Hi Mark, i have listen to d1 - six, d1-twelwe, d1-twelve SE plus 2 d1-reclockers plus d1 server. Total dac is not nice sounding dac it is right sounding dac. At Munich they also used discrete buffer stage, at sound presentation. It is not all in numbers...We listen to the music not numbers. Of course you continue with dac that you know into details. That is the easiest way to know on which step are you at modifying es9038q2m board.
Not just hearing Total dac...hear it in the right system.
Not just hearing Total dac...hear it in the right system.
i have changed the power cord of the dac now. a unshielded cord against an shielded.
the improvement was better then the avcc mod 😉 i think my avcc implementation with lt3042 (vref 3.3v) is not perfect
the improvement was better then the avcc mod 😉 i think my avcc implementation with lt3042 (vref 3.3v) is not perfect
Last edited:
A shielded power cable has more capacitance than unshielded. Sounds like you need more filter capacitance maybe right where the power comes into the board. Probably best to use electrolytics in parallel with ceramics, but you can always try paralleling some film caps too.
Hi andora76, It could be that both our reference dacs sound mostly the same and what sounds different is the apodizing filter on Katana, which may have something to do with it running in master mode. I know on my modded dac I never liked the sound of the apodizing filter, but for Katana it seems to be a good fit. If you try Katana maybe you will find the same thing.
Do you think adding capacitance might solve my hum problem on my Cirrus DAC? Currently I have a 0.47uF MKP 275V safety cap where the AC enters the unit.
Ian, Hard to say. You could try it, but for safety reasons you have to be very careful to use appropriately rated capacitors if connecting directly to the AC line. Very high voltage transients are possible, maybe the worst coming from when lightning strikes a power line. In the US, wall sockets will typically arc over at about 2kV so caps that can safely handle 3kV or so transients may be okay to put at chassis power cord entry points. The amount of capacitance that can be put there may also be limited by safety regulations due the leakage currents that can flow through such caps. So, if thinking of putting caps where the power cable comes in, make sure you use caps that are rated by the manufacturer for that type of use, and that will comply with local safety regulations for maximum leakage currents.
Yup, good advice, that's why I chose what is called a safety cap, intended for filtering mains AC.
The ones I have are exactly like the yellow one below, just a higher value.
Film capacitor - Wikipedia
Safety and EMI/RFI suppression film capacitors
RFI/EMI suppression capacitors
Metallized polypropylene RFI suppression capacitor (MKP) for "X2" safety standard
The ones I have are exactly like the yellow one below, just a higher value.
Film capacitor - Wikipedia
Safety and EMI/RFI suppression film capacitors
RFI/EMI suppression capacitors
An externally hosted image should be here but it was not working when we last tested it.
Metallized polypropylene RFI suppression capacitor (MKP) for "X2" safety standard
Leakage current for the purposes of AC power line caps is not just a function of dielectric leakage, but also of how much current flows from line to ground the given the impedance of the capacitor at the line frequency. Also, how much current flows into the power cord when the device power switch it turned off, etc.
In other words, there is also a limit to what size caps you can put there depending on which of the power leads are involved. Also, one .47uf cap might be okay, but three of them in parallel might not.
Maybe it would make sense to google for something like 'power line filter design' for region where you live. Probably a lot of information will turn up that will tell you what is okay to do and what isn't.
In other words, there is also a limit to what size caps you can put there depending on which of the power leads are involved. Also, one .47uf cap might be okay, but three of them in parallel might not.
Maybe it would make sense to google for something like 'power line filter design' for region where you live. Probably a lot of information will turn up that will tell you what is okay to do and what isn't.
Last edited:
I had a look inside a couple of PC PSUs I had lying around and inside a couple of CD players and they have ceramic disk caps soldered directly across the terminals of the IEC sockets, either 0.33uF or 0.47uF, with ratings of 250VAC or 275VAC.
I'll definitely do some reading on the matter though, as you suggest, mains is not to be trifled with.
I'll definitely do some reading on the matter though, as you suggest, mains is not to be trifled with.
There are different ways of specifying capacitor voltage ratings. In addition, the basic voltage rating may be for normal continuous use, not for occasional higher transients. To be sure a cap is good for that type of service one should read the data sheet for the cap. Remember, they may have to be able to stand a couple of kV or more without exploding or catching fire. Caps for other types of uses don't have the same issues or requirements.
