#avr Logs

Oct 27 2021

#avr Calendar

07:30 AM vishwin60 is now known as vishwin
10:33 AM user____: .
10:33 AM user____: Hi. Is MPLAB X use in the attiny / avr context on topic here? With XC8 as it is installed by default, platform linux.
10:34 AM user____: I am interested in the new xtiny parts which are cheap and have some punch, UPDI only programming, small case (SOIC14 is what I'm most iterested in)
10:35 AM user____: I'll be quiet and lurk for a while for answers. Thanks.
10:48 AM nohit: sure, but i dont think many people use mplab x here
10:49 AM user____: Well, does avr-gcc support for example attiny828 or 848 ... seems not to, not even microchip's "special" avr-gcc release, which seems to be OLD.
10:49 AM user____: (downloaded yesterday)
10:52 AM user____: And I am not at all against bare metal makefile + avr-gcc and or avr-as use, that's where I come from, I am new(ish) to avrs on MPLAB, thus the questions.
11:03 AM user____: Is this channel usually quiet? As quiet as it used to be on freenode? ...
11:04 AM rue_mohr: nobody has realyl been active on irc latley
11:04 AM rue_mohr: channels that used to be boiling over are dead
11:04 AM user____: Right, we are all ghosts. IRC is a ghost party ;)
11:05 AM user____: No need to get depressed by it though. Another way to look at it, is, the hotheads and the asl crowd moved elsewhere.
11:05 AM qu1j0t3: are you saying.... the eternal september... is OVER?
11:06 AM user____: No, it was just misunderstood...
11:07 AM user____: Also, swarms move all the time, the swarm moved.
11:07 AM qu1j0t3: it's always September _somewhere_
11:08 AM user____: No... Only once per year, ~30 days.
11:12 AM user____: slighlty off topic but embedded relevant https://www.theregister.com/2021/10/27/pax_technology_warehouse_raid/
11:12 AM user____: So, anyone here using PICKIT 4 with linux for avr programming?
11:13 AM user____: Re: UPDI minimalist programmer: https://www.electronics-lab.com/usb-updi-programmer-pcb-for-avr-micrcontrollers/
11:14 AM user____: CH330N
11:15 AM user____: What would be better to get to work under linux, possibly with bare metal makefiles: pickit 4 or snap ?
11:18 AM specing_ is now known as specing
11:35 AM user____: Heh, Mouser lists SNAP at about $30 but weight 56kgs ;) Should be 56 grams, I guess.
11:36 AM qu1j0t3: tungsten cube
11:38 AM user____: Well, SNAP seems to be the way to go. It can't do HV UPDI unbricking, otherwise, it does all the Pickit 4 does. No Linux drivers (yet). Seems to be the usual deal of USB connected general purpose Microchip MCU with IOs mapped to programming functions and host loadable firmware.
11:39 AM user____: So it will work with MPLAB X and we can wait for linux drivers.
11:40 AM user____: SNAP is a bare module board, no box. Should be easy for someone to cook up open source firmware for it. It does all Microchip current products sans HV requirement, AVR, ARM, SAM.
11:40 AM user____: Mouser has both on stock.
11:47 AM nohit: user____: have you checked these http://packs.download.atmel.com
11:47 AM rue_mohr: Its a good thing you can use anything to program an avr via ICP eh?
11:47 AM nohit: if any of those have support for your chips
11:47 AM user____: MPLAB X downloads those automatically.
11:47 AM rue_mohr: paralell port serial usb, pata
11:47 AM user____: >> nohit
11:48 AM nohit: yes but for avr-gcc support
11:48 AM user____: rue_mohr: yes but the new parts are a bit iffy in startup entering programming mode.
11:48 AM user____: rue_mohr: there's a HV pulse handshake magic sequence to unbrick.
11:48 AM rue_mohr: oh well dont use them then
11:48 AM user____: They're good and cheap!
11:49 AM rue_mohr: esp if it doesn't work with avrdude
11:49 AM rue_mohr: their not good, you cant program them
11:49 AM user____: The CH330N programmer is all one needs with UPDI. There's a Python driver.
