|
Bloody posted:we have good threads for software things but we don't have anywhere for hardware really so now we do pcie fpgas have hard ip for it and it's fast as hell otherwise for usb, find a mcu that has a decent (lol) usb peripheral or get a ulpi usb xcvr and strap that to a fpga ftdis are probably the most decent option, they have tons of poo poo they can do in silicon -- the ft2232 can do a ton of stuff with its mpsse (i2c, spi, jtag, etc)
|
# ¿ Nov 25, 2014 20:24 |
|
|
# ¿ Apr 29, 2024 11:44 |
|
also avrs suck hth
|
# ¿ Nov 25, 2014 20:25 |
|
Bloody posted:atmegas suck but xmegas are like lower-power more featured msp430s personally i would de-feature msp430s out of existence, i hate the loving things (mostly the fram family) and documentation Bloody posted:pcie sounds cool as hell but hard like how the hell do i write a pcie driver or whatever on the pc side?? on linux or windows? linux pci subsystem isn't half bad, you can write a simple driver in a couple of minutes that'll say hello world upon detecting your device and expose bar0 via an ioctl for you to blink leds at 8Gb/s
|
# ¿ Nov 25, 2014 20:32 |
|
the only thing i'll really give the msp430 fram series is their power draw of like 1.2mA @ 10MHz at 3.3V, without trying to do any lpm poo poo
|
# ¿ Nov 25, 2014 20:33 |
|
Bloody posted:yea debugging sounds stressful. i only really need like 10s of mbits per second so idk pcie feels overkill for that physical layer you need a very expensive oscilloscope generally to confirm / track down SI issues luckily, the channel is mostly built for you on a given motherboard -- if you're building an add-in card, don't gently caress up your short traces to the edge connector, match impedance and you're generally good 10s of mbits, either usb or ethernet i guess. probably usb. or spi
|
# ¿ Nov 25, 2014 21:47 |
|
Bloody posted:how do i spi from a pc modern intel chipsets have a spi controller on them; no idea if you can do dma and such with them, but they exist. unfortunately, they're usually attached to the spi memory that holds your bios, so uh, i wouldn't use that. pretty sure the embedded chipsets though (for like industrial atom) have available spi controllers yeah, you could do a ftdi part for spi, but at that point you may want to just make your device itself the usb endpoint. basically comes down to, integrate usb into your hardware, or interface with an external usb<->* bridge like the ftdi that could give you parallel, spi, etc Illusive gently caress Man posted:one time i soldered an atx 24 pin molex to a PCIE socket to try and power a piece of hardware without needing to shove it in a case. it didnt work. oops Bloody posted:also atmel studio owns because it is visual studio i'll give you that, visual studio owns bones
|
# ¿ Nov 25, 2014 23:00 |
|
proto-advantage will not only sell you a breakout board, you can tell them what chip you want and they will assemble it for you
|
# ¿ Nov 25, 2014 23:23 |
|
Sagebrush posted:yeah it's nothing about soldering it by hand it's that i literally have been too lazy to get the breakouts because i have a bunch of DIP attinys sitting around and tbh i kinda like making dead-bug circuits the only good bug is a dead bug
|
# ¿ Nov 25, 2014 23:38 |
|
ChiralCondensate posted:I (in a team with others) janitor these bits:
|
# ¿ Nov 26, 2014 05:04 |
|
z0rlandi viSSer posted:lmfao who gives a poo poo as long as it works Nah man every cycle counts I finished a FPGA design lately where I had to essentially pick and choose and place each primitive by hand. a more painful asic
|
# ¿ Nov 26, 2014 07:15 |
|
eschaton posted:I'm hoping my miniSpartan 6+ LX ships soon, I should start hacking together a VHDL description of NuBus for it spartan bro gonna do some hdmi overlay type stuff with mine though -- i hope the signal integrity isn't terrible on the board, most kickstarters suck rear end at laying out pcbs and fall over in the transition from maker fun toy to actual product actual quote from a maker i ran into: 'controlled impedance? what's that?' (board had ddr3, usb, ethernet, hdmi) last i heard they were suffering 'difficulties' getting kickstarter stuff sorted
|
# ¿ Nov 27, 2014 01:58 |
|
