Touch I2S-out and Clock-in mods

[hybride]

My reciever is ripped from it's DAC, WIFI and SPDIF circuit and only the processor and controller are running on 1v2 and 3v3 batterie.
What's next? If the Touch would sound better on I2S i more think about a better/other algorithm code in de controller, but that will be hard to prove.
A listening test between them (both on batterie) would be great. Hope someone will do and report...

[JIJ3]

I have both the receiver and the Touch modded identically. I have compared them by feeding the same DVI cable from each one's I2S output into the same input on my DAC. They sound identical on Red Book CD. Since in both cases the clock is generated at the DAC, there really is no reason the two should sound different, since at 100% volume the only DSP either unit does is to convert a FLAC ethernet stream into I2S, which should be mathematically identical. Jitter generated by each unit should not matter, since everything is reclocked at the DAC. Really, they shouldn't sound any different than different hard drives should "sound", as long as everything is error-free. I suppose EMI and power-line hash and such could cause sonic differences, but I have not noticed any. I do use a PS Audio Power Plant regenerator to feed AC to the DAC and the rest of my stereo and plug the Squeezboxes supplies directly into the wall. I use no other Squeezebox modding and the power supplies for both are stock.

Clearly, DAC output of the Touch is audibly far superior to the receiver's. I suspect the SPDIF output is much superior as well, although I have not myself carefully compared them.

I only reason I would propose to mod the Touch instead of the receiver is to receive 88K and 96K files. I do find this to be a valid reason. Be aware that the Touch mod is somewhat trickier than the receiver mod because of the cramped space, but more importantly because the clocks are at double frequency. This requires care at the reclocking flops in the DAC to be sure about all timing relationships. You have to allow for the time it takes for the clock to propagate through the Touch and the I2S signals to return to the DAC. You also have to be more careful with the I2S receiver PCB layout due to the higher frequencies.

If you only listen to redbook, I recommend you enjoy the receiver (with the external clocks-in and I2S-out mods) and not bother with the Touch. However, with all the high rez files becoming available, it is nice to have the flexiblity of accepting them. For instance the new 96K files for McCartney's Band on the Run are fabulous (although I honestly have not listened to them down-sampled to 48K through the receiver).

Jack

[Telrab]

My reciever is ripped from it's DAC, WIFI and SPDIF circuit and only the processor and controller are running on 1v2 and 3v3 batterie.
What's next? If the Touch would sound better on I2S i more think about a better/other algorithm code in de controller, but that will be hard to prove.
A listening test between them (both on batterie) would be great. Hope someone will do and report...

Hi Hybride, how do you connect the batteries to the receiver? I'm curious, because i also used the reciever with 1v2 and 3v3 (lifepo4) batteries and wifi-card removed. I just killed my second receiver (ethernet dead, no lights burning anymore).

Telrab

[hybride]

Hi Hybride, how do you connect the batteries to the receiver? I'm curious, because i also used the reciever with 1v2 and 3v3 (lifepo4) batteries and wifi-card removed. I just killed my second receiver (ethernet dead, no lights burning anymore).

Telrab

I removed the coils under the wifi card and soldered the wires there for 1v2 and 3v3.

I also had a dead reciever! (hot lan chip) I think i was uncarefully with soldering. Now i work like a surgeon with handkerchief with a hole in it :-))

[hybride]

I have both the receiver and the Touch modded identically. I have compared them by feeding the same DVI cable from each one's I2S output into the same input on my DAC. They sound identical on Red Book CD. Since in both cases the clock is generated at the DAC, there really is no reason the two should sound different, since at 100% volume the only DSP either unit does is to convert a FLAC ethernet stream into I2S, which should be mathematically identical. Jitter generated by each unit should not matter, since everything is reclocked at the DAC. Really, they shouldn't sound any different than different hard drives should "sound", as long as everything is error-free. I suppose EMI and power-line hash and such could cause sonic differences, but I have not noticed any. I do use a PS Audio Power Plant regenerator to feed AC to the DAC and the rest of my stereo and plug the Squeezboxes supplies directly into the wall. I use no other Squeezebox modding and the power supplies for both are stock.

Clearly, DAC output of the Touch is audibly far superior to the receiver's. I suspect the SPDIF output is much superior as well, although I have not myself carefully compared them.

