Dac output stage ??

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi Peranders!
Thanks a lot for your concern!
I have used this P-Channel common drain buffer in front of my X-BSOZ and X-SOZ as an experiment, and they works pretty good, and they are actually the buffer from Nelsons ZV4, the only difference is that I have 2,4V bias at the gate. The setup I have shown is just for some experiment since I have never been near a dac before.
From Nelsons text on the ZV4: There is another, more clever way to accomplish this follower,
as seen in Figure 2. Here we use a P channel Mosfet with the
Drain connected to ground and the Source biased up by a current
source, in this case just a resistor. There are several advantages
to this approach:
First, the input can be direct coupled, as it operates at ground
potential, eliminating the input capacitor and bias circuit.
Second, it is self-adjusting in the sense that the Source will
operate at +Vgs where Vgs is the Gate to Source voltage for the
Mosfet, or about 3.5 volts. The amount of current supplied by
the resistor is not critical, and can be quite high if desired, as the
Mosfet will only dissipate that current times the Vgs value. For
the ZVP3310 we will be using, you could run it as high as 50 mA,
which is a lot.
 

Attachments

  • corel.jpg
    corel.jpg
    11.7 KB · Views: 1,203
First test on "no opa´s in analog out"

I have just tested the analog output setup as attached.
My CDP is an Onkyo Integra DX706 modyfied with LCAudio Clock II.

The sound is just so much better, but!!!!
The gain level is unfortunately a little too low for my amps
http://www.diyaudio.com/forums/attachment.php?attachmentid=3072
:bawling:
An other thing is, the balanced operation as shown in the scematic sounds terrible from this DAC, but in unbalanced mode it is just so great.
The filter is is not quite satisfatory a little RF noise is audible at very close distance.
Components in the filter was polystyren caps an byslag resistors.

I have to think of what to do next, but these tests has shown me that something really can be done, and next time i want to try Thorsten Loesch´s filter.
 
Henrik said:
Hi Peranders!
Thanks a lot for your concern!
I have used this P-Channel common drain buffer in front of my X-BSOZ and X-SOZ as an experiment, and they works pretty good, and they are actually the buffer from Nelsons ZV4, the only difference is that I have 2,4V bias at the gate. The setup I have shown is just for some experiment since I have never been near a dac before.
OK, wrong of me. It looked to me kind of backwards. May I ask what the benefits are of this solution instead of having the output related to ground. I get a little "brittish car plus grounded" feeling over this.

I wonder also if the balance will good? Two different mosfet's, they must produce a pretty unbalanced balanced signal? Don't you need more precision in order to get "balanced" signal?
 
Hej peranders!

May I ask what the benefits are of this solution instead of having the output related to ground. I get a little "brittish car plus grounded" feeling over this.

Nelson:
the input system is at virtual ground, we can
swing all the drive current we want without being concerned about
the voltages across the input device, since it will be held very
close to the constant Vgs of the P channel input. No Miller effect,
hardly any gain or capacitance modulation, better performance
I gess this is the explanation on the good sound from this buffer.

I wonder also if the balance will good? Two different mosfet's, they must produce a pretty unbalanced balanced signal? Don't you need more precision in order to get "balanced" signal?
I have done some tests on the BOSOZ as a converter from unbal to bal mode. I found, that with the original passive currentsources (X)BOSOZ sounded a little better than with CCS´s, despite of the poorer produced balanced signals.
The balanced signals produced by the SM5861AP DAC isn´t worth listening to, or it might be the two not diffrential buffers, so I gess there is more to this. Next time I will try to make a diffrential input pair with some CCS´s and drive them with balanced and unpalanced input to see what happens.

Some tests on CCS´s: http://www.diyaudio.com/forums/showthread.php?postid=75141#post75141

Does someone know which type of inductors are the best for Time-Silcing-DAC filters like Torsten Loesch´s? I really woud like to know.
 
Disabled Account
Joined 2002
If I read correctly the input is virtual ground. Are R1 and R11 not too low in resistance ?
Far from experienced/understanding MOSFETS I recently checked Nelson's explanation and he uses 47k5 that is automatically the input resistance of 47k5. In the case of your output stage it is 220 Ohm. Maybe that's why your output signal is low ;)


So logically R7 and R8 can be omitted and R1 and R11 can be 24k if I really understand the working of this P-follower.

Please correct me if I'm wrong !

Regards,

Jean-Paul
 
Hi Jean-Paul

The ZV4 buffer is in it´s originally form inside a feedbackloop in the ZV4, so that is why the inputresistor has the higher value of 47,5K. The 220 Ohm resistor I am using is a gateresistor, which is there to prevent oscillation in the mosfet, nothing more. If you look at the ZV4 scematic and locate R7 an R8, then you will find two gateresistors at 220 Ohm.
R7 and R11 is a part of the lowpassfilter which is there to omit RF noice, they can´t be remooved, on the contrary, mine wasn´t effective enough, that is why I will try out a filter that includes an inductor at 4.5 mH.
R1 and R11 can be 24K, but the value 220 Ohm seems to be the lowest value tht effective prvents the fet from osilation.

I expected low gain (no gain from the follower) when I made this analouge stage, but I a sort of hoped for a little miracle, I got that not for the gain but for the good sound, wich is incredible good, and the gain is just a little too low. I will go for some supersymetric X next time, here I can adjust the gain, but it is a little more demanding in PSU and parts.

Regards Henrik
 
Hi Jean-Paul!

jean-paul said:
So logically R7 and R8 can be omitted and R1 and R11 can be 24k if I really understand the working of this P-follower.

Please correct me if I'm wrong !

Regards,

Jean-Paul


Henrik said:
Hi Jean-Paul

R7 and R11 is a part of the lowpassfilter which is there to omit RF noice, they can´t be remooved, on the contrary, mine wasn´t effective enough, that is why I will try out a filter that includes an inductor at 4.5 mH.

Regards Henrik

I think you are right, R7 an R11 is not essential for the filter, they only attenuates the signallevel, so at the moment I have enough voltage swing at the output. I still need to listen some more to be shure if the good sound still is ok.
But I need to make a new filter as mentioned before.
Regards Henrik.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.