.000003% difference, I'll be very impressed. Do you really think that your ears are comparable in resolving power to a modern Audio Precision analyzer?

 

[andreasmaaan]

Agreed, the op amp is small part of a much larger problem.
I already own a AudioQuest DAC and ODAC, the UDA38 at <12 o clock (subjectively) sounds significantly better than both these DACs even though it probably measure far worse. The user Wushuliu noted also that this DAC does sound nice. it will be a while till I can afford a D50 so spending a few dollars in an attempt to improve it a bit seems like the best thing to, it feels better than abandoning a $250 purchase at least.

How long have you had it? It might be possible to return it. Both Amir and I got very good customer service when we wanted to return our UDA38PRO DACs, and you could point to Amir's measurements as a reason. Particularly if they are indeed now revising the DAC following his review.

Of course, if you like the sound, definitely keep it. Even a very poor-measuring DAC like this is not gonna be that far off transparent (unless you turn the volume up too loud), and you may actually prefer the distortion if it's audible.

 

[Deleted member 3116]

If you want to see what someone with experience has to say about this topic, I'll point you here:
http://nwavguy.blogspot.com/2011/08/op-amps-myths-facts.html

Unless you are exceeding the drive capability of the OPA1612, confirmation bias is, by far, the most likely explanation for the audible differences you describe.

But test equipment measured the OPA1612 and the OPA1622 typical THD+N @ 1 kHz figures of 0.000015% and 0.000018% respectively. If you can hear that 0.000003% difference, I'll be very impressed. Do you really think that your ears are comparable in resolving power to a modern Audio Precision analyzer?

My comparison between 1612 and 1622 was when used as a headphone buffer in the O2 amp, considering how close those distortion figures are its probably related to the difference in output current and how well they could drive my headphones (he400i and HD650). it makes sense the lower current from 1612 could make headphones sound a bit brighter and more ''revealing'', especially the planar, though I tried to compare them at lower volume to mitigate that effect.

For comparison between OPA2134 and OPA1612 in the o2 there was no question that the 1612 was superior in terms of distortion, in this case both were mounted in DIP8 socket making back and forth comparison and blind testing easy... subtle but totally obvious differences.
This is the reason for making the swap in the DAC, Im 100% confident that it had similar improvements but no way to confirm it.

How long have you had it? It might be possible to return it. Both Amir and I got very good customer service when we wanted to return our UDA38PRO DACs, and you could point to Amir's measurements as a reason. Particularly if they are indeed now revising the DAC following his review.

Of course, if you like the sound, definitely keep it. Even a very poor-measuring DAC like this is not gonna be that far off transparent (unless you turn the volume up too loud), and you may actually prefer the distortion if it's audible.

its on sale in a few days too... very tempting

 

[andreasmaaan]

My comparison between 1612 and 1622 was when used as a headphone buffer in the O2 amp, considering how close those distortion figures are its probably related to the difference in output current and how well they could drive my headphones (he400i and HD650). it makes sense the lower current from 1612 could make headphones sound a bit brighter and more ''revealing'', especially the planar, though I tried to compare them at lower volume to mitigate that effect.

For comparison between OPA2134 and OPA1612 in the o2 there was no question that the 1612 was superior in terms of distortion, in this case both were mounted in DIP8 socket making back and forth comparison and blind testing easy... subtle but totally obvious differences.
This is the reason for making the swap in the DAC, Im 100% confident that it had similar improvements but no way to confirm it.[/QUOTE]

Possibly it's differences in output current or distortion you're hearing, but without measuring at the output to the headphones it's not possible to know with much confidence, right?

Since I can like the sound of something this bad, I can only imagine how better the D50 could be.

its on sale in a few days too... very tempting

Can't hurt to find out

 

[maxxevv]

Can you provide some examples of "difficult/impossible to measure" acoustic phenomena that instrumentation cannot detect?

Try "sibilance" .. ?

Frequency response curves of headphones/ earphones that we always see never reflects that. And its something that can be easily picked up with the average undamaged ear.

 

[Wombat]

Try "sibilance" .. ?

Frequency response curves of headphones/ earphones that we always see never reflects that. And its something that can be easily picked up with the average undamaged ear.

Why do you think sibilance can't be measured? It is just a soundwave composite.

 

[maxxevv]

Please provide examples then. I have not seen them published, yet.

 

[dc655321]

Try "sibilance" .. ?

Frequency response curves of headphones/ earphones that we always see never reflects that. And its something that can be easily picked up with the average undamaged ear.

Sibilance is a property of a recording (a bad one) that can be emphasized by a transducer having a boosted response in the frequency range where the effect occurs at (~5-8kHz).

Hint: if you can hear, you can measure it.

Wanna try again?

 

[maxxevv]

As my previous question. Example pkease.

 

[miero]

Hey guys, opamp rolling without a proper measurement is for kids! Get right tools, learn how to use them and post your measurements here.

 

[andreasmaaan]

As my previous question. Example pkease.

Seriously? This is one of the most trivial effects to measure. Here are measurements of a speaker that's too hot in the sibilance region.

And from the body of Atkinson's subjective review (my bold):

"Analog tape hiss in old recordings was a little more audible than I expected—while the 805 D3s were in the system, I was archiving to digital some cassette recordings from the various bands I'd played with in the 1970s—as was the hiss from Jim Hall's guitar amplifier in the right channel of Jim Hall/Charlie Haden. Vocal sibilants were also emphasized to a small degree."

Whether such a speaker actually produces sibilance will depend on whether there is relevant content in the recording in the first place.

 

[maxxevv]

Seriously? This is one of the most trivial effects to measure. Here are measurements of a speaker that's too hot in the sibilance region.

And from the body of Atkinson's subjective review (my bold):

"Analog tape hiss in old recordings was a little more audible than I expected—while the 805 D3s were in the system, I was archiving to digital some cassette recordings from the various bands I'd played with in the 1970s—as was the hiss from Jim Hall's guitar amplifier in the right channel of Jim Hall/Charlie Haden. Vocal sibilants were also emphasized to a small degree."

Whether such a speaker actually produces sibilance will depend on whether there is relevant content in the recording in the first place.

As it is, its a subjective comment.

Where in his measurements does he indicate where and how the sibilance occurs ?

 

[andreasmaaan]

As it is, its a subjective comment.

Where in his measurements does he indicate where and how the sibilance occurs ?

It's not just a subjective comment.

Look at the measurements page. There is an excess of energy in the sibilance region in the frequency response (and a little above, which is probably why he also heard emphasised tape hiss).

His subjective impressions correlate very well with the objective measurements (not that they need to - it's well known that an excess of energy in this region emphasises sibilance).

EDIT: this is what I'm talking about in the measurements:

 

[andreasmaaan]

Try "sibilance" .. ?

Frequency response curves of headphones/ earphones that we always see never reflects that. And its something that can be easily picked up with the average undamaged ear.

By the way, since I just noticed that your original comment referred to headphones, I wanted to clarify a couple of other things about sibilance and headphones.

Firstly, sibilance is a normal property of speech (present in particular, but not only, in the sound made by a spoken "S"). It is not something you want absent from a recording or reproduction. The problem is excess sibilance, where sibilants are emphasised by the recording or by the reproduction system.

On the recording side, excess sibilance can be avoided by ensuring there is no emphasis in the sibilance region (4-10KHz) in the frequency response of the mic or preamp, or the equalisation of the recording. However, as many microphones tend to emphasise sibilance when held in particular ways in relation to the mouth, further sibilance reduction may be necessary. In this case, sibilance reduction is typically effected by use of a de-essing compressor (a compressor that targets peaks in this frequency range without necessarily causing a reduction in the RMS level or frequency response of the recording).

On the reproduction side, excess sibilance can be avoided primarily by ensuring that the reproduction system has a flat frequency response in this region, that there are no transducer resonances, and that distortion is kept as close as possible to inaudible.

Secondly, headphone measurements are not universal in the same way as speaker measurements are, because headphone measurements must be taken on a dummy head/pinnae, which will never be exactly the same as the head/pinnae of a particular person using the headphones. As frequency increases, the differences between the frequency response on any two head/pinnae become greater, because the geometry and composition of the head/pinnae affect shorter wavelength soundwaves to a greater extent than longer wavelength soundwaves. So headphone measurements on a dummy head/pinnae are less reliable in the higher frequencies with respect to any real-world listener (this problem can be avoided, at great pain and expense, by measuring on the head/pinnae of the real-world listener, or by modelling the head/pinnae of the real-world listener and measuring on that).

If a pair of headphones measures flat on dummy head/pinnae after compensation, but still seems to emphasise sibilance to a particular listener, there are four possible explanations (a combination of more than one is possible):

1. the particular recording contains excess sibilance
2. the measurement and/or the compensation curve are wrong
3. the frequency response of the headphones as measured on the dummy head/pinnae is different from that which would be measured if a microphone were placed in the head/pinnae of the particular real listener
4. the listener's subjective perception of what is a correct amount of sibilance is idiosyncratic

EDIT: transducer resonances or high levels of distortion in the sibilance region may of course be additional factors; these too are measurable.

 

[Uda30pro]

Good morning,
I signed up to confirm what was said by Wushuliu.
My Uda30pro works well, there are no distortions, the low frequencies seem very defined.
I hope they send you a compliant dac, I think your Uda30pro was damaged.
I have used all the cheap Chinese dac i2s, the Uda38pro is the one I paid the most but it is the one that works best.

