Thanks Mark,
Here are my boards.
Attachments
Yes, they can play DoP and native DSD.
If you don't like the sound of proper I/V stages, it means you are doing something wrong. This dac can sound really good if you know how to mod it correctly. However, making this dac sound like a $1,000+ dac is a lot work. To get the best sound with single ended tube amps still requires making the dac really good first, then adding the tubes at the end to further shape the sound as desired. Otherwise, all the tubes do is help mask the ugly distortion of a poorly performing dac.
Please tell me what I can do so that my DAC run native DSD.
Maybe need a proper driver? or correct setting in Jriver? I don't understand much on these.
Tks,
Tran.
You need a USB board that comes with working ASIO drivers. As mentioned previously, JLSounds makes one that is probably lowest cost solution. It uses good low jitter clocks and provides galvanic isolation. There is also what is sometimes referred to as Genuine Italian Amanero USB boards. Chinese USB boards don't work for native DSD.
To get the best native DSD sound quality, among other things the dac registers would need to be reprogrammed, but no point bother with that if using single ended voltage mode output. The distortion is already very high that way, 50dB or more worse than it should be, so hard to hear much improvement.
Anyway, once there is a USB board and the ASIO drivers are installed, then just make sure that Windows does not show any default device checks marks in Control Panel > Sound settings for that audio device. If the USB board is not the Windows default device for anything, the ASIO can then work properly. Then you need a program that can play native DSD. HQplayer offers the highest quality conversion to DSD512, but it takes a powerful i7 computer to be able to run the best sounding DSP algorithms.
For lower cost, it is possible to use a cheap Chinese AK4137 board to convert PCM such as CD audio to DSD256, but getting the best sound quality out of it is difficult. I am still working on finding an easier way to do that.
EDIT: By the way, if you meant you didn't like the balanced output that your dac originally had, you are quite right that it is awful. That's not a proper output stage either, and still voltage mode output which is very high distortion. In voltage mode, distortion is around -70dB. In a properly designed current mode version, distortion can be as low as -120dB. It sounds great, but you may never have heard a dac as good as one of these can sound if all the mods are done on it. In that case, no way for you to know just how good it can be.
Last edited:
Thank you, Mark!
You mean that USB board has not to be " Set Default" in " Audio Settings" ?
My mini PC doesn't get any audio device, so when I plug USB DAC in, it becomes " Set Default" automatically.
I will try Amanero board.
PS: I use Output transformer (20K: 600 ohms) for loading that tubes to stepdown the signal.
Please see pic below. The Green circle with check mark inside is bad for audio. It means Windows will really control the device, not your audio application.
You need at least two audio devices to choose from in Windows. Make sure the green circle with the check mark inside is for the audio device you will not use.
Make sure there is no Green circle for audio device you do want to use, also make sure that the audio device you want to use has a working ASIO driver. Set your audio playback application to use the ASIO driver for native DSD.
Also, for best sound quality you should change to using the output stage schematic and AVCC schematic as shown in Post #3003 of this thread. Use OPA1612 opamps, not what is shown in the schematics. In addition, add some parallel film caps to each of the +-15v rails, about 50uf to 100uf total film caps from the +15v rail to ground, and the same 50uf to 100uf total film caps from the -15v rail to ground. Then go from the single ended opamp output into your tube circuit. It will sound 100 times better. Even just doing the AVCC mod will make a huge improvement in sound quality. https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-301.html#post5577605 ...you will be very happy if you try it, I think. Everyone else has found it to be much better.
Attachments
Last edited:
Hi Claude,
Things have been moving very slowly here, but should pick up soon (I hope). We did at least get to hear some DSD512 and the initial impression is that AK4499 sound very good, and very promising. So far, I think it sounds better than I have been able to get Sabre to sound, and it doesn't have that characteristic Sabre sound which tends to be a little on the bright side. Vocals sound better, I think. Bass, too. Saxophones, maybe not. One thing is that the best sound we know how to get today happens to be very dependent on the best sound we can get out of HQplayer. That's true for Sabre and for AK4499, IMHO.
However, AK4499 is very different from Sabre in terms of the care and feeding aspects. The eval board circuitry and the register programming model is quite different from Sabre, which means there is a learning curve with AK4499 just as there has been with Sabre. I have spent some time studying the datasheet and eval board manuals, and learning about how to program it to do what I want. Also learning about the effects of all the various power supplies, decoupling, etc., on sound quality. That has to go slow since the chip is sensitive to power-on and power-off sequencing, among other things. Clocks are the last thing that are allowed to be turned on, for example.
I would also say that like the ESS datasheet, the AKM data sheet is fine for those who already understand the material and are just trying to look up the info they need. The assumptions baked into the document writing in both cases assume to some extent one already knows the chips pretty well. Its the kind of thing that happens when engineers who know the chips well are involved in writing the documentation. Some things seem more than obvious to them because they have been thinking about them and working with them for quite some time. For someone coming in new to ESS products or new to AKM products, it takes some time to get used to the way each tries to communicate what they think is enough to tell you about what they think you probably should need to know. At least the AKM sales rep recently put Jam and me into contact with a Senior Applications Engineer here in the US, so I have someone who is very familiar with AKM's way of doing things when I want to ask a question. It was similar with ESS, when I had I question I could route it through the rep to ESS and get back a written response within a week or so.
Anyway, the bottom line is that I don't have any strong opinion yet as to how good AKM vs ESS can sound. I am also just trying to finish up my Sabre ES9028PRO synchronous mode dac. In the meantime, there are some other communications going on in the background with some people who are just now learning how to get the very best out of Sabre, maybe a little better than anyone has ever done before. Obviously, it will probably take some time before we are at that point for AK4499. Otherwise, no doubt there will be some new products hitting the market soon after AK4499 production has ramped up. Some of the manufacturers in Japan have been working with AKM for a long time, and it remains to be seen how sophisticated their initial dac products will be. My guess would be that it might be quite awhile before we hear the best from the part though, since there are a lot of variables that could affect sound yet to be explored.
In the case of AK4499 eval board, nearly all the power supplies on the board can be jumpered out and external power supplies used instead. No telling what might end up affecting sound there unless some experiments are tried. Also, the eval board doesn't even have a differential summing stage. It is up to designers to come up with their own (although there are suggestions in the datasheet). The I/V opamps are soldered in very close to the dac chip, but I think that may be intended more for safety than anything else. Since I/V stages tend to operate from higher rail voltages than AK4499 can withstand, it is important that not too many dac chips be accidentally destroyed especially while chip production is very limited. That last thing they want designers doing stuff with is the I/V stage, simply due to the risks. Please play around with power supplies first, seems to be the implication. Learn about what you can do that way first, then decide if you need to change the I/V, not the other way around.
That's okay though. I hope that things will start to move faster here before too long. Also, I hope to have more sound say about sound quality of the basic eval board before long too. My initial listening tests using the balanced outputs (that are offset at +2.5v) have been to try some transformers and the balanced input on my headphone amp. I have yet to construct one of the datasheet example differential summing stages, which I expect to have some effect on sound quality.
I also have not tried other reference voltage supplies (think AVCC). For one thing they did a funny ground isolation thing on the eval board for the AVCC supplies. As a result it takes some big electrolytic bypass caps on the AVCC supply to the high side of the ground isolation resistors to minimize bass distortion. The eval board doesn't come with the very largest possible caps. Why? I don't know exactly, but power is supposed to be sequenced on and sequenced off. If big caps take longer to discharge, how to I account for that if the power plug gets pulled? A crowbar circuit? Zener clamps? Don't know.
The dac chips can also produce pops at known times such as when some internal settings are switched, for example. It is up to the designer to make a mute circuit for that if they feel it is needed for their application.
In summary, the chip is complex, it is very different than Sabre, and while it sounds good and very promising, there is a whole lot of work to be done before we know as much as we would like. At least there is no NDA keeping designers from collaborating in places like diyaudio. As soon as eval boards hit the high volume distributors we will probably be seeing a lot people working away on finding out useful things, and or there will be a few blown chips too. Bottom line though, I think AK4499 is probably the best dac chip I have heard so far, and I think people will probably like new, non-Sabre sound better overall.
Things have been moving very slowly here, but should pick up soon (I hope). We did at least get to hear some DSD512 and the initial impression is that AK4499 sound very good, and very promising. So far, I think it sounds better than I have been able to get Sabre to sound, and it doesn't have that characteristic Sabre sound which tends to be a little on the bright side. Vocals sound better, I think. Bass, too. Saxophones, maybe not. One thing is that the best sound we know how to get today happens to be very dependent on the best sound we can get out of HQplayer. That's true for Sabre and for AK4499, IMHO.
However, AK4499 is very different from Sabre in terms of the care and feeding aspects. The eval board circuitry and the register programming model is quite different from Sabre, which means there is a learning curve with AK4499 just as there has been with Sabre. I have spent some time studying the datasheet and eval board manuals, and learning about how to program it to do what I want. Also learning about the effects of all the various power supplies, decoupling, etc., on sound quality. That has to go slow since the chip is sensitive to power-on and power-off sequencing, among other things. Clocks are the last thing that are allowed to be turned on, for example.
