I didn’t know it existed. But don’t get too excited about it. If you look at the data sheet, you’ll find it’s dynamic range is 123dB (even worse than on 24-bit versions of the same product line). This means it’s output is comparable to 21 bit D/A converter. Look at the data sheet again and you’ll find a hint:
- Additional Features: 32-Bit Digital Processing
Now what does this mean? I don’t know. Maybe it does oversampling with 32 bits. This guess is supported by the fact that it’s THD figure is 1/2 of the 24-bit converters of the same product line. Or maybe it’s called 32-bit converter, because it accepts 32-bit input. As I said: I don’t know.
Then you have to remember: 32 bits of this converter are not the same as 32 bits in Cubase. Cubase uses floting-point arithmetic while converter uses fixed-point arithmetic. So there must be conversion somewhere between them. Who does this conversion (Cubase or ASIO driver or audio interface hardware) depends of course on hardware/driver/DAW software combination. It’s very possible that with some combinations audio will be reduced to 24-bits in some point of the chain.
Conclusion: I don’t know if the D/A conversion part is really 32-bit or not. But it doesn’t matter. This is like when companies are selling you 20-megapixel cameras, but “forget” to tell you the lens is capable only to reproduce details which can be captured as well with 5 megapixel camera.