okay i should probably like... make a vague attempt at going to work. Thanks for taking a look at this, everyone. ๐
#circuitpython-dev
ivory yew
tulip sleet
@idle owl I would deinitely say "Bluetooth Low Energy (BLE)" or something like that. Plain "Bluetooth" usually implies classic Bluetooth
idle owl
@ivory yew Hope your vague attempt goes well enough ๐
@tulip sleet Yeah that's why I asked. Ok thank you.
ivory yew
@onyx hinge maybe the bitscope doesn't get a stay of execution. that's much cleaner.
onyx hinge
or do I have those backwards
tulip sleet
@idle owl and "Bluetooth LE" is correct, "Bluetooth BLE" is not
onyx hinge
800khz is "right" isn't it?
tulip sleet
yes
onyx hinge
ok so I labeled them backwards
but .. timing is changing by ~10%, and that's enough to matter
idle owl
@tulip sleet Right like ATM machine. I'll go with the full name and acronym in parens. Thanks!
tulip sleet
there is some trickery at the bottom of neopixel_write that fiddles with the timing values also
onyx hinge
@tulip sleet yeah I looked at that and wondered if it was interrupts -- I introduced disable/enable interrupts indirectly in current_tick() , but getting rid of that didn't fix it
idle owl
@tulip sleet Is it worth using words like "central" and "peripheral"? Or call them something less technical? Or both is what I was thinking? e.g. "This one acts as the peripheral, or some better word here."
I realise people have already done guides with these, so whatever they did probably works, but I figure if I'm going to do this, I might as well do it the best way.
tulip sleet
@idle owl the trouble is that central and peripheral are best used before a connection is set up. Then the important distinction is server and client. A client talks to a remote service on a server. Usually a peripheral provides services, so it's a server, but there are exceptions, like a peripheral talking to the Current Time Service on an iPhone acting as a central
slender iron
@onyx hinge I was looking for BLE pulse ox devices on Amazon
idle owl
@tulip sleet ah ok.
idle owl
@minor plume We do not have a driver for it, and no one is working on one internally. Please move forward with porting the driver, and remember we're here to help you along the way!
@tulip sleet So the controller is the server and the target is the client?
tulip sleet
no, opposite
idle owl
bugger ok
I'm going to come up with something else to call them for this guide then. Because it's not even clear to me ๐
tulip sleet
the UART Service lives on the CPB, and is remote to the phone
idle owl
This is two CPBs.
tulip sleet
the client is the one sending the colors
idle owl
Ok, got it, thanks
tulip sleet
it sends to a remote UART service
tulip sleet
i will get lunch
slender iron
@tulip sleet how is the filesystem stuff coming?
tulip sleet
@slender iron it's really the .ld templating stuff right now. I'm about to get back to it after this morning's reviews and PRs.
slender iron
๐
tulip sleet
atmel-samd works; working on nrf
slender iron
cool, I'm excited for bonding ๐
ionic elk
@meager fog @slender iron let me know if you've got anything immediate you'd like me to work on - otherwise I'll be moving back to implementing the Blackpill and meowbit
slender iron
adding boards sounds good to me @ionic elk . I'm wrapping up my morning reviews now
ionic elk
also @meager fog let me know if you'd like me to pick up one of these https://www.st.com/en/evaluation-tools/stm32f4discovery.html. Makes sense to support since it's a 407
onyx hinge
huh if you re-run an action on a commit, it throws out the old log
manic glacierBOT
ionic elk
@slender iron should I go ahead and merge Displayio?
slender iron
no, did you reply to my comment?
"Let's also make sure that the SPI objects always default to 0,0 as well."
ionic elk
Oh sorry yes the defaults are at 0,0, I fixed both the core SPI module and the Fourwire library
slender iron
ah, there it is
manic glacierBOT
Updating from the discord conversation.
@jepler has a much better analyzer than me. It seems the issue is that the timing is about ~10% faster:
bad (917khz):
good (800khz):
slender iron
@idle owl does your code that hard fault do scanning?
idle owl
This? python if not uart_connection: print("Scanning...") for adv in ble.start_scan(ProvideServicesAdvertisement, timeout=5): if UARTService in adv.services: print("Found a UARTService advertisement.") uart_connection = ble.connect(adv) break # Stop scanning whether or not we are connected. ble.stop_scan()
slender iron
yup
idle owl
Then yes.
slender iron
kk, I just crashed during scanning + usb
idle owl
It does like to crash on reload, which means it goes back into scanning each time.
slender iron
welp, it is happening a lot but the backtrace isn't very helpful. ๐
slender iron
@idle owl what board are you testing with? CPB?
idle owl
Yes
ionic elk
@gentle bronze @slender iron can you glean any significance from this set of steps?
serial_connected () at ../../supervisor/shared/serial.c:43
43 return tud_cdc_connected();
(gdb) s
tud_cdc_connected () at ../../lib/tinyusb/src/class/cdc/cdc_device.h:128
128 return tud_cdc_n_connected(0);
(gdb) s
tud_cdc_n_connected (itf=0 '\000') at ../../lib/tinyusb/src/class/cdc/cdc_device.c:97
97 return tud_ready() && tu_bit_test(_cdcd_itf[itf].line_state, 0);
(gdb) s
tud_ready () at ../../lib/tinyusb/src/device/usbd.h:56
56 return tud_mounted() && !tud_suspended();
(gdb) s
tud_mounted () at ../../lib/tinyusb/src/device/usbd.c:200
200 return _usbd_dev.configured;
This is on a F411 board I've been plugging at for a while. Seems to work ok but USB won't enumerate - doesn't show up on beagle, nothing. But given that other F411 boards like the pyb nano work just fine I'm puzzled.
slender iron
does the 411 let you have different pins for usb?
ionic elk
No, everything's consistent. I've been comparing the blackpill schematic to the PybNano and the Discovery - there aren't any differences in terms of the USB connection other than the ID pin, which isn't part of USB C. I tied the usual ID pin down just in case, but it doesn't help.
slender iron
if the beagle doesn't see anything then it points to a connection problem
that stack trace implies tinyusb doesn't see anything either
ionic elk
As in, physical hardware?
slender iron
ya
ionic elk
Just to clarify, will tinyUSB care about the ID and VBUS pins?
slender iron
I don't think so
ionic elk
hmmmm
slender iron
you could start by getting the tinyusb example going instead of circuitpython
ionic elk
I'm worried they mixed up the pins or something, but since USBC SMT parts don't have a standard layout it's not really possible for me to check
ionic elk
@slender iron not getting anything with Thach's example sketches either, sadly
Calling it a night for now
meager fog
@ionic elk was mobile today - yes meowbit might be fun now cause its has a screen you can use ๐ '407 is also fine...its a subset of '405 but i dunno if we want to bother implementing stuff like Ethernet? at least not yet ๐
manic glacierBOT
@jepler not at all, but I can definitely order a new rev and fit in an RC filter. Currently I don't have an amp, so the audio output is directly fed to the speaker, but I'm thinking about adding an LM386 to the next revision.
I looked at the Bluefruit's schematic, and it seems like it includes an RC filter at the amplifier's input:
Or did I miss something?
manic glacierBOT
That combination of resistor and capacitor in series looks like it is converting the audio signal to be AC-coupled.
