Well... my board is the 4118, I'm using hdam + dxio pro3 USB interface.
Dxio prro3 is a huge improvement in quality vs the direct XMOS weiliang interface .
If anyone bought this DAC, I raccomand to buy it without USB interface.
Inviato dal mio SM-N9005 utilizzando Tapatalk
Last edited:
Hello keres
Thanks for sharing your experiences with the USB interface / warning us
In my post I was referring to the crystal close to the AK4118 on the second version PCB. The optional XMOS board has it's own crystals, for converting USB to spdif. These crystals performances determine quality of the spdif signal the USB board delivers.
The first version PCB, with the AK4113, didn't use a crystal. Yet it displays the sample frequency, but only when it's defined explicitly in the bit-stream by the transport. The receiver in the second version actually compares the incoming frequency with the fixed crystal one, to determine the sampling rate. So the crystal is used to measure the sampling frequency. The master clock is still derived from the spdif signal by use of a PLL and a VCO.
Thanks for sharing your experiences with the USB interface / warning us
In my post I was referring to the crystal close to the AK4118 on the second version PCB. The optional XMOS board has it's own crystals, for converting USB to spdif. These crystals performances determine quality of the spdif signal the USB board delivers.
The first version PCB, with the AK4113, didn't use a crystal. Yet it displays the sample frequency, but only when it's defined explicitly in the bit-stream by the transport. The receiver in the second version actually compares the incoming frequency with the fixed crystal one, to determine the sampling rate. So the crystal is used to measure the sampling frequency. The master clock is still derived from the spdif signal by use of a PLL and a VCO.
Usually USB - SPDIF convertors/modules have 3 different internal oscillators for making clock of SPDIF signal. Stability of these oscillators makes impact on sound quality. Whatever USB module you use the AK4118 of the DAC will extract clock from SPDIF signal. The internal oscillator is used only for sampling rate detection. It is described on page 17 of the AK4118 datasheet.
I thought he/she is going to use TCXO at somewhere else. My suggestion will not make any sense either.
Hi Nosian,
Does it still work and produce nice sound? Was it always as hot as it is now or you touched it for the first time and got surprised that it is so hot?
I know that analogue part of this chip runs with relatively high currents and gets warm because of it. I never tried to touch the chip while DAC was working, but I touched heat sinks of +5V regulators of analogue and digital parts. They were hot but not very much. I could keep finger on it without any pain. Please try it and if one of those is very hot then you can conclude that there is a problem in analogue or digital part. If DAC still works then most probably problem is on analogue side. It might be oscillation of internal OPAs perhaps due to bad soldering and lose of connection of one of decoupling capacitors.
Do you use transformers and output stage instead of OPAs? If YES you need to check if there is equal +2.5V DC offset of each DAC Voutput (+ and -). If is not equal then it is the reason because resistance of transformer for DC is very low.
Hi 730user,
It is hard to predict result. I think it won't reduce HF noise which is hard to eliminate with RC filter with electrolytic capacitors.
In this DAC there are 3 regulators which use common rectifier. I remember that I have seen somewhere advice to use is such cases RC filters at input of regulators because such kind of regulators see HF noise (above 100kHz) in its control pin and try to compensate it creating also noise at input of all regulators. This advise sounds for me logical but I didn't try it although it is not difficult. It just requires to make 3 cuts and solder 3 resistors + 3 capacitors on bottom side of the board.
Hi Nosian,
It doesn't sound good
Hum might be a sign for high power consumption due to damage of analogue part of AK4399. I would use oscilloscope in order to check if there is clean +5V for analogue and digital parts of both AK4399 chips. Each chip has own +5V regulator for analogue part and common one for digital part. It is more convenient to put oscilloscope probe on outputs of regulators from back side of the board because it is very hard to put it directly on small pins of AK4399.
Then I would put oscilloscope probe on resistors 12k connected to Vout(s) of both AK4399 chips to check if there is normal +2.5V DC offsets on all outputs and a test signal. After these measurements you will see troubles (if any).
good luck
Last edited:
Hi 730user,
LT1085 regulators 3x220uf/6.3V Panasonic FM (marked as 10uf on PCB), consequently 2x Elna SilmicII 470uf/6.3V and 1x220uf/6.3V, Elna recommended by Nino, better dynamics and bass response... MJE15034/5G regulators 2x470uf/25V Panasonic FM, consequently 2x1000uf/25V Panasonic FM, of course instead of Panasonic you can use any other low impedance caps, no change for LM7805 and LT1963, latest version here...
