I got a Squeezebox Receiver at Xmas with the intention of modding it & after struggling with trying to get it working directly from my latop, I finally have used another router acting as a bridge between SBR & my laptop. So now I can do mods & listen without having to connect it to my modem/router under the stairs. I'll eventually figure out the settings for correct working direct to laptop.
Anyway, I did some mods which I wanted to report here & also ask some questions.
- First, I wanted to focus on cleaning the PS & removed the 3.3V switching reg (under the WiFi board) & replaced it with a 3.3V LiFePo4 battery feed. I left the 1.2V reg in place as I don't think it feed any sound critical parts.
- I removed the 5V reg that feeds the DAC even though my eventual aim was/is to use an external DAC via I2S & again fed this with 5V battery pack & also at times from a transformer (more about this later). I wanted to hear the on-board WM8501 DAC before I went with an external DAC.
- I removed the non-descript elco coupling caps on the Vout & the signal now comes straight from the Vout pins. Now this has 2.5V bias riding on the signal so either caps or transformer is needed to remove the DC. More later.
- I tapped into the I2S signal which is conveniently located beside the Xilinx chip via 4 22ohm resistors & connected the 4 I2S lines via thin wire a socket made of a DIP8 op-amp adaptor glued to the board facing up. So I can plug in my DIY I2S DACs which have DIP8 socket on the underside of the DAC board. This gives me a nice short length(about 0.5 to 1 inch) for the I2S signal to travel to the DAC (ES9022 bottom of the range Sabre before I use the ES9018 later). Sound through my ES9022 Vout DAC was pretty good.
- Sound wasn't bad from on-board WM DAC through WIMA caps but probably better through the ESS DAC. Neither is up to my reference modded HiFace USB transport.
- So I had an idea of trying some transformers instead of caps. But you can't run 2.5V through a non SE transformer so I connected one leg of the primary to VMID pin on the WM8501 DAC which is a Vref at 2.5V & the other leg to Vout. This cancels the DC across the primaries. Sound was not very good - low & distorted.
- I disconnected the VMID & supplied an external 2.5V from a centre tap of the 5V supply & the sound now soars - best I have heard so far from this device. Quite close to the modded HiFace but just lacks that last bit of clarity & detail that the HiFace delivers without over-emphasing it. To me this sounds like low jitter ie. = very good analogue
- Is the transformer cancelling common noise between Vout & 2.5V? Is this why it seems not to matter if I run from batteries or from a mediocre linear transformer?
I wanted to supply my own external clean 2.5V supply as is stated in the datasheet. When I connect 5V to the DAC & I measure the VMID pin, there's 2.5V on it from an internal voltage divider in the DAC. If I connect an external 2.5V supply to this VMID pin there's no current flow, I reckon because there's no voltage potential. What am I missing here?
Next elements to address:
DAC supply extra de-coupling
VMID de-coupling
Addressing the whole clock area (some ideas about this)
John Swenson probably has done most of this already but I would highly recommend using a transformer in the configuration above. I'm using Sescom Mi-97 line driver transformers 600:600.
Anyway, I did some mods which I wanted to report here & also ask some questions.
- First, I wanted to focus on cleaning the PS & removed the 3.3V switching reg (under the WiFi board) & replaced it with a 3.3V LiFePo4 battery feed. I left the 1.2V reg in place as I don't think it feed any sound critical parts.
- I removed the 5V reg that feeds the DAC even though my eventual aim was/is to use an external DAC via I2S & again fed this with 5V battery pack & also at times from a transformer (more about this later). I wanted to hear the on-board WM8501 DAC before I went with an external DAC.
- I removed the non-descript elco coupling caps on the Vout & the signal now comes straight from the Vout pins. Now this has 2.5V bias riding on the signal so either caps or transformer is needed to remove the DC. More later.
- I tapped into the I2S signal which is conveniently located beside the Xilinx chip via 4 22ohm resistors & connected the 4 I2S lines via thin wire a socket made of a DIP8 op-amp adaptor glued to the board facing up. So I can plug in my DIY I2S DACs which have DIP8 socket on the underside of the DAC board. This gives me a nice short length(about 0.5 to 1 inch) for the I2S signal to travel to the DAC (ES9022 bottom of the range Sabre before I use the ES9018 later). Sound through my ES9022 Vout DAC was pretty good.
- Sound wasn't bad from on-board WM DAC through WIMA caps but probably better through the ESS DAC. Neither is up to my reference modded HiFace USB transport.
- So I had an idea of trying some transformers instead of caps. But you can't run 2.5V through a non SE transformer so I connected one leg of the primary to VMID pin on the WM8501 DAC which is a Vref at 2.5V & the other leg to Vout. This cancels the DC across the primaries. Sound was not very good - low & distorted.
- I disconnected the VMID & supplied an external 2.5V from a centre tap of the 5V supply & the sound now soars - best I have heard so far from this device. Quite close to the modded HiFace but just lacks that last bit of clarity & detail that the HiFace delivers without over-emphasing it. To me this sounds like low jitter ie. = very good analogue
- Is the transformer cancelling common noise between Vout & 2.5V? Is this why it seems not to matter if I run from batteries or from a mediocre linear transformer?
I wanted to supply my own external clean 2.5V supply as is stated in the datasheet. When I connect 5V to the DAC & I measure the VMID pin, there's 2.5V on it from an internal voltage divider in the DAC. If I connect an external 2.5V supply to this VMID pin there's no current flow, I reckon because there's no voltage potential. What am I missing here?
Next elements to address:
DAC supply extra de-coupling
VMID de-coupling
Addressing the whole clock area (some ideas about this)
John Swenson probably has done most of this already but I would highly recommend using a transformer in the configuration above. I'm using Sescom Mi-97 line driver transformers 600:600.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.