XMOS-based Asynchronous USB to I2S interface

[5th element]

That'd be close enough in my book as the 0.47R reference resistor wont be exactly 0.47R, neither will all the relevant connectors/wires be completely accounted for.

I wouldn't mind betting that the temperature at the time, of all the various components, would probably affect the absolute current drawn, as could small variations in the power supply voltage too.

I agree that it isn't perfect, but it's probably safe to say that a 500mA spec'd reg would be more then enough.

I think choosing a reg that could deliver 500mA would be a safe bet.

[m.massimo]

0.47 5%, so a ±5% in the current measurement must be taken into account, at least.
I agree with you that 500mA is a safe bet.

[bismuth]

Hi,
I understand that a regulated PSU is certainly cleaner that the 5V rail from computer USB.
But isn't it overkill to use a shunt reg here ? Isn't the 5V rail only used for feeding onboard three legs regulators in front of the clocks (3V3) and the XMOS chip (3V3,1V8,1V) ?

[barrows]

Originally Posted by bismuth

Hi,
I understand that a regulated PSU is certainly cleaner that the 5V rail from computer USB.
But isn't it overkill to use a shunt reg here ? Isn't the 5V rail only used for feeding onboard three legs regulators in front of the clocks (3V3) and the XMOS chip (3V3,1V8,1V) ?

I have often thought about this myself. I use a different (but similar, XMOS based with lots of onboard regulation/filtering) USB card, and am using a shunt to power it, mostly because that is what I had around to easily build, but it does seem to be overkill considering the onboard IC regulators. Perhaps a simpler supply, using a CRC feeding a 1705 followed by a good quality cap would be just as effective...
Usually shunt regulators are applied directly at the point of load, and not as pre-regulators...
What do others think?

[hochopeper]

A lot of people have reported subjective positive improvements using salas shunt regs. I've not understood the purpose, though looking at the recent Linear Audio - Vol 4 noise graphs , specifically the one available - here, perhaps it is the salas shunt's low noise that is helping here. I recently got shot down in flames (well that's an exaggeration really) for suggesting that depending on the rest of the system (ie feeding an ebay dac kit) that the WaveIO is feeding a salas shunt may be an overly complex and expensive regulator solution to providing a PSU for this device.

I wonder if the change in regulator that people are using also comes with a new independent tx and rectifier? It may be these other changes that are all adding to the perceived improvement.

I ended up deciding on a LT1085 in a pretty simple reg build (build on one half of of OPC's 'the wire' psu board). It is a good but intelligent PSU design by Owen where I've only changed the regulator IC and set it for a fixed voltage output. It has an Nichicon R7 series FPCAP on the output too which can't hurt.

I haven't compared this to anything yet and am currently waiting on a new DAC board to arrive. I have previously only used my WaveIO on usb power when testing with a TPA Opus. So by the time my new DAC comes I'll have no way of doing any meaningful comparison anyway.

[bismuth]

Interesting comparison graph, I think the regulators on the board are LP5900 type with 6,5 uV noise figure. But their PSSR is not very high depending on freq (see datasheet). Maybe it is a good reason for using an 'overkill' shunt pre regulator ?
Or is the 5V used somewhere in the circuit (without additionnal regulator) ?
Aren't noises from clock rail and XMOS chip way higher ? They surely propagate ? Anyone with a good scope to take a look ?

[hochopeper]

The propagation of noise from each part of a circuit should be minimised in Lucian's design, that's why he uses a multilayer PCB

A shunt pre-reg is not the only way to achieve low noise at the input either.

[Lorien]

Originally Posted by bismuth

Interesting comparison graph, I think the regulators on the board are LP5900 type with 6,5 uV noise figure. But their PSSR is not very high depending on freq (see datasheet). Maybe it is a good reason for using an 'overkill' shunt pre regulator ?
Or is the 5V used somewhere in the circuit (without additionnal regulator) ?
Aren't noises from clock rail and XMOS chip way higher ? They surely propagate ? Anyone with a good scope to take a look ?

Yes, interesting graphs! LP5900 has 6.5uV noise figure but it depends in which implementation was measured. I highly doubt that was any XMOS processor around or had common ground with the PC/MAC Even if WaveIO's design has 4 layers, a lot of Vregs and decoupling caps it doesn't mean that there is 6.5uV all over that board! My main goal was to minimize noise as best as possible can but with a 400MHz processor in the center of it and direct GND connection with the host, the goal was (is for newer card) harder to achieve.
Anyway, let's hope that daughter board will have something to say on this as noise will be the first thing to be addressed.
Kind regards,
L

[Magsy]

Please add me to that list for 1x Daughter card

[Lorien]

Done, thank you!
1. hirez69
2. analog_sa
3. m.massimo
4. ozlegend
5. luisbock
6. vitalica
7. palmito
8. gabanyayaya
9. danny_66
10. shoom
11.12. kp93300 X 2
13. lindamar
14. merlin el mago
15. Reo
16. arkadiush1
17. jmltinc
18. zibra
19. tagheuer
20. hochopeper
21. hacker
22. Bemused
23. vanwykca
24. sandor
25. jrling
26. bismuth
27. necplusultra
28. flocchini
29. ichiban
30. edbk
31. Nikon1975
32. pawlus
33. dubai2000
34. henri47
35. presapian
36. Corpius
37. Ryssen
38. carsten.witt
39. Sidiropa
40. Ryelands
41. krautxy
42. 43. 44. 45. Mats x 4
46. ecchecequa
47. 48. lm1no2 x 2
49. Magsy

Regards,
L