Good. People often rationalize intuitions rather than experiment to find out for sure. Of course, experiments take time and sometimes money. One has to pick and choose what to investigate. In the case of dacs, it helps a lot to have good reference dac to listen to, just like it helps to be able to do some measurements with test equipment.
Regarding distance from RPi, a more basic experiment might be to feed the dac from a USB to I2S board. That can take RPi completely out of the equation. If the dac sounds better or merely different than when connected to RPi, that might be taken as a hint that there a problem somewhere that should be investigated.
This is good suggestion. May I know what usb to i2s which has good galvanic isolation to use? Is it Amanero board?
The whole point of Ian's design is to have the I2S line and PI isolated from the DAC.
If have used USB to I2S in the past and based on that experience it turned out that designs taking the USB out of the equation did sound better in my case.
You could use the idustrie designed PI which has a slightly better I2S output ...
Buy a Compute Module Development Kit – Raspberry Pi
If have used USB to I2S in the past and based on that experience it turned out that designs taking the USB out of the equation did sound better in my case.
You could use the idustrie designed PI which has a slightly better I2S output ...
Buy a Compute Module Development Kit – Raspberry Pi
Hopefully Ian one day will make an ak4118 input board (usb,coax,toslink) of RPI footprint so that his reclocker fifopi could be stack directly on it. Something like this on ebay: "AK4118 digital receiver board SPDIF input switch IIS with IIS support XMOS OLED". Then Rpi could be taken out of equation 
he has done that already, but this will be a major step back in sound quality :
Asynchronous I2S FIFO project, an ultimate weapon to fight the jitter
you can just take the I2S output from the AK4118 go from there if you want to go that route
Asynchronous I2S FIFO project, an ultimate weapon to fight the jitter
you can just take the I2S output from the AK4118 go from there if you want to go that route
Since I have been posting results and findings in this thread so far, thought I would mention a problem with clocks and pins. The clock sockets on Iancanada and TP clock boards tend to become permanently deformed by the largest size pins used on whatever clock adapter boards one has. TP clock module boards come with very slender pins, maybe about equal to the smallest or perhaps a bit smaller in diameter than any of the pins that ship with Ian's clock adapter boards. Of the Ian clock boards I have there are two kinds of pins that have been received. The silver ones measure about .021" diameter. The gold ones have a thin and a thicker end. Those measure about .0185" and .024" respectively. After a few insertions and removals of any size pins the sockets become noticeably looser. Probably a good idea to help minimize clock jittter if the same size pins are used on all clock adapter boards. Loose pin connections that are used for clock signals have been observed to increase jitter in some cases, IME.
Last edited:
I noticed a similar problem with the Opamp module pins.
OPA 1612 is the way forward, Bruno Putzeys mentioned it is his favoured Opamp at Munich to someone..
I tried the OPA 1622 and found them to be a PITA to work manually without a solder paste stencil.
I think the thermal pad is so large compared to the pads that it causes the package to lift
and move if excess solder paste is applied to it.
OPA 1612 in a SOIC 8 package with an adapter board is easier, the SQ after the upgrade is immense, many of my reservations over the sound have vanished and the sound is the best I have yet heard in my room.
Bruno Putzeys Balanced preamp, Tom Christiansen Mod 86 Parallel bridged monoblocks (is it a Canada love fest ? ) into naked Quad 989 ESL's and a couple of 12 " dipole subs for some foundation.
Sounding great, looking forward to the NDK clocks and further tweaks.
OPA 1612 is the way forward, Bruno Putzeys mentioned it is his favoured Opamp at Munich to someone..
I tried the OPA 1622 and found them to be a PITA to work manually without a solder paste stencil.
I think the thermal pad is so large compared to the pads that it causes the package to lift
and move if excess solder paste is applied to it.
OPA 1612 in a SOIC 8 package with an adapter board is easier, the SQ after the upgrade is immense, many of my reservations over the sound have vanished and the sound is the best I have yet heard in my room.
Bruno Putzeys Balanced preamp, Tom Christiansen Mod 86 Parallel bridged monoblocks (is it a Canada love fest ? ) into naked Quad 989 ESL's and a couple of 12 " dipole subs for some foundation.
Sounding great, looking forward to the NDK clocks and further tweaks.
For anyone interested in possibly trying film caps for output stage +-15v power filtering to get rid of any remaining opamp graininess, good news that some low priced ones have been found to work well: https://www.diyaudio.com/forums/digital-line-level/314935-es9038q2m-board-448.html#post5793282
Hi Mark,
I received my new TP 3.3v regulators and solederd them yesterday
It already makes a real audible improvement.
As usual, I have a couple of stupid questions for you
I powered them using the dac card using 5v Mpaudio PSU.
I saw that you uses higher voltage (5.7v).
Could you explain me why higher voltage is better?
I saw that you uses 33ohms resistors for avcc and vcca.
Could you explain me what is the use of these resistors and how you chose the ohm value?
Thank you for you kindness and your time.
And the loose pin connections sometimes make bad contact so the clock does not lock.
Good questions, actually.
The higher voltage was an experiment that was done for a couple of reasons. First, I was comparing two regulators one of which was not an LDO. The NewCLassD UWB2 is the non-LDO, and it has a higher dropout voltage than the TP Trident (which uses ADM7150). I wanted to make sure the UWB2 was fully biased on to begin with so I put the socket connector for the AVCC regulator on an adjustable power supply. The second reason is because I have a new friend here locally that has been designing high end analog for 30-years to make his living. He told me that voltage regulators often make analog circuits sound better if they have more than the minimum dropout voltage across them. He said sometimes as much as 10 volts across the regulator is best, but that one has to find the best voltage for each regulator and load.
