This is a review of the Totaldac d1-six DAC and headphone amplifier. It was kindly shipped from europe to me for testing by a member at a high shipping cost of over 200 euros. The d1-six costs 13,500 euros including VAT in Europe and $12,400 for export. The latter translates into US $13,816. This is the most expensive DAC to be reviewed as of this writing.
The d1-six is a two-box solution with the power supply in an external box:
The black plexiglass is meant to give it a touch of class but it is not very successful. I darkened the image above to get rid of the myriad of reflections it shows. The trapezoidal case is not my cup of tea either, in this product or elsewhere.
The yellow OLED display is nice but too small for viewing in a main system configuration. Strange to see 44.1 kHz sampling shown as 44K1. This convention (replacing the decimal point with the unit) is used among engineers sometimes when giving measurements of electronic parts like resistors but I have not see it used in a user interface for non-engineers.
On the power supply, I would have preferred an industrial case that one hides. Not sure how you could ever make it look good stacked on top of the unit or beside it.
Maybe tastes are different in Europe so I am open to be corrected on that front.
The back panel screams "cheap:"
Come on folks. Stamped sheet metal like how you find in a $200 receiver??? What did I get for all the money I spent if it is not a more decent case, one that is machined or something.
And what is with the headphone jack in the back?
Having been sold in Europe, the unit came with a 230 volt-only power supply. Fortunately I have a lab AC generator which I programmed to power it exactly as spec'ed:
Actually I take that back. I initially set it to 50 Hz as it says. But as you see in the measurement section, I changed it to 60 Hz.
The hour meter says 20.1 hours of being on. I left the unit on for that long before doing my listening tests.
A really, really cheap, plastic remote comes with the unit. I didn't take a picture of it as if I showed it to you, half of you would want to jump off a bridge. It is silver looking thing that would be out of place in an asian made audio gear at $300, let alone one at this price. Functionally though, it is response and works well.
Anyway, the unit is not a disaster but far below grade for a high-end audio product where one expects a lot in the jewelry department. Let's hope the objective measurements and listening tests show the value.
By the way, there is a 1000 Euro option to turn this into a streamer. The version I tested as you can tell in the back-panel picture, doesn't have that option.
The Totaldac d1-six is an "R2R" DAC. Translation: "we wanted to build our own DAC because we know more than people who build DAC chips."
DAC Audio Measurements
As I noted above, I started my testing with the unit's AC mains frequency set to 50 Hz:
If you look at the FFT spectrum, we have two peaks, one at 50 Hz and the other at 60 Hz. The 50 Hz is from the unit of course as it is the only device generating it. The 60 Hz is caused by mains leakage of the rest of the test gear (computer and audio precision analyzer). Since this equipment would normally be used in the same phase as the rest of the gear in actual use, I decided to change the frequency to 60 Hz:
We see that there is no ill effect on performance of the unit so I decided to go with it for the rest of the testing.
The unit produces 5 volt output which is more than the nominal we require (a good thing). For fair comparison, I tried to lower the input digital samples/volume control to get it down to 4 volts and that made SINAD actually worse by 2 dB. So I gave it the benefit of doubt and went with the full level for the rest of the tests.
SINAD (signal over noise and distortion) is actually not "bad" for the class, almost matching the Soekris dac1421:
Of course, in the larger scope, the d1-six severely underperforms DACs at a fraction of its cost. It squarely lands in the fourth (worst) quartile of all DACs tested.
We can see the reason for poor performance from see of unwanted distortion spikes to the right of our 1 kHz tone in the FFT graph. You give it one tone, and it generates many, stepping on low level spectrum in the rest of your music. This increases high-frequency energy and potentially generates audible harshness.
We can zoom into this spectrum more with a wideband test of the 1 kHz tone:
So not only do we have a series of unwanted hamonic spikes, but also have a couple of higher power ultrasonic tones around 48 to 50 kHz. If you have any dreams of hearing high-resolution audio, that should be dashed by now as nothing in high-res music has an amplitude as high as these ultrasonic tones. Likely though your speakers don't play these so maybe you are safe.
Yet another analysis shows which harmonics are dominant:
We see that higher order harmonics, i.e. third through fifth are more dominant than 2nd. So much for the assumption that distortion here may be low order and hence euphonic/not audible. Thankfully levels are low to be audible as such but stay tuned for more.
Dynamic range is actually decent:
That should be more than enough for 16 bit and even high-res playback.
You may have noticed a curious thing on the display "Treble FIR Off." I thought this was some kind of reconstruction filter but read the manual and it says it is a compensation for the normal roll off of the DAC itself. So I decided to measure it with and without this setting being on:
What? Yes, there is a roll off without it of about 0.6 dB. But the compensation shoots the other way, raising the level at 20 kHz by 1.2 dB. They couldn't make this flat especially since they are using a fancy FIR filter? Assuming your hearing is very good, then neither setting is good and both act as tone control. For the rest of us older folks, neither matters. Or does it? Read the listening test results.
Next I ran the intermodulation versus level test and skeletons came out of the closet:
What the heck is this? Distortion products that rise to just -40 dB? How would that not stomp on all the low level detail in your music? Let's look at this in detail by a spectrum analysis at -25 dB:
We have our two tones we fed to it at 60 Hz and 7 kHz. In the inset I show how the Audio Precision analyzer measures itself. Compare that to all the unwanted junk in from d1-six.
We see a similar problem in our 32-tone test signal:
It visually doesn't look too bad until you notice the entire graph is elevated. Distortion-free spectrum is limited to 90 dB so not quite there to resolve even 16 bit audio.
The shocker was our THD+N (distortion+noise) versus frequency sweep. Note that this is a wideband test so includes the ultrasonic content I showed above:
I don't think we have ever seen anything as broken as this. Distortion is directly proportional to frequency. I have shown the scope output at 10 kHz for your "sine wave." The thing doesn't resemble a sine wave anymore as the high frequency components have summed with it to become that complex shape. At lower frequencies, you clearly see the "steps" in the waveform. All of this points to no reconstruction filter which we can confirm by feeding the unit a wide-spectrum white noise:
No wonder.
It is not all terrible news. There is some excellence shining through in the linearity test:
This shows accuracy in producing the main tone (all noise and distortion filtered out).
Headphone Amplifier Audio Measurements
I started off with testing the unit at 300 ohm:
Of course, the ills of the DAC bleed through here but even ignoring that, this is hardly enough power. It essentially matches my $250 Topping DX3 Pro in low gain mode. My standard here is 100 milliwatts and the d1-six easily misses that mark with just 16 milliwatts of power.
Switching to 33 ohm load, doesn't improve the picture:
Now we can't even match the DX3 Pro in low gain. What do you want to bet that is because of high output impedance, causing voltage drop?
Indeed that is the case with 37 ohm output impedance. So be careful using this amp with any headphone that has variable impedance as it will act as an EQ (for better or worse).
Listening Tests
Wanting to kill two birds with one stone, I did my testing with the headphone output. I started with my Sennheiser HD650. As the measurements show, there is barely enough power to drive these. It is fine for background listening but not much more than that.
So I switched to Hifiman HE400i which I hooked up to an AB switcher to the d1-six and Topping DX3 Pro. Levels were matched and through Roon player, I could send the same song to both DACs at the same time.
My first test was to see if I could hear the effect of FIR filter on and off. I was surprised that this was very audible. In one setting there were clearly more highs. I checked and that was with FIR OFF!!! This of course is not possible since measurements show otherwise and I should not be able to hear the difference anyway. Sure enough, when I did this test blind, there was no difference at all. I had imagined the original effect.
With lessons learned for the hundredth time to test things blind, I ventured into my AB test with DX3 Pro. Despite all the flaws in the measured performance of d1-six, audible effects are subtle to non-existent. However, with some careful listening, I could hear the accentuated highs through Totaldac d1-six.
Mind you, if you don't do an AB test like I did, you would not have much chance of hearing this effect. Instead your imagination can trigger into high gear thinking you are hearing all kinds of positive effects. Me? I did not hear any of that. No change in soundstage, not increase in PRAT and micro-detail, no nothing. Just a DAC that in "lean back" listening likely sounds like other DACs. So if you are buying this DAC because of some magic it imparts due to its architecture, don't bother telling me about it. Set up a controlled test and then we can talk.
Conclusions
By now, we have such a solid library of product tests that we can predict what happens when someone attempts to sell a concept rather than performance to audiophiles. Lack of measurements is the first warning sign. Totaldac has a single graph that is competing with a long bullet list of audiophile marketing.
The d1-six throws out the most fundamental science in signal processing and digital audio, hoping lay notions of audiophiles compensates. Well, it doesn't. When you put aside all emotions for or against the product, what you are left with is a product that fails to impress across many tests. In some cases like THD+N vs frequency brand new ground is broken, unfortunately on the negative side.
The path to success here is using a proper DAC chip and bringing excellence in the form of implementation. Attempting to reinvent the wheel without design goals and verification is a bad way to go as our test results show. You can't build a new car tire that is not very round and try to make up for it by painting it red and saying it goes fast. Design a custom DAC if you want. But first make sure it brings high-fidelity with it. Don't sacrifice that in order to cater to non-technical audiophile beliefs and myths.
------------
As always, questions, comments, recommendations, etc. are welcome.