The kind of RC filter I'm talking about looks more like this:
the resistor in series limits how quickly the capacitor can charge/discharge, smoothing the square wave of the PWM signal into a triangle(ish) wave of lower amplitude. The product R * C tells you t...
manic glacierBOT
wow thats...kinda terrible. why is that? :)
If you're referring to the VBUS issue, it's because they didn't connect the USB's VBUS to the PA09 pin (presumably so they could expose the PA09 pin, which none of our other boards do), which as far as I can tell is required for the FS USB spec for any device that could be self powered (which is the case if the user had any other powe...
slender iron
@tulip sleet how is the fs stuff coming? did you finish nrf?
tulip sleet
closer; I am just finishing removing CPy-dependent stuff from samd-peripherals. About to submit a PR for that. Just checking some builds.
nrf is really close to building or already does. This is a major refactor
slender iron
how is the peripheral stuff related?
tulip sleet
CALIBRATE_CRYSTALLESS and internal flash config region size setting
slender iron
kk, coolio
I'm going to take a stab at a _bleio.PacketBuffer today
need it for media service
tulip sleet
it is all configurable, and the firmware size region calculation is a lot easier to understand
could you accept my BLE library PR if it's OK by you now? That will help some users.
and fix the nameless advertising
slender iron
ya, its queued up in my tabs to look at
tulip sleet
great
i am paying off a lot of old technical debt right now; sorry, it takes a while
slender iron
ya, it's good to do from time to time. just have to remember the goal of having bonding
tulip sleet
not forgetting; the bonding region is all set up in flash now
slender iron
I tried doing an ANCS demo at a meetup last night and kept getting disconnected
bonding will hopefully help auto-reconnect things
tulip sleet
i am not going to do the .ld refactor on stm32f4 or spresense yet to save time
slender iron
sgtm
meager fog
@tidal kiln ping me when yr around and you wanna try MCP2221 - had good success
slender iron
@tulip sleet this PR of mine is waiting for you: https://github.com/adafruit/circuitpython/pull/2356
meager fog
i also would like to poke into whether we can receive multiple i2c bytes on ft232h - right now its 1 byte per ms i think
but wow MCP2221 is incredibly undocumented, i had to reverse engineer half their protocol based on what packet responses came back ๐
idle owl
@slender iron I grabbed the UF2 you posted yesterday. I verified one more time that it would crash into the hardfault handler on reset (it did), and then loaded your firmware. It hasn't crashed since. Sometimes it took a while, but it was being really consistent for a bit there as well.
idle owl
Thanks! This one was up there on the frustrating scale. ๐
meager fog
BTLE is like that for sure
@slender iron im going to try out the BLE lightbulb once i get thru emails and meetings
so yr thinking is we have these as examples for now...then add as libraries? they feel...librarish
slender iron
I'd be fine splitting them into separate libraries
Its easier to evolve them all in one as they are now but I can see how it'll get unwieldy
meager fog
yah yah - ok its def ok for now!
slender iron
the light has more characteristics than the github documents
I haven't tried poking at them
idle owl
@tulip sleet Now I'm getting this: Traceback (most recent call last): File "code.py", line 93, in <module> File "code.py", line 33, in send_packet File "code.py", line 26, in send_packet File "adafruit_ble/services/nordic.py", line 108, in write File "adafruit_ble/characteristics/stream.py", line 49, in write _bleio.BluetoothError: Unknown soft device error: 3002
Among other things. I changed the code, it's probably a code problem.
tulip sleet
@idle owl that error is "invalid connection handle", so sounds like the connection got broken at an inconvenient time
idle owl
Also getting <BLEConnection object at 20009cf0> object has no service <class 'UARTService'>
tulip sleet
it may be connecting to another BLE device
idle owl
@tulip sleet There aren't any others. It started after I added the switch to be sending a button packet.
I added KeyError to the send_packet try/except at the top and it's catching both of them now.
manic glacierBOT
onyx hinge
@meager fog I'm curious, on CPB did you consider whether to put an RC filter between the PWM outut and that on-board amplifier? I suspect the lack of one is why there's high pitched hiss when playing waves/mp3s (interaction between the pwm carrier and the amplifier's oscillator). with an external amplifier it's better by default but adding an RC filter makes it even better. There's more in the notes on that mp3 PR.
meager fog
@onyx hinge hmm i didnt think of it
if you give me the values you used i could probably wiggle one in
on a PCB revision
onyx hinge
I picked something random out of the junk pile, and it's not in front of me right now. I think that R*C ~= 14kHz. Something that is going to almost fully attenuate the 62.5kHz+ PWM carrier frequency but leave audio intact.
it might have been 470 ohm and 100nF for 21.3kHz
tidal kiln
@meager fog i'm back if you wanna chat MCP2221
tidal kiln
anything new beyond your code updates? going to work those in now and push them to my PR.
meager fog
@tidal kiln nope
repeated oled drawing & MLX90640 handling is really gruesome
so if it still works with your kit, we probably have fully coverage
tidal kiln
yep. will work in your changes and test here then push it up.
awesome work on it BTW.
slender iron
@idle owl the missing service error is the exception I used to throw when the hard fault occurred. (it happens when a disconnect happens while discovering services.) the PR with the fix will raise _bleio.ConnectionError instead because it is clearer
meager fog
@tidal kiln from dark places comes nifty code
tidal kiln
@meager fog my i2c battle arena still works. so looks good. pushing changes to PR.
wow. and no linting needed.
meager fog
@tidal kiln back
ill test your PR
and we can merg
merrrrrrrrrrg
the final breakout will be ~$5 - very nice alternative to mucking with an arduino
meager fog
@carter meerrrged
cedar beacon
What is the advice/status of using YAML files for config in circuitpython?
manic glacierBOT
tulip sleet
@cedar beacon There is no YAML library. There is native JSON support on SAMD51 ("M4") and nRF52840 boards.
cedar beacon
Thanks @tulip sleet
meager fog
or you can do an import of a dictionary - see https://learn.adafruit.com/circuitpython-tv-zapper-with-circuit-playground-express
onyx hinge
Trying to tweak the existing code back into working was super nerve wracking and ultimately futile (I had gotten it to work on SAMD51 but that left 21 broken); the code from the bootloader "just worked" after being copied in, which I take as a good sign. (also, the use ofasm and placement in its own function seem to be key ideas that are likely to make it more robust against future changes, which is great if true)
fingers extremely crossed
meager fog
@tidal kiln hey dya know why i failed to bump blinka on pypi
its not just releasing a new vers?
tidal kiln
looking...
hmmm. it just worked for MCP9600.
all i did was tag the new release
and a few minutes later 1.0.2 is live:
Successfully installed adafruit-circuitpython-mcp9600-1.0.2
@meager fog it looks ok? quick test, fresh venv, pip install blinka, it grabbed 3.2.0
tidal kiln
how are you checking?
tidal kiln
maybe? that looks ok too. yay caches.
or maybe it was super laggy for some reason? and i just happen to be showing up at the right time.
meager fog
๐คทโโ๏ธ
so far the MCP2221 has worked with everything ive thrown at it
i think people considered it flakey cause it didnt have good code. once you send it what it expects, its pretty fast
tidal kiln
nice. glad we decided to work up our own little driver.
meager fog
yah its got some things it does better than the more $ FT232H
im thinking this could be a great chip for using jupyter notebooks as datacollection
tidal kiln
i don't think even the C driver supports multi-byte
meager fog
yeah it doesnt
thus the ???
