Sorry, don't think I addressed your question directly before. Inverting output polarity should not be a problem so long as all channels are the same. Using the circuit on the ES9028 board, if you are saying you have one there to use, as the output stage for a newer ES9038 board that does not have balanced current mode output might be a reasonable thing to try. As I mentioned before, if I were building one from scratch for a DIY project I would try to use high-quality components rather than go for lowest-cost as it can affect sound quality a lot to people who notice such things.
ASIDE: Maybe this would also be a good point to say that I don't want to get to a big side-discussion about what some people can or can't hear, or certain other issues.
I am trying to approach this project from a field-expedient perspective. DIY'ers may not have access to expensive test equipment to optimize designs and make sure all design decisions are fully justified. In the old days before such test equipment was available, engineering sometimes had to be practised in less rigorous ways. Today, and for DIY purposes I feel such methods are okay so long as we try to refrain from making public claims that our design is better than we can prove.
It may be in some cases that I will use better and more expensive, or more quantity of components than absolutely necessary. For example, in the case of the clock upgrade, that I chose to do, to me the experimental results of using the best clock I could reasonably obtain showed unambiguous audible improvement. However, I want to point out that I have done a lot of practising with blind testing on myself over many years and I know quite well it is possible to fool one's self when relying solely on listening tests. For that reason, sufficient care should be taken when relying on such results in the absence of confirmation by measurement equipment. Blind testing can be an extremely useful tool if used appropriately. Listening tests with neither blinding nor confirmation by measurement should always be taken as suspect.
There is much more that could be said regarding the above topics, but I want to leave it here. My aim was to make a disclaimer and I think I have said enough.
If there is a desire talk about how to learn skilled listening or how to use blind testing in practical situations, we could perhaps take it up in another thread in one the other more appropriate sub-forums.
Last edited:
That's fair, Mark. We've had our differences plenty and I wanted to steer clear of personal impressions. I was just hoping to see if it's showing up clearly in a before/after Jtest to show anything pathological going on. Measurements are the levelest playing field we can all work on when we're not in the same room together.
Last edited:
Other way around said: the internal THD compensation does not bring in your test any significant SQ/measurement improvement ... no magic a diyer without NDA must be craving for (of course it would be nice to tinker with it ;-). So good ol' proven analogue measures are obviously the answer and way to go...
I've been reading this tread the last two days and I still have some questions about how to implement this board so perhaps someone here could give me some advice 
Is the I2S connection stable so its worth persuing a USB to I2S bridge board? I wan't this to work with my android to box. My current Arcam rdac works native on usb with that box, so I'm hoping a bridge would be native supported.. I'm a new to the I2S game and certainly to DSD format. It would be fun to try but not really a must, usb support is mostly what I'm after.
Is a usb to coax bridge just as good? On my pc I use a lower sabre chip with 32bit 384kHz and it sounds really nice, so I want the best resolution possible from the android tv box. It has only optical and usb out unfortunately.
Any affordable bridge recommendation would be appreciated! I've also seen bluetooth to I2s board on ebay, will that work?
I bought the version with the screen, but that will only show the pcm/dsd steam, not which source I'm using? I guess I can live with that. Have anyone looked into buying a remote kit for volume and more analoge inputs to make it a more capable preamplifier?
The weird volage divider needed could be a problem since I could not find any kits with KD potmeters, or KD pots in the same form factor as the one the kits used. So a pot swap is not going to be easy on those kits.
My mod plans for this board is dac chip decoupling with a Supercap (0.37F 5.5V) and a 1uF film ppt on the button's pins. I've no idea if that 1uF will do anything^^.
Bought a LT3045 for better regulation and I wonder if placing the output pin between the chip and the crystal would be an improvement to original placement?
I have some mediocre transformers I plan to use to I want to try out current mode and balenced output into a tpa3255 amp. I guess I dont need the dual rail voltage anymore so I hope I can use a 6-7V linear powersupply instead of the +-15V dual rail hassle.
Any advice will be greatly appreciated since im new to the I2S thing!
Is the I2S connection stable so its worth persuing a USB to I2S bridge board? I wan't this to work with my android to box. My current Arcam rdac works native on usb with that box, so I'm hoping a bridge would be native supported.. I'm a new to the I2S game and certainly to DSD format. It would be fun to try but not really a must, usb support is mostly what I'm after.
Is a usb to coax bridge just as good? On my pc I use a lower sabre chip with 32bit 384kHz and it sounds really nice, so I want the best resolution possible from the android tv box. It has only optical and usb out unfortunately.
