RMS results
Hello,
I have recently received the ADAU1761 evaluation board. One thing I have noticed is that the RMS results using the RMS squared envelope and then the square root function work fine for frequencies above 50Hz. When I try an input frequency of 2Hz using the tone input, the results go up and we end up with a 30% error. I have found that if I can the TC of the envelope down to 5, reults are beeter with a 2% error. The problem with that scenario is the response time of the function. Changing the sampling rate did not affect the results either. Any suggestions will be appreciated.
Thanks.
Audio Signal Processors
SigmaDSP Audio Processors
adau1761
David,
I think that I understand what you're doing, but it would be helpful if you could attach a screen shot or the actual SigmaStudio project.
It seems that your issue is that you are getting an inaccurate RMS measurement when you have a 2 Hz input and a fast attack time (high RMS time constant) set in the RMS table. A 2 Hz signal is varying very slowly (0.5 sec period), so I think that setting a fast time constant will track the level of the slow signal too quickly. The algorithm will be adapting to the changes in the signal level more quickly than you want for this low frequency signal. Because the input is at such a low frequency, a slow time constant will more accurately track the slowly-varying level of the input.
Jerad,
Thanks for the response. I have attached two screen shots. One with the TC set to 120 and one with it set to 10. The results when it is set to 10 is much closer to the expected result based on the peak reading. It sounds like I will just need to have the TC set lower. Is there any other method of obtaining an RMS result that would allow a faster reponse time and still mainitain a good result at 2 Hz? Also, where can I get a copy of the schematic for the ADAU1761 eval board?
Thanks,
Dave
If you are trying to measure signals at 2 Hz, why do you need a faster time constant? Is 2 Hz just the lower limit of what you're measuring?
I think that if you need to measure signals at such a low frequency, then the only way to achieve good results with the rms level detectors is by using a slow time constant. If you consider that the period of a 2 Hz signal is 0.5 seconds, so the time between peaks (both positive and negative) will be 0.25 sec. With a 10 dB/s TC, the detector can only move 2.5 dB in that 0.25 sec, but when set to 120 dB/s the detector can react 30 dB in that same time.
Jerad,
I agree with your thought on measuring the TC and the low frequency signal. The system also needs to look at signals out to 10,000 Hz as well. While having the slower time constant is fine, I was just curious if there was another way to improve the response time. If not, that is fine. By the way, where can I find a copy of the schameatic for the ADAU1761 eval. bd?
Thank you very much,
Dave
The evaluation board schematics are included in the board's datasheet, available at the ADAU1761 page: http://www.analog.com/adau1761
Please log in to reply.