Yeah, they are measurements of the linear vs minimum phase option in the AK4399 chip.
I took the pictures from this link Sabre32 | H i F i D U I N O
This link says "The ESS 9018 does not have a minimum phase filter."
I interpret this as an advantage the AK4399 has on the ES9018?
I've been taught minimum-phase fliter without pre-echo is more natural, i.e. closer to real life. Since, well, real life does not have pre-echo.
I don't know if it's true or not that minimum-phase is superior, but I feel like I can listen to my AK4399 DAC's for many hours without my subconscious telling me to turn it off.
When I listen to ES9018 I usually turn it off after 30 minutes or so. I noticed I am doing that subconsciously and I don't know why.
_____
You are asking me If I prefer Filter 01 / 02. Sorry, I don't know for certain if I prefer filter 01 / 02 in the AK4399, since I don't have the necessary controller equipment to change that setting. If I could change it, I would ideally use a randomizer so I don't know which filter is playing, then I would listen and identify. I think direct identifying is in some cases much easer than ABX switching.
Last edited:
To clarify, I mean I would use a randomizer to select the filter 01/02 setting, play music like at least 20 seconds, then identify which filter I am listening to. Then check the selection.
Then a time of silence so the listening mind is reset, then randomize and play the track from the start or a new track. Repeat 20 times all correct, for statistical evidence of the filter difference.
Which filter sounds better is a different question, I suspect minimum phase / zero pre-echo is more pleasant to the human mind, since that seems theoretically correct, however I am not sure.
Then a time of silence so the listening mind is reset, then randomize and play the track from the start or a new track. Repeat 20 times all correct, for statistical evidence of the filter difference.
Which filter sounds better is a different question, I suspect minimum phase / zero pre-echo is more pleasant to the human mind, since that seems theoretically correct, however I am not sure.
Last edited:
Hi Kastor L,
Thanks for your explanation and the link.
If your AK4399 DAC hasn't got software control, the DAC works in parallel mode and the digital filter is the sharp roll-off one. Only in serial mode (controlled by a micro controller) the minimum phase (or short delay) filter can be set.
I can't hear any difference between two at the moment, but that can change in the future. What isn't very convenient either, is the fact that it isn't clear which filter is filter 1 or 2 in our 2xAK4399 DAC. I like your idea of a random switch with a listening pause.
The new AK4490 DAC has even more filter options. It has 2 short delay filters, one with slow and one with sharp roll-off. I think this is a very nice DAC! It's only just released, May 2014 and probably only a few people are testing the device at the moment.
Thanks. I listened to a DAP recently and I had no idea which DAC was inside it, I listened to the sound carefully and decided it was PCM1792 since it sounded most similar to a DAC I heard with PCM1792. I was correct and now I like to try to blind identify like this.
It has a blue LCD panel on the front, but it's very basic, there is a reset button and input select button. I just assumed the designer sets it to minimum somehow while he's making it, that's possible isn't it? I saw a measurement with some high-end roll-off so I assumed it's the minimum. If it's linear then I could be confused.
Ok, sounds exciting.
Last edited:
Thanks for sharing that Info 😎
Just had a peek in the DS I downloaded, and it does look very impressive: 32 bits capability upto 768Khz (I don't think I'll ever hear the difference with a 192Khz mode)...
Hi guys,
I am going to order Tentlabs shunt regulators to use them instead of LT1085 and LT1963 but still thinking should I replace all of them or leave one LT1085 which is used as power supply of digital part of both AK4399 chips? In other words I think that it should be more important to improve clearness of power supply of analogue part of the AK4399 and AK4118 with PLL.
The Tentlabs shunt regulators do not have heat sinks and therefore input voltage must be between 8-12V for 5V regulator and 7-10V for 3.3V regulator. I have 13.5V after rectifier. I can either to remove LT1085 and install 28 ohm powerful resistor or I also can keep LT1085 but change its output voltage from 5 to 10V. I think that the second option should be better. I can install the Tentlabs shunt regulators in place of electrolytic capacitors which are very close to AK4399 chips. Those capacitors are useless in case using shunt regulators. Obviously in this case I will need to provide input voltage for the regulators by one additional wire which I can solder directly to the regulator and not forget to make cut of exiting wire on PCB which goes from LT1085 to analogue power supply input of the AK4399 chip.
