Electronics > Beginners

Recommened SMT output caps for Audio DACs

(1/1)

hamster_nz:
I'm playing around with I2S audio DACs, and noticed that across the output caps there is something small but significant non-linear going on.

With probes at either end of C62 I can see that output of a saw-tooth waveform from the DAC is perfectly flat, but after the 3.3UF cap it has a bow in it.

The scales are the same, there is very constant DC bias over the cap, and there is no load. I would have thought they would have tracked perfectly...

What is going on?

I am also pretty sure I have the wrong sort of cap - currently using some X5R 10V I had to hand. What type, hopefully in an 0805 footprint, would you recommend in the usage? I am interested to get some and see the difference.

(This is just a hobby project, for my own edification, BTW).

ejeffrey:

--- Quote from: hamster_nz on October 23, 2019, 10:45:06 am ---The scales are the same, there is very constant DC bias over the cap, and there is no load. I would have thought they would have tracked perfectly...

What is going on?

--- End quote ---

I'm not sure it is strictly non-linear.  There is some load from the 10 k resistors which should make a highpass filter with a time constant of 33 ms.  With a 10 ms ramp time that will cause a slight bowing.  I'm not sure if it is as much as you are seeing.  This could be made worse if the effective capacitance is reduced due to the DC voltage offset, but I wouldn't expect this to be a problem given the presumably low DC voltage and the relatively large 0805 package size.

I would try doing a distortion measurement instead of a ramp.  That will tell you if you are actually seeing non-linearity or simply a low frequency rolloff.  If this is just an affect of the highpass filter you can just ignore it.  You might want to use a film cap instead of ceramic, but you won't get such a large value in an 0805 package.  The main reason to do this is to reduce piezoelectric effects.  Another option is to switch to a larger electrolytic capacitor such as a 22 uF value.  A better option if you want to keep to small surface mount parts is to raise the output resistance from 10k to 100k.  However once you use a load you may need an opamp buffer as the load on the other end of the cable is likely to still be on the order of 10k.

hamster_nz:
Oh! D'oh! It is most likely phase shift and a slight amplitude drop of the lowest frequencies I am seeing,

I will make some sine signals and see what I get.

hamster_nz:
Took a different approach, and used Excel to see what the change in phase of the fundamental would look like.

That settles it for me. It is just the the low high filter.

Navigation

[0] Message Index

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod