If you can afford it it might be good to have a better supply to swap in now and then and see if cleaner is better for a particular circuit, layout, implementation, etc. If you have nothing to compare with and no instruments to measure with, it can be very difficult or impossible to figure out how well things are working. It can especially be a problem when going for very high SQ and very low distortion. Even measuring harmonic distortion at 1kHz can only tell you so much.
Both the 7815 and 7915 are TO-220 can with good heat sink. The current capacity should have plenty to spare. From what you said, I should try to replace the AVCC regulator.
The I-V buffer also yields more benefit if I can find one ready to apply PC card. I cannot make my own circuit card now.
The ultra low noise regulator is a $50 device which I cannot justify based on your and Markw4's comment.
The DAC card that I have is the one discussed by zek and DRONE7 in another thread.
ES9038 Q2M DAC DSD Decoder Support IIS DSD 384KHz Coaxial Fiber DOP | eBay
Yes, it certainly appears so. In reading more on the ESS Sabre line they continually hint at the importance of AVCC and having that as rock solid as possible being used as a reference one can now see why. When that is complete then the rest will be more incremental improvements.
Now that I am starting to understand that AVCC is actually part of the output, it is now obvious why it is extremely important and needs to be perfect. This aspect changes the game and I experienced it. Hence why ESS refers to this aspect as the analog DAC and they call AVCC the DAC reference level.
Now the question is whether it is better to use a 797 or whether I can parallel a dual 49720. I do not have any extra 797s lying around.
What if we think about it a little? Each channel has two outputs that sit at Vref when there is no sound output. The IV stages use 820 ohm resistors, but let's say they were only 500 ohms. Also, let's say we played a full volume square wave. What happens? When the - output goes low, the opamp for that output has to source current through the DAC to ground. If the opamp output goes from Vref up to 5 volts more than that, then the current flowing through the resistor into the - output and to ground would be 10ma. Similarly, the + output would go high and the opamp for that output would have to go low to sink the 10ma coming from AVCC. No matter the output frequency one side or the other would be sinking 10ma continuously. However music is not full power square waves. And the resistors are not 500 ohm.
In addition to that, even in the absence of the square wave signal, there is probably some quiescent current flowing from AVCC to ground all the time. Can't be much though or it would probably run down the batteries in cell phones.
To speculate that AVCC current is going to be more than LME49720 can handle seems improbable.
In addition to that, even in the absence of the square wave signal, there is probably some quiescent current flowing from AVCC to ground all the time. Can't be much though or it would probably run down the batteries in cell phones.
To speculate that AVCC current is going to be more than LME49720 can handle seems improbable.
This makes sense when viewed from that context but why does increasing the sampling rate increase AVCC current requirements as per ESS?
I am looking carefully at the spreadsheet
ESS Technology :: Downloads
At 0 Mhz, the current consumption kind of follows a linear pattern.
Then as the frequency is increased, There is a non linear pattern even if you remove the static consumption at 0 Mhz.
One thing, this is way beyond me as I have no idea what goes on in those DACS or the circuits used but I'm intrigued as to what is happening with AVCC current and its implications and effects.
ESS Technology :: Downloads
At 0 Mhz, the current consumption kind of follows a linear pattern.
Then as the frequency is increased, There is a non linear pattern even if you remove the static consumption at 0 Mhz.
One thing, this is way beyond me as I have no idea what goes on in those DACS or the circuits used but I'm intrigued as to what is happening with AVCC current and its implications and effects.
Also looked for any evidence of voltage modulation on AVCC power using 192kHz sample rate and FS sine wave at 15kHz and then at 20Hz. At the vertial limit of the scope it didn't show anything below maybe 500 microvolts p-p, or possibly a little less. That was probably virtually all scope and ground noise.