Any suggestions are welcome
Thanks in advance.
I am going to order Tentlabs shunt regulators to use them instead of LT1085 and LT1963 but still thinking should I replace all of them or leave one LT1085 which is used as power supply of digital part of both AK4399 chips? In other words I think that it should be more important to improve clearness of power supply of analogue part of the AK4399 and AK4118 with PLL.
The Tentlabs shunt regulators do not have heat sinks and therefore input voltage must be between 8-12V for 5V regulator and 7-10V for 3.3V regulator. I have 13.5V after rectifier. I can either to remove LT1085 and install 28 ohm powerful resistor or I also can keep LT1085 but change its output voltage from 5 to 10V. I think that the second option should be better. I can install the Tentlabs shunt regulators in place of electrolytic capacitors which are very close to AK4399 chips. Those capacitors are useless in case using shunt regulators. Obviously in this case I will need to provide input voltage for the regulators by one additional wire which I can solder directly to the regulator and not forget to make cut of exiting wire on PCB which goes from LT1085 to analogue power supply input of the AK4399 chip.
Any suggestions are welcome
Thanks in advance.
Valeri100, why not change the transformer?
Use 7V X 2 for digital and 12V X 2 for analog.
This is what I have done for my DAC because Belleson Mini version regulator cannot be mount on heat sink too.
Use 7V X 2 for digital and 12V X 2 for analog.
This is what I have done for my DAC because Belleson Mini version regulator cannot be mount on heat sink too.
Hello Valeri100
The 5V Tentlabs shunt regulators have a max current of 80 mA. As it is a shunt, this means constant 80 mA current draw on the power supply per regulator. Analog 5V supply and digital 3.3V supply are on the same rectifier in you PCB version. Total current would be: 80 + 80 + 100 (for 3.3V) = 260 mA for the shunts. Add the digital supply for both DACS: another 2 x 70 mA max. = 140 mA . 400 mA current is quite a lot, this will result in a huge voltage drop and high ripple in your input voltage. I think this won't work.
Also, the 80 mA limit doesn't match with the 90 mA max analog current of one AK4399, see datasheet. So you need a regulator capable of delivering more current.
If you insist in wanting to use shunt regulators, maybe you can try to use part of the 2x15V supply for analog +5V. You can try to load the power supply with a resistor (you calculated for a certain current), and see how much voltage remains on the big filter capacitor. Sometimes I use a nice tool, PSU Designer II, to simulate the behavior.
My advice would be to use a series super regulator instead. Either way, as you already intended, you must place the regulators very close to the load, because the +5V power supply on your board is of very bad design IMO. Digital and analog ground are not separated and the regulators cannot regulate the load properly. So I think that placing super regulators on the same spot as the LT1085 will not give you the maximum potential of the super regulator.
Two super regs for analog power close to both AK4399s would be a nice mod I think, you solve two problems at once.
Regards, Nino
The 5V Tentlabs shunt regulators have a max current of 80 mA. As it is a shunt, this means constant 80 mA current draw on the power supply per regulator. Analog 5V supply and digital 3.3V supply are on the same rectifier in you PCB version. Total current would be: 80 + 80 + 100 (for 3.3V) = 260 mA for the shunts. Add the digital supply for both DACS: another 2 x 70 mA max. = 140 mA . 400 mA current is quite a lot, this will result in a huge voltage drop and high ripple in your input voltage. I think this won't work.
Also, the 80 mA limit doesn't match with the 90 mA max analog current of one AK4399, see datasheet. So you need a regulator capable of delivering more current.
If you insist in wanting to use shunt regulators, maybe you can try to use part of the 2x15V supply for analog +5V. You can try to load the power supply with a resistor (you calculated for a certain current), and see how much voltage remains on the big filter capacitor. Sometimes I use a nice tool, PSU Designer II, to simulate the behavior.
My advice would be to use a series super regulator instead. Either way, as you already intended, you must place the regulators very close to the load, because the +5V power supply on your board is of very bad design IMO. Digital and analog ground are not separated and the regulators cannot regulate the load properly. So I think that placing super regulators on the same spot as the LT1085 will not give you the maximum potential of the super regulator.
Two super regs for analog power close to both AK4399s would be a nice mod I think, you solve two problems at once.
Regards, Nino
Hi Sunsun22, Hi Nino,
Thank you very much for your valuable advices. I will think a little bit more what to do. I will need to change one transformer if I want to go with the shunts because the one i currently use is not poweful enough.
Best regards
Valeri
Thank you very much for your valuable advices. I will think a little bit more what to do. I will need to change one transformer if I want to go with the shunts because the one i currently use is not poweful enough.
Best regards
Valeri
I read all of the posts here and am surprised that no one has experimented much with the analog supply decoupling to the dac chip.
I had noticed on my dac5 that the decoupling here was at the chip mfr bare minimum suggestion. It would appear that is what is done on the ak boards as well.
I replaced the Silmic II, 10uf with larger Panasonic Pureism, and have noticed a huge change for the better. I think that this may well be partly because of when you have so many of one type of capacitor, it can become a bit colored. Also I think the value, and even the voltage will contribute to the results(lower esr). Something to be aware of is the risk of transferring heat to the dac while doing this, is better to just break(wiggle back and forth) off the Silmics, then lightly, and quickly attach the new leads here.
Another opportunity is the 104 capacitors(bypasses), to replace with something better makes a large difference also. I had started using some mkp1837 everywhere, but have since replaced with some surplus WimaFKP on all of the analog positions(placed underneath).
For the remainder of the 104 caps, I used either some surplus Evox polycarbonate(digital), or the mkp1837.
The rest of the setup is; 4700 uf Mundorf AG, main filter caps, 2 X 100uf Silmic Super Gold +47uf 25v Silmic ARS for analogue decoupling. The opamps are Burson dual modules, and a LME49720 is used for the servo.
I have feeling that these ak boards would benefit from some improvements to the analogue side of things as well, as they seem to be quite similar.
I had noticed on my dac5 that the decoupling here was at the chip mfr bare minimum suggestion. It would appear that is what is done on the ak boards as well.
I replaced the Silmic II, 10uf with larger Panasonic Pureism, and have noticed a huge change for the better. I think that this may well be partly because of when you have so many of one type of capacitor, it can become a bit colored. Also I think the value, and even the voltage will contribute to the results(lower esr). Something to be aware of is the risk of transferring heat to the dac while doing this, is better to just break(wiggle back and forth) off the Silmics, then lightly, and quickly attach the new leads here.
Another opportunity is the 104 capacitors(bypasses), to replace with something better makes a large difference also. I had started using some mkp1837 everywhere, but have since replaced with some surplus WimaFKP on all of the analog positions(placed underneath).
For the remainder of the 104 caps, I used either some surplus Evox polycarbonate(digital), or the mkp1837.
The rest of the setup is; 4700 uf Mundorf AG, main filter caps, 2 X 100uf Silmic Super Gold +47uf 25v Silmic ARS for analogue decoupling. The opamps are Burson dual modules, and a LME49720 is used for the servo.
I have feeling that these ak boards would benefit from some improvements to the analogue side of things as well, as they seem to be quite similar.
Hi phase
Thanks for sharing your experiences with the dac5.
I've experimented some with the analog/digital supply local decoupling, but mainly by using bigger/other electrolytics instead of the 47uF used. It didn't do any good in my setup. But I agree, the HF decoupling (the 104 bypasses) is very basic. I think especially at the reference voltage pins, HF decoupling has a big effect and is worth experimenting with.
I'm just setting the first steps to do so. I've replaced the two 100nF Wimas decoupling VREFH / VREFL with 1 uF, just as is done in the evaluation board (AK4399EQ). Sound has become even more detailed.
I just received 0603 SMT ceramics, which I want to solder directly at the AVDD pin of the DAC. Magnifying glass indispensable! My board doesn't have a proper HF bypass for AVDD, an omission in the design.
I can imagine using 100nF filmcaps instead of ceramic makes an audible difference, but I'm afraid the large size of them makes them also very susceptible of picking up HF noise (antenna effect), which is just you want to reduce, don't you think? The delta-sigma modulator is a big noise generator. The ceramics are very small though, and can be placed very close to the pins.