Bloody posted:Yeah but only if by decent you mean expensive and horribly complex igloo nano is good, i'm using that dev board to prototype for real stuff digilent makes good altera dev boards, the bemicro sdk is a good dev board as well that just plugs right into a usb plug, has a cyclone iv on it for xilinx spartan 6s are the little ones, a kickstarter just finished for the miniSpartan which might be OK. the zynq is pretty legit, you can get get the avnet zedboard and make your own custom linux machine which is legit all of those options are around like $100-$200 i think
|
# ¿ Nov 27, 2014 05:44 |
|
Mido posted:are there x86 core MCUs that have the same kind of features as say a Kinetis K60 series? I've never looked the embedded stuff (atom, etc) comes closest, but even then, iirc its just multiple dies inside and the pch has all the peripherals -- x86 cores interface to any/all peripherals over pcie or dmi one of the intel guys in the thread will probably correct me on that
|
# ¿ Nov 27, 2014 05:46 |
|
Blotto Skorzany posted:can you give me a litany of complaints about them, the fr57xx is one of the chips in consideration for a design i'll be working on in a couple of months sw dude complaints: - lovely toolchain (ccs, mspgcc, gcc-msp430), iar wasn't considered because it's windows only - lovely docs - weird-rear end isa (certain operations very expensive like bit shifts ) also forget about doing intensive math hw (my) complaints: - lovely docs - weird pinouts sometimes - fr5969 has funky errata relating to leakage currents on certain pins - weird rear end fram memory controller timings that changed from stepping to stepping - took them six steppings to get release silicon out which is mildly terrifying - you will need a reset supervisor or otherwise clean reset source, code will corrupt in brownout conditions (the SVS module on the frxxxx parts for some reason is slightly different than other msps) the fr57xx series is older and their first go at the whole fram thing; new designs i'd check out the fr5989, if you can get it (i bought the first batches out of the factory) like i said though, i'll give them props for the absurdly low power consumption and fram reliability
|
# ¿ Dec 1, 2014 07:00 |
|
eschaton posted:how many gates would an FPGA reasonably need to support to emulate a Lisp Machine CPU? the low-end parts from altera or xilinx should be able to do it easy (spartan-6, artix-7, cyclone iv or cyclone v) -- there was a recent kickstarter for a spartan6 devboard which would work, or check out digilent's de-0 or de-1 i didn't see what cpu the machines used on wiki, but for 1980s vintage, i don't think you're running out of gates anytime soon.
|
# ¿ Dec 1, 2014 07:02 |
|
kwinkles posted:i haven't really played around with it too much but i think the quark will have the old FSB architecture and the edison board that has silvermont cores will be the faster and lower power option. i know silvermont has the same IDI interface to the system agent that the big cores have now, and most of the stuff that is on-die will be hanging right off the system agent with IOSF or the intel memory requestor busses. yeah gailelo (whichever one came first) is the og pentium cores on a newer process. do either of those new boards expose gpios via io space? seems like a no-brainer place to put some i/o for x86 so you can just outb() inb() all day long just stay away from outb(0xf5, 0xb2) (probably not a problem on those guys though)
|
# ¿ Dec 1, 2014 07:05 |
|
eschaton posted:the Lisp Machines used custom CPUs specifically designed for ease of Lisp implementation. (which isn't the same as "running Lisp in hardware" as some people claim.) they also had writable microcode, so the lowest-level operations could be made as fast as possible ahh, i see -- i still don't think you'd have a problem with the larger gate count spartan / cyclone devices, sounds like an interesting project though (especially to re from the existing emulators). wonder what the original were clocked at
|
# ¿ Dec 1, 2014 07:46 |
|
i'd be surprised if either the nexys-3 (spartan 6) or nexys-4 (artix 7) wouldn't be enough -- hopefully you can get academic pricing microsemi's a low cost option but i think you'd need the igloo2 to fit everything
|
# ¿ Dec 1, 2014 08:03 |
|
Blotto Skorzany posted:thanks! i'll keep this stuff in mind, esp. the need for a separate voltage watchdog and the toolchain concerns. the ISA stuff is kind of worrying too. cypress fm25vxx series? i'm using those as well, sadly no one makes fram larger than 2mbit. you could always try borrowing time on a linac to really try and induce failures, but fram is radiation-resistant, so... sw complaint/tip: - rtc peripheral on the 5989/6989 is more useful than the one on 5969 oh, i loving forgot (i can't believe this): hw complaint: - msp430 is not jtag compliant. at all. you cannot put it into a chain with other devices. if it's in a chain, it must be the first and only device. the hw responds to any jtag access with a fixed id based on revision of silicion, and you have to twiddle TEST with microsecond timing to put the part into jtag mode. or you can use spi-by-wire which again is not compliant with anything - silicon bsl is uart-based mostly unless you're lucky and can actually purchase the one with an i2c bsl so keep that in mind since you're designing an actual product that will need end-of-line testing / possible field upgrades
|
# ¿ Dec 1, 2014 18:40 |
|
Werthog 95 posted:so if i wanted to make my own, what's the hip good microcontroller for making usb adapters (that doesn't require me to use TI's lovely loving dev environment) pic serious answer: whichever micro someone's already used (assuming it isn't loving horrible) to do something similar since that will save you time and effort unless you don't care about that
|
# ¿ Dec 1, 2014 18:42 |
|
Blotto Skorzany posted:fm24vxx series, the i2c part instead of the spi one. i hope cypress doesn't do something lovely with the ramtron product line (not that ramtron was an especially well-run company before the buyout). i've also tried to get some samples of mram which is another promising tech but everspin doesn't want to play ball without a much higher potential volume than i could ever guesstimate with a clean conscience cypress guys are alright, they just revved the fm25xx parts so i assume they cleaned up the process somewhat. seem to want to support the parts for awhile yet -- what sucks is that the commercial programmer they recommend is $1k, so i had to write my own because gently caress that noise
|
# ¿ Dec 2, 2014 02:29 |
|
longview posted:workin on a little radio project with a PIC24FJ128GA204 right now, 44-pin QFN, 16-bit, up to 96 MHz clock, 3 SPI, 2 I2C, 4 UARTs, most functions are remappable which is new compared to the PIC18s i've used before (thought that was still a high end DSP feature) the remappable pins are sick as hell, love them on the pic24s. i think the only ones that can't really move around crazy like are i2c because you need a specific i/o cell for that another thing i love about pics is that you can program hosed up fuse (config bit) settings and assuming you didn't accidentally set a security fuse, you can fix that with no special programmer (i.e. still isp). avrs are like 'gently caress you, high-voltage time'
|
# ¿ Dec 2, 2014 02:31 |
|
DuckConference posted:been working with one of analog device's fattest audio DPSs (adsp-214xx) for over a year, as many complaints as I have about the libraries and support, the chip and IDE are pretty legit it's like you've never seen the 1984 documentary film 'the terminator', the machines will eventually kill us all
|
# ¿ Dec 2, 2014 02:32 |
|
Blotto Skorzany posted:what do y'all use for lab notebooks? i've been using national computation books but they don't lay flat and tend to fray/dogear more than i like vela workings makes good poo poo; i had work buy a shitload of the series-a2 and series-a2b
|
# ¿ Dec 5, 2014 08:30 |
|
zedboard's probably the best, and its bsp doesn't blow (if you want a zynq anyway) otherwise check digilent
|
# ¿ Dec 8, 2014 02:13 |
|
sund posted:coworker heard i liked jigsaw puzzles and brought me some sweet vendor swag i have that one, it's fuckin' rad Blotto Skorzany posted:one of the new super-SARs, or have you been saddled with the builtin adc on a psoc? ltc and ad have solid high-end adcs, complete with giant $$$ tag
|
# ¿ Dec 9, 2014 08:04 |
|
Olivil posted:it that jim williams's desktop^ yep
|
# ¿ Dec 10, 2014 07:19 |
|
Iirc they shrink wrapped and protected his desk the turbo encabulator is buried in it, never to be seen
|
# ¿ Dec 10, 2014 07:20 |
|
Sagebrush posted:is the EEVblog guy on youtube good or is he a wacko? I watched some of his reviews of oscilloscopes and he seems to know what he's talking about, but he also sounds like he's just about to go off the rails, and I can't tell if he actually is or if it's just, his, infuriating, breathy, australian, ACCENT??? he's a cool dude generally, he does that full time and loves the hell out of what he does
|
# ¿ Dec 19, 2014 21:26 |
|
