[joenetic]

This is a LVDS transmitter, you have to connect this to the FIN1002 which is a LVDS receiver, assuming the chip is even a FIN1002.
This cannot replace the FIN1002

This is what SinePi from IanCanada does, just buy the thing from IanCanada and be done with it.

You knew the circuit of SinePi? It also use LTC6754?

 

[b0bb]

You knew the circuit of SinePi? It also use LTC6754?

IanCanada is on diyaudio, reach out to him there to see about getting a schematic

LTC6754 needing an additional LVDS receiver to convert the LVDS output to LVTTL is too complicated for the SU2.
The only use of the app note is to help LTC sell more of the LTC6754

 

[joenetic]

IanCanada is on diyaudio, reach out to him there to see about getting a schematic

LTC6754 needing an additional LVDS receiver to convert the LVDS output to LVTTL is too complicated for the SU2.
The only use of the app note is to help LTC sell more of the LTC6754

Thanks!

 

[b0bb]

Below is a simplified diagram of how a LVDS receiver can be used to clean up semi-sinusodial signals.
SU2 appears to be using a slightly modified circuit with additional filtering for noise

The voltage at the junction of R1 and R2 is the trigger threshold.
If this voltage changes the duty cycle of the output changes.
This is reflected as a phase change in the complex or s-domain.
Integrated over time, this forms the basis of the phase noise number.

Variations in supply voltage will thus show up as a component of phase noise at the output.
SU2 did not take any additional precautions beyond passive filtering.
I was expecting something better like its own dedicated low noise LDO for this function.

This converter was added as a convenience feature to allow external 10MHz input.
Here is the output of the circuit.
Note the rise time are pretty quick as the input is mostly squarewave in the example
Situation is less ideal with SU2
-FIN1002 specs rise time at 1ns, 10MHz sinewave risetime is 10ns.
-FIN1002 is not operating with optimal input

Additional info here
https://www.ti.com/lit/an/slyt180/slyt180.pdf?ts=1682727261775

 

[b0bb]

Yes, to do so may fix it, but everything settled, no more fun! Really want to know if I can make it better, that's why I bought this used SU2 with low price to mod. I amy also try to put Crystek back to SU2 to compare the sound with external clock, as I remember now 10M clock the sound is better.

There are many other things that can be done to improve SU2, the external 10M input is not one of them
1. NVE GMR isolators, no noisy high frequency carrier wave.
2. LTC4495 + Supercap
3. TI LDOs
4. Polymer bypass caps
5. Additional bypass and filtering around the USB phy.

 

[joenetic]

There are many other things that can be done to improve SU2, the external 10M input is not one of them
1. NVE GMR isolators, no noisy high frequency carrier wave.
2. LTC4495 + Supercap
3. TI LDOs
4. Polymer bypass caps
5. Additional bypass and filtering around the USB phy.

3 and 4 were done, also replaced two bigger 3.3V regulators with LT3045, I found my 5V LPS has about 90uV noise, so will add one more LT3045 for 5V power, PLL IC and circuit are powered by 5V.
I tested supercap powered SU2, but sound like no difference with LPS power.
About NVE isolator, anyone tried it and found significant improvement in sound quality?

 

[joenetic]

Below is a simplified diagram of how a LVDS receiver can be used to clean up semi-sinusodial signals.
SU2 appears to be using a slightly modified circuit with additional filtering for noise

The voltage at the junction of R1 and R2 is the trigger threshold.
If this voltage changes the duty cycle of the output changes.
This is reflected as a phase change in the complex or s-domain.
Integrated over time, this forms the basis of the phase noise number.

Variations in supply voltage will thus show up as a component of phase noise at the output.
SU2 did not take any additional precautions beyond passive filtering.
I was expecting something better like its own dedicated low noise LDO for this function.

This converter was added as a convenience feature to allow external 10MHz input.
Here is the output of the circuit.
Note the rise time are pretty quick as the input is mostly squarewave in the example
Situation is less ideal with SU2
-FIN1002 specs rise time at 1ns, 10MHz sinewave risetime is 10ns.
-FIN1002 is not operating with optimal input


Additional info here
https://www.ti.com/lit/an/slyt180/slyt180.pdf?ts=1682727261775

Thanks b0bb to provide more information to this solution, the HDMI interface of SU2 is also buffer with FIN1002, this is a normal application of LVDS receiver. To use it in sin to square purpose, I found companies like TI, LT.. have LVDS solutions, especially LTC6754, will search more information if replace FIN1002 with it. Since I have no equipment to measure the jitter of the LVDS output, I can only benchmark the result by listening.

 

[b0bb]

I tested supercap powered SU2, but sound like no difference with LPS power.
About NVE isolator, anyone tried it and found significant improvement in sound quality?

It is not the supercap, but the controller that makes the difference. Did you actually use the LTC4425 controller, without it you may not see any improvement.
Generic supercaps with rudimentary controllers like the ones you get on Aliexpress can actually make things worse.

To see the full benefit , run the SU2 at 768k or higher, current consumption increases 4x relative to 44.1k, the LTC4425 come into its own at this point.
Post pictures of what you did with the supercap, implementation details matter.

Some people do not like the edgy sound of the stock Singxers, NVE addresses that

 

[joenetic]

It is not the supercap, but the controller that makes the difference. Did you actually use the LTC4425 controller, without it you may not see any improvement.
Generic supercaps with rudimentary controllers like the ones you get on Aliexpress can actually make things worse.

To see the full benefit , run the SU2 at 768k or higher, current consumption increases 4x relative to 44.1k, the LTC4425 come into its own at this point.
Post pictures of whant you did with the supercap, implementation details matter.

Some people do not like the edgy sound of the stock Singxers, NVE addresses that

I see, how the controller make power better? I measured the power from supercap, when it is discharging, noise is higher than LPS, I cannot take photo to you because I removed the supercap already, I meanly listen to 44.1khz or 48khz from files or Tidal, my Bifrost 2 accept USB or coaxial, I am using coaxial the sampling rate is limited to 192khz, I also no PC to up sample the source data. I don't find SU2 sound edgy, it is warm sound I like it over Digione which is much edgy or dry.

 

[b0bb]

I see, how the controller make power better? I measured the power from supercap, when it is discharging, noise is higher than LPS, I cannot take photo to you because I removed the supercap already, I meanly listen to 44.1khz or 48khz from files or Tidal, my Bifrost 2 accept USB or coaxial, I am using coaxial the sampling rate is limited to 192khz, I also no PC to up sample the source data. I don't find SU2 sound edgy, it is warm sound I like it over Digione which is much edgy or dry.

LTC4425 is a linear controller not a switched mode controller, the extra noise when discharging is a symptom a switched mode controller.
A switched mode controller charges the cap in bursts hence the noise, I think you used the wrong supercap controller.
These are the Aliexpress specials I am referring to.

A linear controller charges the cap constantly controlling the current into the cap, there are very few controllers of this type LTC4425 is one.
You need to use the exact LTC4425 setup I used, an approximation will not give you the desired result.

 

[joenetic]

LTC4425 is a linear controller not a switched mode controller, the extra noise when discharging is a symptom a switched mode controller.
A switched mode controller charges the cap in bursts hence the noise, I think you used the wrong supercap controller.
These are the Aliexpress specials I am referring to.

A linear controller charges the cap constantly controlling the current into the cap, there are very few controllers of this type LTC4425 is one.
You need to use the exact LTC4425 setup I used, an approximation will not give you the desired result.

Actually I didn't use any controller with the supercap, I used a small capacitance supercap directly power SU2

 

[b0bb]

Actually I didn't use any controller with the supercap, I used a small capacitance supercap directly power SU2

That will cause a lot of problems, the uncontrolled charging current spikes will cause instability in the SU2's power rails, you must have a controller to keep an eye on things. Large current spikes also cause a lot of EMI and RF noise, I think you saw some of that.

Details and how you implement them really matter here, supercaps store a lot of energy which can cause damage under fault conditions if not managed properly.

 

[b0bb]

@joenetic , take a look at how I did my supercap, the large groundplane around the yellow supercaps is to keep the setup as quiet as possible.
The board could have been half the size if not.

 

[joenetic]

Yes, I understood supercap's potential dangerous, so I just fully charged it with 5V in external power supply, then plug the supercap into SU2 to test the sound quality, just want to know how it sound, before next step, but I found it has similar or same sound quality to my LPS, I also surprised by the result, I had big expectation on it but disappointed.

 

[b0bb]

Yes, I understood supercap's potential dangerous, so I just fully charged it with 5V in external power supply, then plug the supercap into SU2 to test the sound quality, just want to know how it sound, before next step, but I found it has similar or same sound quality to my LPS, I also surprised by the result, I had big expectation on it but disappointed.

Using it like a fancy battery is not going to work well, the ESR of a supercap is quite large for its capacity, worse than a battery, much worse than a conventional PSU.
Having the controller dynamically manage power allows the supercap to do its real work, buffer large current spikes when digital logic is running under load.
This keeps the voltage on the supercap constant, LTC4425 keeps it varying by no more than 100mV as I mentioned to you before.