|
finally back from easter holidays, my mailbox was overflowing with ebay crap + some circuit boards i've started putting some actual effort into making my own electronics package for one of my ericsson NMT 450 UHF receivers (transmitter electronics will be the next project). i'll be redoing the control electronics, af processing, local oscillator, and discriminator. if you know your radio block diagram that's everything except the first converter (which i might do later, but gotta limit the scope a little). all these things can basically be done independently as separate boards, that limits my project risk since it'll be fairly modular. so far i've made and built PCBs for the local oscillator and discriminator. local oscillator was crap, the use of a CDCE913 isn't terrible (it can't hit every frequency exactly for 6.25 khz spacing but there's a way around that). the problem was using a sine output clock and a 2 layer board. a nice combination of both close-in and broad band phase noise all over the place due to ground bounce. redid that board on 4 layers with a nice vectron clock and better output buffers, should get here next week. discriminator includes the second converter and channel filters, i did a design with some 21.4 MHz filters i found on ebay and an MC3374 as the mixer + limiting amp. i was going to tap the signal and use a tracking PLL as a discriminator but after spending an entire day on it i gave up and made a tank discriminator. that worked, and i found some high Q inductors from murata that might do an ok job for the final layout. as is the board does work with mod wires, but squeezing a few more dB of noise floor out of it might be possible if i improve the crystal filter match. the LC tank will also need temperature compensation so i'll add that. will do a new layout after more experimentation. now for something that actually worked right, the BCD thumb wheels on the front:  the NMT spec had 180 channels, the thumb wheels on the front of both the receivers and transmitters allowed manual channel selection in the base station (normally these would be remotely controlled through an override). two of the digits are 0-9 BCD output, the hundreds digit is a decimal output (10 discrete switch lines). i was able to remove the limit pins on the hundreds digit switch to get the full 0-9 range on that too. wiring up 18 lines to my controller board (not yet designed) seemed kind of silly, so i made two boards that have pads at the same pitch as the switch pads and put an I2C I/O expander in there instead. for the decimal version i added a 74HC147 priority encoder to convert the decimal code to BCD, to the software it looks the same as a BCD wheel. two jumpers set the address of each digit. boards are basically mounted pad-to-pad, for some extra rigidity i added a nice fillet of loctite 3450 epoxy to hold both boards to the plastic frame, pretty stiff assembly now. and i only need 4 lines to read three digits.
|
#
?
Apr 22, 2017 16:55
|
|
|
|
| # ? Apr 28, 2024 12:53 |
|
|
nice
|
|
#
?
Apr 24, 2017 03:19
|
|
|
one of our facilities guys stopped by the other day and mentioned that there's a bunch of hornet nests building up along the top of the building in a location where they can't just reach them from the roof or via a scissor lift. their best solution was to invest in a drone capable of lifting a can of bug spray to squirt the nests but they needed help coming up with a flying bug spray actuator could i please help out with that. it didnt seem like the best idea ever but gently caress it i was like 'yeah sure, just get a spray can trigger handle to actuate the can and a geared-down solenoid that works with the drone's power supply and we'll make that work for you assuming you have an extra RF channel on the drone control to actuate the sprayer' and the guy walked away happy. the project may further involve making a 3D printed adapter connecting the spray handle to the drone's payload mounts and a bunch of other bullshit but it's a kinda fun doohickey so i'm not kicking back on the 'what if we just ignore these dumbass hornet nests that bother nothing' front. i'm also ignoring factors like 'are you actually a good enough drone pilot to fly up to those locations under normal wind conditions' and 'what if the downdraft of the rotors disperses the bug spray' or 'will spraying a bunch of poo poo out of a flying gizmo push you backwards' issues. gently caress it, even if not economical these kind of random-rear end dumb projects are what makes engineering fun
|
|
#
?
Apr 25, 2017 05:05
|
|
|
PDP-1 posted:one of our facilities guys stopped by the other day and mentioned that there's a bunch of hornet nests building up along the top of the building in a location where they can't just reach them from the roof or via a scissor lift. their best solution was to invest in a drone capable of lifting a can of bug spray to squirt the nests but they needed help coming up with a flying bug spray actuator could i please help out with that. do this and post a video also send the video to hackaday
|
|
#
?
Apr 25, 2017 06:08
|
|
|
make sure you're not spraying bees
|
|
#
?