With that, I think we should probably try to get back on topic here. Ian, if any more questions maybe we can take them up in your thread. 🙂
With that, I think we should probably try to get back on topic here. Ian, if any more questions maybe we can take them up in your thread. 🙂
Last edited:
Of course, please continue, I just asked as I thought it was a very similar issue and the information might be useful to others with Sabre boards. Cheers.
Would like to report we have done some more listening tests today with the three dacs, (1) Benchmark DAC-3, (2) my modded ES9038Q2M dac in its present form as described in this thread, and (3) Katana with firmware upgrade.
While there will be more details to describe later in more of a review type format, the outcome of the inter-comparison is that DAC-3 is still clearly the winner, and the other two dacs are declared as tied for 2nd place.
Regarding sound quality, I think it would be fair to say that if one's dac needs can be met with an RPi music player, Katana is a very good dac worth well more than what it costs to buy. Is Katana a $1,000 dac? I don't know exactly, but IMHO it is easily something more than a $500 dac. Again, this is all IMHO, and according to my tastes, so please take it for whatever it may be worth.
While there will be more details to describe later in more of a review type format, the outcome of the inter-comparison is that DAC-3 is still clearly the winner, and the other two dacs are declared as tied for 2nd place.
Regarding sound quality, I think it would be fair to say that if one's dac needs can be met with an RPi music player, Katana is a very good dac worth well more than what it costs to buy. Is Katana a $1,000 dac? I don't know exactly, but IMHO it is easily something more than a $500 dac. Again, this is all IMHO, and according to my tastes, so please take it for whatever it may be worth.
Interesting report, cheers. What would also interest me would be a comparison between the ESS Sabre DACs and some others from Wolfson, Cirrus Logic, Burr & Brown etc.
I just wondered if you had any other DACs available to you and could make any comparisons. I have been trying to find useful comparisons and they just don't exist, as I'm sure you're aware, discussions of what is the best DAC IC always result in a plethora of different answers due to a whole range of factors. It would be truly enlightening if someone were able to compare several of them in the same setup, sadly, very few people are in a position to own multiple DACs based on different ICs and even then, there are all the other factors such as implementation to blur the picture.
I'm still very interested in the Sabre DACs, just concerned at the difficulties in implementing them and how someone like myself would find some of the implementation difficult to achieve.
I'm still very interested in the Sabre DACs, just concerned at the difficulties in implementing them and how someone like myself would find some of the implementation difficult to achieve.
The Katana exercise is showing something more than a tie really. It indicates that without the DSP tricks and THD compensation,, intersample overs, the gains in implementation takes you to a stage where it does not matter much after a certain point.
If one were to list out the features that ALLO had taken the pains and great effort to implement versus the textbook implementation that you had carried out....and it still comes out to a tie. I am sure they expected better than a tie. That clearly indicates that beyond a level of implementation, you are stuck inside a performance window and that other "tricks" need to be carried out to move to the next level. Without these other non textbook implementations, you're as it is called "spinning your wheels".
It's too bad really because they really should have been listening to say your modded board which is "textbook" and then see if they could better it with other "refined" design aspects during development.
I'm not saying that the Katana is bad, no,not at all but I really expected and was hoping and rooting for better better results given the great pains they went through to put out a great DAC.
But the tie is really one test opinion so far.
I remember months back when you opined about the OP amp AVCC and CDSgames was touting the supercap solution and finally laddered decoupling. I wanted to see what the matchup outcome would be. Also your modded DAC is using off the shelf LME 497xx op amps versus Sparkos meticulously matched discrete op amps. You're using the standard chinese board firmware and they went the mile to put in master mode etc. and pi filters coupled with their lowest jitter NDK clock. All this for a tie? Your modded dac does not even carry a decent ground plane either. Again the modded dac is handicapped and it is still a tie. Sigh, really.
One question...is this with the AKM4137? If not, then does the modded DAC with the AKM4137 beat out the Katana?
If one were to list out the features that ALLO had taken the pains and great effort to implement versus the textbook implementation that you had carried out....and it still comes out to a tie. I am sure they expected better than a tie. That clearly indicates that beyond a level of implementation, you are stuck inside a performance window and that other "tricks" need to be carried out to move to the next level. Without these other non textbook implementations, you're as it is called "spinning your wheels".
