#avr Logs

Sep 17 2020

01:56 AM rue_bed2: I been switching to stm32F103
02:03 AM rue_bed2: as the avr line is being killed by microchip
02:03 AM rue_bed2: its still so nice to make a simple little project with a basic avr tho
03:55 AM day: rue_bed2: i still dont quite get what microchip got out of the atmel takeover :/
03:56 AM nohit: they got atmel
04:12 AM nohit: its was the only option really
05:36 AM vmt: stm32f3 gas guzzler omg!1
05:45 AM nohit: and is there something wrong with microchip manufacturing the avrs now?
05:46 AM nohit: at least they brought cheap programmers/debuggers
05:48 AM nohit: and what i have heard, there's some improvements in the chips
05:52 AM nohit: rue_bed2: what do you mean "avr line is being killed by microchip" ?
05:53 AM nohit: has something changed? i recall the ceo/cto of microchip promised that they will keep the avr lines
06:02 AM vmt: well, the old pic vs avr flamewars were not exactly based on rational thinking. now that avr==mcp, avr must be bad, OR then mcp must be doing some black magic fuckery because it's mcp, and we ought to hate them
06:05 AM vmt: though people do get set in their ways, and i understand that. old beards here have probably been using avrs for close to two decades. and that's fine, just that trying to rationally justify their decisions is just an exercise in futility
06:09 AM vmt: not all opinions are equally invalid, of course. just that in mcus often times "better" means "i'm used to this"
09:04 AM rue_mohr: microchip is going to screw it all up, their well on their way
09:05 AM rue_mohr: the policies that I abandoned microchip for haven’t gone away.
09:06 AM vmt: mcp, the ole gas guzzlers
09:06 AM rue_mohr: vmt, you sound like someone with a F150
09:06 AM rue_mohr: do you drive an F150?
09:07 AM vmt: i maintain that americans can't build proper cars. nor ice or electric.
09:08 AM vmt: nor can they design them and get them manufactured in some third world shitholes, either
09:08 AM rue_mohr: I was waiting to cross a street out in the middle of nowhere most of the vehicles were white F150
09:09 AM rue_mohr: I bet you would pull a quick NIMBY if someone tried to manufacture chips next door to ya
09:09 AM vmt: so are you finally willing to admit what kind of an obsolete piece-of-shit gas guzzler the 1284/p is?
09:10 AM rue_mohr: ... you seem to keep mentioning that one chip....
09:10 AM vmt: because that was the subject
09:11 AM rue_mohr: do you want to comment on xmega ?
09:12 AM vmt: i see no real point in xmegas with the availability and tooling out for cortex-m, plus pricing
09:12 AM rue_mohr: oh good we agree on something
09:13 AM vmt: and i would imagine xmega gets slashes off the product line in the not-so-distant future
09:13 AM vmt: slashed
09:13 AM rue_mohr: https://twitter.com/skunkgod1/status/1306596509291405315
09:14 AM vmt: great
09:14 AM rue_mohr: there was like one hickup, the rest, for a rubber chicken, was damn well done.
09:15 AM MrMobius: maybe theyll be making them forever like all the PIC variants
09:16 AM rue_mohr: their going to pull a programming protocol change to all the chips, they wont release the specs for it, the only programmers that will be able to program the new chips will have to be a programmer puchased directly from microchip
09:16 AM rue_mohr: and the chips will only be different by a postfix letter
09:17 AM rue_mohr: at which point china knockoffs will step in and everyone will forget that authentic chips ever existed
09:18 AM nohit: sure, everything is going to shit
09:18 AM vmt: xmegas and a lot of cortex-m variants compete in the same market segment, but there's cmsis for arm and... atmel shit for avr. the way i see it, they should kill the xmega production line and just bite the bullet and focus more on arm
09:18 AM MrMobius: doubtful. that could have happened with other chips and hasnt
09:19 AM MrMobius: as far as chinese knock offs taking over
09:20 AM vmt: rue_mohr: hahahah give me a fucking break lmao. you're pissed at mcp for proprietary shit? rofl. what about debugwire and if you wanted to use it you had to get a fucking atmel dragon/ice/whatever for an atmel(tm) premium price
09:21 AM vmt: but because it's atmel(r)(tm) device, it's okay to pay a premium
09:22 AM vmt: and don't start with the usability of debugwire. it's not the point. the point is that atmel made it strictly proprietary.
