A new Teensy with built-in audio..?

Lately I've been working with a Teensy 4.1 and the audio shield Rev D, and it got me thinking... Have the powers that be ever considered releasing a new version of Teensy with the SGTL5000 audio system built into the board..? The 4.1 might have enough room if some features are removed. For instance, the ethernet and USB host could be removed to make room, or the extra flash pads on the bottom of the board...
 
(even if people over 16y old cannot hear anymore the difference)
After 25 years of various careers as a roadie, recording studio technician, broadcast radio engineer, and general audio enthusiast...I hear *everything*, lol. Or I used to. When I was younger my upper hearing was past 20khz, and hearing tests pegged some parts of my midrange sensitivity down to 2db (two). The audiologist did parts of the test multiple times because she thought I was somehow cheating. But middle age and too many raves have taken some of it away from me.
Thanks for the link to the other thread. Those hifi boards look interesting. Do any other boards/chips out there have anything like the PJRC audio design tool..?
 
IMHO, a super performing MCU, as we have with T4.1, would deserve the best performing audio Codec on the market (even if people over 16y old cannot hear anymore the difference).

Two issues with those chips are their cost and power consumption, not just of the codec chip but also the support circuitry needed to achieve those specs.

The really tough part of that support circuitry is power, and especially grounding. Even with noise in the -85dB range with SGTL5000, we regularly hear reports on this forum where people are experiencing ground loop issues which they can actually hear, which I'm guessing means noise in the -70dB to -50dB range (in the final result reaching their ears as a practical matter of how the board is actually, even if that noise technically didn't come from the codec chip). Even if when those sorts of problems can't be heard, I'm pretty sure they exist but are likely in the -90dB to -70dB range.

I have many times considered making a high end audio shield. In fact, there was one designed years ago with a high end Cirrus chip and its own power supply with the I2S and I2C signals isolated (I believe those isolation chips use capacitive coupling internally). The main problem is that all adds up to exorbitant cost. As I recall, that PCB was pretty large too.

The commercial reality of today's microcontroller and SBC dev board market is a very competitive landscape populated with many low cost products like Raspberry Pi and ESP32. Teensy occupies just a tiny niche in that market, already more expensive than many of the popular competing boards. We have to keep prices reasonable to stay in this market.

While it's fun to dream of a product which could offer -120dB level noise+THD performance and come with the power supply & isolation needed to actually achieve it in practice, the reality is such a large and expensive board would have very low sales. PJRC just can't put the financial investment into that sort of very expensive high-end audio product.

But like the CS42448 boards, I might do more on a DIY design. Maybe...
 
High quality audio codecs are benificial for non-music project too! Like ultrasoon, active noise cancelation and hydrophone projects.
 
Teensy occupies just a tiny niche in that market, already more expensive than many of the popular competing boards. We have to keep prices reasonable to stay in this market.
That's why I stick to additional audio boards. Also as my applications is very special (8-chan data logger, 96-192 kHz, 18-20 bit, ADC only: I use 2 TI TLV320ADC5150)
 
Hallo friends :)

Yes that would be nice. I don't need ethernet for my synth. But I would leave the USB host connection.
 
The VS1053b (MP3/AAC/WAV decoder) is a nice Audiochip, too.
As it can be used for PCM, it can be used for the audio-lib (+ compressed audio optional).
It would be needed to write a "driver" for the audio-lib - I don't think it would be too complicated. But it uses SPI.
 
I think some people would appreciate a Teensy+A. However, there are many configurations that someone might need. It wouldn't address all the different configurations as easily as a separate module does.

My modules are all but ready to release, but I just don't really have a great distribution model in place to sell them. With some encouragement and knowledge that I'm not spinning my tires i would sell them and release the drivers to the community. I've just spent a ton of time on it and I'm frankly just scared of getting burnt. I've been doing market research and a lot of thinking about how to go ahead and market them.

I have a PCM5242 and a TLV320ADC5140 modules ready, working, and they sound great. I didn't do any measurements because I don't have an oscilloscope. The STGL5000 card just didn't have the features and quality that i needed. I needed balanced inputs/outputs and I needed a decent preamp for dynamic microphones. These two modules do very well so far. I have a set hooked up in my game room with some EV112P speakers and my ears really enjoy the sound. Not just a lack of noise, but also just clarity and dynamic range. They stack on top of eachother, and I've successfully passed up to 8 outputs and 16 inputs through the audio library. My driver is written to allow for controlling the device depending on the wire and device address... So you can configure 8 devices with separate settings.

I've discussed the DAC here, but haven't posted much about the ADC yet. Please, encourage me to move forward with some insight into the market if you think that this could be helpful to others.
 
I didn't do any measurements because I don't have an oscilloscope.

You need an audio analyser, not a 'scope, to measure an audio system - 'scopes are mainly 8 bit resolution
with high noise floors and so-so linearity (the front end typically has to work from DC to 100's MHz). Some
'scopes have more resolution, but still its a limiting factor. Turning the probe to x1 may help with audio too,
as that may lose some noise from the probe's divider network.

PC sound card is the most cost-effective way to have an audio analyzer, you probably already have one that's
reasonable, or there are external USB sound cards with good specs at moderate prices. I use a Scarlett Solo
for instance, and there's a range of software including open source for doing audio analysis this way.

I also capture WAV files in Audacity and use Python to analyze them using the soundfile, numpy, scipy.signal
and matplotlib packages.
 
Paul's comments about price are spot on. Once you add all the features that you really want (a better audio codec, extension for more channels than stereo, audio jacks, BT module, LiPo recharging, a case) the increased price drives a lot of folks away.

For example, I'm part of the Tympan team (https://tympan.org) and we've made a T3.6 based audio system with all these features. We'll be doing a T4 based system as soon as the bootloader chip is available. I think it's a cool device that's fully reprogrammable as a Teensy. Great! But, it's not nearly as cheap as anyone would like it to be. As a result, interest is limited, which keeps quantity low, which makes it hard to lower price.

So, yeah, Paul speaks the truth about the very delicate niche that Teensy occupies. Stray out of that tiny box and the market size collapses.

Chip
 
Either will do. These days test equipment is expected to handle 0.001% distortion or better, which is beyond any 'scope.
 
Back
Top