dac I/V convertion with very low distortion

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I asume MiiB that for the PCM1704 you are saying this (attached red circled) has to be change to 2kohm and 870pf as Serg says?

Cheers George
 

Attachments

  • TAGUS%20X.png
    TAGUS%20X.png
    52.8 KB · Views: 468
I didn´t know that you leave sonics i have to check Canalis.

The difference in this implementation is that normaly you see DC servos connected to input of the amplifiers and the noise of the servo opamp will be injected in the main amplifier, so you need to use a audio grade opamp for low noise like opa627 , but in this case the servo is not in the audio path (as Ken Newton has explain in post#561) so you just need to use a opamp with low input offset , the noise is not important (like for exemple max4238 less than 1€ and much beter in dc precision than audio opamps).
This is more economical (opa627 cost more than 20€) and better sonically as the servo is not in the audio path.

It doesn't matter where you connect the servo, it is -always- in the audio
path. Over the years this discussion has come up repeatedly.

The servo has to adjust the -audio- OP DC level and do so by integrating
the audio itself. Every servo works this way, you can connect them wherever
you like but if they are to work, they are in the audio path.

As such they definitely are audible and can often sound worse than a good
coupling cap. I know people who have in fact gone to extreme lengths to
make the servo truly out of the circuit, one worked on heat feedback and
another was a sampling system which made small adjustments at pre defined
time intervals, otherwise being off.

Just socket the opamp and try swapping it - you might get a surpise :)

Sorry, just done the calculations, and yes both values of cap and resistance are -3db at 90-95khz.
Thanks George

George,

The optimum value of the LPF cap will probably be system / DAC dependent.
You might even get away without one, if the I-V is stable as such. The
1704 has low OOB noise.

WRT the 2k OP resistor, I would first try this I-V without an OP buffer.
It may well sound better 'in the raw' so to speak.

The issue with OP buffers is you will be hard pressed to find a design that
preserves the linearity of this I-V's super low distortion.

It will almost certainly require some sort of bootstrapping.
 
George,


WRT the 2k OP resistor, I would first try this I-V without an OP buffer.
It may well sound better 'in the raw' so to speak.

The issue with OP buffers is you will be hard pressed to find a design that
preserves the linearity of this I-V's super low distortion.

It will almost certainly require some sort of bootstrapping.

Hi Zen, I need the lower output impedance of the buffer, (BTW what is the output impedance with and without buffer)? as this needs to push into the Lightspeed (10k'ish) together with whatever the poweramp may be, total can come down to 7k'ish. I also love the idea of direct coupling, hopefully with dc servo. When all is finalized
Cheers George
 
Zen, at the beginning I was reluctant in using a servo, because I am very aware of the problems that conventional servo has . Then I fought of adding a second pole to filter the AUDIBLE noises and distortions produced by the servo integrator , and this work very well, I have lost a lot of time trying to come up with a good solution for this , and I think that I have found it. But if someone with your experience thinks that this particular servo may have an impact in sound , others with less knowledge in analog design may as well feel uncomfortable , so I will add a optional trimmer in the PCB for manual regulation of DC offset.
I have made a lot of tests with this servo, and I can assure you that it has a minimum impact in audio band.
Please take a better look at this.
 
Last edited:
Hi Zen, I need the lower output impedance of the buffer, (BTW what is the output impedance with and without buffer)? as this needs to push into the Lightspeed (10k'ish) together with whatever the poweramp may be, total can come down to 7k'ish. I also love the idea of direct coupling, hopefully with dc servo. When all is finalized
Cheers George

You dont need a buffer just use a 2k5 resistor at the output, this resistor in parallel with the 10k of your atenuator will have the efective 2k out impedance that is require for 2 volts rms out.
 
When you use the Light speed attenuator you sometimes see quite low impedance's hence my suggestion for a Buffer... If I was to make a product.. I would for sure include a buffer... I have often used the simple N-JFeT buffer, with no sonic penalty.

Regarding the servo I totally agree with Sergio.. this way of implementing the Servo has no sonic penalty. This type of servo simply adjust the driving currents on a DC level, and not like most servos by inserting a DC offset in the signal path.

Thanks for the correction on the 1541...This has slipped my mind. sorry.
 
MiiB there is no problem with low impedance at the output of tagus , you can even shunt the output to ground , there will be no damage, there is a good idea to use a attenuator at the output of tagus and then if you want , use a buffer.
Later i will post some ideas for buffers, and the method to include them in the DC servo loop.
I am looking for dual potentiometer with 1k or less for audio use, if someone know any, please let me know.
 
Last edited:
The problem is not in that direction, the problem is the varying gain when we see different impedances in the equipment the DAC is attached to..there you may look into into something with a quite lowish impedance.

I know the current conveyor has no problem with a short, as the output is a current source.
 
smms,
I realized that we need to take care the input impedance next to your circuit.
e.g. SEPP JFET buffer or voltage follower with JFET opamp have high input "resistance" ,but large capacitance.
when the capacitance varies by input signal voltage, the actual IV impedance will vary and cause distortion.
 
Sorry MiiB, my mistake, I understand now.

Thanks Joachim, I will check that.

Shinja I think we can use a cascode that keep the source drain voltage constant, that will remedy the problem , for now I want to concentrate on the pcb, but please see if you can arrange some good buffer and post your ideas.
 
It doesn't matter where you connect the servo, it is -always- in the audio
path. Over the years this discussion has come up repeatedly.

The servo has to adjust the -audio- OP DC level and do so by integrating
the audio itself. Every servo works this way, you can connect them wherever
you like but if they are to work, they are in the audio path.

As such they definitely are audible and can often sound worse than a good
coupling cap.
Therefore you also believe that the whole psu is in the audio path.
I don't have much experience on servos, but the switch that Segio is preparing will allow us to try with and without it.

The optimum value of the LPF cap will probably be system / DAC dependent.
You might even get away without one, if the I-V is stable as such. The
1704 has low OOB noise.
WRT the 2k OP resistor, I would first try this I-V without an OP buffer.
It may well sound better 'in the raw' so to speak.

Absolutely. The analog lp filter has a very detrimental effect on sound, at least on 1704 and 1541 where I have always preferred the dac filterless. This is the beauty of multibit dacs- they don't *require* oob filtering.
Same thing for buffers- the b1 is NOT transparent. I'm still looking for one that truly is.

Another requirement of a good I/v is speed (phase linearity at hf), even some modern opamps fare well in this regard. I think that this circuit is very promising WRT.
 
Polarity is correct.. It works and has nanovolt offset... the buffer in itself has -140dB distortion ans is not visible on the FFT. We have used this buffer in the Paradise, where we have seen it work very very good.. Its easy to implement and requires no selection.

Are you absolutely certain the servo polarity is correct? It appears reversed to me as well.
 
Therefore you also believe that the whole psu is in the audio path.
No. The PS rejection and servo are completely different issues.
You have to think about this a bit more - eventually you will see it.
I don't have much experience on servos, but the switch that Segio is preparing will allow us to try with and without it.
Yes that will be a good exercise.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.