11:49 AM rue_mohr: being able to program them would be a requirement on 'good'
11:49 AM user____: BUT if you fubar the fuses you need HV pulse entry to unbrick.
11:49 AM user____: There are solutions (arduino based), but it's a risk / pain/
11:49 AM rue_mohr: again, then its not good
11:50 AM user____: Honestly, UPDI is back to basics. UART programming on one wire, autobauding. JUST DON'T FUBAR THE FUSES.
11:51 AM rue_mohr: I think your pointing at a peice of **** insisting its good, despite it being horrible
11:52 AM user____: Look, there's one spot of crap on the back side of a large cake... We can work around it? <wink>
11:52 AM rue_mohr: so far, you have said it lets you brick it via fuses, wont start the bootloader right, doesn't have software support, and programming hardware isn't avilable
11:52 AM rue_mohr: and thats were you set the "good" bar?
11:52 AM nohit: if you want to use those new atinys with avr-gcc, download that pack
11:53 AM nohit: its seems to support at least 828
11:54 AM user____: nohit: I am aware how it works and it's already down inside MPLAB, I can locate(1) it.
11:54 AM nohit: ok, you were just asking if avr-gcc supports those chips
11:54 AM user____: Just not clear how I id the xtiny family they belong to.
11:54 AM nohit: that's the way of doing it
11:55 AM user____: Iirc one needs to edit the config files and copy them where avr-gcc expects them.
11:56 AM user____: Also version numbers on packs on that page do not correspond with pack versions reported internally by MPLAB X
11:57 AM user____: Are those xtiny or xmega?
11:58 AM user____: tiny
11:58 AM user____: aka avrxmega3
11:58 AM user____: sigh
12:01 PM user____: https://aur.archlinux.org/packages/avr-libc-avrxmega3-svn/ supported at least in source releases
12:04 PM user____: Right, I just verified avrxmega3 support is in my default distribution's avr-gcc. Devuan Linux "Beowulf" + avr-gcc (GCC) 5.4.0 -- yes it's old
12:09 PM user____: When I'll get around to doing this, is this channel "helpful" should I get stuck? :) Following https://www.avrfreaks.net/forum/atmgea328pb-avr-gcc-osx or similar?
12:10 PM twnqx: well, people here seem to mostly use a variant of a pretty common Makefile for their projects
12:10 PM rue_mohr: if anyone is alive :)
12:10 PM rue_mohr: but I myself stick to the orig avrs
12:10 PM user____: Right, look, I'm in Romania, we're all vampires apparently...
12:10 PM rue_mohr: well, I'm trying to move away from them, #microchip screws things up lots and arm is taking over everything
12:11 PM rue_mohr: china has a promising looking m238 knockoff
12:11 PM user____: I like the small 14 pin SOIC attinys. They solve some problems I used to solve with PICs
12:11 PM user____: Yes there are also STM32F106 knockoffs. GD something?
12:11 PM twnqx: PICs have some weird constraints, like their limited function call nesting
12:12 PM rue_mohr: paged memory, instruction clock dividers, pics suck.
12:12 PM rue_mohr: lots of fancy perphials, but
12:12 PM twnqx: the I/O mapping is neat
12:12 PM rue_mohr: microchip didn't even want to support C
12:12 PM user____: I used PICs for bit banging apps since 1995 or so. Still have some projects.
12:12 PM rue_mohr: and were keeping the programmers topsecret
12:12 PM user____: bknd C did work around the paging well.
12:13 PM twnqx: how did microchip even get as big as they are today...
12:13 PM rue_mohr: paging sucks
12:13 PM user____: No, programming algorithms were open in datasheets.
12:13 PM rue_mohr: so does 4 clocks/instruction
12:13 PM rue_mohr: better than 8051 at 12 tho!
12:14 PM rue_mohr: thats the other thing, 8051...
12:14 PM twnqx: look at 8088! 4 clock cycles for one real cpu tick!
12:14 PM user____: Yes, but that was a financial decision. Scenix did 1 clock ones in 1998. 100MHz 100MIPS PIC
12:14 PM rue_mohr: there are a lot of current chips out that are 8051 based
12:14 PM twnqx: i... have original microchip 8051s DIP40 incoming tomorrow :D
12:15 PM rue_mohr: there are some 14 pin things too eh?
12:15 PM twnqx: dunno, i need them for a snark barker
12:15 PM user____: rue_mohr: yes, Attiny824 for example
12:15 PM rue_mohr: I got a china clock kit with one
12:15 PM user____: Also Silabs and Dallas made MCS51s going as fast as 80MHz
12:15 PM rue_mohr: twnqx, hah, you assembling a snark barker?
12:16 PM rue_mohr: what you using it in?
12:16 PM user____: What is a snark barker
12:17 PM rue_mohr: knockoff sound blaster
12:17 PM rue_mohr: by... uh...
12:17 PM rue_mohr: whatsHisName
12:18 PM rue_mohr: if you dont know if it, look it up, the silk screen is priceless
12:19 PM rue_mohr: next you gonna go a graphics gremlin?
12:19 PM user____: Wow @pcb
12:19 PM rue_mohr: I think I'm at fault for the name, I pushed him to honor the poll he did on the name
12:20 PM user____: Does that implement the MIDI player synth? OPL?
12:20 PM user____: Wow 8 bit ISA card?
12:20 PM rue_mohr: its a deep reverse engineer of the sound blaster
12:20 PM * user____ made such in the 1990s for $work
12:21 PM user____: Do you have anything to plug it into ?
12:21 PM user____: You know about these, right? https://hackaday.io/project/167541-8-bit-isa-prototyping-card https://texelec.com/product/8-bit-isa-prototype-card-v1-0-pcb/
12:22 PM twnqx: i read about those, yes
12:22 PM twnqx: i intend to build upon the snark barker for my own card (only have one slot left)
12:23 PM user____: One slot left in *what*? :)
12:23 PM twnqx: cpu subsystem is a Xi 8088, not sure yet in which configuration (cpu/fpu), that depends on the outcome of tomorrow's test :P
12:23 PM twnqx: oh, a µATX 16bit ISA backplane
12:23 PM user____: Xi 8088 is unknown to me. V30?
12:23 PM twnqx: so, 4 isa slots
12:23 PM user____: https://retrocomputing.stackexchange.com/questions/11001/how-to-use-isa-card-in-modern-pc one learns something new every day
12:23 PM twnqx: no, a semi-faithful ibm XT clone
12:24 PM user____: USB-ISA sounds interesting.
12:24 PM twnqx: i want to figure out if any of my CPUs works with a NEC µPD9008D FPU
12:24 PM twnqx: any 8088, that is - i'll start with a V20
12:25 PM user____: V20 was much better than 8088. I had a XT clone with V20 - very happy with it.
12:25 PM twnqx: but that's not good enough to run a certain PC demo that really needs to be cycle exact, so i hope either my intel 80c88-al2 will do
12:25 PM twnqx: or the oki msm80c88
12:25 PM user____: Well the V20 has turbo/non turbo modes.
12:25 PM twnqx: it has a different IPC
12:25 PM user____: Non turbo is 4.77MHz
12:25 PM twnqx: so it's "too fast" :D
12:26 PM user____: It is drop in pin compatible with 8088, iirc non turbo mode is 1:1?
12:26 PM twnqx: no, even at same freq it's faster at it needs much fewer cycles/instruction on many of them
12:26 PM user____: http://arstech.com/install/ecom-prodshow/usb2isar.html
12:26 PM user____: Ah yes, now I remember.
12:26 PM twnqx: ah, that thing
12:27 PM twnqx: well, i have a few HT310 here...
12:27 PM twnqx: usb host for the XT incoming!
12:27 PM twnqx: it will be hard to get more silly than that.