LT1085 regulators 3x220uf/6.3V Panasonic FM (marked as 10uf on PCB), consequently 2x Elna SilmicII 470uf/6.3V and 1x220uf/6.3V, Elna recommended by Nino, better dynamics and bass response... MJE15034/5G regulators 2x470uf/25V Panasonic FM, consequently 2x1000uf/25V Panasonic FM, of course instead of Panasonic you can use any other low impedance caps, no change for LM7805 and LT1963, latest version here...
Hi, kukynas!
I decided to install a 1 ohm resistor at the output of regulators lt1085 and install capacity 2200x16v Panasonic XPro. They are not expensive and with good parameters. They are used in my minidac pcm1794.
Exit MJE - capacitors Philips 2200x35v. I would bet the same Panasonic XPro, but 16v voltage at them it is not enough. Right?
I decided to install a 1 ohm resistor at the output of regulators lt1085 and install capacity 2200x16v Panasonic XPro. They are not expensive and with good parameters. They are used in my minidac pcm1794.
Exit MJE - capacitors Philips 2200x35v. I would bet the same Panasonic XPro, but 16v voltage at them it is not enough. Right?
Last edited:
+5V analog psu
Hi all
I have some updates regarding +5V analog supply of both AK4399 DACs.
I wanted to make a shunt-regulator-type psu for +5V analog. Each DAC needs approx. 90mA current. I decided to build a Salas Shunt reg, capable of delivering current to both DACs. Advantages of a shunt reg are the very low impedance and the low noise. Because each shunt needs at least 100mA headroom, and all unused current is dissipated as heat, I decided to built one regulator for both DACs. Eventually this didn't work. Because of the low impedance, I got problems with resonance on both DACs . I tried everything, to shield off the wires, but it didn't help.
So, for those of you who want to try a shunt regulator: It probably only works if you use one for each DAC, so you have to built two of them.
I decided to take another approach, and I replaced the LT1086/1085 regulators with 'plain' 7805 regulators. Why? The LT1085 regulators are low drop out ones, which means they can operate with lower input voltage than normal linear regulators. That's nice if we want to be economical with power, because the surplus of voltage x current is wasted as heat. But in this application, we have 10+Volts of input voltage, so LDO isn't beneficial. Because the LDO regulator's pass device is different, it needs quite a big capacitor in order to be stable. That's why the output cap is of so much influence to the sound.
So I decided to use a OnSemi MC7805ACT regulator. It has much better specs in terms of noise and ripple rejection. And it is totally satisfied with a 100u output cap. It has 2V drop-out voltage, which is fine. Sound is better than with the LT1086 regulators in my setup. It's less colored and basses are tight. I like this result best.
There's still some headroom to apply some filtering of HF-noise before the regulator, so that's my next effort. Lately I discovered that it can be very rewarding to do so, when improving my discrete output stage. HF-noise isn't enough rejected by the regulators.
I'll keep you posted.
Regards, Nino
Hi all
I have some updates regarding +5V analog supply of both AK4399 DACs.
I wanted to make a shunt-regulator-type psu for +5V analog. Each DAC needs approx. 90mA current. I decided to build a Salas Shunt reg, capable of delivering current to both DACs. Advantages of a shunt reg are the very low impedance and the low noise. Because each shunt needs at least 100mA headroom, and all unused current is dissipated as heat, I decided to built one regulator for both DACs. Eventually this didn't work. Because of the low impedance, I got problems with resonance on both DACs . I tried everything, to shield off the wires, but it didn't help.
So, for those of you who want to try a shunt regulator: It probably only works if you use one for each DAC, so you have to built two of them.
I decided to take another approach, and I replaced the LT1086/1085 regulators with 'plain' 7805 regulators. Why? The LT1085 regulators are low drop out ones, which means they can operate with lower input voltage than normal linear regulators. That's nice if we want to be economical with power, because the surplus of voltage x current is wasted as heat. But in this application, we have 10+Volts of input voltage, so LDO isn't beneficial. Because the LDO regulator's pass device is different, it needs quite a big capacitor in order to be stable. That's why the output cap is of so much influence to the sound.
So I decided to use a OnSemi MC7805ACT regulator. It has much better specs in terms of noise and ripple rejection. And it is totally satisfied with a 100u output cap. It has 2V drop-out voltage, which is fine. Sound is better than with the LT1086 regulators in my setup. It's less colored and basses are tight. I like this result best.
There's still some headroom to apply some filtering of HF-noise before the regulator, so that's my next effort. Lately I discovered that it can be very rewarding to do so, when improving my discrete output stage. HF-noise isn't enough rejected by the regulators.
I'll keep you posted.
Regards, Nino
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.