I know there are some people interested in what improvement paralleling DAC chips bring to the sound.
I have paralleled PCM 1704 chips in my DAC and I must say that I was expecting more. The output is higher (comparing 1 o'clock to 4 o'clock position on volume knob) but I am not sure if the resolution and dynamics changed much. The cost and time involved in my opinion is not worth it. These were surface mount devices and I had to put one chip on top of the other. There was a gap between the legs so I had to use small pieces of wire to connect them.
From my experience I must say that using separate power supplies for digital and analog section of the DAC brings more improvement than paralleling DACs.
Any other opinions and observations?
Peter
I have paralleled PCM 1704 chips in my DAC and I must say that I was expecting more. The output is higher (comparing 1 o'clock to 4 o'clock position on volume knob) but I am not sure if the resolution and dynamics changed much. The cost and time involved in my opinion is not worth it. These were surface mount devices and I had to put one chip on top of the other. There was a gap between the legs so I had to use small pieces of wire to connect them.
From my experience I must say that using separate power supplies for digital and analog section of the DAC brings more improvement than paralleling DACs.
Any other opinions and observations?
Peter
Paralleled DAC's
Hi Peter (Hpotter),
I can coroborate your findings. I paralled two AD1865N in my NON-OS KWAK-DAC.
Definition was less and the bass kind of boomy. Output was higher of course.
A more intelligent way of paralleling DAC's is using two DAC's in a balanced configuration for one stereo channel. You need two IV converters though. The balanced DAC gives more power or bass-slam .
Hi Peter (Hpotter),
I can coroborate your findings. I paralled two AD1865N in my NON-OS KWAK-DAC.
Definition was less and the bass kind of boomy. Output was higher of course.
A more intelligent way of paralleling DAC's is using two DAC's in a balanced configuration for one stereo channel. You need two IV converters though. The balanced DAC gives more power or bass-slam .
I'm using grade K. Someone on this forum stated that paralleling DACs improves resolution, however it's hard for me to decide if this is what I achieved. The music has more "body" ,but sometimes it can be taken as boominess.
On the current board I can't use full balanced configuration, but that's what I will try in my next project.
On the current board I can't use full balanced configuration, but that's what I will try in my next project.
The three decouple pins got their own separate caps.
Jocko, I still believe that the best preamp is no preamp. My DAC has stepped volume control at the output with max. impedance of 500 ohm and usually 200 ohm. Of course that setup is only good when you use a single source. But since it sounds pretty decent I'm not listening to vinyl anymore.
Jocko, I still believe that the best preamp is no preamp. My DAC has stepped volume control at the output with max. impedance of 500 ohm and usually 200 ohm. Of course that setup is only good when you use a single source. But since it sounds pretty decent I'm not listening to vinyl anymore.
I take back everything I said before against parallel DACs. After modifications I made today my DAC sounds better than ever (I mean way better).
First of all I shielded PCM 1704 chips with pieces of copper (BTW, my original chips where shielded, it's after I installed additional chips I didn't put the shield).
I installed grounded copper plate between digital section (receiver and filter) and DACs (including output stage).
I damped the oscillator with a piece of 1/4" self adhesive rubber.
All the above changes improved the sound by a big margin. The initial boominess is gone, there is much more detail, but most of all there is something that makes all the instruments and voices sound very natural and real. Increased sense of space and 3-dimensionality with air between.
One of the tracks I like to use for testing is "Another Brick in the Wall (Part 1)". I can almost hear what the kids in the background are saying.
I can honestly say now that paralleling DACs make sense and it's worth the effort and time. But it's not just about putting 2 chips together. Small things, seemingly unimportant, like the staff I mentioned above have to be taken care off and then the benefits can be appreciated. My system never sounded better. Highly recommended.
Peter
First of all I shielded PCM 1704 chips with pieces of copper (BTW, my original chips where shielded, it's after I installed additional chips I didn't put the shield).
I installed grounded copper plate between digital section (receiver and filter) and DACs (including output stage).
I damped the oscillator with a piece of 1/4" self adhesive rubber.
All the above changes improved the sound by a big margin. The initial boominess is gone, there is much more detail, but most of all there is something that makes all the instruments and voices sound very natural and real. Increased sense of space and 3-dimensionality with air between.
One of the tracks I like to use for testing is "Another Brick in the Wall (Part 1)". I can almost hear what the kids in the background are saying.
I can honestly say now that paralleling DACs make sense and it's worth the effort and time. But it's not just about putting 2 chips together. Small things, seemingly unimportant, like the staff I mentioned above have to be taken care off and then the benefits can be appreciated. My system never sounded better. Highly recommended.
Peter
HPPlotter
Have you checked that your pieces of copper is actually shielding? I recently did this on a Sony SCDBX940 with a particularly dirty ground and found that they acted as receivers and aerial! In fact, Sony glued copper strips on top of their pulse dacs and they conveyed beautiful square waves at clock frequency. The effect was not due to probe pickup.
Have you checked that your pieces of copper is actually shielding? I recently did this on a Sony SCDBX940 with a particularly dirty ground and found that they acted as receivers and aerial! In fact, Sony glued copper strips on top of their pulse dacs and they conveyed beautiful square waves at clock frequency. The effect was not due to probe pickup.
What kind of shielding did you do with the copper plates?
Like put a wall around the sections?
Do you have any pictures ?
My CS8420 IC is on a separate board, would it be enough to have a separate power supply for it and have it far away from the main board ?
The filter DF1704 is close to the DACs PCM1704 , is it a goog idea to separate the filter from the DACs or just separate digital from analog?
I hope someone has time to replay. Thanks
Like put a wall around the sections?
Do you have any pictures ?
My CS8420 IC is on a separate board, would it be enough to have a separate power supply for it and have it far away from the main board ?
The filter DF1704 is close to the DACs PCM1704 , is it a goog idea to separate the filter from the DACs or just separate digital from analog?
I hope someone has time to replay. Thanks
I installed a 4 x PCM56 board on my non os prototype DAC.
Noise floor was a little better compared to one single chip.
Distortion was not much better.
Then I removed the MSB pot which was installed on only one chip, planning to measure each single chip and compare to the set of 4.
All chips together was not very good, I removed chip nr. 4 and that was much better.
Then I tried 40 different chips in socket nr. 4 and found some that gave good results.
Did the same with nr.3 , now it is already as good as a single very good chip with MSB adjust.
All chips are preselected to give best performance with MSB adjust, most of them are mediocre without the MSB ajust.
What cancels here is mostly the DLE, the linearity of all those chips is very good except the DLE.
One thing gets very clear, if chips are paralleled, the result can be everything.
Wether linearity errors cancel or add up, that is a statistical problem.
Noise floor was a little better compared to one single chip.
Distortion was not much better.
Then I removed the MSB pot which was installed on only one chip, planning to measure each single chip and compare to the set of 4.
All chips together was not very good, I removed chip nr. 4 and that was much better.
Then I tried 40 different chips in socket nr. 4 and found some that gave good results.
Did the same with nr.3 , now it is already as good as a single very good chip with MSB adjust.
All chips are preselected to give best performance with MSB adjust, most of them are mediocre without the MSB ajust.
What cancels here is mostly the DLE, the linearity of all those chips is very good except the DLE.
One thing gets very clear, if chips are paralleled, the result can be everything.
Wether linearity errors cancel or add up, that is a statistical problem.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.