Too much THD with MIXER on ADAU1701
Hi Everybody!
My design is for a subwoofer system consisting of a mixer stage mixing the stereo input feeding an equalizer, with LPF further down feeding the compressor/limiter. When tested using AP equipment, the THD goes to the roof. But, when mixer is disabled THD is acceptable. I wonder how can this be solved knowing mixers introduce harmonic noise that is too bad for the overall noise response.
On the other hand, mixing the stereo signal outside the DSP requires only simple hardware (this has been tested and worked well), but that is not an option since the aim is to do everything inside the DSP, and that seems to add more to the overall cost as well.
Is there any way to limit the THD with mixers involed? Attached is the project file from sigma studio and related AP graphs for THD (with and without MIXER).
Thanks a lot in advance for the help!!!
Regards,
Dan
Audio Signal Processors
stereo
mixer
total harmonic distrotion
noise
SigmaDSP Audio Processors
thd
SigmaDSP
adau1701
Show More
Hello Dan,
The huge increase in distortion suggests that the DSP is clipping with the mixer version of your project. You could verify this by connecting one of the -1701's DACs directly to the mixer output in your signal flow and observing the corresponding output pin with an oscilloscope. Actually inside your signal flow the clipping limit is nominally +/- 16, while the DACs and ADCs clip at +/- 1. Thus the clipping is likely occurring at the output DAC after all the processing, or perhaps inside a high-Q filter. How the mixer contributes to this problem is described below.
A two-input mixer set to 0dB can actually amplify by 6 dB when its two inputs are identical or nearly so. To prevent this, set its control to -6 dB, or substitute a Signal Merger which does this automatically. Another cause of unexpected clipping is any residual DC entering your signal flow via the ADCs. If your scope shows asymmetrical clipping, insert a DC Block filter between each ADC and its mixer input.
Best regards,
Bob
Hello Dan,
Taking a closer look at your distortion plots, it appears the THD is very high for both low-power and high-power levels -- so clipping could not be the cause. Thus it's still a mystery why adding a mixer block would dramatically increase the THD. There's nothing in the mixer algorithm to produce distortion. It merely adds two numbers together and scales the result. To prove this, I added a mixer to the circuit shown at A Poor Engineer's Distortion Analyzer. It produced the same THD reading with or without the mixer.
So what could it be? Since you're testing two prototypes with different hardware -- one with an analog mixer external to the -1701, the other mixing within the signal flow -- perhaps a difference between them is causing the dissimilar THD results. If you're measuring THD with the sub-woofer's power amplifier operating, try disabling it and measure directly after the DAC output filter. This would eliminate any ground loop injecting some amplifier supply current into the signal ground.
Best regards,
Bob
Bob,
Spot on! The cause of distortion is somewhere in the hardware. The 1701 EVB board works fine as tested. We've isolated the dsp itself and tested THD and its fine. Thanks for the suggestion.as always you're help is tremendous!
Best regards,
Dan
Please log in to reply.