Apr 25, 2017 06:56
|
|
|
I made a stupid Android app for capturing timelapses because the first couple I tried on the app store were poo poo. It shouldn't crash anymore I don't think. https://play.google.com/store/apps/details?id=com.technologyisawful.simpletimelapse
|
|
#
?
Apr 25, 2017 16:36
|
|
|
DuckConference posted:I made a stupid Android app for capturing timelapses because the first couple I tried on the app store were poo poo. how hard was it to do all of that? it seems like apps are mostly total poo poo and i'd like to make my own but i have no idea how to go about doing it
|
|
#
?
Apr 25, 2017 19:40
|
|
|
Corla Plankun posted:how hard was it to do all of that? it seems like apps are mostly total poo poo and i'd like to make my own but i have no idea how to go about doing it I'm basically new to Java and OOP and the Android API and the intellij ide so I probably spent 4-8 hours figuring out "oh god what is any of this" Probably spent about 40 hours, but someone who was familiar with all of the APIs and making Android apps probably could have made it in 4 hours.
|
|
#
?
Apr 25, 2017 20:08
|
|
|
Got a wild hair up my rear end and starting looking into emulation. Currently in the process of writing an interpreter for the chip8 in c++. Finished the disassembler yesterday and today got about 10 instructions working on the virtual hardware. Enough for one of the small games to run. After that, I have to figure out how to actually draw poo poo from the memory map to an sdl window using opengl.
|
|
#
?
Apr 26, 2017 04:52
|
|
|
i'm finally gettin somewhere with the cyberpunk 80s motorcycle instrument cluster i got the test display talking to the ARM main processor at a reasonable speed. i'm not super happy with the time it takes to update (25ms to push out the full buffer -- i feel like it has to be possible to do better than that) but it's smooth enough, and i'm really not doing much hard math with that controller. it's mostly just reading from sensors on a loop, making some decisions, and formatting it all for display. i think it'll be usable without too much more tweaking. also have an attiny in the system. it's going to be dedicated to reading the front wheel rpm (speed) and spark coil pulses (engine rpm). i got that reading a hall effect pickup on an interrupt and sending that over 1-wire serial to the ARM. the idea is that i can use the existing cable drive from the front wheel and spin a magnet up inside the instrument cluster instead of having to do the hacky thing most retrofit motorcycle dashboards do and mount a magnet on the wheel itself (blech!). i mocked up and 3d printed a little wheel that spins a magnet in front of the sensor and chucked it in a drill to verify that it works and the screen updates pretty nice and smooth, up to 25mph-equivalent anyway. (drill doesn't go any faster) e: oh durrr i just realized i forgot to multiply the tire diameter by pi. that's why it the reported speed seems unusually slow. durrrrrrrrrrrrrr note to self, fix that  next step is probably firing up the motorcycle with a scope connected to see what sort of pulses i'm looking at from the coil pack, then integrating that into the tiny's code as well. i think i should be able to handle both sets of signals with one tiny85 but worst-case scenario i add another one -- they're like 90 cents after all. oh, and the custom full-size OLEDs finally arrived (see pic). i haven't hooked them up yet but they use the same controller as the testing display so it shouldn't be an issue. 
Sagebrush fucked around with this message at 08:26 on Apr 29, 2017 |
|
#
?
Apr 29, 2017 07:41
|
|
|
more instrument cluster stuff the previous setup was only receiving one byte at a time, just verifying that everything works. in reality the ATTiny needs to transmit more info to the ARM. at least: - speed (0-100mph expected) - engine RPM (0-10000RPM expected) - wheel revolutions (for odometers, approx 865 rev/mile) and i need this data as quickly as possible because i can expect a new trigger as often as every 6 milliseconds. i'm using interrupts but both the input and the communication are timing-sensitive so yeah. gotta packet this data up and minimize the write time. so speed can fit in under 1 byte, engine RPM in 14 bits but just round up to 2 bytes. wheel revolutions is a little trickier but 16 bits won't do it, because that'll roll over at about 75 miles and the bike can go 160 or so on a tank of gas. i could write some kind of rollover-handling code but eh i did it another way. the next available variable is a uint32_t but I don't need to count up to 5 million miles. so i decided to just pretend it's a 24-bit number and write the lower 3 bytes, saving 8 bit-times from the write cycle. won't roll over for about 19,000 miles. i ended up building my own packet protocol with one start byte and 6 data bytes -- plus the occasional extra escape character for when the start or escape bytes appear in the data. i'd never really thought about how bitstreams are handled at a low level like that so it was cool to implement it on a low level and figure out how it works. no stop character or checksumming bc the data length never changes and if i screw up a packet eh whatever it'll re-synchronize on the next cycle. anyway, the long and short of it is i now have a reliable packet handler transferring all the information i need at the resolution i need, and at 115200bps the whole cycle lasts about 660 microseconds -- a tenth of the theoretical ceiling.   (for some reason rigol puts a ? when there's not enough room for the scope to show the decoded value...when you zoom in on the trace it gets it bang-on. quirky) i am finding that i really like bit twiddling, thinking about individual packets of electrons flying around and single memory cells flipping back and forth. somehow it's easier for me to grasp than high-level factoryfactoryfactory code.
