Adau1701 noise (tone) issues

Hello Sebastiaan,

What you are hearing is what is called "Idle Tones". It is a byproduct of Sigma-Delta converters when they have nothing to do, as in no signal to process. If you insert a low -60db signal it will probably go away. They do change with temperature and device to device. They are not there when there is a signal being processed. It is only when you are sending a low level noise floor into the converters.

There are ways to fight these. Since you are using this as a crossover to a speaker you will have a lot of gain so this is why you are able to hear them so easily. Most applications they are way down into the noise floor and very difficult to hear. So you do need to look into your gain staging to make sure you are not adding more gain then needed. I would try to determine the maximum output you will output from your speakers and set the gain in the DSP so that the max volume of the speaker will be close to digital zero in the DSP. But, in your application I would guess you will still be able to hear them. So on to other techniques.

First determine if it is the ADCs or the DACs? The best way to remove these tones is to add a small amount of DC. This is a little difficult with the ADC since they are current input in this part. I think in your case it will be the DACs so I will move to that topic. If it turns out to be the ADCs then let me know.

For the DACs you simply add in some DC to the output using a DC source and a summing cell in your SigmaStudio project right at the output. If you look at the FFT of the output you will see the tones and as you add a little DC you will see the tones move up in frequency until they go up out-of-band. That will be the DC level to use. This will reduce your headroom slightly but I think that will not be a problem.

I am also assuming you are AC coupling the output of the DACs.

Dave T

Dear Dave

Many thanks for your suggestions. I have a similar (the same?) problem. A distinct hiss / noise at 1.2khz which becomes less noticeable, when turning up the volume (the music covers it). I have the following configuration:

Within Sigma Studio I've implemented the following configuration, follwoing your suggestions:

Did I use the right components? What settings do you suggest for the in the DC element (level and format)? The way I've implemented it I was unable to identify any effect on the noise.

Thank you again for you help or any suggestions

aightgenosse

Hello aightgenosse,

You basically used the right components but there are some thing I cannot tell from looking at the screenshot.

First, I would place a DC Block on the input from the ADC. The ADC will have some offset and this can cause issues with thumps when muting and also if you EQ up the low end, then the DC can get really high.

Then you are adding WAY too much DC. It is set to "1" in 5.23 format which is a 0DBFS signal! I am surprised you are getting anything out. The DC you need on the output will be much smaller. I think you must be using an older version of SigmaStudio since the DC cell does not have a button to choose HEX instead of a fixed point real number. You should be using a hex number for the DC offset.

To set the DC offset, you really should be looking at an FFT of the DSP output. There you would be able to see the idle tones and watch them go up as you add DC. Then you can raise the DC until they are up out of the audio band and that will be your number to use. I would add a small margin since it will be slightly different for each part.

One other thing to check for: Are you using Double Precision version of the crossover algorithm? This can also cause some low level noise and graininess in the low end.

One more question: When you mute the outputs, does the noise go away? (not the mute that is before the crossover)

Then go into the register settings and activate the Mute in the Audio DAC section of the register controls. Does it go away then?

Dave T

Hi Dave

Thank you sooooo much for your help! We've been fiddling with this for over two years and the last issue we're encountering is the hiss. Once we get it under control we can finish our project.

Can I supply some additional info, so you get a clearer picture?

I've just recently updated Sigma Studio to version 4.2. The DC sources are changed and the setup looks like this now:

With which HEX values should I start and how do I increment them?

The crossover is changed to single precision now.

All the mute-checkboxes (before and after the crossover) have zero effect on the hiss.

Since the hardware is in my friends basement I'll be able to check if muting the Audio DAC will have an effect no earlier than next Monday.

Thank you so much & kind Regards

aightgenosse

PS: I've read that the Sure ADAU1701 board can be fussy regarding the power source. Do you think it would be worthwhile using batteries as a power source to eliminate this a possible source of the hiss?

Hello aightgenosse,

I apologize for not proof reading my post more closely. The crossover should be double precision, Single precision is the one that can cause issues with low frequency EQs. single precision would be OK if you had the crossover set to a fairly high frequency like 4-5K or higher but I am guessing you have the crossover point lower.

It is not a bad idea to use a battery to test for a power supply issue. However, you should still have a regulator even with a battery for power. It would be good if you could share your schematic. Is it a switching regulator? What frequency?

Also, I am assuming you are using a PCB that you designed? Is it a 4-layer board with the internal layers as power and ground planes?

If I get a chance today I will go into the lab and experiment with DC values. I have a busy day planned so it might not happen. Start with zero and then start with like 1F then 2F, and see if you hear a difference. Actually, start with something higher like 7FF. That should move the idle tones up out of band. Then if this is the cause then the value can be optimized.

Dave T

Hi Dave

No worries. I'll switch back to the double precision crossover.

We use the following PSU:

http://webcache.googleusercontent.com/search?q=cache:6RIJZjSV9AAJ:store3.sure-electronics.com/ps-sp11115+&cd=5&hl=de&ct=clnk&gl=ch

Then we use the following module to get 3.3v DC out of 24DC:

https://www.aliexpress.com/item/LM2596S-ADJ-DC-DC-Switching-Step-Down-Module-Power-Supply-Converter-12V-24V-New/32763404013.html?spm=a2g0s.9042311.0.0.48c24c4dDqipCM

And finally we use the ADAU1701 board from Sure:

https://www.ebay.com/itm/ADAU1701-Audio-Digital-Signal-Processor-APP-DIY-2in-3out-control-APM2-DSP-board-/112267187788

No need to rush with the HEX values, I won't be able to test them before next Monday.

Thanks a lot & kind Regards

aightgenosse

Hello aightgenosse,

These two power supplies can very well be part of the problem. The switching frequency of the 24V supply is 25kHz and the 3.3V regulator has a switching frequency of 150kHz. Both of these are really low and are in the band for the oversampling Sigma-Delta ADCs. I would look for power supplies that are either linear or switching at frequencies above 1MHz. Then you still have to filter them well.

So do some experiments using lab DC power supplies or batteries.

Dave T

Hi Dave

Thank you again for your helpful suggestions. Should we identify the PSU as the culpable, would something like this (only for the ADAU1701 board) be more suitable?

https://www.audiophonics.fr/en/linear-regulated-psu/linear-power-supply-board-dc-lt1083-25v-to-30v-6a-p-10900.html

Best Regards

aightgenosse