PDA

View Full Version : Super regulator measurements



seanadams
2005-07-21, 12:05
I have been testing a modified Squeezebox2 using Andrew's super regulators - there is a definite improvement at the RCA outputs when a super reg is used to power the DAC chip:

http://www.seanadams.com/sb2_super_regs/

Also trying some hacks to the oscillator and s/pdif circuitry - I've been fiddling around with some jitter measurements and will have more data on that soon. Replacing the external PS with a linear supply, although it doesn't appear to have much impact on the DAC, *does* reduce jitter measuably (from about 3.5ns to 1ns pk-pk). This simple THD+N test is not expected to reveal any effects of jitter at the DAC output though.

Triode
2005-07-21, 12:53
Sean,

Interesting - I take this as official sanction to discuss SB2 mods on this forum...

I'm most interested in spdif output mods and would be interested in what you have been considering. Experience moding my CD transport has lead me to the conclusion that jitter does matter and the psu for at least the clock is critical... [at least into a DAC with minimal jitter reduction]

Quick question: other than not playing 48K sampled stuff, would removing the 12MHz clock cause any issues [does any logic rely on it oscillating?]

On my CD transport I also found benefit reclocking the spdif line with a flipflop just before the output buffer. Is the output of your Xilinx clocked anyway to avoid any benefit of this?

Adrian

seanadams
2005-07-21, 13:03
I tried a whole bunch of things including

- disabling 12.2880 osc (yes it's okay to do this - 32/48Khz will just not work)
- reclocking s/pdif with a discrete flip-flop
- dedicating a linear regulator to the oscillator
- moving the s/pdif signal off of the 74hcu04
- using external 5v linear supply

Will have more detailed measurements later, but basically the easiest tweak (changing PS) seems to account for the vast majority of the improvement. below 1ns we are approaching the lower bound of what I can measure - SB2's output actually measures lower jitter than the reference signal generated by the analyzer.

Michel Fombellida
2005-07-21, 13:44
Hi all,

My tweaked SB2 is on its way back to me from the tweaker.

He added an external linear PS, as far as I know at the moment (but I'll know more as soon as I get it) this PS made with high quality components generates a new 5V to replace the orginal one, plus another 5V and a 9V all generated and regulated independantly. They feed directly some different components inside the SB2 (they have been decoupled from the original source).

Then he also added a transformer on the S/PDIF output (Lundhal transformer) and 2 word clock inputs (one at 11.2896Mhz and one at 12.288Mhz) to be fed from an external clock. They replace completely the internal crystals.

More info when I have it back. As I have a second SB2 I should be able to compare the two.

Michel

Yannzola
2005-07-21, 17:21
Sean,
Did you try powering the SB with batteries? A la Vinnie Rossie? I'd be curious to know how this compares jitter-wise to Andy's super regs PSU mod.

y.

Vinnie R.
2005-07-21, 20:13
Sean,
Did you try powering the SB with batteries? A la Vinnie Rossie? I'd be curious to know how this compares jitter-wise to Andy's super regs PSU mod.

y.

Yes, I am using a 12V SLA battery. A LT1086CT-5 5V regulator is used for the 5V that feeds the SB2's main voltage input. I am also feeding the 12V off the battery directly into the SB2, as a replacement for the 14V internal switching supply that feeds the critical 5V and 3.3V internal regs that power the DAC, SPDIF, etc. I have plenty to post about all of this soon, but for now let me say that the improvement is amazing! I have modded the analog output stage as well, and the results are telling me that an external dac is not needed now! :-)

Slim Devices certainly has a nice product here (at an affordable price), but it can be taken to a whole new level of performance with mods, which I am planning to offer soon. Now, back to listening for me...

seanadams
2005-07-22, 10:18
Sean,
Did you try powering the SB with batteries? A la Vinnie Rossie? I'd be curious to know how this compares jitter-wise to Andy's super regs PSU mod.

y.

I tried a battery on the internal supply - did not measure any change for this test.

Tom Alves
2005-07-22, 13:51
Did you try powering the SB with batteries?The following might be a wild guess but there again it might be rellevant. Yannzola, you know my view that I prefer my SB2 to my old CDS2. I had the whole of my mains supply to the hi-fi upgraded by RKR himself. (For those who don't know of Roy and his mains wizardry he can be found on the Pink Fish forum) I have long pondered whether my findings have been as they have becasue of the levelling affect of decent power supply. I was (one of) the first to try the SB1 with one of Andy Weekes PSUs and am convinced that a clean steady supply of current is essential to get the most from the SB especially if you use the internal DAC. So maybe a pure DC supply might be the edge that you're looking for.

I'll be interested to hear how people get on. And if ever any of these mods make it into SB3 I'll be buying that as well.

Andrew L. Weekes
2005-07-23, 08:14
Out of curiosity, where is the jitter measurement being made, is it demodulated at the analogue output?

Just based on my quick calc's, in a 20k bandwidth, 3.5ns would equate with only 67dB dynamic range, which I'm sure isn't the case, 1nS would only be around 78dB.

I wonder if the analyser's jitter is also wrapped up in that measurement somehow?

Are you having fun learning the new toy anyway :-)

Andy.

NealG
2005-07-23, 15:24
Hmmmmm.... big :)

Sean have you listened subjectivley yet? What do you hear?

Patrick Dixon
2005-07-24, 01:05
I think he's just got to go out and buy some audiophile amps and speakers ... and some wanky wire to connect them all up ;) ;)

seanadams
2005-07-25, 16:57
Out of curiosity, where is the jitter measurement being made, is it demodulated at the analogue output?

Just based on my quick calc's, in a 20k bandwidth, 3.5ns would equate with only 67dB dynamic range, which I'm sure isn't the case, 1nS would only be around 78dB.

I wonder if the analyser's jitter is also wrapped up in that measurement somehow?

Are you having fun learning the new toy anyway :-)

Andy.

That's a pk-pk measurement and really says nothing about the distribution or the spectrum of the jitter. I'm only considering it useful as a rough relative measure for quickly comparing different setups.

The dscope manual describes in not much detail what the hardware is doing (page 155):

http://www.prismsound.com/downloads/dS3_Operation_Manual_A4nc.pdf

You can see the demodulated jitter content in the audio band in the FFT view - it is basically flat, and you can see the level drop slightly with the linear PS. I don't have a lot of time at the moment to do more tests but when I have more data I will post it.

seanadams
2005-07-25, 17:03
Hmmmmm.... big :)

Sean have you listened subjectivley yet? What do you hear?

Yes, it sounds great, but I honestly don't feel qualified to talk about subjective differences - I'd be hard pressed to tell the different between 320K MP3 and lossless PCM, so I rely on my equipment and on feedback from those with better stereos and better ears. Possibly too much time spent listening the my Honda's 2KW sound system as a teenager. :)