12:28 PM twnqx: oh, i also have an original IBM CGA card for that machine
12:28 PM rue_mohr: heh
12:28 PM user____: again this link has a lot of good links https://retrocomputing.stackexchange.com/questions/11001/how-to-use-isa-card-in-modern-pc
12:28 PM rue_mohr: do you have a CGA monitor?
12:28 PM twnqx: the problem is
12:28 PM user____: rue_mohr: I think a modern lcd with vga to cga cable works
12:28 PM twnqx: the interesting demos use NTSC for color artefacts
12:28 PM twnqx: 1024 colors out of a CGA!
12:28 PM user____: Oh, noes.
12:29 PM user____: I know about those.
12:29 PM rue_mohr: lcd might
12:29 PM twnqx: no
12:29 PM rue_mohr: the sync rates are out to lunch tho
12:29 PM user____: rue_mohr: the more CNC/industrial ones do. I know, I was amazed to see it work.
12:29 PM twnqx: you need at least a linedoubler
12:29 PM user____: Not ntsc colors, CGA RGB mode, not analog.
12:30 PM rue_mohr: and unless you pull fancy stuff, composite doesn't work
12:30 PM twnqx: that's still only 14.... MHz
12:30 PM rue_mohr: well, never did for me
12:30 PM twnqx: yeah, i have issues getting it to work, it's not as compatible as i would like
12:30 PM user____: Yeah, finding a working color crt monitor to accept crazy hv rates will be "fun"
12:31 PM twnqx: i hope to solve that with a small fpga for linedoubling, and emulating artefacts
12:31 PM user____: In the 1990s I fried a bw monitor with an overclocked Hercules card. The HOT blew up after dozens of hours of torture with high H rates.
12:31 PM twnqx: it's not much more than sampling 4 pixels (16bit) and generating "the right" artefact color from it, then buffer the line
12:31 PM user____: It grew a hole in the side, spitting blue sparks.
12:31 PM twnqx: :S
12:32 PM rue_mohr: you know, there is the graphics gremlin?
12:32 PM twnqx: yes
12:32 PM rue_mohr: it works with modern monitors
12:32 PM user____: Later models did not permit the crazy H rates for a reason. See above for reason.
12:32 PM twnqx: but that's a pure linedoubler
12:32 PM twnqx: afaik there's nothing out there that tries to generate artefact colors
12:32 PM twnqx: except that strange guy who tries to build a complete CGA in an FPGA
12:32 PM user____: Aren't there cheap? line doublers from the 1990s somewhere? Ready made box?
12:33 PM twnqx: tpns, but. it. is. not. enough.
12:33 PM rue_mohr: heh, yea tell me if you can find some HPGL optimizer code too eh?
12:34 PM user____: https://www.artisantg.com/TestMeasurement/68330-1/JDR-Microdevices-JDR-PR10-16-Bit-Prototype-ISA-Board ooh shiny. The ones I used in the 1990s were slightly smaller and had no buffers on them
12:34 PM twnqx: yeah, and having the '688s on too is really useful
12:35 PM twnqx: also looking forward to february
12:36 PM twnqx: 2022 must be the year of the 8088, given it's *4
12:36 PM user____: ;)
12:36 PM twnqx: so i'll get fresh ones from the fab then!
12:36 PM twnqx: (actually already paid for them...)
12:37 PM twnqx: kind of crazy they are still being made, must be some military contract in play
12:37 PM user____: Thanks for the memories. I have to move and cook a bit.
12:37 PM user____: twnqx: it says surplus nos.
12:37 PM rue_mohr: I wonder how small an 8088 could be shrunk to using modern processing
12:38 PM twnqx: https://www.mouser.de/ProductDetail/Renesas-Intersil/CP80C88-2Z says sopmething different :D
12:39 PM twnqx: that's the matra-harris - harris - intersil - renesas military semiconductor chain
12:40 PM twnqx: they even have a die shot in the data sheet
12:42 PM twnqx: the milspec are crazy expensive, got a NOS quote of 140€ earlier this year
01:24 PM user____: re
01:24 PM user____: Oh I saw a CDP1802 space grade price, close to 20000 $, years ago
01:25 PM user____: Probably every transistor on the die was verified by a separate virgin.