Any affordable bridge recommendation would be appreciated! I've also seen bluetooth to I2s board on ebay, will that work?
I bought the version with the screen, but that will only show the pcm/dsd steam, not which source I'm using? I guess I can live with that. Have anyone looked into buying a remote kit for volume and more analoge inputs to make it a more capable preamplifier?
The weird volage divider needed could be a problem since I could not find any kits with KD potmeters, or KD pots in the same form factor as the one the kits used. So a pot swap is not going to be easy on those kits.
My mod plans for this board is dac chip decoupling with a Supercap (0.37F 5.5V) and a 1uF film ppt on the button's pins. I've no idea if that 1uF will do anything^^.
Bought a LT3045 for better regulation and I wonder if placing the output pin between the chip and the crystal would be an improvement to original placement?
I have some mediocre transformers I plan to use to I want to try out current mode and balenced output into a tpa3255 amp. I guess I dont need the dual rail voltage anymore so I hope I can use a 6-7V linear powersupply instead of the +-15V dual rail hassle.
Any advice will be greatly appreciated since im new to the I2S thing!
impulse60, I have a low-cost USB to SPDIF board coming from China, so haven't tried that yet. I would route it into the SRC4392 input and out of there to the DAC SPDIF in. The SRC4392 can support up to two I2S ports and four SPDIF/AES inputs. The Chinese SRC board I am using has TOSLINK and coaxial SPDIF inputs that are software selectable by writing to SRC4392 registers. It would take a little work to wire up the other unused inputs on the board.
Otherwise, if no SRC then USB to SPDIF conversion then to the DAC SPDIF input should work pretty well.
Transformer IV conversion will probably adversely affect performance compared to opamps. There is a reason boxes like Benchmark DAC-2 and Crane Song Avocet use opamps rather than transformers. But if one happens to like the sound effect of transformers then no problem I guess.
It turns out there is no magic bullet with these DACs to get a big sound quality improvement with one or two easy mods. It takes some work and attention to detail to compete strongly with some of the better commercially available products.
As far as remote mods go, there are probably a few ways to control volume. The volume pot is actually read by an A/D inside the microcontroller which then programs the DAC chip accordingly. Most of the volume control is between maybe about +1.5v and ground, with closer to ground giving more volume. A motorized pot, or a little DAC, or maybe a well-filtered PWM output from a microcontroller could potentially do it.
Similarly, the jumpers on the board to select inputs and reconstruction filters are microcontroller inputs. Installing a jumper pulls an input down to ground. One could do the same with relays, little switching FETs, or maybe outputs from another microcontroller. They could be opto-isolated or not.
Don't know how far back you read, but I would suggest starting with the following post and reading the references given there: http://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-58.html#post5399799
EDIT: Could you provide a link to the DAC board you will be using? It might help if we could see which one it is.
Otherwise, if no SRC then USB to SPDIF conversion then to the DAC SPDIF input should work pretty well.
Transformer IV conversion will probably adversely affect performance compared to opamps. There is a reason boxes like Benchmark DAC-2 and Crane Song Avocet use opamps rather than transformers. But if one happens to like the sound effect of transformers then no problem I guess.
It turns out there is no magic bullet with these DACs to get a big sound quality improvement with one or two easy mods. It takes some work and attention to detail to compete strongly with some of the better commercially available products.
As far as remote mods go, there are probably a few ways to control volume. The volume pot is actually read by an A/D inside the microcontroller which then programs the DAC chip accordingly. Most of the volume control is between maybe about +1.5v and ground, with closer to ground giving more volume. A motorized pot, or a little DAC, or maybe a well-filtered PWM output from a microcontroller could potentially do it.
Similarly, the jumpers on the board to select inputs and reconstruction filters are microcontroller inputs. Installing a jumper pulls an input down to ground. One could do the same with relays, little switching FETs, or maybe outputs from another microcontroller. They could be opto-isolated or not.
Don't know how far back you read, but I would suggest starting with the following post and reading the references given there: http://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-58.html#post5399799
EDIT: Could you provide a link to the DAC board you will be using? It might help if we could see which one it is.
Last edited:
Thanks for all the feedback Markw4! I just bought this one: ES9038 DAC DSD Decoder Support IIS DSD 384KHz Coaxial Fiber DOP+TFT LCD | eBay.
The transformers might perform worse, but I've become a fan of no caps in the audio chain lately. I also needed a reason to try the transformers. My goal is to make something that sounds different to my modded Arcam Dac. The headphone dac I've cobbled together has a sabre chip directly coupled to parallel op amps, and it sounds glorious to me. Compared it to 220usd Creative dac and mine sounded slightly better. I belive that some sound aspects is not represented with just THD measurements.
