The SB3 is an awesome machine, so much so that it's replacing my super-modded CD3.5. (shock, horror!)
As it stands, the SB is powered from a 5V/2A wallwart and sounds pretty good. Surprisingly good in fact. Ok, it's not as refined, detailed, open or dynamic as the 3.5, but with a little attention I think it could be pretty amazing.
With an already very low level of jitter ( ~ 65ps), and given that I don't have an external DAC, I'm tempted to ignore digital/reclocking options for now. The analog stuff, however, is gagging to be changed. First up, the DAC.
The factory standard DAC is the Burr Brown PCM1748E (although there are some reports of people having a KE fitted at the factory) which is the little brother of the PCM1748KE. Main differences are THD-N levels of 0.003% and 0.002% respectively, a SNR of 100dB and 106dB respectively and a dynamic range of 100dB and 106dB respectively.
Looking at the PCM1748 datasheet gives some idea of possible mods. Based on my noodlings with the CD3.5's DAC (based entirely on Martin Clark's Acoustica pages) I reckon the first things to do would be to upgrade the power and decoupling on the DAC.
The datasheet recommends that the digital (Vdd) supply is derived via linear regulator from the analog (Vcc) supply. Analog rail is 5V, digital is 3.3V. I reckon that I could leave the 5V / regulator / 3.3V rail in place, but break the 5V link to the Vcc pin. That would allow fitment of a Flea for the analog rails; I did this in my CD3.5 and the difference is not subtle... in the SB3 I would imagine that breaking the noisy link between digital/analog will have even more sonic benefit.
The second thing that jumps out at me is that the PCM1748 has a common voltage pin (pin 10) brought out for decoupling that is biased to Vcc/2. The typical application here is to use a 10uf tant and I wonder if replacing that with a big oscon would reap benefits. Alternatively, adding a 0.1u film in parallel may improve things slightly.
The analog output opamp is a NJM2041. It sits between the DAC and the output RCA sockets and would appear to add some gain in addition to buffering. Replacing this with something like an AD8620 should, in theory, have sonic gains. There may be some issue with regard to phase reversal here, but I'd need to check. Some posts on the internet suggest bypassing the opamp entirely, removing it from the SB3 but this leads to a phase reversal and a drop in output power, giving approx 1V pk-pk. My primary concern with this mod is that the DAC may not be happy driving the buffer/TA stages of my preamps via a lengthy, capacative interconnect and, depending on input impedance of whichever pre I use (62, starfish, ad815) the performance could drop. I'm thinking that maybe building a little buffer circuit from opa627s or similar high-performace opamps could be a good option here.
Output coupling caps are 10u SMD jobbies. Ugh. Whilst fiddly, it may be possible to replace these with something a little nicer, perhaps bc128s or (if I can make 'em fit) some film caps. This warrants investigations I reckon.
Has anyone done these or other mods? Any input greatly appreciated - hopefully we can take this thread to interesting places
Cheers,
Carl
As it stands, the SB is powered from a 5V/2A wallwart and sounds pretty good. Surprisingly good in fact. Ok, it's not as refined, detailed, open or dynamic as the 3.5, but with a little attention I think it could be pretty amazing.
With an already very low level of jitter ( ~ 65ps), and given that I don't have an external DAC, I'm tempted to ignore digital/reclocking options for now. The analog stuff, however, is gagging to be changed. First up, the DAC.
The factory standard DAC is the Burr Brown PCM1748E (although there are some reports of people having a KE fitted at the factory) which is the little brother of the PCM1748KE. Main differences are THD-N levels of 0.003% and 0.002% respectively, a SNR of 100dB and 106dB respectively and a dynamic range of 100dB and 106dB respectively.
Looking at the PCM1748 datasheet gives some idea of possible mods. Based on my noodlings with the CD3.5's DAC (based entirely on Martin Clark's Acoustica pages) I reckon the first things to do would be to upgrade the power and decoupling on the DAC.
The datasheet recommends that the digital (Vdd) supply is derived via linear regulator from the analog (Vcc) supply. Analog rail is 5V, digital is 3.3V. I reckon that I could leave the 5V / regulator / 3.3V rail in place, but break the 5V link to the Vcc pin. That would allow fitment of a Flea for the analog rails; I did this in my CD3.5 and the difference is not subtle... in the SB3 I would imagine that breaking the noisy link between digital/analog will have even more sonic benefit.
The second thing that jumps out at me is that the PCM1748 has a common voltage pin (pin 10) brought out for decoupling that is biased to Vcc/2. The typical application here is to use a 10uf tant and I wonder if replacing that with a big oscon would reap benefits. Alternatively, adding a 0.1u film in parallel may improve things slightly.
The analog output opamp is a NJM2041. It sits between the DAC and the output RCA sockets and would appear to add some gain in addition to buffering. Replacing this with something like an AD8620 should, in theory, have sonic gains. There may be some issue with regard to phase reversal here, but I'd need to check. Some posts on the internet suggest bypassing the opamp entirely, removing it from the SB3 but this leads to a phase reversal and a drop in output power, giving approx 1V pk-pk. My primary concern with this mod is that the DAC may not be happy driving the buffer/TA stages of my preamps via a lengthy, capacative interconnect and, depending on input impedance of whichever pre I use (62, starfish, ad815) the performance could drop. I'm thinking that maybe building a little buffer circuit from opa627s or similar high-performace opamps could be a good option here.
Output coupling caps are 10u SMD jobbies. Ugh. Whilst fiddly, it may be possible to replace these with something a little nicer, perhaps bc128s or (if I can make 'em fit) some film caps. This warrants investigations I reckon.
Has anyone done these or other mods? Any input greatly appreciated - hopefully we can take this thread to interesting places
Cheers,
Carl
