with balanced output DACs like the AK4399, both phases will be at half the supply voltage i.e. 2.5V
have you read the datasheet, understand that the switched capacitor filter output stages are internal CMOS op amps with typical op amp limitations like V offset from manufacturing tolerances? - in fact they are rather poor op amps for DC performance
OK, a quick scan doesn't find a DC offset spec - that's a fail for the datasheet, typical CMOS op amp numbers will be milliVolts, even 10s of mV are possible if the designers only cared about AC spec - all the application circuits are AC coupled
OK, a quick scan doesn't find a DC offset spec - that's a fail for the datasheet, typical CMOS op amp numbers will be milliVolts, even 10s of mV are possible if the designers only cared about AC spec - all the application circuits are AC coupled
Last edited:
2.5V sure, but 2.500V? Some trafos are sensitive to even a few mV of DC and as jcx points out CMOS opamps aren't known for their DC stability, quite the opposite in fact they're the last choice of designers where DC stability is concerned.
Thanks abraxilito, I did think I could see a low pass filter in the diagram even with my limited knowledge.
So I will put 10K resistor in series with the cap and remove all other resistors
Now, do I need a cap on each of the Dac outputs to avoid this DC offset? I have no idea what this is but I do understand the principle of series caps blocking DC, would this mean 4 caps? and would they need to be audio grade?
Stu
So I will put 10K resistor in series with the cap and remove all other resistors
Now, do I need a cap on each of the Dac outputs to avoid this DC offset? I have no idea what this is but I do understand the principle of series caps blocking DC, would this mean 4 caps? and would they need to be audio grade?
Stu
Ah OK, when someone above said there was a DC path (as a good thing), I assumed you understood they were saying the transformer should be applied between output and ground, giving a large offset which would of course saturate the transformer. My misunderstanding.
In terms of offset due to offset from the opamps, I contacted Jensen about this a few years back to find out what one of their transformers could tolerate before distorting. The discrete stage I was using had up to 10mV offset. Jensen told be the transformer would be happily with much more than this. I can't imagine any DAC opamp would produce enough offfset to be a problem.
I know 'cause I've measured it. Not a problem. Also, a 3-5K snubbing resistor is a good idea. You can tune that on an O'scope if you have one. No need for the series resistors.
The Jensen iron is nice, but I like the Cinemag even better. I have a good collection of transforms to use on DAC outputs.
Its going to depend on what your trafo can tolerate. Myself I always include the cap and don't notice anything bad about the sound of a cap, but I'm a heretic. I don't use 'audio grade' caps anywhere. If you decide to include the cap then you don't need 4, only 2 - one per trafo. It doesn't matter which side of the trafo primary winding you include the cap on. I'd think something like 220uF/25V would be in the right ballpark.
Aren't fig.19 and 21 of the datasheet dc coupled ? Only fig. 20 shows ac coupling. The datasheet also says: "The AK4399 has full differential SCF outputs, removing the need for AC coupling capacitors", so the AKM engineers must be pretty confident on the dc offset match.
In any case, if you introduce a capacitor in there, you need to check with a scope how it interacts with the transformer and the dac output impedance, to guarantee flat frequency response.
Hi Stu, as mentioned, the AKMs were designed not to need coupling caps, so the best thing to do to be sure is to measure the + and - outputs. They should be the same, within a millivolt. If so, forget about the caps. If not, we can investigate further.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.