Hi all,
this will be my first post here and I'm interested to see what all you other people here think about this schematic and the project itself.
Short background: My band is running out of analog outputs on our digital mixer. Of course we could buy an additional stagebox, but this is unnessesary expensive at the moment. The mixer has AES digital output, but all off-the-shelf converters are way to expensive for our budget. I browsed the web and found plenty (affordable) ICs capable of converting AES to I2S or similar, which then could be converted to analog.
Application: Extension of outputs on a digital mixer for live applications. No High-End studio use. Should be way more affordable then a complete stagebox --> BOM-Cost < 25$
Theory of Operation: The AES input is isolated by the transformer and converted to I2S by the DIR9001. I2S is converted to analog by the PCM1789. The output is attenuated to +4dBu and low-pass filtered (PCM1789 is delta-sigma) by the OPA1612. There is a digital power supply (3.3V) for the DIR9001 and the digital part of PCM1789 and an analog supply for the OPA1612 and the analog part of PCM1789.
Since I haven't designed much audio-electronics up to now, I'm interested in your opinions and improvemts you would do the this circuit.
I attached the schematic, if you want to directly check the KiCad project, check out my github repository: https://github.com/sschwaab/aes-dac
Thanks for all thoughts!
this will be my first post here and I'm interested to see what all you other people here think about this schematic and the project itself.
Short background: My band is running out of analog outputs on our digital mixer. Of course we could buy an additional stagebox, but this is unnessesary expensive at the moment. The mixer has AES digital output, but all off-the-shelf converters are way to expensive for our budget. I browsed the web and found plenty (affordable) ICs capable of converting AES to I2S or similar, which then could be converted to analog.
Application: Extension of outputs on a digital mixer for live applications. No High-End studio use. Should be way more affordable then a complete stagebox --> BOM-Cost < 25$
Theory of Operation: The AES input is isolated by the transformer and converted to I2S by the DIR9001. I2S is converted to analog by the PCM1789. The output is attenuated to +4dBu and low-pass filtered (PCM1789 is delta-sigma) by the OPA1612. There is a digital power supply (3.3V) for the DIR9001 and the digital part of PCM1789 and an analog supply for the OPA1612 and the analog part of PCM1789.
Since I haven't designed much audio-electronics up to now, I'm interested in your opinions and improvemts you would do the this circuit.
I attached the schematic, if you want to directly check the KiCad project, check out my github repository: https://github.com/sschwaab/aes-dac
Thanks for all thoughts!
Attachments
Keep in mind that the nominal level of a mixing console is typically some 20 dB below its maximum level. For an analogue mixing console, that would be 4 dBu nominal with a 24 dBu clipping level, for a digital mixing console, -20 dBFS nominal level.
I don't know what capacitors you intend to use, but large-value ceramic capacitors are usually class 2. Those distort a lot, are microphonic and can unintendedly act as piezo tweeters. Class 1 ceramic capacitors such as NP0 a.k.a. C0G have none of those drawbacks, but are only available in relatively small values.
I don't know what capacitors you intend to use, but large-value ceramic capacitors are usually class 2. Those distort a lot, are microphonic and can unintendedly act as piezo tweeters. Class 1 ceramic capacitors such as NP0 a.k.a. C0G have none of those drawbacks, but are only available in relatively small values.
Thanks for your input! The analog output should be connected to a PA power amp. I searched the web for some datasheets and most specified ~5 dBu as the level required for maximum output. That was the reason why I went with +4dBu.
As for the caps: For the filter caps in the OPA1612 I planned to use film caps (would C0Gs be equally good?). For the 22u output caps I think I have to use (bipolar) aluminum electrolytics because of the large capacity required.
For decoupling I thought about X7R MLCCs, would it be better to use (where possible) C0Gs?
Thanks!
As for the caps: For the filter caps in the OPA1612 I planned to use film caps (would C0Gs be equally good?). For the 22u output caps I think I have to use (bipolar) aluminum electrolytics because of the large capacity required.
For decoupling I thought about X7R MLCCs, would it be better to use (where possible) C0Gs?
Thanks!
C45 can better be C0G than X7R, I'm not sure about the rest. Is the clock coming from the AES receiver clean enough for the DAC?
Good catch, C45 as C0G sounds reasonable.
The DIR9001 has an internal PLL with jitter of 100 ps max.
Do you know any rules of thumb which jitter is normally considered acceptable?
The DIR9001 has an internal PLL with jitter of 100 ps max.
Do you know any rules of thumb which jitter is normally considered acceptable?