I only reason I would propose to mod the Touch instead of the receiver is to receive 88K and 96K files. I do find this to be a valid reason. Be aware that the Touch mod is somewhat trickier than the receiver mod because of the cramped space, but more importantly because the clocks are at double frequency. This requires care at the reclocking flops in the DAC to be sure about all timing relationships. You have to allow for the time it takes for the clock to propagate through the Touch and the I2S signals to return to the DAC. You also have to be more careful with the I2S receiver PCB layout due to the higher frequencies.

If you only listen to redbook, I recommend you enjoy the receiver (with the external clocks-in and I2S-out mods) and not bother with the Touch. However, with all the high rez files becoming available, it is nice to have the flexiblity of accepting them. For instance the new 96K files for McCartney's Band on the Run are fabulous (although I honestly have not listened to them down-sampled to 48K through the receiver).

Jack

Many, many thanks for sharing this experience, very useful.

In de past i'll tried SACD en DVD audio. While these recordings where better and with higher samplerate it couldn't add what i was missing in redbook. I Iearned that the key of making music from sound, even with redbook, is commonly a DAC issue. It is incredible difficult to make a good analog sounding DAC. I succeeded in that area and now i enjoy optimum musicality from Redbook.

[jkeny]

Hi Hybride, how do you connect the batteries to the receiver? I'm curious, because i also used the reciever with 1v2 and 3v3 (lifepo4) batteries and wifi-card removed. I just killed my second receiver (ethernet dead, no lights burning anymore).

Telrab

Yes Telrab, I killed one Duet also & I suspect it was using separate 1.2V & 3.3V power supplies too. I wonder if the sequence or timing that these PS come up in is important?

Hi Hybride, glad to see your mods are still working good. I'm going to look into taking the I2S off the Duet & also looking at incorporating a Sabre DAC inside the case fed directly with I2S from the Duet.

[omainik]

I am EXTREMELY pleased with the result. The real key to this is bringing in low-jitter clocks from the DAC to the Touch. Jitter generated inside the Touch is now unimportant, since all I2S signals are reclocked near the DAC chip (I use Tent Clocks and pico-gate LVC logic).
Jack

How do you recognize the right clock frequency on the DAC side for reclocking?

[JIJ3]

I use a D-flip flop as sync discriminator. WS is connected to the clock input and the master clock is connected to the D input. When WS and MC are synced, the output of this FF is steady. If out-of sync, there are pulses on the output. This output toggles another FF which controls the source of the master clock. The output of this FF changes state until the MC and MS are synced.

Works great.

Jack

[omainik]

I use a D-flip flop as sync discriminator. WS is connected to the clock input and the master clock is connected to the D input. When WS and MC are synced, the output of this FF is steady. If out-of sync, there are pulses on the output. This output toggles another FF which controls the source of the master clock. The output of this FF changes state until the MC and MS are synced.

Works great.

Jack

Great!

I used a slightly different configuration on a Squeezebox Duet, only for Redbook. Clock Injection of the clock responsible for 44K1 Data. I take out of the Squeezebox only the Spdif signal. An Cs8416 receiver is drived by clock an works in a sync with the squeezebox. BCK and WS are derived from the clock through division; the Data line comes out of the cs8416 spdif receiver.

I am really pleased with the result.

I want to try a solution with the Squeezebox Touch, which is able to switch between 44k1, 48k, 88k2 and 96kHz. NOS direct to TDA1541 DAC.

I should switch the crystal and the division for generating WS and BCK to switch between 44.1 and 88.2 respectively 48 and 96 kHz. Could be tricky to realize without a microcontroller.

The first d-flip flop drives a binary counter. the LSB drives the switch between frequencies, the next bit drives the factor switch for the division. So the Circuit Cycles in a definite way through all possible frequencies.

That should do the trick?

Which DAC IC do you use?

[JIJ3]

Yes, that should work. Be aware that the Touch requires double frequency clocks (22.5792 and 24.576 MHz). I don't use SPDIF, so I can't say whether you will run into problems with the receiver chip at the different frequencies.

I also use a TDA 1541. I can switch between OS (for redbook) or NOS.

Jack