i looked at the packets to see what it seemed to be doing
ds said 65535 in/out
there's an internal counter for bytes-sent
you can see it in status packet
tidal kiln
good eye. i saw some of those codes that weren't in the C source, so wasn't sure what to make of them.
tidal kiln
that easy? ok. will try that next time.
meager fog
๐ค
tidal kiln
yah, if this little thing can be put out there for $5, it's a super easy way to tether I2C sensors off USB
meager fog
yah
right now people do computer->arduino->sensor which means you can't easily tweak settings
plus of course added complexity, and slow UART
what if, you skip the whole arduino step - read right into your compy for instant graphing & data collection
tidal kiln
totes
meager fog
and since im doing stemma/qwiic - its a great mix
tidal kiln
and it's a wee little board thats not much bigger than the USB connector
meager fog
chain all the sensors you want
tidal kiln
so looks like magic
meager fog
no soldering
yeah its magic - thanks to blinka
like ideal demo i think would be running kinematics
stuff that arduino doesnt do so great at
main meteor
Is this going to be a new product? I'm intrigued.
tidal kiln
maybe?
graceful heart
any updates on https://github.com/adafruit/circuitpython/issues/1050 ? I think this could be super useful ๐
not just for making bluetooth mice/keyboards, but for other HID devices. the MS surface dial is an HID device, for example.
tulip sleet
@graceful heart we are ironing out the last bugs for BLE hiD, and we need to implement bonding so you don't have to pair every time. That is in progress right now.
marble hornet
Iโd greatly appreciate some help with burning the bootloader to an atsamd21e18. Iโm using a blink edu, I also have a blink edu mini here too, and have burned uf2 bootloaders into samd51 chips countless times before, with mostly no hiccups. However this time Iโm trying to burn the trinket bootloader onto a bare atsamd21e18 and Iโm having some odd issues.
Once I get the chip connected and I flash it using jlinkโs utility
โโโ
Loadbin /absolute/path/to/file 0x00
โโโ
And the upload verified and works. However when I power cycle after disconnecting the debugger the board will not register over usb
The d13 led pun flashes like normal. Iโm on Mac but when I plug it into a windows machine it recognizes there is a usb device but it can not resolve what it is. I can only assume there is a bug in the bootloader firmware?
Any and all help would be appreciated. Please ping me๐
that last conclusion might be wrong as i tried flashing some old version of the bootloader and i'm getting the same result
marble hornet
It is failing to find the device descriptor
When connected on windows this is what is in the device events log in the device manager
Looks like the right vid/pid
manic glacierBOT
solar whale
For anyone interested -- homebrew now installs the latest gcc (9-2019-q4-major) for the Mac https://github.com/ARMmbed/homebrew-formulae -- seems to work OK
river quest
200! https://github.com/adafruit/Adafruit_CircuitPython_Bundle/blob/master/circuitpython_library_list.md
slender iron
Hi all, our CircuitPython community meeting is tomorrow at 11am Pacific / 2pm Eastern here in our Discord voice chat. Everyone is encouraged to attend. The notes doc with more info is available here: https://docs.google.com/document/d/1ldHCkUp7F9_pI76nMbmEpiWUzQbxCCy38CRS_zv3UBA/edit?usp=sharing <@&356864093652516868>
@marble hornet What are you expecting to happen that doesn't?
marble hornet
I'm expecting the board to register as a trinket in bootloader mode.
and I'm seeing nothing.
meager fog
@graceful heart its being worked on - no ETA but we're getting closer ๐ its much harder than expected ๐ฎ
tidal kiln
@main meteor for your intrigue: https://youtu.be/afOqJdf2MlU
ivory yew
@tidal kiln that is super cool
tidal kiln
@ivory yew thanks. yah, hopefully it will be. credit to ladyada though for figuring out the final under the hood magic incantation.
ivory yew
Ladyada is a master of deep magic.
meager fog
thats the only thing i can do
i cant like, tie my shoes or eat on a regular schedule
deep packet inspection โ๏ธ
ivory yew
A whole mood.
marble hornet
@slender iron any thoughts ?
onyx hinge
woof, getting "down" headers AND the debug header on the stm32f405 feather took way too long. but now, with stacking feather headers in a breadboard, I can get it right side up and with debugging.
(and involved sanding a few thou off the down pin headers, possibly because the debug header was not exactly where it needed to be .. or maybe they really aren't meant to fit at the same time)
slender iron
@marble hornet not many. a restart on mac may help if it shows on windows. a usb sniffer can be helpful when debugging a new board
marble hornet
Okay, I will look into getting one. In the mean time: it is just the chip on a qfn breakout not a new board Per se. And supporting electronics in a Breadboard
slender iron
is the usb running through a breadboard?
@graceful heart you should be able to do your rotary BLE HID with the latest CP master and the latest library
start with this example: https://github.com/adafruit/Adafruit_CircuitPython_BLE/blob/master/examples/ble_hid_periph.py
graceful heart
thank you! @slender iron I'll give that look!
slender iron
@prime flower forum post for ya when you have time: https://forums.adafruit.com/viewtopic.php?f=60&t=159594
marble hornet
@slender iron yes the usb is going through bread board
marble hornet
i'll try soldering it
prime flower
@slender iron Will take a look at it during my AM support, thanks!
ionic elk
@onyx hinge I've been using the stacking headers to elevate mine:
onyx hinge
oh that would have worked better, since the plastic is on the top side that way
ionic elk
You also mean the F405 feather, right? Or do we have a new prototype coming along ๐
ionic elk
Lol I think the only difference is the 7 has ethernet or something
onyx hinge
incidentally this shows how far I got with "learning" the stm32 last week
ionic elk
that we aren't even going to do for a whiel
yep just ethernet
what are you using to debug?
@onyx hinge st-util or openocd?
onyx hinge
@ionic elk I have a j-link. using JLinkGDBServerCL as gdbserver on linux
dfu-util -a 0 --dfuse-address 0x08000000 -D build-feather_stm32f405_express/firmware.uf2 oh gee why doesn't the device work after this. sigh.
onyx hinge
did somebody do the uf2 bootloader?
ionic elk
For the most part so far I've done everything with an ST Link and st-util/openocd
onyx hinge
does that upload firmware any faster than dfu-util?
ionic elk
I haven't worked on it
It uploads it... very fast? I'm not sure. It takes like 2s
is dfu util faster?
onyx hinge
no, it's over 1 minute
onyx hinge
https://github.com/mmoskal/uf2-stm32f oh hey this has files for the feather
something called feather_f405 anyway
I think it needs .ld file magic in CP that is beyond me though
the uf2 bootloader locates at 0x08000000, so circuitpython has to move up by 32kB
ionic elk
doesn't stm32duino have one too?
