low pass filter for DAC output

Status
Not open for further replies.
Jump to Latest
#61
The line stage will actually work on a single supply, If you want a split supply that's possible too but it means more caps. Single supplies are the best use of caps - I generate a negative rail with a paralleled pair of ICL7660s when I run from a 3.6V battery supply. As you have plus/minus 5V, we can treat that as a single 10V supply as it means just one cap bank between the rails.
 
#65
If you did have amps with balanced input the trafo wouldn't be totally necessary, but still desirable. As it appears your amps have unbalanced, then its non-negotiable.

No its to go from balanced (the output of the AD815) to unbalanced. Its also very handy to get the right voltage levels 🙂
 
#72
What about FDNR filters? They seems to be interesting (less critical components,no interaction with each other )
 

Attachments

  • FDNR.png
    FDNR.png
    58.5 KB · Views: 221
#73
Yes I looked at that kind of active filter when I was considering how to build my elliptic. It needs more opamps than the form I eventually chose and also depends more critically on the opamps' GBW. Seeing as the opamps with the highest GBW tend to be current feedback types (at lowest cost too) I looked into making it work with those for a while before giving up in favour of the less complex (in active components) schematic I'm showing on my blog.
 
#75
Looks to be a Bessel (phase linear) type. Very good for phase, very poor for stop-band rejection. It might suit a 16X oversampled DAC design where the designer was mainly concerned about phase distortion (i.e. having no overshoot on a step function).
 
#77
I no longer have a day job in anything related to audio. I know its 'received wisdom' that phase is very important but I've not found it so in practice. Whereas when I worked in the biz (this is going back over a decade) I'd have chosen linear phase over sharp slopes, these days I choose the opposite, based on listening. If you take a look at what loudspeakers generally do to phase then you might begin to wonder why it should be a big concern in electronics.
 
#78
I know the cross over has issues with phase anyway. So I don't use a cross over two- way and three way system. I am a full ranger and that is why nagging you with these phase question. Which design will you go with if you have to chose one with most steep attenuation after pass band and minimal effect on phase?
 
#80
abraxalito said:
Cut off frequency's a subjective judgement call. I aim for -50dB at 24.1kHz, which is the first frequency for images to appear. With a filter around 11th order, this puts the corner frequency somewhere between 17 and 18kHz. If you want more of the audio band to listen to you can relax the image rejection (which I chose just on a whim) or you can increase the filter order.

As regards the steepness - again its a trade-off. Steeper filters have better rejection but poorer phase response. Eliminating the images improves the dynamics, subjectively, results in a lower perceived noise floor. A shallower filter will probably sound better than no filter at all.
Click to expand...

Greeting to Hangzhou!

I am using the PCM1795 DAC and am upgrading it. It is effectively a 24bit DAC and its output has a 24dB/Oct LP filter with a corner frequency at just above 100kHz. I can change it to have a lower corner frequency at just below 60kHz that will result in 14dB better attenuation. For an example, the existing LP has 86dB attenuation at 1.4MHz, and if I change it, it can have 100dB attenuation at 1.4MHz.

There is no signal attenuation at 20kHz but the phase is increased from 25.7 degree to 41.3 degree!

Would you do it? 41 degree is a lot! Can we hear it?

Regards,
Bill
 
Status
Not open for further replies.
Top Bottom