[JoshK]

I am busily working on a couple other projects atm, and I can't seem to find where I put the chassis that I was going to use for the SB transplant  , but I thought I'd sketch a thread out with some ideas/brainstorming and information gathering/sharing. 

First we have the conceptual diagram of the SB's power distribution.



I logically group this as 5V for peripherals and 14V for analog section which is currently being taken off the multiplier.  I'd like to experiment with feeding the analog section with its own power supply.  There is a lot of motivation already written for doing so, as noted in Pat's thread** as well as others.

Has anyone seen pictures of the board that point to where this multiplier is?  [I don't have the board in front of me, and plus pictures are worth a thousand words]

Has anyone sketched out the PSU flow on the board?  I am pretty sure someone has already thought of this problem and people have probably already done this experiment.  My goal is not to be the first but to also try it.  I will observe what others have done but still plow my own path. 

With the experiments done with the Felix power filter, my plan is to incorporate a Felix before the PSU, but possibly also some variant after in the passive filtering.  I might continue to scroung the surplus sites looking for an appropriate dual voltage switcher to experiment with and compare to linear supplies.  My a priori hypothesis is that a felix plus switcher o/p's the linear with all grouped together (based on some of this group's empirical feedback) but that when the analog portion is split from the grouped supply and done seperately, a linear might prove useful. 

I already have a pretty much top component implementation of a 5V linear to compare all results with. 

It has always been my assumption that all these uber power supplies were kind of wasted effort since everything was going through a lousy multiplier.  If we take the multiplier out of the analog PSU then maybe it won't prove to be wasted. 

** Pat's thread

[JoshK]

Here is one of the first posts I read on the topic... from the horse's mouth.

Sean Adam's on super regulators and jitter

Sean's original post with power diagram

[art]

Hmmm.....reading the first thread sounds a lot like what I am trying to do.

With only 0.75" x 1.5" to work with, it isn't easy. The supply for the clock has to have low 1/f noise, and very low output Z to keep it quiet once things get going. And not take up any space. (Getting any 2 together is not hard. Getting all 3 at once is.)

For those of you who haven't found out yet:

You need to turn the display off when it is playing!


The noise on the 3.3 V rail is horrible to begin with. It is only worse with the display on. What makes it even more annoying, I have some linear supplies that actually perform worse than the stock switcher. So, buyer beware!

Pat

[Den]

Hi Josh.   I think Vinnie used to substitute 12V battery power for the 14V.  You might want to ask him where he plugged into the circuit.

[JoshK]

I was aware of that, but I am really not trying to make the lab a forum for circumventing the pros on this board.  I am more interested in sharing mutual discoveries among the crowd who wish to learn and do for themselves.  This may seem like a subtle difference, but its there none-the-less. 

If Vinnie wishes to share his experience he can do so, but I am not going to try to illicit his experience and then share that with everyone freely.  Either we learn it for ourselves (which could be a different way to skin the same cat) or he shares by his own accord. 

My assumption though is that those who would wish to participate in this thread are a different target audience then Vinnie's potential customers, but there is probably a small overlap as well.

[ToddSTS]

Hi Josh.   I think Vinnie used to substitute 12V battery power for the 14V.  You might want to ask him where he plugged into the circuit.

That is what I was thinking.  I have a an analog only RWA modded SB3, so I don't have the sweet battery power mods.  But IIRC those mods included a power connector mod that allowed both a 12V line and a 5V line to be fed from the battery pack into the squeezebox.  Not sure if he fed that 5V straight into the dac or how he split it up.  I've been wanting to send mine back to complete the mods to full battery mode, but with a new house and a new wife I just don't have the funds at the moment.    So I guess that puts me into part of the overlap that JoshK is talking about. 

Todd

[Den]

I appreciate your point about learning for yourself, but the other point of not stepping between Vinnie and potential customers. . .he dropped SB mods from his offerings more than a year ago, probably two years ago.

[Den]

I appreciate your point about learning for yourself, but the other point of not stepping between Vinnie and potential customers. . .he dropped SB mods from his offerings more than a year ago, probably two years ago, and, sure, he's free to post in any thread he wants to, but maybe he does not monitor threads outside his own forum.

[Den]

oops.  quasi double post.

[JoshK]

Good to know, I was not aware of that.  I don't read many of the vendor forums. 

Still, if he is willing to share then we can ask him.  Some other pro modders have shared some tid bits of experience on the SD forum, from what I saw from skimming.

Anyway, it shouldn't be hard at all.  We just need to find the trace on the board where the analog split off from the multiplier is, cut it and provide our own PSU.  If done correctly, in theory there is no reason why it couldn't be easily reversed with jumpering the cut in the trace. 

I was just throwing it out there to see if anyone had traced out where that multiplier was.  I'm willing, but I haven't cracked the case on my SB in a while and probably won't have time to for a bit. 

What Pat says about the 3v3 supply though brings up the bigger questions I have.  I am assuming the clock's supply is derived from the 3v3 supply?  That would make sense, I did'nt know what all "IO, logic" implied or where the clock supply was coming from. 

I have more leanience (sic?) in room, as I plan to use a much larger case, not the original SB case.

[NewBuyer]

...For those of you who haven't found out yet:

You need to turn the display off when it is playing!


The noise on the 3.3 V rail is horrible to begin with. It is only worse with the display on...

