TAS2781EVM: 16bit i2s Sine wave 50hz missing outp/outn

Hi Jose, 

So the issue is the speaker output, and IV-sense TX are dropping for some time and then coming back up. Can you monitor the interrupts using the "IRQ" button in PPC3? I would like to see if the device is reporting any issue

Yes, IV-sense from SDOUT is in 2's complement

Jose Paza1 said:

with the "mono left channel" selected, does it mean that it will disregard the data on the R channel?

Yes, based on your settings left channel is slot 0, slots are 16 bits, and the frame starts on the low->high transition of FSYNC. this would mean the device is only recording the first 16 bits at the start of each frame and ignoring the rest. 

How does your output voltage waveform look? or the IV-sense data? is it a 50Hz sine wave during the period where it is playing? from the screenshot it is hard to tell, but is it otherwise functional besides the periods when the audio is dropping? I am wondering if there is some issue with your audio format.

Regards,
Arthur

Hi Art,

thanks for the clarification about the mono left and the 2's complement on the IV. at this moment I need to resolve the drop the the output drive before processing the IV data. The sine wave seems to be at 100HZ when being transmitted over the i2s interface.. I will make some adjustment to trick the output 50HZ. 

I have captured when the device IRQ line right about when the output drop. looks like the device is detecting some i2s error. I couldn't workout the errors as I monitor the i2s lines as well and it seem like the signals are perfectly square. Do you think I need to reduce the frame rate to mitigate such issues?. at the moment my setup is just connecting wires from micro to the 2781 and I'm suspecting that the invalid error in the i2s is dues to this. but the waveform capture says otherwise.

BTW the L/R channel data is only use when the "slot Select Config" is in stereo? and the Outp and outn have their own load?.

Hi Jose, 

I agree that your clocks look too be exactly 48kHz and 32 SBCLKS/FSYNC.

It looks like your scope is showing some digitized output of the clocks. in reality you may be getting some inductance/reflections on the clocks (SBCLK in particular). can you try using a more solid electrical connection such as twisted pair cables with one ground connection per signal?

We have seen similar issue like this before when using these fly-wires, and if there was something fundamentally wrong with your clock frequencies/ratios then the device would never be functional at all, this sort of instability makes me think it is a SI issue. 

Regards,
Arthur

Hi Art,

I have directly plug the micro development board to the EVM 2781 but the dropout didnt improve.

I believe this can be improve on the board that im going to start. Its also noted that I have switched the IO vdd to 3.3V to match the micro, that would have made the rise times less ideal.

I will be developing this board that about to make base on TAS6582-Q1 for more output current.

I have put a EVM TAS658? and the software to TI and I havent heard back. could you please follow up to expedite the request?.

once again thanks for the assistance..

...

Jose

Hi Jose, 

I have raised your request to the marketing team to grant access ASAP

are you able to capture an analog waveform of the Clocks? Can we see if there is some obvious distortion?

Regards,

Arthur

interesting update just found out now... if I set the ppc3 volume to >-3db with amplifier level 11dbV and full 16bit swing in the I2S data from thye micro, the dropout disappears. for higher adjustment of the volume and amplifier theres is no dropout at all..

I was planning to use these 2 controls for current adjustment..hmmmm that just reduces the range of current adjustment if that is the case..

Hi Jose, 

that is unexpected, can you save the PPC3 file and share it with me to check?

Regards,
Arthur

see attach

Hi Jose,

I do not see the attachment. can you try attaching again, perhaps use a different browser? E2E can have issue with attachments using some browsers.

Regards,

Arthur 

TAS2781_2Ohm_nodrop-vol0db_modsADV-TI.zip

ok I have zip the file.. and it seem to have attached..

thank you Jose,

I Will check it today

Regards,
Arthur

Hi Art,

thanks..

BTW I have the current reading from the 2781 with no dropout.. the waveform is quite what we expect and it is perfect for our application, except the step seen on close to zero which is a minor thing. Is there a possibility that the distortion can be reduced? it will just increase our harmonic content reading while current charging the switchgear product which this will be used. Just wondering if it is a setting or the output LC circuit I need to tune.

jose

Hi Jose, 

this is still an unexpected behavior. this device does not require any LC filter for audio quality. the only time we would recommend a filter is when EMI becomes a concern. 

Apologies I did not get to setup the EVM yesterday with your clocks and PPC3 File. I will be able to check today. I do see that your ppc3 file is in "Tuning Mode" is that what you are using to get these current readings? I am wondering if we can put the device into "ROM Mode" which will disable all DSP processing and could be a clue to where these dropouts are coming from?

Regards,
Arthur

Hi Jose, 

Just to confirm, this is a current sent waveform, correct?

Also, what is the load being used for this test. this looks extremely similar to the waveform of a sinewave through a triac. my point is this seems like an issue with the load. this could be verified if you intead record the V-sense signal. is the Vsense a clean sine?

Regards,
Arthur

also, I checked your PPC3 file sing the same I2S timings, and was able to get a clean I-sense recording. 

Hi Art,

yes this is the current waveform. The load I have used is 2.2ohm high power resistor and inline is a pie LC network. The vsense is not a clean sine wave, it is sine wave super impose to a PWM.

we dont have a triac as a load.

I always switch to ROM mode. The PPC3 soft3ware whenever you save will always save default to tuning mode for some reason (issue). I always configure my EVM in ROM mode.

could you please send thru your current reading waveforms? btw I will be away for couple of weeks starting tomorrow. ty

Just bypass the LC pie network..

current waveform improve a lot.. will be revisiting this filter or maybe as you mention doesnt require a filter at all..