How big is the bootloader? that could be very problematic for our internal flash filesystem
I'll do some research
Ideally we want it out in one of the 128kb sectors at the end of the flash
onyx hinge
yeah it looks like it occupies all those 16kB sectors at the start
ionic elk
yeah that's no bueno. Means no internal flash, like at all
onyx hinge
the other thing I was trying to figure out is, can I enter the DFU bootloader from software? I didn't find an answer yet. (I recall this was possible on the first USB DFU device I played with, some Atmel AVR)
ionic elk
Though I guess that's less of an issue for feathers, to be fair. Basically only affects discovery boards and maybe you just don't care for those anyway since they have a built in st-link. Biggest issue would be off brand boards without flash or a programmer built in
It'd be nice if we could just shift it out to one of the big sectors but we'd need some serious magic to get that to load up properly
onyx hinge
this is totally a detour for me
anyway, I can upload firmware via gdb now and it is faster than dfu
Transfer rate: 44730 KB/sec, 14313 bytes/write.
ionic elk
I am not totally sure about a boot into DFU... it requires the boot pins, maybe there's some kind of thing that'd let you persist that past reset but I kinda doubt it
how fast is it for you? For the cp build? just for my reference
onyx hinge
how long does "make" take from "clean"?
user 0m21.882s
https://community.st.com/s/question/0D50X00009Xkh7wSAB/jumping-to-the-bootloader-via-software-does-not-permit-dfu-mode some people have these notes about entering DFU via software reset without boot0 but it's going a bit over my head
OK OK back to nRF audio playback rate, which is where I intended to start the day ๐
manic glacierBOT
ionic elk
@onyx hinge sorry I meant how long does it take you to flash your finished binary for the Cpy build - mine takes about 5s to flash. If it takes a lot longer than that and starts getting on your nerves, you could try the ST link as another option.
ionic elk
Yeah that sounds fine I wouldn't worry about it then
onyx hinge
groan, whole-computer lock-up while running gdb on a nRF
I had hoped those would be gone with my upgrade earlier this year but no such luck
idle owl
@slender iron The only way around the new error you introduced is to import _bleio and then except _bleio.BluetoothError as far as I can tell. It's better than the hardfault handler, but I really feel like I shouldn't be importing _bleio. You didn't base it on anything so I can't except something general to catch it.
idle owl
For context, I'm consistently getting that error if I soft-reload/reset the receiver while the controller and the receiver are connected and packets are being sent from the controller to the receiver.
slender iron
@idle owl we can make the ble library provide them as well. using general errors is less good because it catches more than you want and is less clear. Is your code that is doing the sending checking to see if the connection is still connected?
meager fog
@tidal kiln fyi https://www.silabs.com/products/interface/usb-bridges/classic-usb-bridges/device.cp2112 i didnt catch this one
idle owl
@slender iron I'm not sure what sending checking code looks like. There is a lot of checking going on in this code.
slender iron
can you link me to it?
idle owl
Having the BLE library provide it is at least one step better than requiring an import of _bleio.
Yeah hold on.
slender iron
ble is tricky because the connection can be lost at any time
idle owl
I assume it's something in the controller code. That's what's crashing.
is the controller.
meager fog
@tidal kiln its fairly cheap, about same price as MCP2221 - clearly they're meant to compete ๐
oh nvmd "Not Recommended for New Design"
its about to be killed. we can skip
tidal kiln
๐ฆ is there a suggested replacement?
slender iron
@idle owl the uart_connection.connected is what I was thinking
idle owl
@slender iron So what you were asking about is in there then?
slender iron
yes
idle owl
Ok
slender iron
but I wonder if it should be checked in send_packet too
all of the sleeps could build up to cause the .connected to be far away from the use
idle owl
the tiny sleeps are debouncing for button packets, and there were no button presses when I got it to crash to this error. The final time.sleep is required to avoid a bad checksum error from sending too fast.
how would I add it to send_packet?
meager fog
@tidal kiln nope - that chip is EOL
idle owl
if uart_connection.connected at the beginning?
idle owl
I thought that's what it does right now. Dan wrote that for me, I only technically understand it.
slender iron
as it is it is reactive by catching exceptions
you could also be proactive and check that the connection is still valid
idle owl
I'm not following what you want me to do. I checked to see if not uart_connection.connected and it's never getting to that bit in send_packet when it fails this way.
all I did was print
but it did not print
Traceback has send_packet in it. So I guess it's trying? I must be checking in the wrong place.
ivory yew
Checking the connection first always leaves the chance of the connection dropping by the time you interact with it.
idle owl
I don't understand how to write the code. I understand the concept, I think.
ivory yew
It seems to me when interacting with a library where almost any method could throw a connection error at any time you need to be reactive and use try catch.
idle owl
I have one, but the issue is that Scott introduced a new error that is in _bleio and would require importing that to catch it. It's _ so I'm hesitant to be importing it.
ivory yew
Should probably alias it in the Python ble library
slender iron
@ivory yew I've been thinking about the best way to handle ble disconnects. It'd be nice if there were only a few spots to catch the exception and check connected
ionic elk
does anyone know where micropython stores pin information for pyb.SCREEN()? I'm trying to figure out what the Meowbit screen is from their micropython port but I can't seem to find anything about the screen on the board level.
ionic elk
Scratch that I found it.
ionic elk
yep blegh thanks for looking though
man micropython does not really give a dang about keeping its definitions in one place huh
kinda dreading that that'll keep showing up...
stuck elbow
tbh micropython doesn't really give a dang about much anything that is not pyboard
ionic elk
lol
Well, if anyone happens to have a reverse engineered schematic of the Meowbit it'd save me some dredging
stuck elbow
and you can't really blame them, it's what puts bread on their table
ionic elk
yeah true
stuck elbow
I think you can steal the pin mapping from the makecode arcade bootloader configuration
ionic elk
Does that have peripheral info? The MCU map I've got, its the peripheral maps I need.
Where would I find that?
nvm you're right, it's all in here https://github.com/KittenBot/uf2-meowbit/blob/fat/boards/meowbit/board.h
river quest
News from the future ... "There was an issue with your delivery"
drowsy geyser
That's how AppleTV works...
meager fog
@tidal kiln ahhh hidapi isnt a trivial pip install - wanna keep hacking on it?
sudo apt-get install libusb-1.0-0-dev
river quest
"CircuitPython snakes its way to Scott's TV"
sterile bronze
lurking
old smelt
Lurking
drowsy geyser
Also lurking.
tidal kiln
@meager fog to remove the requirement for hidapi? and just use libusb? write our own hid read/write?
turbid radish
lurking today
tidal kiln
lurking
timber mango
lurking today
meager fog
@tidal kiln i mean you need to apt-get install libusb so that hidapi can build
stuck elbow
lurking
meager fog
@tidal kiln thats' done in the travis setup
drowsy geyser
BTW, huge hugs to everyone who did Blinka! I am now controlling a Sphero RVR with a Raspberry Pi because of it!!!
tidal kiln
hmm. ok. let me look...
ruby atlas
rebooting
gilded cradle
That's great to hear @drowsy geyser
meager fog
mystic anvil
lurking
river quest
200 CircuitPython Libraries!
https://github.com/adafruit/Adafruit_CircuitPython_Bundle/blob/master/circuitpython_library_list.md
https://circuitpython.org/libraries
The Binho Nova is a Multi-Protocol USB host adapter with support for I2C, SPI, UART, 1-WIRE, SWI protocols all in one slim, robust package, and now it supports CircuitPython BLINKA!
https://support.binho.io/learn-and-grow/binho-+-circuitpython
CircuitPython slithers to the pages of MAKE Magazine!