01:25 PM qu1j0t3: or it's 98% markup
01:26 PM qu1j0t3: 9800%
01:26 PM * qu1j0t3 goes to lunch
01:26 PM user____: Nono.
01:28 PM user____: https://chipscapes.com/products/the-rca-1802-microprocessors-in-space
01:30 PM user____: You know the cosmac elf online javascript simulator? I like it. Someone should make an avr one like that <wink>
01:33 PM user____: 1802s were used in equipment we built at my 1st work place, I'll always remember that...
01:45 PM user____: References to space rad hard 1802s at bottom http://www.decodesystems.com/cosmac/
01:48 PM user____: Was there ever a rad hard space grade Atmel part? I know some mil and automotive ones might qualify anyway.
01:49 PM user____: http://ww1.microchip.com/downloads/en/DeviceDoc/00002200C.pdf hmm rad hard and space qualified parts
01:50 PM user____: AtmegaS128 ?
01:51 PM user____: I guess I answered my own question. I'll be quiet now and read.
02:00 PM user____: I kid you not, Avnet has 115 AtmegaS128's https://octopart.com/atmegas128-md-hp-microchip-119990788
02:03 PM twnqx: and then you call them and find out "oops, must be an error"
02:03 PM user____: Depends where you call from, I guess. For me, from Eastern Europe, they'll just "shoot back".
02:04 PM user____: I have not seen a price anywehre for these parts, not even the plastic encased atmegas128
02:04 PM user____: Been looking for 15 minutes.
02:07 PM user____: Interesting for resurrecting old computers: https://fjkraan.home.xs4all.nl/digaud/arduino/ppread/
03:42 PM exp: wow I've been trying to integrate some stuff with home assistant platforms
03:42 PM exp: and dear god the state of them is diabolically bad
03:43 PM exp: my favourite so far, home assistant has a graph option for its dashboard, that is the single worst graph renderer i've ever seen
03:43 PM qu1j0t3: :[
03:43 PM exp: it seems to pick about 6 points at random and then just draws lines between them
03:43 PM qu1j0t3: screenshot?
03:43 PM qu1j0t3: o
03:43 PM exp: hold on let me re-add the card so you can see
03:46 PM exp: qu1j0t3: so this is the default: https://i.imgur.com/eu5RbSt.png
03:47 PM exp: but i discovered it has a 'show more detail' button, which makes it look like this: https://i.imgur.com/JteaYEi.png
03:47 PM exp: can you guess what the underlying data looks like?
03:47 PM exp: it looks like this: https://i.imgur.com/5ImyL6I.png
03:48 PM exp: that graph is also available in the UI, but it occupies about 20% of screen width and can't be horizontally aligned, so it's also unusable!
03:48 PM exp: no wonder so many people export this data straight to influx and use grafana
03:54 PM qu1j0t3: wow.
03:54 PM exp: the entire thing feels like this
03:54 PM exp: barely hacked together nonsense
03:55 PM exp: MQTT in general i'm very sceptical about
03:55 PM exp: it's a product that doesn't have a purpose and can't scale effectively that i can see
04:45 PM rue_mohr: looks like solar power data
04:57 PM exp: rue_mohr: check out the last link to see the raw data
04:57 PM exp: it's just a fridge :-p
04:59 PM rue_mohr: oh
05:05 PM exp: yeah not very thrilling
05:05 PM exp: can't believe that someone would ship that though
05:06 PM exp: then again this same software thinks power is only measured in W/kW and recommends you file bugs to remove this weird "VA" measurement ☺
05:24 PM rue_mohr: kva will reflect the phase angle
05:24 PM rue_mohr: if they said kw they would probably be wrong
05:55 PM exp: rue_mohr: indeed but the web ui literally rejects VA as a form of power measurement
05:55 PM exp: which is quite amusing
05:56 PM exp: having said that i don't see any competitor that can natively convert W to even kW without manual hacks
11:18 PM specing_ is now known as specing