Thanks for the tip about trying different electrolytics without desoldering every time, good idea, I think I'll try it.
Thanks for sharing your experiences with the dac5.
I've experimented some with the analog/digital supply local decoupling, but mainly by using bigger/other electrolytics instead of the 47uF used. It didn't do any good in my setup. But I agree, the HF decoupling (the 104 bypasses) is very basic. I think especially at the reference voltage pins, HF decoupling has a big effect and is worth experimenting with.
I'm just setting the first steps to do so. I've replaced the two 100nF Wimas decoupling VREFH / VREFL with 1 uF, just as is done in the evaluation board (AK4399EQ). Sound has become even more detailed.
I just received 0603 SMT ceramics, which I want to solder directly at the AVDD pin of the DAC. Magnifying glass indispensable! My board doesn't have a proper HF bypass for AVDD, an omission in the design.
I can imagine using 100nF filmcaps instead of ceramic makes an audible difference, but I'm afraid the large size of them makes them also very susceptible of picking up HF noise (antenna effect), which is just you want to reduce, don't you think? The delta-sigma modulator is a big noise generator. The ceramics are very small though, and can be placed very close to the pins.
Thanks for the tip about trying different electrolytics without desoldering every time, good idea, I think I'll try it.
By the way: what are the Panasonic Purism caps? I've never heard of them (only Panasonic FC and FM types), are they good?
I admire your patience using the smt capacitors, and I know they can be better in certain places, especially digital places.
The Pureism capacitors came from fleabay, the ones I used were 100uf, just because I had them around, I would think that 22uf to 47uf would be just fine. Another to try here might be Rubycon ZLH.
From what I understand, the Pureism were Panasonics attempt at audio capacitors, came in some bluray players. They sound like a smoother FC, or a more linear Silmic maybe.
Another thing to try is to place an X2 capacitor across the incoming AC. This helped a lot to lower the noise, picked up more detail as a result.
The Pureism capacitors came from fleabay, the ones I used were 100uf, just because I had them around, I would think that 22uf to 47uf would be just fine. Another to try here might be Rubycon ZLH.
From what I understand, the Pureism were Panasonics attempt at audio capacitors, came in some bluray players. They sound like a smoother FC, or a more linear Silmic maybe.
Another thing to try is to place an X2 capacitor across the incoming AC. This helped a lot to lower the noise, picked up more detail as a result.
Last edited:
Is it possible to use the 10pin xmos usb module for the dual mono ak4399 dac?
disconnect the I2s signal from ak4113 and connect to the input of ad4399 to accept dsd
input. (manually wire the 4 output signal from the module to the main dac pcb) I presume
the dual ad4399 with software control is not connected for parallel mode.
tnx.
disconnect the I2s signal from ak4113 and connect to the input of ad4399 to accept dsd
input. (manually wire the 4 output signal from the module to the main dac pcb) I presume
the dual ad4399 with software control is not connected for parallel mode.
tnx.
The first step for me😀 is the power supply .
Actualy for me 😀 the best results are with the SALAS Reg Shunt but it's my opinion .
After if you are not Happy ...... change the capa
Serge
Thanks,
it's quite a challange...Lately I was thinking: maybe the antenna effect is not that much problem here, because the caps are for decoupling. So any noise picked up, they are supposed to lead to ground. Anyway, it works in your case, so maybe I'll try too.
systeme: I don't know the 10pin XMOS module, but in order to work 'out of the box' with the 2xAK4399, it must only have SPDIF output.
But you can disconnect the AK4113 and use I2S input, of course. Then only your USB input remains. But if you want to play DSD as well, both DACS must be set into DSD mode by the micro controller software. So you must program a new micro controller to make the switch possible.
Another thing: Filtering of DSD output is also different, see datasheet. You need at least 30dB/octave attenuation and cut-off at 50kHz. This can be implemented with a 3rd order LPF in combination with the DACs' internal filter. PCM filter on this dual AK4399 PCB is 2nd order with 120kHz cut-off.
Last edited:
Hi lomonosoff: I did try it at the beginning but since then I didn't, at the beginning it sounded bit boring, I'm still waiting for other parts to arrived so I'll give it a try once everything received...
- Status
- Not open for further replies.