Home of the Squeezebox™ & Transporter® network music players.
Page 1 of 2 12 LastLast
Results 1 to 10 of 41

Hybrid View

  1. #1
    Junior Member
    Join Date
    Nov 2011
    Posts
    13

    Squeezebox Touch Hardware Mods

    Hi there ! I just reach this board few days ago thinking to find what i was looking for, hardware modifications for the SBT.
    Is there anything here ?
    I just found few posts about this but not a lot....

    Hope there will be DIY enthusiats here !

    Cheers.
    Olivier

  2. #2
    Hi! I've done a fair bit on mine but the last time I tried discussing it on this forum, it turned into a bit of a nightmare argument over commercial mods and general perceived problems with suppliers which really only told 1 very biased side of the story.

    On my own unit, I have solid polymer caps on the SPDIF buffer chip, Main 5v, clock switch logic chips, both 5v DAC rails. Low noise regs for the DAC 5v and 3.3v. The 5v supply for the touch screen sense have been removed from the DAC rail and attached to the main 5v rail. I have a low noise linear PSU with 80va transformer and 44,000uF low esr smoothers. The DC blocking caps in the audio path have been removed. On the clocking side, I completely reverse engineered the clocks. I have 2 external low jitter clocks which are attached directly to the switching logic (buffer logic removed which made a significant difference). I'm now looking are bypassing the switching logic too! The clocks are temporarily mounted external to the touch (on the back) and connected via coax with grounded only at the source. I'm also running soundchecks TT3 but with the display enabled.

    The audio quality from FLAC and WAV is shocking!! by shocking I mean unbelievably good!!

    What are you thinking about and how far do you want to take it?????





    I should point out my wireless works fine with these mods!!!! (not that I'm currently using it of course!!)
    Last edited by UV101; 2012-02-19 at 12:50.

  3. #3
    Senior Member
    Join Date
    Jun 2006
    Posts
    1,690
    Great work. So I take it the boards that are installed are the low noise regulators, right?

    When you talk about bypassing the switching logic, what are you referring to? Getting rid of the functionality or putting it off on another board?

    As far as I know there are two chips involved, a clock mux and a flip-flop. If you are only going to play one family of sample rates you can get rid of the mux (and one of the clocks), but I don't see how you can get rid of the flop, its the primary method by which the low jitter clock is used, by reclocking the data coming out of the processor.

    I've done a lot of testing of different muxes and flops and found that the ones already in the Touch are the best sounding ones of all that I tried (and I tried a lot).

    There are actually two clock muxes, one for the "music" path and one for the "effects" path. If you are not using the effects at all you can cut the clock signal going to that mux thus decreasing the load on the clock source which does slightly help. (if you do that either remove the effects mux or tie the input to ground, you do NOT want to leave a floating input)

    Exactly how those clock boards get connected to the Touch is fairly critical. You can have a significant impact on noise and jitter by different ground connections. Using the shield on the coax is actually a very good way to do it. If it were me I would tie the shields on the two coaxes together and connect that to the Touch groundplane as close to the signal connections as possible. The power supply connection gets tricky. Its best if you have a separate power supply for the clock boards (not just a separate regulator, a complete separate power supply). With the two separate boards it gets a little tricky. I would put a very low impedance connection between the ground planes of the two boards, a nice wide flat braid would work or a strip of copper etc. I would connect the tied together braids from the two coaxes to the center of this connection between the boards. The "gnd" connection from the clock PS also connects to this point.

    This meets two important criteria for a good connection of an external clock: there is one and only one ground connection between the Touch and the clock board, and that the return current is in close, low impedance proximity to the signal. Using the coax shield for this, and tying them together meets this nicely. Because you have two clock boards it makes things a little tricky but by making a low impedance ground connection between them and connecting THAT to the Touch you can still maintain a really good connection.

    This doesn't mean that what you have now won't work, its definitely a significant improvement, but that by doing it this way you can make it even better. I once added clocks to a SB3 with long ribbon cable with no termination etc, the signal integrity at the far end was horrible, but it STILL sounded way better than stock.

    John S.

  4. #4
    The boards inside the unit are my low noise regs. The circuit is based heavily on one of the commercial designs but there are a few changes such as additional noise filtering on VREF. The layout of the positive reg has been designed so that an oscillator and cap can be fitted directly to the board itself effectively changing it into a decent clock. This is what is currently attached to the rear of the unit. I was supprised how well it works even with the clocks external on coax.

    The replacement PSU supplies 5v low noise regulated for the main player supply rail but it also supplies 11v unregulated to the 2 internal regs and clocks (drop out it about 5v for these regs).

    With respect to the clocking, The connectivity details are here. U14 is removed and the 24.576Mhz feeds into pad 6 and 22.579Mhz into pad 8. Pads 12&13 are linked to preserve the 2nd clock feed (presumably to the the "effects" area? John are you able to confirm this is the right way round?)



    The original clock digram I removed from the previous thread is here



    As the oscillators have an enable feature,I plan to try eliminating the Texus Instruments logic U15 & U33 and run the clock direct switched on as required likie this.



    I'll try changing the grounding arrangment on the clocks. Currently they are source gounded at the PSU. I didnt try with the gnd only at the touch end of the coax but I will try to see if there is any improvement.

    Cheers,
    Last edited by UV101; 2012-02-20 at 12:51.