Make: Vol. 71 FIX OUR PLANET featuring CircuitPython! On page 82 is a CircuitPython powered edge-lit LED heart by Geek Mom Projects. AND! In TOOLBOX, the CircuitPython ruler from Digi-Key and Adafruit!
https://makezine.com/
tidal kiln
@meager fog ah, gotcha, for travis, yah, ok need to this:
https://learn.adafruit.com/circuitpython-on-any-computer-with-ft232h/linux#install-libusb-5-2
river quest
ADABOX 14 in the wild
Approx 4,000+ CircuitPython devices in the wild ๐
sterile bronze
I do have a hug report only today, lurking for the rest.
river quest
Adafruit joins the LoRa Alliance
https://learn.adafruit.com/using-lorawan-and-the-things-network-with-circuitpython
https://blog.adafruit.com/2019/12/05/adafruit-joins-the-lora-alliance-loraalliance-lora-lorawan-adafruit/
meager fog
@tidal kiln try it!
river quest
Engineered Sandwich is a small rechargeable freestanding light-emitting sculpture. A translucent 3D-printed lens press fits between two hand-finished wood buns. A circuit python controller randomly changes RGB LEDs, continuously projecting a cool gradient on all sides of the lens and through a series of slots at the top of the piece.
http://www.extrasleepy.com/#/engineered-sandwich/
https://www.instagram.com/p/B5yclN6h7pD/?igshid=1tuxoudkzvu2t
On Wednesday, Dec 11th at 8pm EST, Particle's Senior Manager of Developer Relations, Brandon Satrom will be on ASK AN ENGINEER!
https://twitter.com/BrandonSatrom
https://www.adafruit.com/ask
18th is ADABOX unboxing
Show and Tell and AAE on 25th, and Jan 1st!
Twitter
DRAFT IS HERE!
https://github.com/adafruit/circuitpython-weekly-newsletter/blob/gh-pages/_drafts/2019-12-10-draft.md
turbid radish
Yes, thanks for anyone who gives us Newsletter info
fierce girder
Just lurking today
inland tusk
Just lurking for hug reports
drowsy geyser
Not sure if you caught it earlier, but hugs to everyone who worked on Blinka! I'm able to control my Sphero RVR with a Raspberry Pi because of it.
idle owl
@drowsy geyser Added to the notes! Are you text only or do you want to read it off yourself?
drowsy geyser
Text only today. Sorry!
idle owl
No worries, simply wanted to know ๐
turbid radish
key-cad vs. k-eye-cad......
slender iron
๐
marble hornet
kai-cad ?
onyx hinge
(I just copied that code from somewhere else ๐ ๐ )
stuck elbow
"the other Fritzing"
onyx hinge
๐ for blinka section
errant grail
<late and lurking>
idle owl
turbid radish
haaa haaa haaaa haaaa
pastel panther
Amazing alliteration!
gilded cradle
Very alliterative
solar whale
@pastel panther I see what you did there ๐
pastel panther
๐
tidal kiln
is this a new attendance record?
idle owl
@tidal kiln Not quite, I think
inland tusk
pretty close
idle owl
"I caused the problem and then fixed it by copying someone else's code" kind of sums up software engineering.
gilded cradle
I think the record happened in February because I kind of remember it.
pastel panther
@onyx hinge Once you use stencils, you'll never go back. They're soooooooo much a better use of your time
and pony up for stainless because kapton is kinda meh
onyx hinge
yeah I went cheap this time. so if I hate it I'll buy metal and not just give up..
old smelt
I have to duck out for a 2:00 meeting. Have a great week, all!
ivory yew
I think my metal stencil from PCBWay was like $10.
pastel panther
๐ @old smelt
@ivory yew how big? Most breakout sized things are around $10 from OSH Stencils
onyx hinge
(make sure the neopixel fix is in it?)
ivory yew
Not sure but you can see it in the background of this picture.
inland tusk
@solar whale Could use the interupts for MCP23017
ruby atlas
terminal pixels
ivory yew
This is much bigger than most breakout/devboards.
solar whale
@inland tusk not sure how that would work, but will look at it
It does not talk directly to the RPi GPiO pins. Not sure how to trigger
inland tusk
@solar whale The chip has two interrupt pins which can be bound as one. Take a look at the datasheet.
solar whale
OK -- will do
inland tusk
Thanks
errant grail
Thanks!
pastel panther
@ruby atlas probably some good ideas to steal from PixelBlaze and or possibly FadeCandy:
https://www.bhencke.com/pixelblazegettingstarted
https://learn.adafruit.com/led-art-with-fadecandy/intro
onyx hinge
ah yeah I bet fadecandy was what I was trying to think of
pastel panther
the visualizations @ruby atlas mentioned reminded me of pixelblaze
think we need a PyxelBlaze
or a BlinkaBlinker
or somesuch
ruby atlas
yeah they're not particularly fancy - yet.
idle owl
Dropping off. Need to prep some things for dinner later. Cheers!
pastel panther
@minor plume let me know if/when you want some pointers for writing a library for the MAC3010X
lone sandal
prime flower
@tidal kiln I raise you a magnetometer scatter plot ๐
Need to do some hackin' to make it plot in real-time though, 3D space is fun to scatter plot
this'll be interesting for ML down the road
tidal kiln
nice. ill be curious to see how you do live plotting.
oh yah, was it just a solder bridge?
manic glacierBOT
manic glacierBOT
timber mango
@meager fog @tidal kiln I scrolled back and saw you mention "not recommended for new design" but was that for the CP2112? I could not find that, it even has a fresh Catalina driver. However, the MCP2221 is EOL and has a replacement with one diff: faster UART. See https://www.microchip.com/wwwproducts/ProductCompare/MCP2221/MCP2221A
manic glacierBOT
sterile bronze
@tulip sleet @idle owl @solar whale Over the past week and a bit I've been working on the Bluefruit equipment and thanks to all of you this has been a very positive experience. My most recent work with Bluefruit has been to extend the learn project written by @split ocean . I have written additional code for use with the Feather nRF52840 and to use it as a client so that it advertises a color, similar to that CPB. My final step with that was to make the on board NeoPixel to display the color that is be advertised. I was able to get this to work but not by using the learn guide "Introducing the Adafruit nRF52840 Feather". I found that on the Pinout page it states to use PIN_NEOPIXEL to access the rgb NeoPixel but this didn't work. I used other coding that did do the job and works well. I'm sharing the code to all and maybe the learn guide can be updated. Also if JP wants to add the nRF52840 to his learn guide for Hide and Seek since the Feather nRF52840 can also be used as a hidden item. (I'm going to read and work through the guides that have been written so I can use GitHub better and in the future, thanks)
idle owl
@sterile bronze PIN_NEOPIXEL may be Arduino. It should be board.NEOPIXEL for all CircuitPython board with a built in NeoPixel.
Congrats on figuring it out!
stuck elbow
@tulip sleet would you have a moment later? I took a stab at that USB HID LED support, but I suspect tinyusb doesn't support it fully
unfortunately the tinyusb source code is just a string of three-letter acronyms to me
solar whale
@sterile bronze Glad it is working for you and thanks for the example. I agree that the guide is a bit confusing regarding the pin names -- As @idle owl pointed out, those are used for Arduino. Just FYI, you can check the pin names at the REPL by doing import board dir(board) that will list all the pins. also you can look at the boards pins.c file -- for the feather_nRF52840_express https://github.com/adafruit/circuitpython/blob/master/ports/nrf/boards/feather_nrf52840_express/pins.c
sterile bronze
@solar whale thanks for the link to the GitHub files that is a great help. I keep learning more proving that an old dog can learn new tricks.
solar whale
Glad to help a fellow "old dog"
tidal kiln
@timber mango the A variant of MCP2221 is what is being used, just being lazy when talking about it. not sure about the CP2112 EOL thing though.
timber mango
@tidal kiln Ah yes, thanks.
idle owl
@tulip sleet Please let me know when you're around, I added the rest of the error handling to my code, and commented it, and I would appreciate another look at it.
tulip sleet
@idle owl I am back; going to dentist at 1:30
idle owl
@tulip sleet Sent the files. Controller is what has the added error handling around send_packet because it's failing to the new _bleio error.