To go further we would someone need to volunteer to add a current shunt resistor in series with AVCC. A DVM might give something reasonable if checked in both AC and DC millivolt modes. Have to look at the manual for the meter, some can do AC+DC RMS millivolts on one setting over much of the audio band. Some also have an optional 4-1/2 digit mode.
Or, differential scope probing might be best. Various ways to do that.
To go further we would someone need to volunteer to add a current shunt resistor in series with AVCC. A DVM might give something reasonable if checked in both AC and DC millivolt modes. Have to look at the manual for the meter, some can do AC+DC RMS millivolts on one setting over much of the audio band. Some also have an optional 4-1/2 digit mode.
Or, differential scope probing might be best. Various ways to do that.
I'm now looking at volume control solutions and it suddenly became expensive with balanced stepped volume control! The usual solutions have tons of pcb traces and a lot of electronic components. If we are aiming for 120db+ signal to noise ratio I dont think that will happen with more components added.
I therefore wonder if I can manage with the software included volume control. Have any of you actually heard any differences at different volumes?
I therefore wonder if I can manage with the software included volume control. Have any of you actually heard any differences at different volumes?
Definitely best to use the software volume control as the DAC does some tricks inside to optimize S/N at different volumes. The volume goes up as the pot moves CW towards ground. CCW on the pot should be connected to the + pin on the board. If volume control seems to be all at one end an easy fix is to put a resistor in series with the + lead. Roughly the same resistance as the pot works okay for me.
Last edited:
Also, you can try a pot with type "C" (anti-log) characteristic. At least for ver. 1.04 and 1.06 boards it gives you smooth and comfortable curve.
C taper should work okay. However, if you look at the taper curves for most pots it turns out it costs too much to actually make them log or anti-log so they typically have two different linear resistance sections in series as a way of very roughly approximating a smooth curve. There is an abrupt change in rate of change of volume that is audible when the wiper crosses over where they meet at some point in the shaft rotation. Sometimes I find it annoying if I happen to be operating the pot around that point, don't know about anyone else.
Anyway, I didn't have an anti-log in stock so I just used a linear that I had.
Anyway, I didn't have an anti-log in stock so I just used a linear that I had.
Heard back from app support re 'master mode I2S.' Might be worth another go, but if it didn't just work and I needed to look into why a bit deeper I would probably need a fairly fast logic analyzer on hand. Only have a slow one now, good enough for I2C but not high-speed I2S/PCM. That being the case my inclination is to leave it for now and work on getting ready for some distortion testing and maybe trying out distortion compensation.
Regarding distortion compensation, so far I am not aware of anybody exposing that capability to non-NDA agreement holders. If I find it helps a lot I would probably have to see what ESS might say about potentially making it end user accessible somehow or other.
Regarding distortion compensation, so far I am not aware of anybody exposing that capability to non-NDA agreement holders. If I find it helps a lot I would probably have to see what ESS might say about potentially making it end user accessible somehow or other.
I finally got around to getting some perfboard and am laying out two channels holding 3 ICS each. Each channel will have a dual 49720 for AVCC and the other two for IV section.
Each channel will have its own Sulzer LowZ low noise reg and transformer.
(This cleans the parts box out) Believe it or not the power supplies was removed from a modded POOGED Philips CDB-482 from the late 80s. Recycled for use 30years later and the Power supply is still relevant!
The one thing I am contemplating is whether I should lower the resistance for the filter and uP the capacitance. I'm thinking thermal noise on AVCC due to 10K. Any ideas?
Each channel will have its own Sulzer LowZ low noise reg and transformer.
(This cleans the parts box out) Believe it or not the power supplies was removed from a modded POOGED Philips CDB-482 from the late 80s. Recycled for use 30years later and the Power supply is still relevant!
The one thing I am contemplating is whether I should lower the resistance for the filter and uP the capacitance. I'm thinking thermal noise on AVCC due to 10K. Any ideas?
The current is not determined by the value of the I/V resistor but by the internal resistance of the DAC. Which should be not all that different btw.