TLV320AIC3106EVM-K: Query related to AGC of the codec.

satheesh kumar32

Part Number: TLV320AIC3106EVM-K
Other Parts Discussed in Thread: TLV320AIC3106

Hello!!

I am working on TLV320AIC3106 EVM, I configured the EVM by executing the following commands sequentially.

w 30 02 00
w 30 07 80
w 30 13 04
w 30 11 0F
w 30 0C 40
w 30 1A D0
w 30 1B 78
w 30 67 C4
w 30 68 CC
w 30 1C BA
w 30 22 18

I am testing the codec by subjecting it with sinusoid of 1KHz.

without AGC in picture I am getting pretty good SINAD,

but once I enable the AGC SINAD deteriorates drastically

I guess the deterioration in SINAD is due to the glitches in time domain.

I will be grateful if someone help me out of this.

over 6 years ago

0 Steve-Wilson over 6 years ago

TI__Mastermind 30220 points

Satheesh,

what is the input level for the AGC tests?

have you adjusted the max gain, hysteresis, or enabled clip stepping?

best regards,
-Steve Wilson

0 satheesh kumar32 over 6 years ago in reply to Steve-Wilson

Expert 1045 points

Hello Steve,

In our experiments input to the codec was varied from 200mV(peak) to 1mv(peak),
The figures in the query were taken when the input was at 200mV(peak)-1KHz tone, the amplitude is maintained at a constant level(200mV-peak)..
We are powering the EVM using external power supply, as in the forum in response to similar query, someone from your side raised doubts about the USB powering.
To best of our understanding hysteresis has a role to play when the input is near noise threshold,
but in this experiment we are sufficiently above the noise threshold. We also enable clip stepping.
We have explore different combination of maximum gain, attack and release time.
Attack and decay time were fine tuned by subjective evaluation, by giving speech as input to codec.

Our concern is that if the codec output is given as input to say a FM modulator and the modulated signal is demodulated using a standard FM de-modulator, the SINAD obtained is very poor,
and that is because of the deteriorated SINAD at the codec output in the presence of AGC.

0 satheesh kumar32 over 6 years ago in reply to satheesh kumar32

Expert 1045 points

A gentle reminder

0 satheesh kumar32 over 6 years ago in reply to satheesh kumar32

Expert 1045 points

A gentle reminder

0 Steve-Wilson over 6 years ago in reply to satheesh kumar32

TI__Mastermind 30220 points

Satheesh,

I apologize for the extended delay in my response.

I was finally able to set this up and I do experience issues with the AGC algorithm,

can you please confirm that the issue is resolved when you Disable the AGC? then we can start addressing the AGC coefficients

best regards
-Steve Wilson

0 satheesh kumar32 over 6 years ago in reply to Steve-Wilson

Expert 1045 points

Hi Steve
When agc is disabled, codec provides expected performance as stated before

Regards
Satheesh Kumar S

0 Steve-Wilson over 6 years ago in reply to satheesh kumar32

TI__Mastermind 30220 points

Satheesh,

Let's adjust your AGC coefficients.

Can you change your Attack and decay times? currently your decay time is 1000x your attack time, lets bring it down to 10x and see how that does.
w 30 67 C4 # attack time = 10ms multiplication factor = 2
w 30 68 C0 # decay time = 250ms multiplication factor = 1

best regards
-Steve Wilson

0 Steve-Wilson over 6 years ago in reply to Steve-Wilson

TI__Mastermind 30220 points

Satheesh,

Have you tried the changes?

Reducing your max gain is the next parameter to change, I was able to reduce the distortion just by reducing the max gain, so it would seem that there may be some overshoot going on.

best regards,
-Steve Wilson

Audio

Audio forum

TLV320AIC3106EVM-K: Query related to AGC of the codec.