TAS5711: Biquad calculation

nic siebenborn

Part Number: TAS5711

Hi folks,

I am developing with the TAS5711 and I need to fill the coefficiense of the biquads. As my soundguy does not know Biquads I wrote a tool to convert from any filter entered into Biquad (high-pass, EQ etc.).

For my formulas I refered to this post and therefore the formulas from this document https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/6/0027.Biquad-Filters-Application-Note.pdf

Unfortunatly it does not work. For example when entering a high pass with fc=500Hz the high sounds are also significantly more silent.

My Question therefore is: is there another document for the biquad calculation? The TI tools do not help me as they eather do not support Tas5711 or just give me some numbers but I need the formulas.

thanks and regards
Nic

over 7 years ago

0 Shawn Zheng over 7 years ago

TI__Mastermind 24835 points

Hi Nic,
The formula in that document should work. In that post, you could find a excel file which can be used to caluculate the notch filter taps.
e2e.ti.com/.../649805
You can compare it to the description in section 1.3 in the that document. I believe it helps your understanding.
Best regards,
Shawn Zheng

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

Thanks for the confirmation
but still the question remains how it comes that all frequencys are more silent when using an high pass.
kind regards
Nic

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

regarding the excel sheet: In Cell P5 it says N6+1 (N6 contains a) but in the document on Page 7 it says b0=a (so the +1 is missing)
How comes?

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,
They have the different definition. The data in the excel is the final efficients, while the 'a' in the document is not.

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

Hi,
where can I find the definitions? I can add the +1 for the notch filter but I also need the definitions for the other filters, so I can implement them.

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Nic,
Again the document shows the formula for the filter co-efficients calculaltion. And the excel file actually shows the same calculation as the document basically. It's a example that how to caluculate the co-efficients by the formula from the document. The definition of the symbol 'a0'/'b0'... are not the same between them. 'a0' in the excel file is the final co-efficients, but the one in the document is not.
Essentially they are the same thing. Please spend some time correlating them two.

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

Hi,
I can not find the documentation on which the formulas in the excel sheet are based.
I need formulas to calculate from the filter propertys (like bandwitch etc) the correct biquad values I can send via I2C.
Can you help me find them ?
In the Appnote there is no word about the +1 the same is valid for the TAS datasheet.

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,

Let's take the example of the Notch filter. Please find the formula in the section 1.3 in the attached document.

4162.Biquad Filters Application Note.pdf

The formula in the excel shows the same info as this document.

Best regards,

Shawn Zheng

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

Hi Shawn,

I have implemented the filters. Now I want to check them and wanted to calculate the example at the end of the document. I miss the information, which sample frequency TI has used for the EQ example calculation of the appendix.

kind regards
Nic

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,
This expample in the appendix in that document is based on 48kHz sampling rate calculation. But please neglect this example, becasue it only for AIC Codec device. There is some difference in data format between TAS5711 and Codec. So the final result for the register values are different.
Best regards,
Shawn Zheng

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

Hi,

due to the lack of possebilities to verify my results I just started to meassure with my oszi.

I get results which are ok for my BTL speakers. But for my PBTL speaker, I do not see a difference in amplitude.
Can you tell me how I set the filter for a PBTL device? I set the registers (PWM etc) according to the datasheet, but in there is not written how to set the filter parameters for PBTL.

Regards Nic

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Please refer to this post in wiki for PBTL configuration.
e2e.ti.com/.../2833.tas57112731-configuration-on-pbtl-mode

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

I calculated a Butterworth 1 HP with Fc=50Hz and Fs=44100Hz
With the formula from the app-note I get
k=88199,627
Wc=314,16
and the filtercoef
a1= -0,9929015
the TI GDE for TAS 5711 gives the content of register 0x2a with
007F951D03806AE300000000007F2A3B00000000
this means a1 =0x007F2A3B
divided by 8388608 it is 0.9934763

can you please tell me how this difference appears?

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,

There is a typo in the document for the 'a1' calculation in 1st order Butterworth Filter. It should be a1=(k-wc)/(k+wc), not (wc-k)/(k+wc).

Best regards,

Shawn Zheng

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

Hi Shawn,

I have a simmular Issue with the EQ Equation and I broke it down so I can say that my calculation for d is different. Can you tell me the correct formular?

kind regards
Nic

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,
Could you please help to clearify your question? Which type of the filter are you refering to?

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

I mean the Equalizer filter 1.2 in the document and my only difference to the Eva Kit is in variable b1 so it must eather be the equation for b1 or for d

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,

I think there is no problem in the formula for 'd' calculation in section 1.2. Please check the attached excel file, which could help your understanding.

EQ.xlsx

Best regards,

Shawn Zheng

0 nic siebenborn over 7 years ago in reply to Shawn Zheng

Prodigy 160 points

In the formula of the document it says b1=d*(1+a) and in the Excel file it says =d*(1-a)
so there is a difference
Is the Excel file correct?
Can you give me Excel files for the other filters like variable Q?

0 Shawn Zheng over 7 years ago in reply to nic siebenborn

TI__Mastermind 24835 points

Hi Nic,
Do you mind communicating by email? Could you please help to close this question and let's communicate by email? My email address is shawn(dash)zheng(at)ti(dot)com.

Audio

Audio forum

TAS5711: Biquad calculation