Asynchronous I2S FIFO project, an ultimate weapon to fight the jitter

[nicoch58]

there are a lot of revision ,if you want info better ask direct to Salas...

[1audio]

I'm not sure what you are trying to contribute. The Salas regulator is a good design for a number of tasks. It is not quiet enough to get the best performance from a low phase noise oscillator. The voltage stability over time and temperature can also shift the fundamental frequency of a precision oscillator, possibly enough to affect some applications.

The simple regulator above does not have the lowest output impedance and a number of other limitations but it has very high line noise rejection and a very low noise floor both important for getting the most performance from a very high quality crystal oscillator and it does it using 3 active components. Depending on its application more parts can get small improvements but at some point the improvements do not propagate through to the regulated circuit.

[qusp]

Demian, i think i have the right transistors here to try your circuit and as i have said, i intend to. but i'm wondering can you recommend any other small signal transistors in case? its been a while since i used the 2n3904 etc and they are probably well buried in my parts bin. I know i have bc860/50. also i have the 431 in soic8, but i have plenty of other low noise references too, i will toy with some others. is the schematic on the previous pages the latest or should i search the widget thread?

[1audio]

I used the 2N3904 because its in my bin and has lower noise than a 2N4401 but not by much. Almost any small signal transistor will work with the best possibly 3 dB better than the worst. The low impedance of the reference (470 uF) means that the current noise is not significant. The TL431 is used more like a servo to correct for the changes in Vbe due to load and temperature, it is not the reference that the system is correcting to in the audio band.

The schematic is current I believe. There is some potential for lower output Z with a higher bias current in Q1, but it won't affect a crystal oscillator and it will add more power consumption. Let us know how it goes.

[qusp]

thanks for the info, is there any advantage to be gained from making the 470u as low impedance as possible? or any old polymer ok? i take it the 1u on the output is film? so i take it an LM317 could be used in the same way to servo the vbe as the 431? not that I would, just thinking it through, I tend to learn more from doing. after knocking this up on perf i like the idea of a really tiny version using the sot23 431, BC850/60 and smd everything. i figure it could be made pretty small; allowing more ideal placement

[TheShaman]

It'd be nice to see small PCBs for Demian's clock regs.
Group buy material, IMHO!

[qusp]

well that would be up to Demian and is not my intention; i'm sick to death of group buys tbh and i'm still not finished the ones i've ended up running more than i'd hoped

[1audio]

The cap isn't too critical. Lower ESR won't make a lot of difference since the internal noise of Q1 will be much higher. The output cap can be whatever as well if you are driving an XO. A small bypass ceramic would also be required for an XO.

[iancanada]

Originally Posted by TheShaman

It'd be nice to see small PCBs for Demian's clock regs.
Group buy material, IMHO!

That's good idea.

Originally Posted by 1audio

The cap isn't too critical. Lower ESR won't make a lot of difference since the internal noise of Q1 will be much higher. The output cap can be whatever as well if you are driving an XO. A small bypass ceramic would also be required for an XO.

Hi Demian,

I did some simulation to your reg and shown pretty good result. I'm thinking about design a prototype PCB for real evaluation. But it should be under your promise.

Ian

[1audio]

Attached is a zip of he expresspcb file I used to test the design concepts. It incorporates an input regulator to handle short circuit protection etc. the regulator circuit supporting two dip oscillators and the transformer coupled output to prevent the system grounds from undoing all the work.

I won't have a chance to create a schematic for it but by careful inspection that should not be necessary. Its offered as a starting point, not as an end in itself.

The basic regulator was first published by someone else (I have lost the reference) and my addition is the TL431 servo.

Don't trust this, study and verify (I may well have made mistakes) and build on it.