Jonny 290 posted:godfuckingly fast opamps and components agilent fabs their own frontend components on an InP process for their 90000 series, 32ghz of true analog bandwidth (that's the 5th harmonic of 12.5Gb/s and third of 21)
|
# ¿ Dec 19, 2014 22:57 |
|
and the probe tips are ceramic and break really easily, which is sad because they're really goddamned expensive
|
# ¿ Dec 19, 2014 22:57 |
|
Tin Gang posted:also while testing junk I realized the old adjustable dc power supply I have is rated for up to 6 amps apply it to your nipples
|
# ¿ Dec 23, 2014 00:03 |
|
bobbilljim posted:what do you call a decorative component on your 'board? Bloody posted:Lmfao
|
# ¿ Dec 23, 2014 06:48 |
|
All this talk of clock speed, and no mention of edge rate 10Mhz tells you the period of the signal, doesn't tell you poo poo about the edge rate of the signal; if you've got some shiny new parts driving 10mhz on a tiny rear end process, you could have near sub-ns rise times, which is a frequency much higher than 10mhz here's a good explanation: http://www.freelists.org/post/si-list/3dB-or-Knee-Frequency,3 Si-list is legit, even if it's sometimes full of grey beard fights, but these guys design 100G+ telecom backplanes, so I listen
|
# ¿ Dec 26, 2014 21:10 |
|
fritz posted:it's going to have power @ all frequencies? it's hard for us to make perfect square waves in nature (impossibru), so in reality the edges do have non-infinite slope. smaller, faster transistors can drive these edges incredibly fast compared to larger, slower transistors from last decade ill try to sketch something when I'm back home
|
# ¿ Dec 29, 2014 21:48 |
|
Sagebrush posted:nah they were yellow. sparkfun switched to red after that incident. they put up a blog post too iirc where they played the hurt victim and tried to shame/guilt fluke for you know, trying to enforce their poo poo. Fluke ended up sending them some real ones though. I've got a 8 year old 87 III, it owns bones. i don't remember what the 87 v adds over it (just a TC input I think)
|
# ¿ Jan 1, 2015 17:42 |
|
longview posted:Anyway you should all get the Keysight U1272B or better instead of Flukes, I don't care for the Fluke 289, it's slow to power up and the menu system involves a lot of navigation to get to common modes. this thread reminded me i don't know that much about fluke's lineup of dmms (i've just always used the 87 iii since i was a wee intern), and that higher model #s don't mean better the 287 and 289 seemed 'better' than the 87, but i got the vibe that they were clunkier to use menu-wise -- the 87 you just clunk the knob to whatever and probe things like 1 second afterward; thanks for confirming that vibe the 179 seems ok, but it looks like the 87 v retails for like $350, which for something that has a lifetime warranty and will last at least a decade, if not more, is a loving steal basically, this thread should go get $350 and buy a 87 v
|
# ¿ Jan 2, 2015 00:10 |
|
hobbesmaster posted:so when you have a file which is literally called "raw binary file" you should just stream it bit by bit into a flash to be loaded later right they don't have a sync word like 0xAA995566 from xilinx land? i miss altera a little bit
|
# ¿ Feb 23, 2015 20:12 |
|
eschaton posted:are there any examples you can point to of good or bad Verilog, VHDL, and so on? fpga tooling on linux is only good for running headless builds imo; if you need to use the gui at all, stick with windows. linux is a second-class citizen (as it should be)
|
# ¿ Feb 23, 2015 20:13 |
|
|
# ¿ Apr 29, 2024 11:44 |
|
sund posted:i cut my vacation short a day to assist with some qualification testing that we ultimately failed and spent two weeks poring through code and manuals, staying late making one off builds that i knew didn't make sense just to satisfy someone else's theories and pestering support engineers at another company to figure out why their product was flaking out when we were trying to using it. finally stole the scope from the hardware department today and found out the power bus drops half a volt below the absolute minimum voltage required for the flaky part when it performs a certain operation. three people assuring me it was rock solid and double checked, while hinting that i was probably toggling a non-existent reset line. loving up (non-exotic) power supplies is inexcusable in my book you have to basically ignore your datasheet entirely, or have dumb-rear end system-level requirements
|
# ¿ Feb 23, 2015 20:15 |