Really looking forward to the wrapper lib you want to write, because the error handling in this code is beyond ridiculous.
tulip sleet
You can put the try with a catch-all except in send_packet() instead of in the main program. So change except (OSError, KeyError): to except: and add the pylint comment
idle owl
oh.
I thought I tried that. Probably didn't.
I'll test it now.
That works! Much better, thank you.
meager fog
@timber mango yes we are using A rev of silicon
tulip sleet
@stuck elbow I'll take a look but may not have time for a few days. Does the example help any? https://github.com/hathach/tinyusb/blob/master/examples/device/hid_composite/src/main.c. I know it doesn't do anything but the setup might help. The INVALID report type is pretty suspicious.
stuck elbow
@tulip sleet I will be away for the rest of the week anyways, and there is no hurry
tulip sleet
sure, it's in the PR's so I'll review the open ones from time to time.
stuck elbow
that example doesn't actually look at the report type, I suspect nobody ever tested it
there are some TODOs in the tinyusb code around that area too
idle owl
@tidal kiln Sorry to bug you, but if you have time to reply to the LED animations review, I'd appreciate it. I managed to back myself into a corner with needing it for a guide. I wasn't thinking it would grow like it did, or that it would take a while to get reviewed (that's on me). Thank you for reviewing it!
tidal kiln
@idle owl done
idle owl
Thanks!
manic glacierBOT
@jepler I looked at the Feather M4 Express output on a logic analyzer and it looks good to me. (No DAC connected.) It clocks at 1.04 MHz for this file. I have some new DAC boards on the way in case I broke mine.
river quest
lots of good python on hardware in this one, thanks everyone who sent in links / issues / PRs and more! https://www.adafruitdaily.com/2019/12/10/200-circuitpython-libraries-binho-ble-and-more-python-adafruit-circuitpython-pythonhardware-circuitpython-micropython-thepsf-adafruit/
slender iron
@stuck elbow it does look like HID set report isn't fully implemented
since INVALID is always reported. Does the data change at all?
stuck elbow
it doesn't support multiple interfaces, but that's probably not required
it does
I compiled that code and used it in my keyboard and it works
the report is always 1 byte long
slender iron
that sounds right. it's length is defined in the hid descriptor
and if there is only one report, then the report id is omitted
idle owl
@slender iron @tulip sleet Are we still able to do beta.1 today?
stuck elbow
@slender iron are you saying it thinks the report is invalid because there is no id?
slender iron
@stuck elbow I don't think it knows its invalid. I think it should be OUT
not sure how feature reports are handled though
slender iron
@idle owl it'd be nice to see if #2363 could be merged in first but I'm not sure if @onyx hinge is around today
idle owl
Agreed. Fair enough.
slender iron
@stuck elbow ya, I think its a typo. but I'm not sure how feature reports are handled (or what they are)
stuck elbow
hehe, I have no idea how USB works, it's going to be fun
slender iron
the usb complete book is a pretty good primer
stuck elbow
thanks for the pointers, this is really helpful, by the way
slender iron
HID is definitely more complicated than other classes
the xfer_cb is called through a function pointer
iirc
stuck elbow
yeah, in the CFG_TUD_HID
slender iron
(my brain is fresh on hid from doing it for ble)
stuck elbow
I wonder if I should just make a pull request to tinyusb changing that to OUTPUT and see what the other devs say :P
slender iron
thach would probably be fine with that
downloads the spec to see how feature reports work
stuck elbow
the golden labrador school of development
uh, sorry, I didn't mean to make you do it
onyx hinge
@slender iron to answer your question on #2363: I don't know how the same delay numbers work on M0 and M4, but the boards I tested work. The code is not code I can say I understand, it's code I lifted from the uf2 loader.
I think the hand-wave explanation is, with caches disabled on the m4 parts, they perform slower
a 180MHz board, I'm not sure if the delay values would need to be tweaked or not. But would that be different than with the previous code?
slender iron
@stuck elbow glancing at the spec doesn't explain how feature reports are transferred
@onyx hinge ya, the old code would suffer from it too. it would be nice for the new code to work better too. I think we can merge as-is and follow up later
onyx hinge
I investigated some implementations of neopixel that change systick during neopixel driving, but I don't think that's a good fit for circuitpython. those said they worked in the presence of "overclocking"..
I think it can be macro-ized to allow a compile time different delay on the hypothetical 180MHz boards
slender iron
well limor was hoping to allow dynamic overclocking I think
we can cross that bridge when we get there
onyx hinge
or rewritten to use dma'd pwm which is what nRF does
slender iron
yup
that would be good for getting neopixel off of the cpu
@onyx hinge anything else close to done that you want to get in?
onyx hinge
@slender iron nothing that is currently a filed PR. I have a simple fix for nRF playback rate bug but didn't actually PR it yet. So it depends how motiviated/time constrained you are by other things. Totally get if you want to go ahead asap
@slender iron we could dispatch to one of two copies of the code (nominal speed and overclocked speed), or clock down during neopixel writes too
onyx hinge
Thank you ! @slender iron
slender iron
#2342 is waiting for you too
onyx hinge
oof thanks
yeah I started to look at that and tried to factor MOAR and it spiraled out of control
onyx hinge
also I think it might have broken the spresense port but I didn't have a board to check on at the time. now I do.
slender iron
@onyx hinge how close is mp3?
we probably have one more release we'll do before the end of the year
onyx hinge
@slender iron now that "protocols" is merged, mp3 needs to be rebased
so let's not try to do it today
onyx hinge
I think mp3 is pretty solid but it makes people see bugs that exist in audio more generically
onyx hinge
okay
ionic elk
@onyx hinge @slender iron am I correct in interpreting that you are considering dynamic Neopixel timing? I was working on that module last week using the DWT, which should be shared between the samd and stm32 ports, but I put it off because it was having annoying variance. I can revisit it as a new PR if you two want to brainstorm a bit.
slender iron
@ionic elk later but not now
ionic elk
k
onyx hinge
yay, objective achieved: I can now add mp3file without touching anything in audiocore!
idle owl
@slender iron Why does the code check to see if any existing connections are providing a UARTService before the while True:? It is only outside the loop until it enters the loop, and it hasn't scanned or connected outside the loop. I'm confused by the inclusion of this code before the loop. ```python
See if any existing connections are providing UARTService.
if ble.connected:
for connection in ble.connections:
if UARTService in connection:
uart_connection = connection
break```
How would it have any existing connections?
idle owl
hmm. ok
idle owl
Thank you.
manic glacierBOT
onyx hinge
@ionic elk DWT sounds like an interesting way to do sensitive timing. did you start to have positive results?
ionic elk
@onyx hinge nah lol
but it "should" work
eventually
The primary issue seemed to be that the instructions to turn the GPIOs on and off were taking a long time, but compensating by subtracting an equivalent amount of time from the delay loops wasn't working as it should have.
I was extra confused because the Arduino neopixel library uses ideal timings, but I don't know how that could possibly work when just turning the pin on and off takes .3us which is a neopixel low bit by itself.