09:23 AM vmt: not to mention atmel obsoleted a lot of those icds to push new ones out to the market
09:24 AM vmt: atmel akbar
09:30 AM gurki: well at some point they had to consolidate their avr/pic stuff ...
09:31 AM gurki: tbh: im surprised this happens so late after the acquisition
09:32 AM nohit: but does it matter
09:32 AM nohit: you cant still get the old chips if that's what you want to use
09:32 AM nohit: *can
09:32 AM vmt: now it doesn't. use cortex-m and stop buying those f150s... i mean avrs
09:34 AM nohit: and learning a new microcontroller is fun
09:54 AM vmt: yes we need more microcuntrollers
09:56 AM nohit: sure, why not
09:57 AM vmt: yes
09:57 AM twnqx: just use x86. modern ones are SOCs anyway, with megabytes of internal memory
09:57 AM nohit: the choice isnt always yours so it shouldnt be something to be afraid of
10:01 AM vmt: +1 for modern x86. i think we should all spin up n layer boards
10:02 AM vmt: actually fuck it, just deadbug the sucker
10:03 AM vmt: then you can run linux, kubernetes restful apis and the whole nine yards. now we're cooking with gas!
10:08 AM gurki: vmt: im horrified by the fact that embedded is in the process of transitioning to that right now
10:09 AM gurki: need more iot!
10:10 AM vmt: well, i think in general diminishing returns have hit the "technology" sector hard, and it has more or less stagnated in a lot of areas
10:12 AM TechChristoph: well some of us use avrs some of us armcortex
10:12 AM TechChristoph: who cares about aslong both groups have fun with their device
10:20 AM nohit: you wont find that here
10:23 AM merethan_w: When Atmel puts stuff in their datasheet as PortD and PortE, are these two distinct modules on silicon? I´m investigating hardware damage and wonder if damage to ¨port D¨ can/may affect other peripherals in this section.
10:25 AM merethan_w: Obviously breaking a mux will, but I mean, say, damage to USB more likely to break SPI in the same port.
10:25 AM merethan_w: More likely in the same port than the neighboring port.
10:56 AM Rab: merethan_w, https://zeptobars.com/en/read/atmel-atmega8 https://zeptobars.com/en/read/atmel-avr-at90usb162
10:58 AM Rab: I can't answer your question, and I think it probably depends on the type of damage anyway: ESD, overcurrent, etc. I am also no expert at physical chip layout. But I see a bunch of pins in both those parts which are surrounded by similar structures with large features, and I suppose those are the port pins. There are other, fewer pins which seem to have dedicated functions i.e. power.
11:00 AM Rab: So what that suggests to me is that the power drivers and pin protection are at the individual pin level, with the port-level logic being somewhere inside the chip. So for many failures I would expect the pin involved to be damaged, but maybe not adjacent pins, and maybe not the logical port or adjacent ports as a whole.
11:02 AM Rab: Of course if ESD makes it past the protection diodes, there is no telling what it might damage in the chip.
11:03 AM Rab: Unfortunately I can't find any die photos with the bond wires still attached, or a block diagram superimposed.
11:06 AM Rab: I have not seen a lot of AVR failures, and the ones I have seen were related to soldering overtemp rather than I/O trauma. If you want to know whether a conservative failure-tolerant design would partition functions e.g. across opposide sides of the chip, I suspect it's not worthwhile.
04:56 PM LeoNerd: I have a scaled integer that stores a current reading in multiples of 0.1mA... is there a good name for that unit? The only thing I can think of is decimiliAmps
04:56 PM LeoNerd: dmA
04:56 PM LeoNerd: Which kinda sucks
04:57 PM vmt: for the end user?
04:57 PM LeoNerd: Oh it's all internal code.. I just like to keep the units on it so I'm clear this is "scaled and calibrated", and not a raw ADC measurement
04:59 PM vmt: i'd just call it frac or something and be done with it
05:06 PM vmt: however if this is for a typedef, i've never really understood typedeffing for semantics and i don't think i ever will
07:35 PM Rab: LeoNerd, it's either 0.1mA or 100uA. I think you can call it whatever you want.
08:33 PM rue_mohr: vmt, funny, all the way thru, I was able to use parallel port programmers
08:34 PM rue_mohr: I didn't dig into T10, cause .. I just cant find a project with only 4 signal lines
08:34 PM rue_mohr: er 3
11:19 PM day_ is now known as day