Serious note: we almost did not do this review because of the extreme high cost of shipping and insurance. If you as much as me value occasionally dipping our toe into high-end audio, please donate some money so that I can pay for the return shipping as opposed to just the owner: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/
The d1-six is a two-box solution with the power supply in an external box:
The black plexiglass is meant to give it a touch of class but it is not very successful. I darkened the image above to get rid of the myriad of reflections it shows. The trapezoidal case is not my cup of tea either, in this product or elsewhere.
The yellow OLED display is nice but too small for viewing in a main system configuration. Strange to see 44.1 kHz sampling shown as 44K1. This convention (replacing the decimal point with the unit) is used among engineers sometimes when giving measurements of electronic parts like resistors but I have not see it used in a user interface for non-engineers.
On the power supply, I would have preferred an industrial case that one hides. Not sure how you could ever make it look good stacked on top of the unit or beside it.
Maybe tastes are different in Europe so I am open to be corrected on that front.
The back panel screams "cheap:"
Come on folks. Stamped sheet metal like how you find in a $200 receiver??? What did I get for all the money I spent if it is not a more decent case, one that is machined or something.
And what is with the headphone jack in the back?
Having been sold in Europe, the unit came with a 230 volt-only power supply. Fortunately I have a lab AC generator which I programmed to power it exactly as spec'ed:
Actually I take that back. I initially set it to 50 Hz as it says. But as you see in the measurement section, I changed it to 60 Hz.
The hour meter says 20.1 hours of being on. I left the unit on for that long before doing my listening tests.
A really, really cheap, plastic remote comes with the unit. I didn't take a picture of it as if I showed it to you, half of you would want to jump off a bridge. It is silver looking thing that would be out of place in an asian made audio gear at $300, let alone one at this price. Functionally though, it is response and works well.
Anyway, the unit is not a disaster but far below grade for a high-end audio product where one expects a lot in the jewelry department. Let's hope the objective measurements and listening tests show the value.
By the way, there is a 1000 Euro option to turn this into a streamer. The version I tested as you can tell in the back-panel picture, doesn't have that option.
The Totaldac d1-six is an "R2R" DAC. Translation: "we wanted to build our own DAC because we know more than people who build DAC chips."
DAC Audio Measurements
As I noted above, I started my testing with the unit's AC mains frequency set to 50 Hz:
If you look at the FFT spectrum, we have two peaks, one at 50 Hz and the other at 60 Hz. The 50 Hz is from the unit of course as it is the only device generating it. The 60 Hz is caused by mains leakage of the rest of the test gear (computer and audio precision analyzer). Since this equipment would normally be used in the same phase as the rest of the gear in actual use, I decided to change the frequency to 60 Hz:
We see that there is no ill effect on performance of the unit so I decided to go with it for the rest of the testing.
The unit produces 5 volt output which is more than the nominal we require (a good thing). For fair comparison, I tried to lower the input digital samples/volume control to get it down to 4 volts and that made SINAD actually worse by 2 dB. So I gave it the benefit of doubt and went with the full level for the rest of the tests.
SINAD (signal over noise and distortion) is actually not "bad" for the class, almost matching the Soekris dac1421:
Of course, in the larger scope, the d1-six severely underperforms DACs at a fraction of its cost. It squarely lands in the fourth (worst) quartile of all DACs tested.
We can see the reason for poor performance from see of unwanted distortion spikes to the right of our 1 kHz tone in the FFT graph. You give it one tone, and it generates many, stepping on low level spectrum in the rest of your music. This increases high-frequency energy and potentially generates audible harshness.
We can zoom into this spectrum more with a wideband test of the 1 kHz tone:
So not only do we have a series of unwanted hamonic spikes, but also have a couple of higher power ultrasonic tones around 48 to 50 kHz. If you have any dreams of hearing high-resolution audio, that should be dashed by now as nothing in high-res music has an amplitude as high as these ultrasonic tones. Likely though your speakers don't play these so maybe you are safe.
Yet another analysis shows which harmonics are dominant:
We see that higher order harmonics, i.e. third through fifth are more dominant than 2nd. So much for the assumption that distortion here may be low order and hence euphonic/not audible. Thankfully levels are low to be audible as such but stay tuned for more.
Dynamic range is actually decent:
That should be more than enough for 16 bit and even high-res playback.
You may have noticed a curious thing on the display "Treble FIR Off." I thought this was some kind of reconstruction filter but read the manual and it says it is a compensation for the normal roll off of the DAC itself. So I decided to measure it with and without this setting being on:
What? Yes, there is a roll off without it of about 0.6 dB. But the compensation shoots the other way, raising the level at 20 kHz by 1.2 dB. They couldn't make this flat especially since they are using a fancy FIR filter? Assuming your hearing is very good, then neither setting is good and both act as tone control. For the rest of us older folks, neither matters. Or does it? Read the listening test results.
Next I ran the intermodulation versus level test and skeletons came out of the closet:
What the heck is this? Distortion products that rise to just -40 dB? How would that not stomp on all the low level detail in your music? Let's look at this in detail by a spectrum analysis at -25 dB:
We have our two tones we fed to it at 60 Hz and 7 kHz. In the inset I show how the Audio Precision analyzer measures itself. Compare that to all the unwanted junk in from d1-six.
We see a similar problem in our 32-tone test signal:
It visually doesn't look too bad until you notice the entire graph is elevated. Distortion-free spectrum is limited to 90 dB so not quite there to resolve even 16 bit audio.
The shocker was our THD+N (distortion+noise) versus frequency sweep. Note that this is a wideband test so includes the ultrasonic content I showed above:
I don't think we have ever seen anything as broken as this. Distortion is directly proportional to frequency. I have shown the scope output at 10 kHz for your "sine wave." The thing doesn't resemble a sine wave anymore as the high frequency components have summed with it to become that complex shape. At lower frequencies, you clearly see the "steps" in the waveform. All of this points to no reconstruction filter which we can confirm by feeding the unit a wide-spectrum white noise:
No wonder.
It is not all terrible news. There is some excellence shining through in the linearity test:
This shows accuracy in producing the main tone (all noise and distortion filtered out).
Headphone Amplifier Audio Measurements
I started off with testing the unit at 300 ohm:
Of course, the ills of the DAC bleed through here but even ignoring that, this is hardly enough power. It essentially matches my $250 Topping DX3 Pro in low gain mode. My standard here is 100 milliwatts and the d1-six easily misses that mark with just 16 milliwatts of power.
Switching to 33 ohm load, doesn't improve the picture:
Now we can't even match the DX3 Pro in low gain. What do you want to bet that is because of high output impedance, causing voltage drop?
Indeed that is the case with 37 ohm output impedance. So be careful using this amp with any headphone that has variable impedance as it will act as an EQ (for better or worse).
Listening Tests
Wanting to kill two birds with one stone, I did my testing with the headphone output. I started with my Sennheiser HD650. As the measurements show, there is barely enough power to drive these. It is fine for background listening but not much more than that.
So I switched to Hifiman HE400i which I hooked up to an AB switcher to the d1-six and Topping DX3 Pro. Levels were matched and through Roon player, I could send the same song to both DACs at the same time.
My first test was to see if I could hear the effect of FIR filter on and off. I was surprised that this was very audible. In one setting there were clearly more highs. I checked and that was with FIR OFF!!! This of course is not possible since measurements show otherwise and I should not be able to hear the difference anyway. Sure enough, when I did this test blind, there was no difference at all. I had imagined the original effect.
With lessons learned for the hundredth time to test things blind, I ventured into my AB test with DX3 Pro. Despite all the flaws in the measured performance of d1-six, audible effects are subtle to non-existent. However, with some careful listening, I could hear the accentuated highs through Totaldac d1-six.
Mind you, if you don't do an AB test like I did, you would not have much chance of hearing this effect. Instead your imagination can trigger into high gear thinking you are hearing all kinds of positive effects. Me? I did not hear any of that. No change in soundstage, not increase in PRAT and micro-detail, no nothing. Just a DAC that in "lean back" listening likely sounds like other DACs. So if you are buying this DAC because of some magic it imparts due to its architecture, don't bother telling me about it. Set up a controlled test and then we can talk.
Conclusions
By now, we have such a solid library of product tests that we can predict what happens when someone attempts to sell a concept rather than performance to audiophiles. Lack of measurements is the first warning sign. Totaldac has a single graph that is competing with a long bullet list of audiophile marketing.
The d1-six throws out the most fundamental science in signal processing and digital audio, hoping lay notions of audiophiles compensates. Well, it doesn't. When you put aside all emotions for or against the product, what you are left with is a product that fails to impress across many tests. In some cases like THD+N vs frequency brand new ground is broken, unfortunately on the negative side.
The path to success here is using a proper DAC chip and bringing excellence in the form of implementation. Attempting to reinvent the wheel without design goals and verification is a bad way to go as our test results show. You can't build a new car tire that is not very round and try to make up for it by painting it red and saying it goes fast. Design a custom DAC if you want. But first make sure it brings high-fidelity with it. Don't sacrifice that in order to cater to non-technical audiophile beliefs and myths.
------------
As always, questions, comments, recommendations, etc. are welcome.
Serious note: we almost did not do this review because of the extreme high cost of shipping and insurance. If you as much as me value occasionally dipping our toe into high-end audio, please donate some money so that I can pay for the return shipping as opposed to just the owner: https://www.audiosciencereview.com/forum/index.php?threads/how-to-support-audio-science-review.8150/