Yesterday I connected UTC A20 transformers directly to AK4399 outputs. The result exceeded expectation!!! Sound stage become deeper and wider. I also noticed bass improvement. I did not make A-B test but it is clear without it. I said that result exceeded expectation because I did the same two years ago with DAC based on CS4398 and did not noticed any improvement. I don't have an explanation about that.
Generally it makes sense to use transformers in output stage. I see only advantages:
1. No OPAs
2. No negative feedback in amplifier
3. It is very simple
4. Power supply can be simpler (no need for separate power supply for OPAs.) In my case one transformer less.
5. With a good transformer the DAC sounds very good
6. Galvanic isolation between DAC and pre-amplifier. It is valuable bonus for me because the DAC is directly connected to Nelson Pass F5 amplifier which doesn't have input capacitor. Thus I don't need to worry that in case if for any reason an OPA of the DAC will be broken and DC will go directly to my amplifier.
I will stay with this mod and will think what to do next.
Generally it makes sense to use transformers in output stage. I see only advantages:
1. No OPAs
2. No negative feedback in amplifier
3. It is very simple
4. Power supply can be simpler (no need for separate power supply for OPAs.) In my case one transformer less.
5. With a good transformer the DAC sounds very good
6. Galvanic isolation between DAC and pre-amplifier. It is valuable bonus for me because the DAC is directly connected to Nelson Pass F5 amplifier which doesn't have input capacitor. Thus I don't need to worry that in case if for any reason an OPA of the DAC will be broken and DC will go directly to my amplifier.
I will stay with this mod and will think what to do next.
Last edited:
Sounds to me that you are getting a similar result that I am getting with tubes.
I can do A B comparison because I have got op output and tubed output working at same time.
I just use two sets of interconnects going to the same amp and select different inputs at amp.
Op amp sounds clearer but more digital and shouts a bit.
Tube output fills the room and bigger soundstage still verry clear but with more warmth.
I think it's just as clear it's just the extra bass and depth on tube output.
To me the tube output sounds more natural.
Hi kukynas
Of course you can hear differences with changing opamps and/or decoupling caps.
If you want to stick with opamp output, here some suggestions::
The opamp used in the analog output stage has a combined function. It does some amplifying, filtering and adds the two differential signals from the dac to one output signal.
The filter used is called a 2nd order differential multiple feedback low pass filter. This type of filter is commonly used in cd-players output stages. The values of the resistors and capacitors determine the gain, the cut-off frequency and the filter characteristic. In general it's not easy to design these filters, but you can e.g. use Texas Instruments Filter Pro software.
If you want to change cut-off frequency (here set to 109.5 kHz) or characteristic of the filter (e.g. to Bessel), you must design your own filter; you'll end up with different values for C and R.
If you are content with your design, you can even improve sound, for example, with using matched resistors and capacitors, to improve the filter performance. Or you can replace the capacitors with polystyrene types or polypropylene. Also resistor material alters sound character (metal film or carbon film). As do different types of decoupling capacitors and super regulators in power supply...
You could also experiment with vulejovs' approach. The filter then becomes of 1st order, and has only 20dB roll off / decade. I don't know if this is enough to attenuate HF out of band noise, but you can try. If it works, then simpler is better.
Anyway, I'm glad to help if I can, let me know.
My tube output still comes out louder than op amp output so I just adjust the volume on my amp.
The main thing I notice is the imaging is better on tube output.
The sound seems to come from between and around the speakers more.
The op amp output is a smaller sound coming from the speakers more if that make sense.
I think it's the bass performance of the AK4399 IC. I got better bass and depth also with my discrete output stage.
Valeri100: Congratulations with your results! I have no experience with transformer output, but I can imagine the advantages. One more point to add to the list: Because each DAC is operating in mono mode, channels are combined and so are the outputs. Ability to drive a load is therefore doubled. That could also contribute positively to the results when driving a transformer, I think, especially because only one of the two differential outputs is used with transformer output.
Last edited:
analog psu for AK4113
I'd like to inform you about my latest mod. I separated analog and digital power supply of the AK4113.
The analog supply pin is powering the PLL, which reconstructs the clock signal of the transport (I'm using Squeezebox Touch with Toolbox). A voltage controlled oscillator is used to generate the clock. To improve jitter reduction, the power supply needs to be as clean as possible, in order to keep the VCO stable.
Because digital and analog powersupply is the same, digital noise can enter analog supply pin, despite of the ferrite beads.
The mod is simple.
I bought a new super regulator of Belleson, the super power mini (partsconnexion). It's almost as small as a regular TO220 package, but can only handle 225mA max, and even less with higher Vin-Vout difference. But in this case it's perfect, analog supply draws only 5mA typical.
It's very low noise, only 4-5 micro Volts. It's a 3.3V regulator you need. Analog and Digital voltage may not differ to much, max 300mV. So I had to check the voltage of the digital regulator, but the difference was just fine, 20 mV.
I removed the 47uF ROE capacitor and the ferrite bead. The regulator then could take the place of the cap. The Vin of the regulator comes directly from the big filtering cap of the supply (with a wire). Close to the chip I soldered a decoupling cap on top of the little smd 100nF ceramic.
Sound quality improved a lot. More detail, sound is more fluid and transparent, voices are much better. I even thought of vinyl when listening, it's very nice.
I think, whatever output stage you have, this mod is an improvement. But performance is , to some extend, also dependent on quality of the incoming digital audio signal, I think. You can also try to use a common 3.3V regulator first, but a super regulator has better performance.
Regards, NinoSimona
edit: The same mod can be done with the newest PCB with the AK4118
I'd like to inform you about my latest mod. I separated analog and digital power supply of the AK4113.
The analog supply pin is powering the PLL, which reconstructs the clock signal of the transport (I'm using Squeezebox Touch with Toolbox). A voltage controlled oscillator is used to generate the clock. To improve jitter reduction, the power supply needs to be as clean as possible, in order to keep the VCO stable.
Because digital and analog powersupply is the same, digital noise can enter analog supply pin, despite of the ferrite beads.
The mod is simple.
I bought a new super regulator of Belleson, the super power mini (partsconnexion). It's almost as small as a regular TO220 package, but can only handle 225mA max, and even less with higher Vin-Vout difference. But in this case it's perfect, analog supply draws only 5mA typical.
It's very low noise, only 4-5 micro Volts. It's a 3.3V regulator you need. Analog and Digital voltage may not differ to much, max 300mV. So I had to check the voltage of the digital regulator, but the difference was just fine, 20 mV.
I removed the 47uF ROE capacitor and the ferrite bead. The regulator then could take the place of the cap. The Vin of the regulator comes directly from the big filtering cap of the supply (with a wire). Close to the chip I soldered a decoupling cap on top of the little smd 100nF ceramic.
Sound quality improved a lot. More detail, sound is more fluid and transparent, voices are much better. I even thought of vinyl when listening, it's very nice.
I think, whatever output stage you have, this mod is an improvement. But performance is , to some extend, also dependent on quality of the incoming digital audio signal, I think. You can also try to use a common 3.3V regulator first, but a super regulator has better performance.
Regards, NinoSimona
edit: The same mod can be done with the newest PCB with the AK4118
Attachments
Last edited:
Thank you NinoSimona I will try this upgrade.
I have noticed you are using squeezebox touch.
I have got a few of these on my watch list on eBay.
Do you use it through coax into your DAC.
Hi NinoSimona,
Many thanks for your detailed description. I also planned similar mod but using another shunt regulators from Tentlabs. But first of all i will try XMOS USB module. As far as I understood in this case AK4118 with PLL is not used and clock is not derived from incoming SPDIF signal but it will be received by the AK4399 from the XMOS card which will be connected via I2S interface. Theoretically jitter will depends only instability of the oscillators used in that card.
Perhaps in this case influence of clearness of power supply will lower than in case of useing SPDIF. However reference voltage (on VREFHR and VREFHL pins) must be clean as much as possible.
Many thanks for your detailed description. I also planned similar mod but using another shunt regulators from Tentlabs. But first of all i will try XMOS USB module. As far as I understood in this case AK4118 with PLL is not used and clock is not derived from incoming SPDIF signal but it will be received by the AK4399 from the XMOS card which will be connected via I2S interface. Theoretically jitter will depends only instability of the oscillators used in that card.
Perhaps in this case influence of clearness of power supply will lower than in case of useing SPDIF. However reference voltage (on VREFHR and VREFHL pins) must be clean as much as possible.
Dave: Yes, I use the coax connection Valeri100: I see. Unfortunately the AK4118 as well as the AK4113 does only receive spdif. So in fact the signal received from the XMOS pcb is spdif. That is a little devious, but that's how it works imho. Better would be a direct I2s from the XMOS to the AK4399, I agree. So optimizing the receiver performance is also worthwhile when using usb input. But you can bypass the AK4118, but then you can't use spdif anymore. btw. the crystal nearby the AK4118 is used to sense sampling freq.
Last edited:
Hi NonSimona thank you for your expllination.
I also considered getting this dac kit that also uses AK4399 but uses a CS8422 .
Its also on ebay item number 121282422522.
The advantage that I could see of this dac pcb is that its 32/192 but its not dual mono.
I am not sure but I think it might be this dac pcb that I got confused with that controls volume but im not 100% sure on that.
I chase this dac because its dual mono and a lot more people seem to be using this dac pcb.
I am not experianced with pcb layouts znd logic what do you think of this dac pcb ?
Thank you again.
Hi NinoSimona I also changed the blue oscon cap near the optical input for a oscon sepc cap same value.
I cannot tel you if it made any difference because I do e this when putting together PCB.
I just had one in my draw so used it because I read of people changing normal oscons for sepc oscon with good results.
Hi NinoSimona,
As far as I understand, with the newest PCB it is necessary to remove LT1963 and put the super regulator of Belleson instead of it? Am I right? LT1963 has noise 40 micro Volts which is almost 10 times higher than the Belleson. Thus it might be noticeable.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.