MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

New Teensy 4.0 - phenomenal performance vs cost
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author New Teensy 4.0 - phenomenal performance vs cost
TuskSE
Look at this madness!
https://www.pjrc.com/store/teensy40.html

ARM cortex-M7 running at 600Hz
Hardware float and double calculations
I2S and S/PDIF digital audio
Teensy 3.2 pin compatibility
$20 (!)

Only downside is the 100mA power draw.

That aside: as far as I understand, this is much more power than anything used in mutable instruments lineup. Just gotta figure out what I actually need all that power for...

I feel the potential for an Ornament & Crime type thing with digital oscillators, filters, etc.

Also, If I understand the chip data sheet right, the PWM outputs are clocked at 150MHz - I haven’t done the sums, but maybe this is sufficient for high resolution and fast CV output without the need fro an external DAC?[/s]
jorg
I have been developing some FX algorithms that could benefit from a fast FPU. This looks pretty interesting.

Teensy 4.0's CPU is claimed to be very friendly to both overclocking, and clock speed changes, without disrupting peripherals. That is aimed at letting you manage power by adjusting clock speed. You might even dial it up or down to match the amount of code in your algorithm.
guest
yeah, that thing is crazy. and 100mA seems low to me for that much power. its probably better than the computer im using right now.
jorg
If you power it with an on-module switcher, you might expect to draw about 36mA from the +12V rail. Not much at all!
fuzzbass
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!
toneburst
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!


It's not 100% pin-compatible with Teensy 3.x, so it probably won't work as a drop-in replacement, anyway.
pld
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!

The faster core can run a deep learning model that infers predictive parameter sets, so the module interface gets more streamlined and intuitive as ever more apps are added, rendering manuals obsolete. Tada!
Sandrine
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!

hihi

I like the i2s has SAI capability, codecs are nice on SAI. This is a really interesting powerful dev board for audio I/O. I wonder how the Arduino IDE support is for that?
Timmy
toneburst wrote:
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!


It's not 100% pin-compatible with Teensy 3.x, so it probably won't work as a drop-in replacement, anyway.


The T4 pins are compatible-enough to potentially allow it to be a drop-in replacement for the T3.1/3.2 in the original 14HP O&C design. However, it's not that simple, and there are a myriad of tricky timing issues etc that mean a LOT of changes, many of them involving a lot of trial-and-error, and hence time, would be required to get a T4-enabled O&C to work. That's not speculation: mxmxmx has had access to a pre-beta and then a beta T4 for several months, and started to investigate the retrofitting issues. He got as far as eventually getting the beta T4 to boot, but only after disabling probably about half the code (and the display etc). But the real showstopper is that it now appears that the onboard ADCs, on which O&C modules rely, are actually a lot less precise in the T4 than they are in the T3.2, rendering it useless for quantisation duties. Thus, a T4 upgrade to existing O&Cs is unlikely to be feasible, and almost certainly not effort-effective. Plus, it the T4 won't physically work with the microO&C variants, that use connections to pads underneath the T3.2. Those pads are completely different in the T4, so an adaptor board would be required. Just doesn't seem like it is worth the effort, given the architectural limitations inherent in the O&C design anyway. Can't make a silk purse out of a sow's ear.
Timmy
pld wrote:
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!

The faster core can run a deep learning model that infers predictive parameter sets, so the module interface gets more streamlined and intuitive as ever more apps are added, rendering manuals obsolete. Tada!


We've been talking to Elon about his NeuraLink] thing... https://neuralink.com
mxmxmx
Timmy wrote:
toneburst wrote:
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! :bananaguitar:


It's not 100% pin-compatible with Teensy 3.x, so it probably won't work as a drop-in replacement, anyway.


fwiw, it's entirely pin-compatible (in terms of O&C) so that's not really the issue; all the needed signals are there (and plenty more); just the 1062 ADC might be a little less suitable, i don't know. also, it does/should boot up (there were issues with the PMIC / power-up sequencing in the early phase of beta testing, ie when powering the thing from 12V / some DC|DC converter, but that's been resolved. at least it worked here, eventually)

so porting the code should be entirely doable. chances are the pay-off is going to be very slim though, as was pointed out, mainly because of existing bottlenecks (the OLED/DAC shared SPI bus). with the 1062, there's a few more options to disentangle the DAC/OLED now, but then we're already talking about new/different hardware; i've already cooked up something IoTesque, but progress (re firmware) has been very slow (due to lack of time/motivation/skills).
SphericalSound
Timmy wrote:
toneburst wrote:
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!