Also, my latest merge is attempting to inject lines from mpconfigboard into py/proto.c???
<<<<<<< HEAD:ports/stm32f4/boards/stm32f411ce_blackpill/mpconfigboard.h
#define FLASH_SIZE (0x80000)
//etc etc
=======
const void *mp_proto_get_or_throw(uint16_t name, mp_const_obj_t obj) {
//etc etc
>>>>>>> master:py/proto.c
@onyx hinge you wrote this new proto.c file, right? Any ideas? I didn't even know a merge could do that.
onyx hinge
what the !?
yes I wrote that file, but I have no idea what git thinks it's trying to do
prime flower
Jupyter Notebooks + CircuitPython for Computers
Wait, that's a fun GIF. Here we go...
@tidal kiln Seems like JupyterLab has some issues to be worked out in regards to displaying animated/interactive plots without first converting the graph to HTML first. Which loses the interactivity and the real-time-ness of it.
or I'm just bad at JupyterLab and Matplotlib ๐คทโโ๏ธ
ionic elk
Oh man I'm in a git hole now D:
Welp nevermind github knew what to do just fine. it was just my local git that was freaking out
onyx hinge
@ionic elk oh phew, you recovered all your work?
ionic elk
@onyx hinge yeah I think the issue was that there were commits in master via my PYB nano branch that actually deleted old versions of the new PR's files. So my git wanted to delete them but couldn't and was flipping out. But Github's commit system had no problems with it at all? Who knows.
tidal kiln
@prime flower i hacked up something that sort of worked, but i didn't use that animate feature. i'll DM it to you.
onyx hinge
OK mp3 rebased and re-tested (on nRF pwm out). It LGTM
solar whale
@indigo wedge got my serpente (USB-C) board today -- works great!
indigo wedge
Hurray :)
onyx hinge
well that's weird
#0 reset_into_safe_mode (reason=reason@entry=NORDIC_SOFT_DEVICE_ASSERT)
at ../../supervisor/shared/safe_mode.c:85
#1 0x00056fb4 in softdevice_assert_handler (id=<optimized out>,
pc=<optimized out>, info=<optimized out>) at common-hal/_bleio/Adapter.c:58
#2 0x00024a44 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(this is working on the "write file via usb" crash)
at ../../supervisor/shared/safe_mode.c:86
86 if (current_safe_mode > BROWNOUT && reason > BROWNOUT) {
(gdb) where
#0 reset_into_safe_mode (reason=HARD_CRASH)
at ../../supervisor/shared/safe_mode.c:86
#1 0x00054dd6 in HardFault_Handler () at supervisor/port.c:168
#2 <signal handler called>
#3 0x000276ee in gc_free (ptr=<optimized out>) at ../../py/gc.c:673
#4 0x08029000 in ?? ()
``` multiple personalities@onyx hinge want to sneak mp3 into the release? I haven't started it yet
onyx hinge
@slender iron sort of! I feel a bit icky about that change to fat that ended up being needed for nRF!
the one I commented on
slender iron
looks
onyx hinge
it turns out that nRF needs the (read) buffer to be word aligned but we don't provide that in the general case. mp3 triggers it frequently.
but you can trigger it just by doing an unusual read
I am positive this is one of those things you will feel I tackled at a wrong layer
slender iron
that doesn't surprise me. tinyusb does copying for that reason
onyx hinge
fat USUALLY does it via a copy too, but when you have at least 1024 bytes of contiguous read it doesn't.
other than that one thing, it has little scope to interfere with anything so I'd feel comfortable tossing it in even if it turns out to need some rework. and we know we have people who are excited about mp3 reaching master, let alone a release. but excitement isn't a reason to hurry
switching gears to work on a 3d model for a friend but I'll keep an eye on discord notifications
slender iron
its not clear to me what its doing
onyx hinge
.. or will in a minute
onyx hinge
initially, rbuf just points at buf
it's advanced as the read is partially done
there's a fast-case for when a read is at least a full sector /* Read maximum contiguous sectors directly */
slender iron
at what point does it need to be aligned? in fatfs?
onyx hinge
in this case, it reads directly to "rbuff", but there's no alignment guarantee for it
so my change just stops "unaligned rbuff" from hitting this fast path
slender iron
mem_cpy can't handle misaligned copies?
slender iron
ah
onyx hinge
full traceback at https://github.com/adafruit/circuitpython/issues/2332
the alignment check makes it skip down to where it does a disk_read into fp->buf instead, and then that is copied into rbuff with mem_cpy
fp->buf has an alignment guarantee
slender iron
would the better fix be to make spi_flash_read_data handle misaligned buffers?
onyx hinge
I liked this approach in part because it avoided any new allocations
there's already an aligned buffer available to us
slender iron
it costs a copy though
onyx hinge
now, there could be other users of spi_flash_read_data who will be sad
wouldn't fixing spi_flash_read_data also do a copy?
slender iron
read_data would only need to read a word to align the buffer and then do its normal thing
I was thinking you'd do an extra small aligned read
onyx hinge
ah okay, to just a uint32_t on stack or something
onyx hinge
OK, I'll add notes to 2332 then
slender iron
k, thanks!
onyx hinge
(I'll reopen it too)
slender iron
๐
onyx hinge
@slender iron what if the read is the last 3 bytes of flash?
will the chip return junk in the 4th byte and all is okay?
slender iron
you can still read the last three into a buffer of 4 bytes
onyx hinge
oh you think it can read "just three" ?
meager fog
@onyx hinge ok ready for your next project!?!?
onyx hinge
@meager fog I think so! Want me to look at stm32 audio or do you have a different direction?
(are you excited for mp3 in beta1? I am!)
meager fog
i ammm!
your next project is to make a zune
well, zune like ๐
MP3 player for pygamer
ivory yew
... yes.
onyx hinge
okay, I think I see what you're asking for. That'll get me into some stuff I don't know yet which is always awesome
ff/rew by time .. oof, that'll need some stuff we don't have yet:)
onyx hinge
okok
onyx hinge
so the "deliverable" is some python code, probably some graphics, and whatever core changes turn out to be needed?
meager fog
but basically - demo the mp3 playback with a simple UI
ya
you will fnd out very fast if you need to fix anything
onyx hinge
yup no doubt
meager fog
especially with buttons/screen/mp3 all at once
but also, great demo ๐
i dont think you'll be able to play mp3 from SD
onyx hinge
we know there are problems updating the screen while playing audio, I worry they are not going to be simple
meager fog
but, try it
onyx hinge
actually I did that successfully, maybe you missed hearing it
meager fog
oh ok well gr8, use Sd
onyx hinge
that was on pyportal, not sure if it makes a difference
meager fog
yeah this isnt going to be huge amt of UI
onyx hinge
pyrocker? pytunez?
orchid basinBOT
meager fog
lol
onyx hinge
is the logo a snake in a coccoon?