I would also say that like the ESS datasheet, the AKM data sheet is fine for those who already understand the material and are just trying to look up the info they need. The assumptions baked into the document writing in both cases assume to some extent one already knows the chips pretty well. Its the kind of thing that happens when engineers who know the chips well are involved in writing the documentation. Some things seem more than obvious to them because they have been thinking about them and working with them for quite some time. For someone coming in new to ESS products or new to AKM products, it takes some time to get used to the way each tries to communicate what they think is enough to tell you about what they think you probably should need to know. At least the AKM sales rep recently put Jam and me into contact with a Senior Applications Engineer here in the US, so I have someone who is very familiar with AKM's way of doing things when I want to ask a question. It was similar with ESS, when I had I question I could route it through the rep to ESS and get back a written response within a week or so.
Anyway, the bottom line is that I don't have any strong opinion yet as to how good AKM vs ESS can sound. I am also just trying to finish up my Sabre ES9028PRO synchronous mode dac. In the meantime, there are some other communications going on in the background with some people who are just now learning how to get the very best out of Sabre, maybe a little better than anyone has ever done before. Obviously, it will probably take some time before we are at that point for AK4499. Otherwise, no doubt there will be some new products hitting the market soon after AK4499 production has ramped up. Some of the manufacturers in Japan have been working with AKM for a long time, and it remains to be seen how sophisticated their initial dac products will be. My guess would be that it might be quite awhile before we hear the best from the part though, since there are a lot of variables that could affect sound yet to be explored.
In the case of AK4499 eval board, nearly all the power supplies on the board can be jumpered out and external power supplies used instead. No telling what might end up affecting sound there unless some experiments are tried. Also, the eval board doesn't even have a differential summing stage. It is up to designers to come up with their own (although there are suggestions in the datasheet). The I/V opamps are soldered in very close to the dac chip, but I think that may be intended more for safety than anything else. Since I/V stages tend to operate from higher rail voltages than AK4499 can withstand, it is important that not too many dac chips be accidentally destroyed especially while chip production is very limited. That last thing they want designers doing stuff with is the I/V stage, simply due to the risks. Please play around with power supplies first, seems to be the implication. Learn about what you can do that way first, then decide if you need to change the I/V, not the other way around.
That's okay though. I hope that things will start to move faster here before too long. Also, I hope to have more sound say about sound quality of the basic eval board before long too. My initial listening tests using the balanced outputs (that are offset at +2.5v) have been to try some transformers and the balanced input on my headphone amp. I have yet to construct one of the datasheet example differential summing stages, which I expect to have some effect on sound quality.
I also have not tried other reference voltage supplies (think AVCC). For one thing they did a funny ground isolation thing on the eval board for the AVCC supplies. As a result it takes some big electrolytic bypass caps on the AVCC supply to the high side of the ground isolation resistors to minimize bass distortion. The eval board doesn't come with the very largest possible caps. Why? I don't know exactly, but power is supposed to be sequenced on and sequenced off. If big caps take longer to discharge, how to I account for that if the power plug gets pulled? A crowbar circuit? Zener clamps? Don't know.
The dac chips can also produce pops at known times such as when some internal settings are switched, for example. It is up to the designer to make a mute circuit for that if they feel it is needed for their application.
In summary, the chip is complex, it is very different than Sabre, and while it sounds good and very promising, there is a whole lot of work to be done before we know as much as we would like. At least there is no NDA keeping designers from collaborating in places like diyaudio. As soon as eval boards hit the high volume distributors we will probably be seeing a lot people working away on finding out useful things, and or there will be a few blown chips too. Bottom line though, I think AK4499 is probably the best dac chip I have heard so far, and I think people will probably like new, non-Sabre sound better overall.
Last edited:
Hi Mark
Many thanks for this very detailled and complete answer!
I understand the learning curve and that will make it somewhere more entertaining: don't we want DIY and trials... and learn from it
All you say sounds very promising, this is possibly the way to go either with this evaluation board, or if too expensive with a chinese one. But in either case no doubt AKM will have many followers and a big community to complement your seaches and works, something I am interested to contribute to aswell.
Again many thanks for all this
Claude
Many thanks for this very detailled and complete answer!
I understand the learning curve and that will make it somewhere more entertaining: don't we want DIY and trials... and learn from it
All you say sounds very promising, this is possibly the way to go either with this evaluation board, or if too expensive with a chinese one. But in either case no doubt AKM will have many followers and a big community to complement your seaches and works, something I am interested to contribute to aswell.
Again many thanks for all this
Claude
Nice amp tcq.
Is this a 'spud' amplifier, ie a single stage amp?