|
|
#
?
May 1, 2017 01:01
|
|
|
Sagebrush posted:more instrument cluster stuff it's because the factoryfactorybuilderfactory stuff obfuscates what's going on
|
|
#
?
May 1, 2017 01:11
|
|
|
Sagebrush posted:
woah where/how did you source a custom OLED display, and what was the minimum order quantity?
|
|
#
?
May 1, 2017 01:44
|
|
|
ah it's only semi-custom tbh. it's a standard size, but they only had them in green. i talked a talk about being a university professor and doing electric vehicle work and they spun up a line and made two in amber (god's own color). $40 each at that production rate, it just took like 2 months to arrive. i did get one fired up with the little driver board for the mini OLED and they are plug and play but quite dim -- i need a much beefier power supply to achieve full brightness. so i gotta work that out next.i am now up to the point of successfully reading a simulated tachometer and speedometer pulse simultaneously, processing the data into RPM and speed, and sending it to the main controller with what looks like zero glitches. i'll do a short running average on the main controller and that should be all the filtering and post-processing i need. as a side effect, i added another feature: since i know the wheel speed, the road speed, and the motorcycle transmission's gear ratios, i can make a good estimate of what gear the bike's in. normally gear indicators use sensors in the transmission and are almost impossible to retrofit. woot
|
|
#
?
May 1, 2017 07:22
|
|
|
Sagebrush posted:as a side effect, i added another feature: since i know the wheel speed, the road speed, and the motorcycle transmission's gear ratios, i can make a good estimate of what gear the bike's in. normally gear indicators use sensors in the transmission and are almost impossible to retrofit. woot isnt the moment you need to have that information right when you shift? it won't be able to know what gear you're in if the clutch is in and you're able to rev arbitrarily. i guess you could have a clutch switch signal so you know when that's pulled, and not change the display until the clutch is let back out and it's able to figure it out again. but isnt that moment when the clutch is in and you're changing gears the moment youd probably want to know what gear you're in/going into? i guess if you don't ever look at it while shifting it doesnt matter! 
|
|
#
?
May 2, 2017 01:38
|
|
|
anyone know good tutorials on audio filter design?
|
|
#
?
May 2, 2017 01:48
|
|
|
Raluek posted:isnt the moment you need to have that information right when you shift? it won't be able to know what gear you're in if the clutch is in and you're able to rev arbitrarily. i guess you could have a clutch switch signal so you know when that's pulled, and not change the display until the clutch is let back out and it's able to figure it out again. but isnt that moment when the clutch is in and you're changing gears the moment youd probably want to know what gear you're in/going into? you are right that it won't be accurate when the clutch is disengaged and you're moving, but actually in the moment of shifting it's not super important to know what gear you're in (imo). it's a sequential gearbox, so each shift action will just put you one gear higher or lower, and the motion is always the same...so just knowing what gear you're in before the shift is the important info. the shift also takes about 500ms if you do it right so the moment of uncertainty is very brief. the main reason i would like one is because i'm a compulsive top-gear checker. whenever i'm riding along in 5th i keep poking at the shifter with my toe, checking yea, yea, still top gear, yeah, yeah. all motorcyclists are either compulsive top-gear-checkers or compulsive turn-signal-cancellers (they don't shut off automatically) or both. and yes there is a clutch switch and a neutral switch, so i can probably wire into those and add some logic to avoid the undefined behavior entirely.
|
|
#
?
May 2, 2017 02:07
|
|
|
I like to think that I'm chill enough that if I was a motor cycle man that I wouldn't be either. I wouldn't even care if I was screaming along in second gear revving the tits off it. yolo, if you know whaT i mean
|
|
#
?
