Might have fried the output of Axo


#1

Hello!
I was going to try a new patch and in a rush of enthusiasm I forgot to disconnect the Axo from the USB power connector. There was a buzzing sound as i plugged in a TRS into the input and I might have touched the GPIO pins closest to the three glued caps in the corner.

Now there's only a weak digital noise on one channel and a very weak output on the other one.
The emulator shows signals both on the input and output correctly. : /
However, the headphone output works as it should thank god!

How are the headphone outputs and main outs connected to each other software wise?


Strange bug with flashing red led and low voltage report
Connecting audio output to an amp with shared power supply (same Ground)
SPI/I2C OLED display
#2

you have remembered to split the L/R output channels?
can you try to just send the output to LEFT (rather than both the left and right) and see if you are still getting no/distorted output. then try the RIGHT on its own. (i.e. audio/out left)
( I quite often forget this, when I'm unplugging and plugging in boards :smile: )

Ive just dug thru the code... so this needs verifying with @johannes .

Axoloti has two output buffers for left and right, these are 'delivered' to the ADAU1961 codec.
as far as i can tell, its then the ADAU1961 internal mixer that then outputs to the TRS outputs and the headphone.
i.e. I don't think its possible to send different data for the headphone vs the main outputs, its more a mixing function of the codec

hey while I'm here though, a small question for @johannes , am i correct that on the data sheet for the ADAU1961, you can also take (analog) input directly thru to the output via the mixer , p33 has L/R and aux inputs on the mixer.
also is mono output connected somewhere (MONOOUT)

EDIT: btw, just be careful, if you are planning to drive something from the headphone output, Johannes has been very clear the headphone output should only be used for headphones, and can cause damage if you try to drive speakers etc.
again. again Im sure he can supply more info.


#3

I don't remember reading that; can you tell me what is allowed and what not?
so far I always used the mini-jack out as line-in for my sndcard . are you saying that is wrong?
Also, mini-jack is not the same as headphones-out (actually, my headphones are all 1/4" ).


#4

it is designed for headphone output, it was documented in hardware topic

sorry i don't have more details... just remembering being told not to use it for anything other than headphones... perhaps @johannes can supply more details, and possible risks.


#5

I don't know much about electronics (so dont take advice), but as far I know it is fine use the minijack out as a line-out. Many speakers have a build in amp (including computer speakers) and also use line level in, so simply saying 'don't connect to speakers' might be inaccurate.
Connecting speakers directly could cause damage, but it would be stupid to do that anyway. just dont use low-impedance equipment (such as speakers)


#6

According to @johannes himself:

The headphone output is biased at 1.65V. That means its ground connection is not at ground. Connecting headphone out to other gear will short circuit this bias, result in U3 (the audio ADC/DAC) getting hot, and probably strongly degraded output quality. It does not damage the ADC/DAC chip as far as I have seen.

You have to use a Di-box with ground lift option to get normal levels etc. Since i damaged my Axoloti's audio out driver (IC4) already :grin: i use headphone out with this one
Sounds nice.

Johannes also said capacitor mod implemented to headphones out is possible.


#7

That's right, there is only one stereo pair of ADC/DAC.

Yes, analog in can be mixed into the output without passing through ADC/DAC. Could be used for true zero-latency monitoring, dry/wet mixing, or analog bypass.

MONOOUT is used to generate the capless headphones virtual ground.

Using the minijack as line out is fine ONLY if the device connected has its own (isolated) power supply, and is not connected to the computer directly or indirectly.
So, connecting powered speakers is fine, but not if they're powered from USB or have another input connected to a pc soundcard or pc line-out.


Direct analogue path?
#8

Thanks for all your replies!

I'll be connecting my Axo to the line input of my Korg KM402, the Axoloti will probably be powered from the computer or a USB battery pack.
The mixer is connected to the computer from my Firewire soundcard.

Ok so I'll check out a cap modification to get rid of the DC bias then..


#9

Hello!
Did some quick measurements with my Caltec multimeter since I was a little bit confused about the whole DC offset on the headphone output subject..
These are the values I produced from measuring first on the headphone output, then on the big TRS output, first with no signal, then with a sinus tone of approx. 4kHz from the axoloti. The tone is straight from the oscillator without any processing.

(quiet)
Right headphone output: 9mV AC, -0.1mV DC
Left headphone output: 7mV AC, 0.3mV DC

Test tone:
RCH: 296mV AC, 0.5mV DC
LCH: 295mV AC, 1 mV DC

Big TRS output:
(quiet)
Right (Ring) output: -22mV AC, 7mV DC ?!
Left (Tip) output: 645 mV AC 7mV DC
Test tone:
RCH: -21mV AC, 7mV DC
LCH: 300mV AC, 292mV DC <- double checked this, the DC component seems to be linearly proportionate to the frequency (lower when frequency is lower and up to 600mV at higher frequencies).

Another thing, a headphone amp is supposed to drive a relatively low-impedance load (mine are approx. 85 ohms), and the line input of my mixer is in the order of 10k, so I don't see why it would be bad for the headphone output to be connected to the line input..
I totally would get it if there was a case of a 1.65V DC bias tho..


#11

That's probably measured using headphone "ground" as reference which is not ground but has a 1.65V DC bias towards general ground.

This topology was chosen to achieve high quality (capacitor-less) headphone output.


Why the DC bias on headphone out?
#12

facepalm Yes.
Note to self, don't try to do smart stuff in the middle of the night when you should be sleeping.. Yet again.

OK I'll start a new topic on capacitive decoupling for all the main output-burners out here! ^_^


#13

Definitely something similar happened to my axo core. Very weak output with a lot of noise.
Right output is almost dead, left output has more signal but also more noise.

Axo was connected to an amp that was suddenly disconnected and reconnected,
in that moment a burst occurred and the main output was gone...

However, the headphone output seems to be working without noise.
Summarizing I'm very very sad..and since I got a bit lost in the electronics in this thread,

So what is the cheapest and safer way to use the headphone output as a line out?
Help, please.


#14

@ndlopes
Axoloti Hardware

3.5mm (1/8") mini-jack is a headphone output. Use only for headphones (headphone jack is electrically biased at 1.65V.)


#15

So apparently IC4 (the line output driver) broke down due to a voltage spike. This causes IC3 (just behind the line out socket) to overheat (be careful not to burn your finger when you check this!), dropping the analog supply voltage and then the audio chip would not function anymore (but that seems not to be the case on your board).

This mod would remove power to IC4 and remove the DC bias from the headphone out and couple that into the line out socket:

step 1 : cut the power to IC4

step 2 : cut three tracks on the bottom side.
(blue are copper tracks on the bottom side, red are copper tracks on the top side)

step 3 : connect one wire and two capacitors:
For the capacitors I'd suggest 10 microfarad, 16V electrolytic types. More voltage rating does not hurt. 22uF or 47uF would also be good values, it only makes a difference for low-impedance loads.

step 4 : add two resistors to bias the capacitors properly
I suggest 10kiloOhm resistors, but anything between 2.2kOhm to 22kOhm would do the job fine.

This mod bypasses the ground loop break circuit. Ground loop noises are most likely when you power Axoloti Core by USB from the same PC that is also involved in the audio path somehow (like soundcard to the same mixerdesk...).

BTW: the 'S', 'T' and 'R' markings mean sleeve, tip and ring. The top row goes straight to the headphone, line-out and line-in jacks, the bottom row connects to the circuitry. This is to allow circuit hacks, like re-purposing the 1/4" jacks to footswitch or footpedal inputs, breaking out to dual mono jack inputs and outputs, or adding a relais for true bypass...


Control transfer failed: -7 - need a little bit of help
#16

Hey bro,sorry for your loss. Till be using one of these:


Its cheap and will do the trick.
I tried out a few capacitor based configurations but without satisfying results.
Good luck!


#17

My axoloti lives again :smiley:


#18

Alright!what method did u use?


#19

The hack proposed by @johannes.
Used 10k resistors and 1mF 35v Caps.
After all I have a big electronics store in my neighborhood.
Now I've a bit of digital noise (probably due to the USB connection) and it seems that the output is a bit lower...
But completely useful and stable. :smile:


#20

Hey @johannes, how'bout if one is feeling friskey and would like to attempt replacing the headphone amp IC? It seems this component (TPA6132A2) is available from a retailer near me..


#21

QFNs can be tricky. With a hot air soldering station and flux, this can work. Avoid a lot of solder on the center pad - if there is too much there, the component will float on the center pad, not making connections to the signal pads.