I remembered that on this linked page there are two posts by Sean Adams (one at top of page, another about mid-page) about the SB3 noise floor being lowest when the display is at FULL brightness...

[art]

You can measure a direct connection between pin 14 of the clock/SPDIF chip to the (+) terminal of the 16 V, 220 uF cap on that rail.

It also goes to the big FPGA right next to it. Draw your own conclusion.

As for full brightness................

What the hell was he measuring????? I am measuring the noise on the rails, specifically the +3.3 V. No, you gotta turn the damn display off.

Pat

[Wayne1]

The 14 volts from the switcher goes to a series of regulators for the analog section. The main reason to have the 14 V there is to derive 9 volts to power the analog output ICs. There is a resistor in the line from the switcher to the "BFC". If you can find that and remove it... well I think you should be able to work out the rest.

There are many ways to surpress "noise" in power supplies. A CLC circuit is one way. Jensen 4-pole caps are another. CRC's will also lower ripple and noise.

Keep in mind that most regulator circuits do rely on feedback. If there is noise being added after the reg, it can come back to haunt you 

No, you gotta turn the damn display off.

A grounded shield can be installed behind the display that will reduce the radiated noise.

[jhm731]

There are many ways to surpress "noise" in power supplies. A CLC circuit is one way. Jensen 4-pole caps are another. CRC's will also lower ripple and noise.

Gee, you left out your Bybee upgrade and all those cute little pucks. 

[randytsuch]

Hi Josh
A while ago, I traced through how power routes on a SB2, and posted it in a thread here
http://www.audiocircle.com/index.php?topic=45330.msg447603#msg447603

So, the multiplier you are looking for is on the display card in a SB2.  I never looked on a SB3, but I would imagine it is around the display, since the main SB2 board layout is very similiar to the bottom of the SB3 board.

One of these days I will finish my SB2 project, have been sidetracked again.  I did finish my taxes this weekend, but now I have a project going on my pool that is taking priority, since it is warming up here.

BTW, I noticed you were just talking about improving the 9VDC for the analog output.  I don't plan to use the SB2 analog output, so I dont care about it, but I think it would help to do something about the 5VDC used for DAC power.  I was going to build a little shunt reg based on a thread by Thorsten, that I could mount close to the dac.

Randy

[tanchiro58]

I was going to build a little shunt reg based on a thread by Thorsten, that I could mount close to the dac.

Hi Randy,

Does this make your SB2 sound much better than the linear PS and Felix+switching PS? Why should the shunt reg be mounted close to the DAC? Any information would be appreciated. Thanks.

Tan

[JoshK]

To tell you the truth, I am really only interested in the digital portion of the SB2 also.  But I am interested enough to know how the old analog supply worked and its limitations because I may try to work with it at some point.  I have also considered tapping into the SB2 and using a twisted pair dac or similar in the same box without the SPDIF interface.  So I'd need to know where to tap the I2S from and if it needs to be offset.  John Swenson's experiments and sharing are invaluable here.

[TomS]

To tell you the truth, I am really only interested in the digital portion of the SB2 also.  But I am interested enough to know how the old analog supply worked and its limitations because I may try to work with it at some point.  I have also considered tapping into the SB2 and using a twisted pair dac or similar in the same box without the SPDIF interface.  So I'd need to know where to tap the I2S from and if it needs to be offset.  John Swenson's experiments and sharing are invaluable here.

I wanted to try something similar lashing up the SB3 I2S to the I2S bus on my Doede Douma DDDAC1543 (inside one box).  When I poked into it, I was led to believe it wasn't possible, or at least not trivial without some gate logic/timing work.  I might rather try it with a lifted Duet receiver module now, so at least I wouldn't have to deal with the display/ps, but same question applies.  I just didn't travel too far down that road as I moved on to strictly SPDIF digital connection to the DCX.  Maybe there is hope for this? 

[sts9fan]

I have also considered tapping into the SB2 and using a twisted pair dac or similar in the same box without the SPDIF interface.

I am also thinking about getting my SB to feed I2S straight into a Twisted Pear Opus or upcoming Buffalo DAC.

Kris

[JoshK]

To tell you the truth, I am really only interested in the digital portion of the SB2 also.  But I am interested enough to know how the old analog supply worked and its limitations because I may try to work with it at some point.  I have also considered tapping into the SB2 and using a twisted pair dac or similar in the same box without the SPDIF interface.  So I'd need to know where to tap the I2S from and if it needs to be offset.  John Swenson's experiments and sharing are invaluable here.

I wanted to try something similar lashing up the SB3 I2S to the I2S bus on my Doede Douma DDDAC1543 (inside one box).  When I poked into it, I was led to believe it wasn't possible, or at least not trivial without some gate logic/timing work.  I might rather try it with a lifted Duet receiver module now, so at least I wouldn't have to deal with the display/ps, but same question applies.  I just didn't travel too far down that road as I moved on to strictly SPDIF digital connection to the DCX.  Maybe there is hope for this? 

I believe what is needed is a bit shifting circuit (logic/timing) to go from left justified I2S to right justified, or maybe its the other way around.  Twisted Pair fortunately has a metronome CB that does just that plus some reclocking.  So I think the harder parts are done if you trust TP's boards.   

Basically I think all the pieces are there, but I am really out of my element here and I could easily crash and burn on this project, but it would be worth the learning experience in my mind.