May 2, 2017 02:35
|
|
|
C.H.O.M.E posted:anyone know good tutorials on audio filter design? what kind of filter: speaker crossover, eq, synth, formant, etc? analogue (active/passive) or digital (convolution, fir, iir)? also, what's your current level of audio fundamental knowledge?
|
|
#
?
May 2, 2017 08:43
|
|
|
Sagebrush posted:the main reason i would like one is because i'm a compulsive top-gear checker. whenever i'm riding along in 5th i keep poking at the shifter with my toe, checking yea, yea, still top gear, yeah, yeah. all motorcyclists are either compulsive top-gear-checkers or compulsive turn-signal-cancellers (they don't shut off automatically) or both. yeah I drive a car and I frequently check the tach to make sure I'm in sixth
|
|
#
?
May 2, 2017 13:34
|
|
|
ynohtna posted:what kind of filter: speaker crossover, eq, synth, formant, etc? analogue (active/passive) or digital (convolution, fir, iir)? I have some knowledge of audio, mostly analog, but I have at least seen digital filters in school but that was years ago. I have taken some analog and RF classes, can read a BODE blot, I can read a pole-zero diagram. have taken classes on integral transforms and I am brushing up on FFT now. I am going in for a digital filter design interview and I need to at least work through designs of some FIR and IIR digital filters before that happens. Wikipedia has good info and one simple example of an IIR and FIR filters but it's not exactly a tutorial.
|
|
#
?
May 2, 2017 16:18
|
|
|
C.H.O.M.E posted:I have some knowledge of audio, mostly analog, but I have at least seen digital filters in school but that was years ago. cool stuff. i make many trips to this well when revising digital filters: https://ccrma.stanford.edu/~jos/filters/ good luck! e: also http://www.analog.com/media/en/training-seminars/design-handbooks/MixedSignal_Sect6.pdf ynohtna fucked around with this message at 17:17 on May 2, 2017 |
|
#
?
May 2, 2017 17:11
|
|
|
ynohtna posted:cool stuff. very cool, thanks!
|
|
#
?
May 2, 2017 18:00
|
|
|
luv 2 see 2 m8s help 1 another out
|
|
#
?
May 3, 2017 01:21
|
|
|
echinopsis posted:luv 2 see 2 m8s help 1 another out keep your sex life in your e/n thread
|
|
#
?
May 4, 2017 13:11
|
|
|
im addicted to spreading my diseaSE 
|
|
#
?
May 6, 2017 07:27
|
|
|
I got this for like $10 at valvoline world headquarters when they were clearing out old buildings   Need to figure out if they work
|
|
#
?
May 7, 2017 20:45
|
|
|
moron izzard posted:valvoline the famous xnem character
|
|
#
?
May 8, 2017 20:10
|
|
|
oh man I want that HP thing anyway I think instead of trying to reverse engineer the Atari BIOS ROM I'll write my own replacement, especially since I don't have any documentation, tools, or entry points for it my game currently doesn't use it for anything beyond booting and hardware exception handling (which it does badly) so being able to split support routines and memory management off into a separate binary would be nice
|
|
#
?
May 9, 2017 00:16
|
|
|
i sorta want babby's first logic analyzer but i need something that will support differential input (for CAN lol). what are my options, yospos
|
|
#
?
May 9, 2017 02:10
|
|
|
Super-cheap but effective: Get a bootleg Saleae from China for like $15 and convert it to single-ended with an amplifier.
|
|
#
?
May 9, 2017 02:26
|
|
|
got the 486 working again w/ dos6.22 and win3.11, and just bought a bunch of poo poo from ebay to see if i can break it again: - 8x4MB SIMMs to max out the 32mb system capacity - swap out the 486 SX 25 with a 486 DX2 66 - the existing CPU is SMT but there's a separate socket2 port for dropping in an upgrade (and i didnt want to deal with lovely fans on pentium overdrives) - an actual sound blaster(R) 32(TM) ISA card - gonna toss a couple of the current 1MB SIMMs into there - an IDE CDROM to replace the proprietary-via-soundcard mitsumi 2x kid cad is gonna really be blazing once i get this puppy upgraded ps: it's also already got a working 10mbit isa ethernet card (netgear ea201 iirc)
|
|
#
?