I hope to achive a good noise floor with the balenced outputs and less taxing on the psu with skipping of the op amps. I'd also like to hear the complete system without any opams. With the transformers I can skip the op amps in the amplifier aswell.
Regarding the jumpers, doesn't it require 2 jumpers to trigger one of the features? Or do I remember it wrong? Most of the kits has input relays so I hoped I could use those on the jumpers. That could be a problem if I suddenly needed 2 relays to trigger simultaneous.. I could create something in Arduino but its been a while since I've played with that.
What specs is that usb to SPDIF capable of? I'd like to achive something better than 24bit/192kHz that I currently have with the Arcam rDac. Could you provide a link to the board you bought? So the I2S has to go through the optical or coax on this board? (I thought for a moment that some of the pins also was some sort of an input^^).
I've have read the hole thread, but it's a lot to digest
The transformers might perform worse, but I've become a fan of no caps in the audio chain lately. I also needed a reason to try the transformers
. My goal is to make something that sounds different to my modded Arcam Dac. The headphone dac I've cobbled together has a sabre chip directly coupled to parallel op amps, and it sounds glorious to me. Compared it to 220usd Creative dac and mine sounded slightly better. I belive that some sound aspects is not represented with just THD measurements. I hope to achive a good noise floor with the balenced outputs and less taxing on the psu with skipping of the op amps. I'd also like to hear the complete system without any opams. With the transformers I can skip the op amps in the amplifier aswell.
Regarding the jumpers, doesn't it require 2 jumpers to trigger one of the features? Or do I remember it wrong? Most of the kits has input relays so I hoped I could use those on the jumpers. That could be a problem if I suddenly needed 2 relays to trigger simultaneous.. I could create something in Arduino but its been a while since I've played with that
. What specs is that usb to SPDIF capable of? I'd like to achive something better than 24bit/192kHz that I currently have with the Arcam rDac. Could you provide a link to the board you bought? So the I2S has to go through the optical or coax on this board? (I thought for a moment that some of the pins also was some sort of an input^^).
I've have read the hole thread, but it's a lot to digest
I've read the last pages and noticed a couple of things I missed the first time.
Do you think it necessary to change software on those to make them perform well?
They looked nice but I'm a afraid that they need better regulators aswell, and buying multiple ultra low noise LDOs quickly add up. I'd like to order the bridge soon since it will probably take 3 weeks to receive any board.
I've been tempted to upgrade the oscillator, but I'm not sure how easy the new one would be solder back on. I've read what you wrote on the process and it seemed like something I could manage with a bit of luck. So did you hear a difference from just the clock upgrade?
Do you think it necessary to change software on those to make them perform well?
They looked nice but I'm a afraid that they need better regulators aswell, and buying multiple ultra low noise LDOs quickly add up. I'd like to order the bridge soon since it will probably take 3 weeks to receive any board.
I've been tempted to upgrade the oscillator, but I'm not sure how easy the new one would be solder back on. I've read what you wrote on the process and it seemed like something I could manage with a bit of luck. So did you hear a difference from just the clock upgrade?
Last edited:
@impuls60, The board you are getting looks like the one I have except for the extra display board and rotary encoder. Don't know what those will do for you, maybe you can let us know when you find out.
Regarding the Sabre DAC you already have and like, I think you would probably find it is more or less junk compared to what this one can be if you do all the mods I suggest, including the addition of a small headphone amp if you want to have comparable quality there. However, you would be in for a lot of work and some more money. Depends what you would like to do.
Anyway, this is what I'm trying to do here: help people who want to do a DIY DAC mod project producing very high sound quality performance at the lowest cost we can figure out. Performance and sound quality first, cost second. I am also hoping that people who want to learn how to do some of these things in a DIY context will have a chance to do that along the way.
Regarding the Sabre DAC you already have and like, I think you would probably find it is more or less junk compared to what this one can be if you do all the mods I suggest, including the addition of a small headphone amp if you want to have comparable quality there. However, you would be in for a lot of work and some more money. Depends what you would like to do.
Anyway, this is what I'm trying to do here: help people who want to do a DIY DAC mod project producing very high sound quality performance at the lowest cost we can figure out. Performance and sound quality first, cost second. I am also hoping that people who want to learn how to do some of these things in a DIY context will have a chance to do that along the way.