Regards.

 

[Deleted member 3116]

In conclusion: replacing the ''faulty'' OPA1612 didnt change anything, there is still some very low noise/hiss in that channel. The layout ,location and decoupling caps for each channel's op amp is different, that channel's configuration just isnt agreeing with 1612 I guess.

The 2134 were put back in and it'll be left that way, another JFET input op amp would likely work fine, such as OPA1642, but it isnt worth the effort.

Also using this for the first time with the lossless volume control at max and analogue attenuation in a speaker amp worked fine, no noticeable distortion.

This DAC will be kept to compare with the D50 at some point, to see how much measurements actually influence what we hear.

Since the sound of something objectively bad can be so good I wonder how the objectively good D50 will sound,
both D30 and D50 measure very similarly but the D30 has been reported to not sound as good so maybe its other aspects of the D50 design that makes it good and not necessarily the measurements, most of differences in measurements you see on here are well beyond audibility so they shouldnt matter and imply all these DACs sound identical but most folks on the internet would not agree.

 

[amirm]

Good morning,
I signed up to confirm what was said by Wushuliu.
My Uda30pro works well, there are no distortions, the low frequencies seem very defined.
I hope they send you a compliant dac, I think your Uda30pro was damaged.
I have used all the cheap Chinese dac i2s, the Uda38pro is the one I paid the most but it is the one that works best.

Regards.

Good morning and welcome aboard. While the company initially offered to send me another unit, they did not offer it again at the end and I did not want to push them. Was happy to get my refund and run.

I do agree that given the soldering issues on mine, there could be other problems in the unit I tested. The distortion was quite audible so I have to think others could hear it too if they had the same performance as mine.

 

[Deleted member 3116]

Good morning and welcome aboard. While the company initially offered to send me another unit, they did not offer it again at the end and I did not want to push them. Was happy to get my refund and run.

I do agree that given the soldering issues on mine, there could be other problems in the unit I tested. The distortion was quite audible so I have to think others could hear it too if they had the same performance as mine.

you mentioned they measured 10dB better performance, was this the same unit or a different one? either way it indicates a problem

 

[amirm]

you mentioned they measured 10dB better performance, was this the same unit or a different one? either way it indicates a problem

It is a different unit with different equipment and software measurement than mine.

 

[Soniclife]

By the way, since I just noticed that your original comment referred to headphones, I wanted to clarify a couple of other things about sibilance and headphones.

Firstly, sibilance is a normal property of speech (present in particular, but not only, in the sound made by a spoken "S"). It is not something you want absent from a recording or reproduction. The problem is excess sibilance, where sibilants are emphasised by the recording or by the reproduction system.

On the recording side, excess sibilance can be avoided by ensuring there is no emphasis in the sibilance region (4-10KHz) in the frequency response of the mic or preamp, or the equalisation of the recording. However, as many microphones tend to emphasise sibilance when held in particular ways in relation to the mouth, further sibilance reduction may be necessary. In this case, sibilance reduction is typically effected by use of a de-essing compressor (a compressor that targets peaks in this frequency range without necessarily causing a reduction in the RMS level or frequency response of the recording).

On the reproduction side, excess sibilance can be avoided primarily by ensuring that the reproduction system has a flat frequency response in this region, that there are no transducer resonances, and that distortion is kept as close as possible to inaudible.

Secondly, headphone measurements are not universal in the same way as speaker measurements are, because headphone measurements must be taken on a dummy head/pinnae, which will never be exactly the same as the head/pinnae of a particular person using the headphones. As frequency increases, the differences between the frequency response on any two head/pinnae become greater, because the geometry and composition of the head/pinnae affect shorter wavelength soundwaves to a greater extent than longer wavelength soundwaves. So headphone measurements on a dummy head/pinnae are less reliable in the higher frequencies with respect to any real-world listener (this problem can be avoided, at great pain and expense, by measuring on the head/pinnae of the real-world listener, or by modelling the head/pinnae of the real-world listener and measuring on that).

If a pair of headphones measures flat on dummy head/pinnae after compensation, but still seems to emphasise sibilance to a particular listener, there are four possible explanations (a combination of more than one is possible):

1. the particular recording contains excess sibilance
2. the measurement and/or the compensation curve are wrong
3. the frequency response of the headphones as measured on the dummy head/pinnae is different from that which would be measured if a microphone were placed in the head/pinnae of the particular real listener
4. the listener's subjective perception of what is a correct amount of sibilance is idiosyncratic

EDIT: transducer resonances or high levels of distortion in the sibilance region may of course be additional factors; these too are measurable.

Just want to say thanks for posts like this, your posting some great stuff.