It's not 100% pin-compatible with Teensy 3.x, so it probably won't work as a drop-in replacement, anyway.


Can't make a silk purse out of a sow's ear.


Ohh, that´s a pity. I was drooling with the idea of my o_C v4.0 Dead Banana

But thanks for the clarification!
joem
It'll be interesting to see what new modules take advantage of the Teensy 4, and how they take advantage.

I want to get my hands on one to see what it can do video-wise. Really curious if the PXP can be used in some useful way with composite video signal generation. Trying to parse the datasheet but pixel pipeline might be a bit beyond me right now.
Roni
Sandrine wrote:
I wonder how the Arduino IDE support is for that?


Teensyduino is an add-on for the Arduino IDE and brings full support for all Teensy revisions. It works quite well.
cackland
joem wrote:
It'll be interesting to see what new modules take advantage of the Teensy 4, and how they take advantage.


Curiosity has me too.
Dogma
Sandrine wrote:
fuzzbass wrote:
OK but this does not mean we get another twelve functions for o_C. The manual is already 60 pages long! Rockin' Banana!

hihi

I like the i2s has SAI capability, codecs are nice on SAI. This is a really interesting powerful dev board for audio I/O. I wonder how the Arduino IDE support is for that?


So does that mean an update to the rll sandy? That would be amazing
kallikak
That's an amazingly powerful little device. I just wish they'd make it just a tiny bit bigger so all pins were breadboard friendly and avoid the surface mount pads underneath.

Ken
weasel79
kallikak wrote:
That's an amazingly powerful little device. I just wish they'd make it just a tiny bit bigger so all pins were breadboard friendly and avoid the surface mount pads underneath.

Yeah i'm kinda bummed there is no .5 version from the get go or at least some info if there will be a version at some point with more pinouts.

that being said - anybody knows a vendor in the EU that has it in stock currently?
mos6502
This is exactly what I needed. Thanks PJRC smile
Sin_Phi
Was wondering if the 4.0s extra horsepower would even be useful to the sorts of applications the O_c is suited to. Was thinking you would get more value out of it by targeting audio processing applications with a real ADC/DAC. The onboard ADC should still be decent enough to read pots, but wouldn't really use it for CV or audio unless you are going lo-fi.

I was previously toying with that idea for a 3.6, but paused that project when I saw the 4.0 coming. Picking that up again. There are going to be some fun technical issues with (the WM8731 especially) for sure.

I have another thread out there I need to dig up, but was interested in what people saw as desirable applications from the 4.0 that they couldn't get out of a 3.2/.6. Personally I want to play around with some reverbs, maybe something of a stereo erbverb. A lot of potential to do some of the more heavy algorithms like granular or physical modeling.

As to the lack of a .5 version, I2C IO expanders and 4051 analog switches are your friend. Your part count may go up, but it is a lot nicer to route 2-3 wires to the IO board than 16. I actually prefer the FFC pads to the SD card holder if you are building your own solution.
pld
Sin_Phi wrote:
Was wondering if the 4.0s extra horsepower would even be useful to the sorts of applications the O_c is suited to.

IMO horsepower not as much as the extra storage, at least for "pitch" use cases (even if the FPU is nice...) LFO-y/continuous signals that's perhaps different, but they aren't what it's suited to given the single SPI. In a fit of "why not?" I made an STM32F405 riser for the o_C but for lack of a higher purpose it fizzled out once I'd written drivers and base framework.

Even factoring out the screen, IIRC the app that comes closest to maxing out the 3.2 is Piqued (4x envelopes), but that's more due to almost zero optimization effort.
quarterturn
I've been reading some of the discussion threads over on PJRC related to Teensy 4.0 and music, and I realize the math for things which would take advantage of the computing power is waaaay over my head. Stuff like phase vocoding. I have a hard enough time understanding stuff like convolution for filtering, forget stuff like pitch correction.
But I certainly look forward to kits or finished products using it.
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 1
Powered by phpBB © phpBB Group