Last edited:
I'm not sure how many LDOs will be necessary. So far I only used 2 linear regulators on the DAC board. I have been using a Silent Switcher to power the DAC itself, but I am hoping to get the headphone amp and DAC both on one lower cost switcher. There may be a need for several capacitors to be added to the cheap headphone amp kit. I don't use some of the stock parts that come with it.
USB to SPDIF board I will be trying looks like this: CM6631A Digital Interface Module DAC Board USB to IIS SPDIF Output 24Bit 192K 325646485555 | eBay
It also has I2S.
Headphone amp kit I started with looks like this: 1PCS LME49720NA+LME49600 headphone amplifier KIT | eBay
It looks quite a bit different now, especially from the bottom. I will try to post some pictures soon to help fill in how the project is progressing.
It also has I2S.
Headphone amp kit I started with looks like this: 1PCS LME49720NA+LME49600 headphone amplifier KIT | eBay
It looks quite a bit different now, especially from the bottom. I will try to post some pictures soon to help fill in how the project is progressing.
SRC4392 looks like this: Free Shipping! 1pc SRC4392 asynchronous Upconversion with USB coaxial optical I2S Decoder board-in Integrated Circuits from Electronic Components & Supplies on Aliexpress.com | Alibaba Group
This piece is an important one even though it is one of the more expensive items. The SRC4392 chip itself only costs about $15 new, but I don't have a better source for a board to easily integrate in with project as we are doing at this point.
Maybe later when everything is settled and done, somebody might want to look at designing a PCB to include only what is needed, or which can be populated only as desired. For now, we are still experimenting and figuring out an architecture for the whole thing on a modular level. Or, at least, I am.
This piece is an important one even though it is one of the more expensive items. The SRC4392 chip itself only costs about $15 new, but I don't have a better source for a board to easily integrate in with project as we are doing at this point.
Maybe later when everything is settled and done, somebody might want to look at designing a PCB to include only what is needed, or which can be populated only as desired. For now, we are still experimenting and figuring out an architecture for the whole thing on a modular level. Or, at least, I am.
Last edited:
I like the ambition level on this project!
What I meant with LDOs adding up is that I see 2 more regulators and a new clock on the SRC4392 board. I guess those would have to be of the same quality that of the DAC?? I read somewhere that the in slave mode the dac relies on external clock..dont know if it applies here.
I've started looking for crystals but shipping alone here in Norway is 25usd. I got free shipping at 50usd so I might order a crystal for that board also since I then dont have to pay for shipping. Do you think its a 100Mhz crystal utilised there too?
What I meant with LDOs adding up is that I see 2 more regulators and a new clock on the SRC4392 board. I guess those would have to be of the same quality that of the DAC?? I read somewhere that the in slave mode the dac relies on external clock..dont know if it applies here.
I've started looking for crystals but shipping alone here in Norway is 25usd. I got free shipping at 50usd so I might order a crystal for that board also since I then dont have to pay for shipping. Do you think its a 100Mhz crystal utilised there too?
The SRC4392 uses standard audio clock frequencies. For 192kHz, 96kHz, 48kHz, output it would use 24.5792MHz. For 44.1kHz and multiples of that, the clock would be 22.5792MHz. The SRC chip can have two clocks attached and select one or the other under software control depending on desired output sample rate.
For 221kHz out which is my target, the clock would be 27.000MHz. It should be a good quality crystal clock with low jitter. I will have to do a clock upgrade on that board too, but not sure yet if it needs to be quite as good as the DAC clock, or even exactly synchronous with the DAC clock (although that might help achieve best possible performance).
Should be okay though with a good very stable local clock if running with the DAC as I2S master. The SRC has a PLL that should lock onto the DAC BCLK frequency and keep the samples pipelined as needed. For the incoming BCLK to be low jitter and high quality, the I2S connection probably should be very short distance and or carefully treated as a transmission line. We will see how it looks, sounds, and performs when we get to that point.
For 221kHz out which is my target, the clock would be 27.000MHz. It should be a good quality crystal clock with low jitter. I will have to do a clock upgrade on that board too, but not sure yet if it needs to be quite as good as the DAC clock, or even exactly synchronous with the DAC clock (although that might help achieve best possible performance).
Should be okay though with a good very stable local clock if running with the DAC as I2S master. The SRC has a PLL that should lock onto the DAC BCLK frequency and keep the samples pipelined as needed. For the incoming BCLK to be low jitter and high quality, the I2S connection probably should be very short distance and or carefully treated as a transmission line. We will see how it looks, sounds, and performs when we get to that point.