It's too bad really because they really should have been listening to say your modded board which is "textbook" and then see if they could better it with other "refined" design aspects during development.
I'm not saying that the Katana is bad, no,not at all but I really expected and was hoping and rooting for better better results given the great pains they went through to put out a great DAC.
But the tie is really one test opinion so far.
I remember months back when you opined about the OP amp AVCC and CDSgames was touting the supercap solution and finally laddered decoupling. I wanted to see what the matchup outcome would be. Also your modded DAC is using off the shelf LME 497xx op amps versus Sparkos meticulously matched discrete op amps. You're using the standard chinese board firmware and they went the mile to put in master mode etc. and pi filters coupled with their lowest jitter NDK clock. All this for a tie? Your modded dac does not even carry a decent ground plane either. Again the modded dac is handicapped and it is still a tie. Sigh, really.
One question...is this with the AKM4137? If not, then does the modded DAC with the AKM4137 beat out the Katana?
Last edited:
AKM4137 in DSD output mode is part of my present implementation. It would be cheap and easy to do if the chip were just added to the dac board design. That will stay as part of the dac for now.
However, when I said there is a tie I do not mean they sound the same. They do not sound the same. I think we have both just hit the same bottleneck having to do with internal ES9038Q2M interpolation filter DSP horsepower. We have both got the distortion, jitter, and other problems down to the point where the filter DSP horsepower matters and is holding both of us back. How would anybody know in advance that is how it would end up? Virtually nobody doesn't use the built-in ESS filters.
There is one interesting thing though. ES9038PRO has a feature where if you want parallel all the dac channels and just run the chip in stereo mode then you can double the number of available FIR interpolation filter taps. It is just a reallocation of DSP resources. Now we are talking about the possible next step. I am thinking about getting a board like you have only in kit form so I don't have to take it apart first, and with a 9038PRO chip in it. Only thing holding me back is the stupid interconnection of all the AVCC pins for both left and right channels. I would have to lift up at least half the AVCC pins to separate them. What a pain! Yes, I could do it but you know virtually nobody else would because of the off-putting nature and risks of the tiny work. So if I did do it, it would only be because it would make it easier for me to investigate better interpolation filtering without having to get into ASICs. That would make it a very attractive option.
However, when I said there is a tie I do not mean they sound the same. They do not sound the same. I think we have both just hit the same bottleneck having to do with internal ES9038Q2M interpolation filter DSP horsepower. We have both got the distortion, jitter, and other problems down to the point where the filter DSP horsepower matters and is holding both of us back. How would anybody know in advance that is how it would end up? Virtually nobody doesn't use the built-in ESS filters.
There is one interesting thing though. ES9038PRO has a feature where if you want parallel all the dac channels and just run the chip in stereo mode then you can double the number of available FIR interpolation filter taps. It is just a reallocation of DSP resources. Now we are talking about the possible next step. I am thinking about getting a board like you have only in kit form so I don't have to take it apart first, and with a 9038PRO chip in it. Only thing holding me back is the stupid interconnection of all the AVCC pins for both left and right channels. I would have to lift up at least half the AVCC pins to separate them. What a pain! Yes, I could do it but you know virtually nobody else would because of the off-putting nature and risks of the tiny work. So if I did do it, it would only be because it would make it easier for me to investigate better interpolation filtering without having to get into ASICs. That would make it a very attractive option.
Last edited:
Just tried it. It works about the same as the more expensive AK4137 board so far as I can tell. Again, the main difference is that it is I2C/PCM/DSD input only, no SPDIF, TOSLINK, or AES.
What does the AK4137 do - upsample the input then feed it into the DAC? Is the main purpose of this jitter reduction?
Sorry if this is something that has already been answered.
Could one of these be used to provide TOSLink and Coax inputs to an AK4137 board?
WM8804 S/PDIF TOSLINK to IIS Converter I2S Codec Project integration DIY | eBay
Sorry if this is something that has already been answered.
Could one of these be used to provide TOSLink and Coax inputs to an AK4137 board?
WM8804 S/PDIF TOSLINK to IIS Converter I2S Codec Project integration DIY | eBay
Last edited: