Isn't there some filtering in with the op-amps?
Where did you join the I²S so as to maintain the differential inputs for the 1305?
Not that I can see. The TDA1305T datasheet states clearly that only a 1st-order LP filter needs to be implemented:
"Connecting external capacitors CEXT1 and CEXT2 between FILTCL and VOL and between FILTCR and VOR respectively provides the required 1st order post filtering for the left and right channels (see Fig.1). The
combinations of RCONV1 with CEXT1 and RCONV2 with CEXT2 determine the 1st order fall-off frequencies."
The internal RCONV = 2k; with CEXT=1nF this gives a -3dB point of 79.6kHz. In the DACMagic implementation this is followed by a second passive 1st order RC filter (R=680, C=1nF) with a higher cut-off frequency of about 234kHz. This is where I stole my outputs from.combinations of RCONV1 with CEXT1 and RCONV2 with CEXT2 determine the 1st order fall-off frequencies."
The opamp buffer does not appear to perform any additional filtering.
Directly after the 1315, before the combinational logic. Unfortunately, the 1315 does not tristate and always outputs a signal, so it had to be isolated. The tricky part was to locate where to do this on the PCB. After some headscratching it turned out that only 3 PCB traces needed to be cut, all on the top side. See attached pictures for my birdsnest test setup.
I added a SN74HC157 to beef up the signal level, and also use this to switch between the RbPi/HifiBerry and CD inputs. Works nicely; just let me know if you would like more details.
Attachments
Passive output stage
Would be interesting to compare experiences for the DACMagic!
I used 2.2uF WIMA MKPs and was pleasantly surprised.
Will expand to differential setup later, needing 4 caps; suggestions for better, reasonably priced output caps are welcome.
Isn't the LiTe Ah a current-output DAC? This would require an I/V stage as well, not such a straightforward mod.
Would be interesting to compare experiences for the DACMagic!
I used 2.2uF WIMA MKPs and was pleasantly surprised.
Will expand to differential setup later, needing 4 caps; suggestions for better, reasonably priced output caps are welcome.
Last edited:
Op-amp filtering
Hi IanAs,
In reply to your question:
Since I have been intending to learn Spice for quite some time, I took this opportunity to model the output stage and simulated it. The opamp does add what looks like another 1st-order LP filter, at a frequency close to the preceding passive RC LP filter. Not really important I guess, but wanted to mention it anyway.
Hi IanAs,
In reply to your question:
I said that "the opamp buffer does not appear to perform any additional filtering". This is not entirely correct; there is a small capacitor in the feedback loop which I missed...😱
Since I have been intending to learn Spice for quite some time, I took this opportunity to model the output stage and simulated it. The opamp does add what looks like another 1st-order LP filter, at a frequency close to the preceding passive RC LP filter. Not really important I guess, but wanted to mention it anyway.
Attachments
I own a DACMAGIC 2 that I'm not using anymore, and a Pi3 with HifiBerry DAC+ Pro XLR, that I'm not using anymore...
Anything interesting I could do combining them?
Maybe not for a great DAC, but at least a "conversation piece".
I've been itching to mod/mess around with the DACMAGIC 2, since it's so big and roomy.
Anything interesting I could do combining them?
Maybe not for a great DAC, but at least a "conversation piece".
I've been itching to mod/mess around with the DACMAGIC 2, since it's so big and roomy.
The DAC+Pro does have an I2S output header, so you can do the same mod as described above.
You will no doubt notice an improvement in clarity, detail and space over the normal SPDIF connection. In case of the DAC+ Pro it would be interesting to know how the Dacmagic compares to the Digiberry unit sound wise as well; after all, they are separated by 20 years of development...
As stated, the biggest improvement by far came from bypassing the output stage. You definitely should do this! If you go balanced with some nice large film caps (or trafos), together with the RBPi and separate power supply this will nicely fill up the empty space inside the Dacmagic. 😉
In addition I did similar mods to the power supplies and decoupling as described in the HiFi News article. This further improved the sound quality considerably - to a level that I am very happy with. It sounds so good to me in fact that I even suspended my search for a new DAC!
You may be in for a surprise here. Always interesting to hear other's experiences, so please keep us updated.
You will no doubt notice an improvement in clarity, detail and space over the normal SPDIF connection. In case of the DAC+ Pro it would be interesting to know how the Dacmagic compares to the Digiberry unit sound wise as well; after all, they are separated by 20 years of development...
As stated, the biggest improvement by far came from bypassing the output stage. You definitely should do this! If you go balanced with some nice large film caps (or trafos), together with the RBPi and separate power supply this will nicely fill up the empty space inside the Dacmagic. 😉
In addition I did similar mods to the power supplies and decoupling as described in the HiFi News article. This further improved the sound quality considerably - to a level that I am very happy with. It sounds so good to me in fact that I even suspended my search for a new DAC!
You may be in for a surprise here. Always interesting to hear other's experiences, so please keep us updated.
Attachments
I recently bought a dacmagic 1 and I'd like to follow John Westlake's suggestion by replacing the TDA1315H internal clock generator; the output pin 41 (ground 41) towards the 74HC04N filter should be bypassed by inserting a new clock generator board (es.:tentlabs type); it would also be necessary to interrupt the output of said filter towards the two TDA1305T, connecting the further outputs of the new generator board to the input pins 4 (ground 15) of the aforementioned chips; if someone had already made these changes, I would need to better identify the bypass points and any track interruptions; just to integrate some information appeared on the network, the dacmagic 1 has the same symmetrical dual differential topology of the dacmagic 2, implemented with a further analog filtering order ...