May 9, 2017 06:20
|
|
|
ugh I remember those proprietary CD-ROM drives there's a photo on wikipedia of a sound blaster with 4 separate CD-ROM headers (Mitsumi, Sony, Panasonic, and generic). you have to enable the right one with jumpers. 3 of them are 40-pin headers. lol
|
|
#
?
May 9, 2017 06:29
|
|
|
Progressive JPEG posted:got the 486 working again w/ dos6.22 and win3.11, and just bought a bunch of poo poo from ebay to see if i can break it again: ok but where's the gravis ultrasound?
|
|
#
?
May 9, 2017 15:40
|
|
|
Olivil posted:ok but where's the gravis ultrasound? They're really expensive for some reason, I'm gonna guess there's some audiophile bs going on with them? I just wanted something that would be easy to find drivers for. The card its got now is some kind of off-brand clone that I'm p sure didn't ever really work right, even when the system was new
|
|
#
?
May 9, 2017 16:37
|
|
|
the motorcycle project owns my dream is to build a 6502-based homemade computer of some sort. you can order brand-new CPUs and 6522 VIAs straight from Mouser and there's lots of instructions and projects online such as this engineer's instructions on building a basic 6502 computer/crash course in system architecture
|
|
#
?
May 10, 2017 07:50
|
|
|
moron izzard posted:I got this for like $10 at valvoline world headquarters when they were clearing out old buildings I am literally supporting something of this era at this point. . . the guy who bought it way back when is now writing code in *shudder* excel to do what he bought it for. New era hardware (the stuff he had it for controlling) would run like $500. . . He's spent a year now trying to make it work in excel to keep the old hardware.
|
|
#
?
May 12, 2017 14:49
|
|
|
while waiting for PCBs i decided to strip down my NMT 450 transmitter to clean out the rust and get measurements so i can start planning the board layouts to reiterate this is a 50W continuous 450-460 MHz FM transmitter used for analogue mobile phones from the 1980s, designed and built in sweden front panel, i removed all the front panel connectors to get the cable harness out, added the fuse holder since this series transmitter presumably had external fusing  back of front panel, the BCD channel selector is one of the I2C mods i did a while back. there's a stud mount 16V zener diode in the bottom right corner that protects against over/reverse voltage  PA low pass filter/coupler, removed for cleaning and modification. this cleans up the output from the PA to remove harmonics. the mounting hardware for this was pretty rusty after 10-15 years in a basement. i'm pretty sure it's mostly made of aluminium oxide, but there's probably some beryllium in there too. internally it's a sandwich of two white ceramic plates with a super thin PCB inside, then a fairly thick silver plated something outer shell. the input is the BNC on the left, there's an SMB near the middle which is a -40 dB output coupler that goes to the front panel for monitoring, and the stuff near the output is a directional coupler (~-40 dB bidirectional). i removed the original rectifier diodes and interface board from the directional couplers and routed + ordered a new board that uses two AD8307s to get better dynamic range. plan is to use the forward for power levelling and reverse power mostly for informative measurements so matching isn't super critical. the original diode detectors were fine but they have a limited measurement range and regulating output power over more than a decade (e.g. 5-50W) becomes very difficult since the detector gain changes significantly  under the can is the actual PA assembly, only modification is replacing the 0.5 ohm current sense shunts on the PAs with a 10mOhm ones to reduce losses and adding a little extra supply decoupling. parts remaining: local oscillator modulator af processing exciter amplifier with gain control various interconnect and filter boards main controller should keep me busy for a while, but some parts can be shared with the receiver
|
|
#
?
May 14, 2017 12:08
|
|
|
planning out the graphics for my motorcycle instruments, and i figured i'd make a simulator why not, so i can work out the exact pixel dimensions and prototype animations and poo poo top and bottom are LED strips for the tachometer/idiot lights/night mode/etc and the four circles are MFD buttons. Processing is great for hacking out graphical stuff like this really fast but it's also real lovely at handling accurate bitmapped fonts for some reason so i'm still working on that Sagebrush fucked around with this message at 09:53 on May 15, 2017 |
|
#
?
May 15, 2017 07:16
|
|
|
|
| # ? Apr 28, 2024 12:53 |
|
|
continuing to work on the new 2020 thing https://soundcloud.com/danwarren/2020-chapter-3-version-10 in this chapter we meet america's deathwish and get to laugh at donald trump's dick Trig Discipline fucked around with this message at 07:41 on May 15, 2017 |
|
#
?
May 15, 2017 07:35
|
|