Audio Shield w Teensy 3.6 only produced a click on init

Status
Not open for further replies.
W

weasel

Active member
hey there
tried hooking up the audio board for some fft analysis within my big diy synth today.

i fail to get anything into or out of the audio board though. i bought the pinned version of the 3.6 board, the Rev C of the audio board and soldered pins on the audio board. here's pictures:

IMG_8337.jpgIMG_8336.jpg

i tried several of the hardware test including tonesweep and the karplus guitar synth, all i get on both headphone and line out is a couple milliseconds short click/burst once the test projects post the "setup done" message.

the teensy is working fine with an OLED (the three cables on the lower pins).
the fft example projects runs fine and analysez the sine wave just right, but zero on a connected line in.

teensy is powered via macbook pro usb.
 
Try running the synth guitar example, and then measure the DC voltage on each of the 3 test pads underneath the headphone jack.

The DC voltage is an indication whether the STGL5000 chip initialized properly and turned on its analog circuitry.
 
thank you! all measurements down in that corner give me 0v, not a good look i assume.

i was supposed to measure between VGND and L/R respectively, correct?

IMG_8338.jpg

edit: measuring teensy GND to audioshield L / R testpoints shows around 1.6v each.
 
Safe to assume nothing is connected to pins 18/19? And they are properly soldered connected and not shorted. If that is the problem Paul's test will show voltage.

I ran that sample on T4 with RevD audio and it failed startup - I had 18/19 wired to unpowered i2c display pins on another T4 to sniff i2c, that killed the i2c commands to STGL5000 I assume as it start up mute.

Removed those pins and it worked, and it worked on another T4 with that adapter as well.
 
thanks i just tried (again) to unplug all OLED and other wires, no difference in behaviour. all i get is a short click on booting up, then silence. voltages are shown as described above.
 
weasel said:
edit: measuring teensy GND to audioshield L / R testpoints shows around 1.6v each.
Click to expand...

Yes, this is the correct way to measure. 1.5 to 1.6 volts means the SGTL5000 did turn on its analog circuitry. That is a sure sign the SDA and SCL signals are connected properly, and power is good, and it is generally working (but not getting digital data).

The next step is to look at the I2S signals. The first step is to look at the DC voltage (again with the negative lead on GND) on the LRCLK, BCLK and MCLK pins (pins 23, 9, 11). Again, measure these while the synth guitar example is running. These 3 clocks should measure approx 1.65V, because they are clock signals with ~50% duty cycle. Try measuring on the audio shield side, to check that Teensy's signal really is reaching the audio shield. If you can measure frequency, they should be 44.1 kHz, 2.8 MHz, 11.3 MHz. The frequency mode on most handheld multimeters is limited to about 20-100 kHz max, so don't be surprised if a multimeter can't read them. A logic analyzer or oscilloscope is usually needed to see these signals. But just a quick DC voltage measurement can tell you if the pin is 1.65V or zero or 3.3V (and the near-zero case where the voltage fluctuates when you touch the wires is a sure sign of a disconnected wire).

Again, the analog output voltage confirms your audio shield did initialize. So now it's just a matter of confirming whether it's really getting the 4 I2S signals needed to output (pins 22, 23, 9, 11). It is running and when it receives all 4 of those signals, it very likely will play sound.
 
thanks a lot for the thorough troubleshooting guide, i'm on EU time hence the late response.
will check the data and audio connection w an oscilloscope in a few and report back

just wanna emphasize again that the inital click is very reliably producable. to me as a novice it sounds as if on completion of the startup commands something gets shorted. the clicky burst is actaully pretty similar to a classic "short" sound - i am pretty familiar with those.

i am starting to suspect those damn little dupont pinheaders. iirc the stackablle ones i used for the audioshiel might even have been the cheapest china tier..
a related question on this: i've read that the auddio pinheader on the shield is form compatiblwe with the pc/atx audio frontplate connector, like these cables:
https://www.amazon.com/Alphacool-Ph...ension+cord+pc+internal&qid=1579785591&sr=8-9
but, the way i mounted the audio shield underneath the teensy, i can't really use those, right? unless i would give me even more dupont unreliabilty by stacking two stackablle pinheaders?
or maybe if i remove the extra pins from a 90 degree 2x5 pin thing..? guess htat should work. oh man i just like thinking out loud on a forum, answering myself.

also, i know this is not the place. but man you are absolutely killing it with your products mr. stoffregen. although i wish more people would contribute to the audio patching environment, i am a heavy axoloti user, but hey i am sure that is on the list for whenever you get tired of dominating the market with your latest hardware.
 
Last edited:
yes i triedd both headdphone and line out. photos of the board are posted above?

i just measured 3v, 1.5v, 1.5v DC on LRCLK, BCLK and MCLK pins (pins 23, 9, 11)
strugling to find the right trigger settings on my scope to get reliable frequency readouts.

edit:
9 is at 2.8MHz,11 at 11 Mhz, 23 my scope doesn't show a frequency but looking at the graph and horizontal resolution it should be around 44k

whats next? i gguess we are narrowing it down to the actual DAC not working?

edit 2: clocks at p9 and 11 pulse between 0v and 3ish v as expected, but pin 23 pulses betweenn what looks like 3v and 3.3v, llooks like a much smaller amplitude with a big DC offfset.
i realized theres a 3v pin right next to p23 but the sollder points are definately clean and far from touching.

edit 3: measuring pin 23 on the teensy directly gives the correct 0-3v pulse amplitude. since my solder point actually look ok i assume there's something broken on the audio shield pcb?
yeah there seems to be a short between pin23 and the adjactant 3v pin ("24") on the audio shield. again, soldering looks clean but i will resolder now.
IMG_8341.jpg
 
Last edited:
for the record, i am ggiving up on this, after sucking out the first two pins i realized theres no way i will get this 14pin-header off the pcb again. short between pin23 and 3v remains. funny enough, when the guitar runs and i measure the resistance with a multimeter, the "short beep" sometimes changes between p23/3v and p23/gnd.

i'm putting some pinheaders on a teensy 4 and respective audio shield now, see if i get somewhere with that one haha.

edit. lol. awesome strumming pattern.
 
Last edited:
Must admit, I'm stumped too. We do test every audio shield here, which does do a test on both its ADC and DAC sections. How it could create 1.6V but not work is a mystery to me.

I've asked Robin to follow up. We'll send you another audio shield. All I ask is that you'll let us know here on this thread if it works or does the same thing. I'd really like to get to the bottom of this mystery!
 
hey guys
i really appreciate the effort, no need to send me a replacement though, i wrote it off under "manufacturing reject".. if you guys really wanna find the cause of it i'm happy to do further measurements or send it back to you. will reply to the email robin.
but i'm pretty sure that either i broke the pcb layers while soldering, or it was broken already by that random one off manufacturing mistake.
 
I would like to understand what went wrong.

We do a bed-of-nails test on every PCB. That test does have the STGL5000 play sine waves, which are captured by the test fixture and run through a FFT to make sure they're really the correct frequency & phase (or very close). So every audio shield should be good when shipped. If a bad one managed to get shipped somehow, I really want to learn what happened and make sure it never happens again.

It's also possible it could have been damaged in shipping or while soldering. But the fact that you got the DC voltage on the analog outputs shows the board was partially working, which is a strange result for the type of damage that could normally happen. Very mysterious...
 
Status
Not open for further replies.
Back
Top