.. who later emerges, a beautiful buttersnake
ok can I unsee that idea now?
ivory yew
Buttersnake? Hahaha. I can't breathe.
meager fog
๐ง 
tulip sleet
@slender iron the very first sentence of the release notes said "This is release 5.0.0 beta.0." I fixed that on github
slender iron
haha, oops. ๐ thanks!
tulip sleet
np, thanks for doing the release!
slender iron
waiting for the release files to do the posts and website merge
why is it a draft now?
tulip sleet
I thought it was a draft, sorry; make it a release.
slender iron
I publish because then github adds the tag and triggers the build
tulip sleet
i hope it doesn't trigger twice
slender iron
looks like I was able to cancel it
ah, our file count is higher because we added korean
and two boards
idle owl
@slender iron Thank you!
slender iron
np @idle owl. good to keep the release train rolling
orchid basinBOT
raven canopy
@slender iron working on migrating adabot to actions. wrt the bundle update script, i've kind of hit a roadblock on adabot's email. repo secrets aren't available in Python as env vars, and non-public emails aren't available through the API (rightfully).
the only real way i can see to do it, and keep the email secret, is to encrypt it into a file in the repo, and then decrypt on-the-fly.
thoughts?
slender iron
@raven canopy you should be able to provide secrets. you have to be explicit in the yaml to tell it to provide the secret to the step
@peak thicket the image demo is done in arduino and not circuitpython I believe
raven canopy
scrolls up. sees the with: ๐คฆ
slender iron
np ๐
raven canopy
thanks!
tidal kiln
is there anything like a helper lib for playing tones? like what's in the CPX lib, but more stand alone
raven canopy
omg. i had previously used env: to pass in a secret in the library workflow. double ๐คฆ... this is what happens when you stop working on something in the middle of it.
@tidal kiln there's this, but its def not as complete as the CPX lib. https://github.com/adafruit/Adafruit_CircuitPython_SimpleIO/blob/master/examples/simpleio_tone_demo.py
tidal kiln
@raven canopy thanks. that looks like it would work OK actually.
or i can at least borrow heavily from it. and might need to adapt for CPB.
manic glacierBOT
onyx hinge
hmph "I shoulda tested"
.. playing mp3 back in stereo on samd dac doesn't work. the way it handles stereo is different than nRF, where I did most of my testing, because L/R get separate DMA channels.
tulip sleet
do you mean I2S DMA?
onyx hinge
I think I mean DAC (headphones on pygamer)
tulip sleet
oh, never mind, DAC
I wonder if there's another mode
do you mean the L and R need to be in different buffers, or just that you set up two DMA transactions? I have a vague memory of a way to set an increment for DMA so the data bytes/words don't have to be consecutive.
so L/R could still be interleaved
onyx hinge
I don't know the specifics yet. I recall that there is special handling in the audio sample API ("single channel") that I think is related. I only just now tested this combination, and it doesn't work, so now I need to figure out what's up.
I assume that stereo works, generally, though I'll try a wave file to be sure
onyx hinge
huh a wave file with these specs shouldn't be a problem ... should it ? bwv-854-abr-fm.wav: RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, stereo 44100 Hz
(it doesn't sound right)
manic glacierBOT
manic glacierBOT
quasi pulsar
Iam just starting out so having very basic doubts , lets assume I made a school project in circuit python kit on adafruit kit , if I want to make it as a product what would be next steps
for example : Adafruit ItsyBitsy M0
ionic elk
@stuck elbow did you mention you had worked with the Meowbit before?
main meteor
At a minimum, you could just buy ItsyBitsy boards, load your code on them, and resell them. But I could be misunderstanding what you have in mind.
quasi pulsar
@main meteor , I mean how would you make it like a finished project with circuit enclosed in plastic
@main meteor lets assume my project has components more than what I need , how could I reduce the components I dont need from my end product
main meteor
Oh, you're thinking of a custom board and enclosure, that's somewhat more advanced.
ionic elk
@quasi pulsar You could start by looking at the schematic for the Itsybitsy or a feather, and get familiar with what's included. In general, the strict minimum that a processor needs is the processor itself with all the right pins connected to power/ground/left alone as appropriate, capacitors on the power lines, a set of crystal oscillators (which generate a clock signal that the processor uses) and typically a voltage regulator to make sure the voltage coming in is appropriate. The ItsyBitsy is pretty stripped down already so it's probably close to the minimum already.
main meteor
You could also look at the Trinket, which is even more minimal/small.
ionic elk
Then you'd need to learn how to lay out a circuit, with Kicad or Eagle or whatever. Then you need to make sure you know how to upload firmware to it with something like JTAG, without the use of an Arduino bootloader, which won't be on there by default (though you could add it later)
Or, as @main meteor says, you can just take one of these minimal boards and stick it on an expanded PCB you design. There's no shame in that, tons of people do it to save time/money. I've seen itsy bitsys/etc mounted in 3d printers, drones, all over the place.
ivory yew
It also super depends on your project. There's lots of things out there that use the dev board (itsybitsy/whatever) and just have all the peripheral components on another circuit board and connect them together.
haha jinx, @ionic elk.
ionic elk
lol we on the same page. When I did my first big project I was very obsessed with doing it the "right" way, which lead me to spend a ton of time switching over to STM32 and new "professiona" toolchains that didn't even save us anything. It'd probably have been way cheaper if I'd just mounted an arduino on there for the first 500 units.
ivory yew
Here's an example: The let's split keyboard uses an Arduino Pro Micro and the PCB for it just holds the keyswitches and diodes: https://mehkee.com/products/lets-split-pcb?variant=5227242749980
@ionic elk yeah, I had similar thought process when designing winterbloom sol, but ultimately I did want to learn to make my own self-contained boards so.
main meteor
I've been known to do exactly that, like this project where I just put a Teensy on top
ionic elk
Ofc, doing it entirely yourself is a good learning experience. But you should be mentally prepared to mess up a couple times if you do.
ivory yew
Yep. First revision of Winterbloom Sol didn't work at all. I forgot one of the power connections and totally wired the programming header to the wrong pins.
ionic elk
It's so easy when you're new to it to accidentally forget a power line, or not read up totally on a boot setting, or whatever. It can be kinda painful, especially if the board literally blows up on you.
main meteor
There's also this one where I implemented an Arduino-like circuit along with my own customizations
ionic elk
I had a project where we didn't connect VBAT, and had to manually fix like 15 boards with an x-acto and magnet wire. So yeah - leave yourself space for a couple screw-ups
ivory yew
Is that all through-hole? It's so pretty.
ionic elk
I like to call these puzzle-piece engineering
ivory yew
lol
The beta boards have a teeny bodge because KiCad's footprint for tac switches isn't quite what I expected.
ionic elk
I think it'd actually be rad if you could take that even further - swap the 0.1" connectors for something super duper low profile and design them to be ordered in bulk with a bootloader already attached
you could probably make a business out of that... hmmm....
orchid basinBOT
orchid basinBOT
main meteor
Thank you, yes, those are through-hole.
The last boards I made that needed a bootloader flashed used an ordinary 0.1" rectangular header pattern. For the prototype boards, I'd solder on a header as I was flashing them a lot. For the production boards, I used pogo pins and the same pads. These days, I'd look at TagConnect as it's compact and headerless.
I've also gotten fond of spring loaded pogo pin clips for the 6-pin FTDI style connectors
swift arrow
wooooohoooooo adabox 014 is in da mail!
manic glacierBOT
slender iron
@tulip sleet you around?
tulip sleet
yah - I was out for hours but am back.
slender iron
I'm confused about the current ordering of things in the fs vs what they are now
does the existing nrf linker already have a bleconfig area?
tulip sleet
not really - there's a mention in the comments: ...
hold on for link
but I don't think that's even right.
slender iron
I agree the goal is to have the fs in the same spot as before
and if we have flexibility to change the ble config size that is a bonus
slender iron
how many bonds can we store in 32k?