The resistor is a similar thing. My friend told me it often sounds better if some, but not too much of the power supply current is dumped into a resistor, although he didn't know why. With these 3.3v regulators, a 33-ohm resistor will draw 100ma, which is more than the dual ES9038Q2M AVCC pins will draw. It should be enough to get the regulators working somewhere in the middle of their range. Also, in this case I suspect it may have helped by making the source impedance of the power supply look lower to the dac chips. They see the voltage regulator and the resistor as being in parallel so the power supply impedance they see with the resistor (up to pretty high frequencies) has to be less than 33-ohms and it stays mostly resistive. Voltage regulators themselves usually tend to start looking inductive at higher frequencies and as seen by the load (the dac chips).
The reason for not raising the adjustable power supply voltage too high in this case was to avoid heating put the voltage regulators too much. The 33-ohm resistors dissipates around one third of a watt, so I used and 1/2-watt resistor. The regulator has current passing through it too, which is equal to the resistor current plus the dac chip AVCC current. That current is multiplied by the voltage drop across the regulator to calculate how much power the regulator is dissipating. That power causes heating. Even though the regulators are operating within their specifications, they don't have heat sinks and I wanted to play it safe and not get them too hot.
The actual voltage I started with was 5.7v because 5v wasn't enough to turn on the 3.3v UBW2. The best sounding voltage I used was 6.7v. Both regulators sounded best with the higher input voltage. The best overall sound quality was with both the resistor and with the higher voltage. Therefore, I would recommend that TP Trident ADM7150 be run maybe at around 7v input. 8v might be okay too if heating is not an issue, but I didn't check into that (NOTE: I was told some older Tridents used ADM7154 which is limited to 5v input, so better check the number on the regulators you have). I also recommend to use the 33-ohm resistor or some similar value, and 1/2-watt rating or higher. All the above applies to AVCC only, don't know if it makes any difference for VCCA and DVCC.
Incidentally, people with Buffalo dacs using the more recently made Trident-AVCC regulators may also have ADM7150 chips on them. They might sound better with 7v too, but someone would need to try it. Just don't try it with ADM7154.
Last edited:
Usually, the Power Supply Rejection Ratio (PSRR) gets better, when there is a higher dropout voltage on the LDO. Check for example the datasheet of ADM7150, there in Figure 28 it can be easily seen. ==> This might have as well audible effects, as the impurities of the input supply will be better rejected by the regulator, in case there is higher Vdropout.
As the LDO is a regulator, and if there is quasi "no-load", then the output can be "jumpy": there could be too small resolution of the control over the output level. (Too much juice is pumped from input to output, even with the slightest change in control). If you increase the load, then the control has more headroom for resolution, therefore it can better regulate and calm down the output voltage.
In other terms, you can see more noise on the supply without load, and less noise, when there is load.
Unlikely to the be the cause in this case since the ADM7150 was fed from a well regulated linear power supply. There was very little if anything to reject via PSRR.
True, but many people have become overly focused on noise as the only factor that really matters. In many cases noise is not what matters most. There are several other attributes of voltage regulators that can be measured, so if we want to theorize about possible cause and effect, maybe good to consider those other attributes as possibly important factors too.
Hi to all
Yesterday I mounted all the boards the I recveived (FifoPi + dual ES9038 DAC + ESS Controller + Standard I/V)
I have 2 different problems:
1) If i mount the ESS controller on the Non isolated GPIO of the FifoPi the controller doesn't work. At all. The LED of the controller remain completed turned off (during the startup of the raspi, on the led of controller i can see some little light (some point flash) but when the startup procedure is finished the LED becames black). Instead if I mount the controller directly on the DAC (on top of the FifoPi) the controller seems to work; The light is on on the led and also if I stream music the controller show the right information on the audio stream. COULD BE A PROBLEM ON THE FIFOPI ? ANY SUGGESTIONS ?
2) From the I/V Board I'm not able to ear anything. Also in the case that the streamer seems to work properly, the ESS Contraller (mounted on the DAC directly) show the audio stream info, all the led on fifopi and DAC became green during the streaming, from the I/V Board doesnt exit enithing. ANY SUGGESTION ?
I'm using Raspberri Pi 3 with DietPi with MPD client and also Jriver Istalled
I'm using a PC power supply (waiting the LifePO4) to prower the FifoPI with 5V and 3.3 V , To power the DAC with 3.3V and to power the I/V board with +12V 0 -12V
Yesterday I mounted all the boards the I recveived (FifoPi + dual ES9038 DAC + ESS Controller + Standard I/V)
I have 2 different problems:
1) If i mount the ESS controller on the Non isolated GPIO of the FifoPi the controller doesn't work. At all. The LED of the controller remain completed turned off (during the startup of the raspi, on the led of controller i can see some little light (some point flash) but when the startup procedure is finished the LED becames black). Instead if I mount the controller directly on the DAC (on top of the FifoPi) the controller seems to work; The light is on on the led and also if I stream music the controller show the right information on the audio stream. COULD BE A PROBLEM ON THE FIFOPI ? ANY SUGGESTIONS ?
2) From the I/V Board I'm not able to ear anything. Also in the case that the streamer seems to work properly, the ESS Contraller (mounted on the DAC directly) show the audio stream info, all the led on fifopi and DAC became green during the streaming, from the I/V Board doesnt exit enithing. ANY SUGGESTION ?
I'm using Raspberri Pi 3 with DietPi with MPD client and also Jriver Istalled
I'm using a PC power supply (waiting the LifePO4) to prower the FifoPI with 5V and 3.3 V , To power the DAC with 3.3V and to power the I/V board with +12V 0 -12V
Thank you Mark.
It is a very clear explanation.
I already ordered 33ohms 1w resistors.
I will solder them soon and I'll make some test using higher voltage after checking if the TP are ADM7154 or ADM7150.