I've heard some pretty darn good SET amps and was involved with design of
one that used a 45 OP tube. Stunning sound with the right speakers.
T
Hi Terry,
01A, 26 and 45 are the best sounding triode tubes that USA have manufactured. And my system uses all 3 of them
That photo you see is buffer tube after DAC chip. They include 01A/26 tube (switchable) along with Tango ISO NP-126 OPT. DAC board get inside it.
Tube 45 SET amp is unbeatable if we have enough high sensitive speaker.
Last edited:
Hi Terry,
01A, 26 and 45 are the best sounding triode tubes that USA have manufactured. And my system uses all 3 of them
That photo you see is buffer tube after DAC chip. They include 01A/26 tube (switchable) along with Tango ISO NP-126 OPT. DAC board get inside it.
Tube 45 SET amp is unbeatable if we have enough high sensitive speaker.
Hi Tran,
I don't own a SET amp lthough have been tempted many times, however,
helped a good friend with his development of a world class 45 amp.
His speakers are Goodmans Axiom 80 (with OB subs) and for what they do
right, ie; ndrange, they are incredible speakers. I know a few people now
playing with the Ruilitt
Do you have the Sabre DAC board inside the tube buffer base ?
Your DAC setup sounds interesting, are you driving DAC into passive I-V
-> tube -> Tango transformer.
T
Well it's nice to see you have changed your mind on transformers.
Has Jam has been schooling you up a bit?
The whole VREF (AVCC) setup is strange. With 0 ohms for R14_x, the LF
distortion is lowest for a given electro cap size but there is also the distortion
hump at 5k, which the 10 ohm decoupling R appears to fix.
AD817 based regs for VREF are a simple Jung SuperReg knock off, for the
most part, wher the Reg OP bootstraps the opamp power supplies. But
AD817 has significant OP noise. They were obviously focusing on speed and
transient response with these supplies.
Also the whole 'gull wing' VREF grounding arrangement is questionable.
I think as previously mentioned (1audio), looks like there will be
considerable optimization possibilities and most likely AKM don't even know
the real performance potential of this DAC. Case in point their AK5394 ADC,
people have attained performance *far exceeding data sheet specs.
So it's fair to say this is all a good thing - we have a DAC here with unknown
potential, but it will take some work to unearth that potential.
I haven't even delved into OP voltage swing configurations and how this DAC
tolerates it. J Westlake was alluding on another thread, that the resulting
distortion is due to internal resistor non linear voltage coefficient on Sabre.
Has AKM improved this? An ability to tolerat *some voltage swing on OP
will open up many possibilities.
Crowbar (brute force) can work but you can also get around ramp up time
on very low noise RC attenuation networks with a schottky diode bypass on
the R before the C. The schottky forces the C high to within 0.2 volts of desired voltage, ie; enough to get things running, then the remainder 0.2 or
so volts 'floats' up to desired spec with full RC noise attenuation and the diode is out of the circuit.
None of this should be required with proper grounding design. It's not
rockets science to keep ground current loops separate.
Onward and upward so to speak
Do you know the projected price of this DAC chip?
T
I’m new to this party so was wondering if anyone could post a picture of this output stage to help me with layout please. Also I’m assuming I can use opa1602 or 1612 instead of the lme49720
Tim
Last edited:
Hi Tim,
I was asked to provide output stage layout instructions using through-hole components. My preference would probably have been to use SMD, which I think might have actually been easier to construct. In any case, the package I put together can be found at: Dropbox - Output Stage Instructions.zip
Please keep in mind that other modifications are very important too. In particular, the AVCC circuit is critical as the dac chip has zero PSRR at the AVCC terminals, and any defects in AVCC regulation are quite audible. If you would like more info on that aspect of construction, please let me know and I will try to see what I can find for you.
I was asked to provide output stage layout instructions using through-hole components. My preference would probably have been to use SMD, which I think might have actually been easier to construct. In any case, the package I put together can be found at: Dropbox - Output Stage Instructions.zip
Please keep in mind that other modifications are very important too. In particular, the AVCC circuit is critical as the dac chip has zero PSRR at the AVCC terminals, and any defects in AVCC regulation are quite audible. If you would like more info on that aspect of construction, please let me know and I will try to see what I can find for you.
Well it's nice to see you have changed your mind on transformers.
Has Jam has been schooling you up a bit?
Yes. He even gave me some little transformers to try that were used on a commercial product. Unfortunately, they were not as clean as the differential receiver in my headphone amp, which I am not sure is as clean as it can get. So far, I still see transformers as a sometimes necessary evil.
Link?
Sorry, don't know yet.
Agree with much of your commentary.
cheers
