Just noise with Teensy 3.6 and Audio Adaptor Board

Status
Not open for further replies.
Here it is:
IMG_2249.jpg
 
I have checked and double checked every single connection.

+ 3,3 V is connected to a pololu regulator (3,3 V)
GND is connected with GND of the regulator and the teensy

All other connections are as follows:

Code:
LRCLK pin 23
TX    pin 22
SCL   pin 19
SDA   pin 18
VOL   pin 15
SCLK  pin 14
RX    pin 13
MISO  pin 12 
MCLK  pin 11
SDCS  pin 10
BCLK  pin 9
MOSI  pin 7
MEMCS pin 6
 
I have figured out that the wires are the problem. I think they are too long. Sometimes the board plays the test files until I touch the wires.

Peter
 
I had a similar problem with a T3.2 and an audio adaptor connected by wires that were about 3 inches long. It seems that there can be some clock skew and cross coupling in those wires, especially without a good low inductance ground between the two boards. This causes the clock and data to get out of sync, resulting in noise. Adding some copper foil under the wires and separating each one cured the random rude sounds. Kapton tape holds the wires to the copper foil, which is grounded in multiple places. Other sound anomalies were cured by replacing the cheap SD card with a true Class 10 card. The project has worked perfectly ever since. Pictures are before and after the added ground. BarryBox_4_x.jpgBarryBox_5_x.jpg
 
I have added the pullup resistors as recommended, and with them the audio board works better. But sometimes I still get the noise. I think I will solder the audio board on the top of the teensy.
 
Maybe small series resistors on the MCLK and BCLK outputs could also help?

Same sort of trouble we've seen many times with OctoWS2811... before PJRC started selling the board with 100 ohm resistors and connectors for using CAT5 or CAT6 cable. The fast edges of the clocks and long wires can cause a lot of crosstalk.
 
Please post photos of how you've actually connected the wires. Show (don't tell) how you've really connected everything.
 
So this is the actual Setup:

20170107_231403.jpg

Problem seems to be Pin 11 - MCLK.
Adding a 1K Resistor improved the stability a lot. (chaos when touching the resistor pins - same as before)
Im sorry I could not find the 10K Resistors. It stopped working with a 130K Resistor.

then I bridged Pin 11 from Teensy and Audio Board with a piece of a Guinness can....
(ok... in fact, i used one of the wires of a resistor.... i'm still not in office, so i have to improvise)

problems were gone then.
 
MCLK Fixed for me

So this is the actual Setup:

View attachment 9331

Problem seems to be Pin 11 - MCLK.
Adding a 1K Resistor improved the stability a lot. (chaos when touching the resistor pins - same as before)
Im sorry I could not find the 10K Resistors. It stopped working with a 130K Resistor.

then I bridged Pin 11 from Teensy and Audio Board with a piece of a Guinness can....
(ok... in fact, i used one of the wires of a resistor.... i'm still not in office, so i have to improvise)

problems were gone then.


Made an account just to say that the MCLK fix worked for me. Sounded raspy and like the sample rate was constantly changing. Adding the 1K on MCLK fixed it entirely. I also have pull ups (1K) on the I2C bus as well, but I'm not sure if they are doing anything at the moment.

Thanks!
 
I struggled with this too. I used 300mm jumper wires that I had laying around with a teensy 3.2. and an audio adaptor perfectly fine for ages. As I was now testing a Teensy 3.6 I got very odd crackling, and general noise as well. I could faintly hear the actual audio in the background though.

Adding the 1K on MCLK totally fixed it.
 
The BCLK and MCLK signals are RF @ 2.82 and 11.29MHz. It should be self-understanding to treat such signals in an appropriate way, keeping the wiring as short as possible to prevent parasitic inductance and to reduce the EM stray field. 300mm jumper wire is definitively not state of the art, and though a series resistor might cure several symptoms by damping the inductive effects of this wild wiring, it is not the optimal way to go since other problems like phase jitter might arise. Even when only prototyping a project, one should always do it already with a CE and/or FCC certification of the final commercial product in mind.
 
Using short wires is the best way.

But I can tell you, PJRC will soon make a revision to the audio shield to add a 100 ohm resistor to the MCLK signal.
 
noise t3.6 and audio adapter board

Using short wires is the best way.

But I can tell you, PJRC will soon make a revision to the audio shield to add a 100 ohm resistor to the MCLK signal.

I have the same noise issue I think - hissy/whistling digital noise on input, with an audio shield stacked directly on top of a T3.6. Again, on a 3.2 it works ok.

I have a rev C audio shield - does this have the MCK resistor?

TIA.
 
I have a Devin Townsend Ocean Machine Delay & Reverb "Pedal", which can (among other things) record 1 minute 21 seconds of looped audio that you can play with and record (mix) into. Should I do a quick tear down and see what it uses for memory?

It was manufactured by Moor and designed in Shen Zen, so I have no idea what to expect inside as they make a lot of pedals, amps, and audio gear.
 
I picked up a Teensy 4.1 & audio board rev D to begin experimenting with FFT. But I jumped ahead and already soldered the header pins to the Teensy, so now I'm stuck using jumper wires. But I had an idea...

Can shielded wire be used, like a section chopped out of an old RCA cable..? And then ground only one end of the shields (to avoid ground loops). I'm assuming twisted pair would be useless, since the connections aren't differential.

Which connections are the highest bandwidth and therefore (?) most sensitive to crosstalk and EMI..? I'm an old broadcast radio guy, but new(ish) to these little electronical widgets the kids are all playing with nowdays. :p
 
Yes, wire can be used to connect those signals the needed locations. Shielded wire isn't necessary. Ordinary wire is fine if you keep the length under ~4 inches.

MCLK is the most sensitive signal. Rev C and Rev D audio shields have a 100 ohm resistor which allows the shield to work when a normal wire is used. The Rev B and earlier shields do have issues with a long wire on MCLK with Teensy 4.0 & 4.1, and to some degree also Teensy 3.6.
 
Status
Not open for further replies.
Back
Top