  5. #5
    Senior Member
    Join Date
    Jun 2006
    Posts
    1,690
    On the taking out the mux and using the clock enables, it MAY work if the output is tri-stated when disabled. BUT you now have the capacitance of the disabled output in parallel with the load, this may or may not cause a problem.

    With this scheme you give up the ability to set the effects sample rate different than the music sample rate. Thats why there are two muxes in the first place. At it's worst this will just cause the effects sounds to sound strange. If you always have effects turned off it probably should not be a problem at all.

    With the two clocks directly connected together I would probably put a small resistor on each output (say 22 to 47 ohms) and connect the other ends together, this will help isolate the oscillators.

    John S.

  6. #6
    Quote Originally Posted by JohnSwenson View Post
    On the taking out the mux and using the clock enables, it MAY work if the output is tri-stated when disabled. BUT you now have the capacitance of the disabled output in parallel with the load, this may or may not cause a problem.

    With this scheme you give up the ability to set the effects sample rate different than the music sample rate. Thats why there are two muxes in the first place. At it's worst this will just cause the effects sounds to sound strange. If you always have effects turned off it probably should not be a problem at all.

    With the two clocks directly connected together I would probably put a small resistor on each output (say 22 to 47 ohms) and connect the other ends together, this will help isolate the oscillators.

    John S.
    Thanks for the contribution John :0)

    In the diagram I had included a couple of 10r resistors to isolate the oscillator outputs. I knew it wouldn't be straight forward but I think it might be worth trying.

    I now understand why there are 2 mux (U15 & U33)chips. I couldn't work out why there were 2 independently switchable system clocks but now I understand.

    So what exactly does the effects clock do and when would it come into play? I see what you are saying with my conceptual clock, there would only ever be 1 clock feed albeit, could be sent to both area's.

    A couple of quick wins..........
    For those using the SB internal DAC, you can bypass the DC blocking output caps C88 & C91 10uF 16v SMT caps below the DAC which has been previously discussed elsewhere. I've removed mine and linked the pads. Another thing I have discovered is that the 5v SOT23-5 reg (U21) not only supplies the DAC with its 5v, it also supplies the sense side of the touch screen. On the front of the main PCB near the narrow screen connector, there is a 10r resistor (R179) this is the supply from U21 to the screen. It is possible to remove this resistor and connect the supply direct from the device input supply via 10r. The result removes the screen sense from the DAC regulator. I'm not sure how much of a audible difference this would make on its own but I would suggest it can only have a positive effect on the DAC supply.







    On the subject of PSU regulation, can anybody identify these components?
    I know they are 3.3v and 5v regs but I'm interested in the spec. I believe there may be significantly better low noise versions available with similar pin outs. The pin marked NR seems to be conveniently placed near to GND for some of the better spec regs that use a bypass cap on this pin for very low noise operation!



    I've got a few Nat Semi low noise regs on the way to try and hopefully a donor unit to fit them in!!! (eh Andy!!)

  7. #7
    Thanks, I'll see how I get on when the regs arrive. I'll try with lower value solid polymer and see how I get on.

  8. #8
    Right so I'm now in receipt of andyjh's touch and some 20uV and 40uV noise (very low) regs that should swap out with the originals.

    What I plan to do is run his stock unit against my modded unit to start with as this will give me a cast iron comparisson of the hardware changes i've made so far. (could really do with a 3rd to leave standard for the duration lol) What I'm hoping to hear is a a significant gap which will narrow as the mods are applied to Andy's touch. I'll update the thread as I make changes and compare the results.

    This is what I intend to do

    1: Baseline - compare with identical software settings to my own modded unit
    2: Increase input voltage to 5.4v to allow dropout for 5v regs connected directly to the input supply (thus avoiding one of the internal SMPS)
    3: Work on the 5v rail for the internal DAC
    4: Work on the 3.3v rail for the clock and associated components
    5: Look at the clocking - See if I can improve on the layout (as compared to my external clocks)and get the clocks inside the case again
    6: SP/DIF - along side I will be looking further back before the buffer to see if there is anything else further back in the circuit that can be improved
    7: External PSU
    8: External Wifi Antenna??

    That should keep me busy!!!

    **I'd better tell Andy the plan eh!!!
    Last edited by UV101; 2012-03-02 at 15:37.

  9. #9
    Member
    Join Date
    Feb 2006
    Location
    West Sussex
    Posts
    45
    Sounds like a good plan mate, lol

    Watching with interest
    Lounge - SB Touch with TT3.0, DC cap mod, XBMC ASRock, Proceed AVP2, CinePro 3K6, Monitor Audio GR20's, 10TB unRAID
    Workshop - SB3, T-Amp, JBL C1's
    Office - SB Touch, CLabs 2.1

  10. #10
    Junior Member
    Join Date
    Mar 2012
    Posts
    7
    Hi guys

    Currently i am using modded SB Duet, but in the near future I will probably buy and mod SB touuch a little bit.

    I would like to ask for your opinion - what do you think about changing stock clocks for Crystek CCHD-957 and power them up using two LIFEPO4 batteries?
    This battery can supply 3.3V so exactly matches crysteks requirements.
    Crystek clocks need maximum 25ma of current, so one battery could power a clock for approximetely 100h, so its nice.

    What do you think about this idea?

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •