|
Microcontroller advice is definitely going to be biased to whatever someone has used lately, there are many options, and most of them will work just fine. Personally, I'd start to steer away from anything 8-bit, I think those are all dying. There are some good low power ARM M0 dev boards from ST, or the ESP32 is cool and gives you some potential future WiFi capability. A little bit more challenging to reduce power usage, but it's doable
|
#
?
Jan 26, 2020 19:20
|
|
|
|
| # ? May 9, 2024 08:57 |
|
|
I'm getting to the point where I think I should stop writing unnecessary blocking code. I've seen the tutorials that use state machines to implement blinking LEDs or Button debouncing, but I'm having some trouble imagining how to implement something more complex. I think I'm having trouble wrapping my head around it because It feels "wrong" for me to use global variables for all my subroutines. Does anyone have any suggestions for lessons or videos to help me apply state machine (or other non-blocking programming methods) beyond a simple linear task?
|
|
#
?
Jan 26, 2020 19:46
|
|
|
What do you want to do?
|
|
#
?
Jan 26, 2020 19:47
|
|
|
KnifeWrench posted:I would never trust my alarm to something that isn't plugged into the wall, sleep mode notwithstanding. Is there a reason you're trying to make this battery operated? A simple wall wart could obviate all of your power concerns, at least for your initial prototype. As figured out, the flipping design is what was driving me to a battery powered device rather than a wired one where the cable would get twisted. I could use some sort of circular connector like a barrel jack that can spin when plugged in maybe? Something fancier like a slip ring as mentioned seems like it's getting pretty complicated for what I was imagining. As long as the battery in this thing can last a reasonable amount of time (a week, minimum) and I can give some indication that the battery is low (less than 24 hours remaining I guess), I'm pretty OK with using a battery powered alarm clock. csammis posted:The MSP430 by Texas Instruments is the MCU I use in power-sipping applications because of its extremely low current sleep states. Even the little ones would be able to handle your IO requirements without a problem. Ah you're right, I didn't mention the piezo buzzer I was planning on using. I'd like the LEDs to be on all the time. Would using PWM or something to have them very rapidly blinking so that it's imperceptible unless you wave the clock around save me significant power? It will be a 24 hour display. The 1000mAh source is maybe on the low side. The idea behind high capacity rechargeable AAs instead would mean that should the batteries get low, I can swap them out instantly rather than having to plug it in and avoid the cable problem.
|
|
#
?
Jan 26, 2020 19:56
|
|
|
Cojawfee posted:What do you want to do? I still need to write out all my requirements for my next project. Roughly I have a blower that should turn on and off based off of time and/or temperature. Also it needs to be able use mqtt for remote changes to the timer and temperature settings that should be able to take effect during a blower cycle. A future iteration of this project would be to include a series of diverter valves and the ability to queue up different things that need to be... Uhh... Blown for a certain amount of time.
|
|
#
?
Jan 26, 2020 20:12
|
|
|
Imagine recommending a TI microcontroller to a beginner, what a monster
|
|
#
?
Jan 26, 2020 20:51
|
|
|
ArcticZombie posted:As figured out, the flipping design is what was driving me to a battery powered device rather than a wired one where the cable would get twisted. I could use some sort of circular connector like a barrel jack that can spin when plugged in maybe? Something fancier like a slip ring as mentioned seems like it's getting pretty complicated for what I was imagining. They make those little magnetic disconnect USB cables, you could just power it via a 5V USB power supply and use one of those cables so it pops off if it gets too twisted. You'd have to remember to plug it back in when it did, but it'd be a lot easier than designing some slip-ring... thing. ArcticZombie posted:Ah you're right, I didn't mention the piezo buzzer I was planning on using. I'd like the LEDs to be on all the time. Would using PWM or something to have them very rapidly blinking so that it's imperceptible unless you wave the clock around save me significant power? It will be a 24 hour display. The 1000mAh source is maybe on the low side. The idea behind high capacity rechargeable AAs instead would mean that should the batteries get low, I can swap them out instantly rather than having to plug it in and avoid the cable problem. I'm also dealing with LED displays right now. PWM will save you power but probably not as much as you'd hope. How big a display are you thinking of using? The small ones (like, up to 0.8" tall digits I think) use one red LED per segment, but the larger ones (ones I'd consider "clock-sized") use two LED's or more and have to be driven at correspondingly higher voltages, using more power. Shame Boy fucked around with this message at 21:43 on Jan 26, 2020 |
|
#
?
Jan 26, 2020 21:34
|
|
|
Also keep in mind that if you need PWM that means your microcontroller can't go into as deep a sleep mode since it needs to leave the oscillator running, which uses even more power, etc. Might wanna build out some stuff on a breadboard and measure the current consumption before you commit one way or another to batteries.
|
|
#
?
Jan 26, 2020 21:44
|
|
|
Splode posted:Imagine recommending a TI microcontroller to a beginner, what a monster That’s what I learned on and what I know can be dropped to single/low double digit microamps without breaking a sweat, and Code Composer Studio isn’t total poo poo  I still use the MSP430FR series for most of my hobby projects. Professionally I work mostly with M4F cores that require a shitload of thinking and preparation to drop into a low-leakage wake up state and actually stay there for an appreciable amount of time.What’s a good hobbyist option these days for ULP?
|
|
#
?
Jan 26, 2020 22:04
|
|
|
A total beginner isn't really ready to worry about ultra-low-power devices. If you need to be thinking about microamps, you're into a space where you have to start characterizing stuff like the passive draw of power supply components and that's just too much to start with. Even just talking about Code Composer Studio is too far. If you only need to get down to milliamps, Arduino stuff with the prewritten sleep libraries will do fine.
|
|
#
?
Jan 26, 2020 22:11
|
|
|
ArcticZombie posted:Ah you're right, I didn't mention the piezo buzzer I was planning on using. I'd like the LEDs to be on all the time. Would using PWM or something to have them very rapidly blinking so that it's imperceptible unless you wave the clock around save me significant power? It will be a 24 hour display. The 1000mAh source is maybe on the low side. The idea behind high capacity rechargeable AAs instead would mean that should the batteries get low, I can swap them out instantly rather than having to plug it in and avoid the cable problem. If the room it's in is really dark at night you could save a huge amount of power (relatively) by dimming it based on a light sensor. Would also make it more comfortable to sleep next to without being too dim during the day. LED displays will be the majority of the power draw regardless, you could mitigate by e.g. dimming the display more or turning it off as the battery drains to ensure that you have enough power left to actually signal the alarm. I'd also consider going for the DS3231 for the RTC functionality, it consumes more power but it's very accurate compared to a typical RTC (~60 seconds per year for 0-40 degrees C). Get a ~1F supercap for the RTC backup if you expect the battery to ever run out, then it won't lose time even if the rest of the circuitry turns off. For the DS3231 the supercap can be charged to 5V even for a 3.3V VCC so a normal silicon diode + series resistor from the 5V supply is all you need for a charger.
|
|
#
?
Jan 26, 2020 22:14
|
|
|
CopperHound posted:In the vaguest of senses, I want to remove all delays/noops from any of my code that isn't real time sensitive. Have you drawn out state diagrams? I find state machine code somewhat inscrutable and difficult to follow/write when I'm approaching them from a blank slate. State diagrams make a lot of sense to me though, and are easy to translate to code
|
|
#
?
Jan 27, 2020 00:09
|
|
|
ante posted:Have you drawn out state diagrams? I’ve used auto code to make FSMs from a state diagram / flow chart because at this point in my life / career, gently caress writing the 10247th FSM, I’m just going to draw it / document it and then let a computer spit out code. Buuuut...I’ve always had this setup with some elbow grease and am wondering now if there’s a good open-source / quick-turn solution to spit out C or Verilog from some kind of YAML/XML input? Have used MagicDraw in the past.
|
|
#
?
Jan 27, 2020 00:32
|
|
|
csammis posted:That’s what I learned on and what I know can be dropped to single/low double digit microamps without breaking a sweat, and Code Composer Studio isn’t total poo poo Excluding Arduino which is a no-brainer, my recommendation would be STMicro. It's got a free, open source tool chain and CubeMX is a godsend when you're just starting out. Atmel has very nice datasheets, and the Arduino legacy can help. TI micros are a running joke at my work though, the punchline generally being that the errata sheet is longer than the datasheet.
|
|
#
?
Jan 27, 2020 03:53
|
|
|
movax posted:I’ve used auto code to make FSMs from a state diagram / flow chart because at this point in my life / career, gently caress writing the 10247th FSM, I’m just going to draw it / document it and then let a computer spit out code. There's a XML FSM standard: https://en.wikipedia.org/wiki/SCXML. The Qt SDK includes an editor. I don't know about C/Verilog translators.
|
|
#
?
Jan 27, 2020 13:05
|
|
|
I just learned an important thing: I knew that solder creeps at room temps but never thought about the fact that it will do that when screwed down to a terminal block. What I learned is that you should never fully tin a wire being screwed into a terminal block because it will creep and loosen. Just the bare minimum tip to keep the wire together. TMYK 
|
|
#
?
Jan 27, 2020 18:24
|
|
|
If you have stranded wire, use ferrules. They're neato.
|
|
#
?
Jan 27, 2020 18:28
|
|
|
insta posted:If you have stranded wire, use ferrules. They're neato. Yea I definitely should start. And I always could use a reason to buy another crimper.
|
|
#
?
Jan 27, 2020 18:34
|
|
|
I've had good luck with the cheap Chinese 4-jaw crimper for ferrules.
|
|
#
?
Jan 27, 2020 20:57
|
|
|
insta posted:If you have stranded wire, use ferrules. They're neato. How have I just discovered that these exist this late into my life??? This is a godsend for stranded wire at screw terminals.
|
|
#
?
Jan 27, 2020 21:07
|
|
|
sharkytm posted:I've had good luck with the cheap Chinese 4-jaw crimper for ferrules. If you arent using a MIL-DTL-22520 qualified DMC AF8 on NAVAIR QPL listed ferrules don't even speak to me. Just kidding this costs $22 shipped from the USA, comes with 1200 ferrules, goes down to 22AWG which is small enough to mostly cover the stuff I do. CarForumPoster fucked around with this message at 21:52 on Jan 27, 2020 |
|
#
?
Jan 27, 2020 21:49
|
|
|
CarForumPoster posted:If you arent using a MIL-DTL-22520 qualified DMC AF8 on NAVAIR QPL listed ferrules don't even speak to me. I need to get a good deal on one of those. (The AF-8, AFM-8, HX-4). I recently found out the AF-8 can crimp coax pins; no more soldering!
|
|
#
?
Jan 28, 2020 02:07
|
|
|
babyeatingpsychopath posted:I need to get a good deal on one of those. (The AF-8, AFM-8, HX-4). I recently found out the AF-8 can crimp coax pins; no more soldering! I have an HX-4 because I needed a big hex die for a project years ago. They can be dirt cheap on eBay. $100 for an AF8. Problem is finding the crimp barrels you need for anything other than the most common connectors.
|
|
#
?
Jan 28, 2020 17:53
|
|
|
CarForumPoster posted:I have an HX-4 because I needed a big hex die for a project years ago. They can be dirt cheap on eBay. $100 for an AF8. Problem is finding the crimp barrels you need for anything other than the most common connectors. We've got the crimpers at work, but when there are three people crimping on two different projects on opposite ends of the shop, it gets to suck. Having an AFM8 of my own would be choice. That and whatever that environmental splice crimper is (for M81824/1-xx splices). I wonder if there's an HX-4 die... fake edit: oh snap there is! Hmmm, might have to invest in one of these.
|
|
#
?
Jan 29, 2020 01:36
|
|
|
babyeatingpsychopath posted:We've got the crimpers at work, but when there are three people crimping on two different projects on opposite ends of the shop, it gets to suck. Having an AFM8 of my own would be choice. That and whatever that environmental splice crimper is (for M81824/1-xx splices). I wonder if there's an HX-4 die... I bought my HX4 w/die for $40 on an eBay auction in ~2009
|
|
#
?
Jan 29, 2020 01:41
|
|
|
Not really sure where to put this but I'm sure there's a reasonable cross-pollination in this thread Electronic weather stations. There are a bunch out there, mostly from Oregon Scientific and La Crosse and AccuRite, in the range of $15-250 with rain gauges and wind indicators etc Are there any models that provide a serial out, or similar, that someone recommends? It seems like they change the model number every 90 days but they all have the same basic internals and wireless specs. If there's a thread for this I apologize, I looked but didn't find anything. Arduino etc of course has temp/humidity (I think the famous 3 wire H-11 or whatever is like $0.20) stuff you can buy but looking for a prebuilt unit. Bonus points if it has some sort of open source thing where I can import all the data and pipe it to prometheus or something.
|
|
#
?
Jan 29, 2020 08:49
|
|
|
Hadlock posted:Not really sure where to put this but I'm sure there's a reasonable cross-pollination in this thread Davis Vantage Vue and Weewx. But it's not cheap. I've been running this setup for several years if you have questions. (There are good cheap alternatives to the Davis branded serial and Ethernet adapters too.)
|
|
#
?
Jan 29, 2020 16:25
|
|
|
Hadlock posted:Not really sure where to put this but I'm sure there's a reasonable cross-pollination in this thread I have one of the all-in-one Accurite stations. The display/receiver has a USB port. Someone figured out it appears as a HID device and did some work to decode it in Linux on a RPi.. http://www.desert-home.com/2014/11/acurite-weather-station-raspberry-pi.html My station and receiver is exactly the one pictured on the first page. Also, there are some libraries out there to receive data from Accurite (and other brand) sensors with a RTL-SDR dongle.
|
|
#
?
Jan 29, 2020 16:58
|
|
|
stevewm posted:
Along those lines, you can use a cheap RTL SDR (or any other) with a homebuilt antenna to receive weather satellite images directly.
|
|
#
?
Jan 29, 2020 17:31
|
|
|
Sagebrush posted:the MOSFET you've got is fine. these are the key parts Thank you for again for this. I tried signaling the mosfet both with and without the shifter (so at 5V and 3.3V) and they both worked as expected, they both had the same current and amperage running through them when enabled. And I even made a diagram to show what I was doing. Eventually I want to get the whole thing put into that diagram since the project doesn't currently have a good diagram like that, though it may become a total rat's nest of a picture before I'm done... E: There's no resistors in my diagrams for the LEDs, but I'm not actually using bare LEDs, I'm using these switches that have a resistor built in 
FISHMANPET fucked around with this message at 07:52 on Jan 30, 2020 |
|
#
?
Jan 29, 2020 19:54
|
|
|
Tie your sources to ground and your gates to the GPIOs
|
|
#
?
Jan 29, 2020 20:35
|
|
|
Hmm I'll have to check what I've actually got wired up, I made that at work from my notes so I may have mixed up my notes vs actual implementation. E: Yup, my notes wore backwards, I updated the diagram. FISHMANPET fucked around with this message at 07:52 on Jan 30, 2020 |
|
#
?
Jan 29, 2020 21:05
|
|
|
babyeatingpsychopath posted:We've got the crimpers at work, but when there are three people crimping on two different projects on opposite ends of the shop, it gets to suck. Having an AFM8 of my own would be choice. That and whatever that environmental splice crimper is (for M81824/1-xx splices). I wonder if there's an HX-4 die... Iwiss makes a $35 crimper for M81824 / Raychem / D436 / whatever splices. We just recently bought one to try out, and it passes both go/no-go and wire pull tests with flying colors. I've got one in my Amazon shopping cart waiting for next payday. https://www.amazon.com/dp/B07T19FBXX/ If you do pick up an AFM8, try to find one of the inexpensive universal positioners too. I've seen some off-brand ones on Amazon for like $20, and I think the DMC ones are still around $100. They're not perfect for everything, but still incredibly useful and a lot cheaper than buying a pile of individual ones.
|
|
#
?
Jan 29, 2020 21:07
|
|
|
So, I want to switch a 30a, 110v normally open contactor with microswitches. Any single switch turns on the contactor, and they all have to be open to turn it off. I'm assuming I could have a 12VDC system of several microswitches wired to a pair of busses, using a DC control/AC switch SSR to switch the coil side of a normal industrial contactor, right? e: drawing at a 2nd grade level for some relative clarity 
Hypnolobster fucked around with this message at 16:14 on Jan 30, 2020 |
|
#
?
Jan 30, 2020 16:02
|
|
|
Hypnolobster posted:So, I want to switch a 30a, 110v normally open contactor with microswitches. Any single switch turns on the contactor, and they all have to be open to turn it off. Yes. I like SSRs for stuff this because theyre usually optoisolated already.
|
|
#
?
Jan 30, 2020 16:41
|
|
|
CopperHound posted:I'm getting to the point where I think I should stop writing unnecessary blocking code. I would caution you that state machines are not synonymous with non-blocking/asynchronous execution. You can write a 'blocking' FSM pretty easily, and in fact the simplest way to bang one out in C is going to block. One of the ways I like to write FSMs without blocking everything else is to encapsulate the state in an update function that gets called from an ISR (assuming your choice of micro and your firmware architecture don't make doing anything beyond setting a flag in an ISR a Very Bad Idea). I can expand on this a little bit if you like, but the general idea is probably obvious.
|
|
#
?
Jan 30, 2020 17:13
|
|
|
CarForumPoster posted:Yes. I like SSRs for stuff this because theyre usually optoisolated already. Followup, if I'm running the DC control wire a significant difference (like 50 feet), do I need to think about voltage drop/load on the DC side of the SSR? It's hard to find a spec for that, but I'm assuming that I could pretty easily use 20 gauge wire for the microswitches and just run the DC side at 24v or so and not have any problem.
|
|
#
?
Jan 30, 2020 17:19
|
|
|
Hypnolobster posted:Followup, if I'm running the DC control wire a significant difference (like 50 feet), do I need to think about voltage drop/load on the DC side of the SSR? It's hard to find a spec for that, but I'm assuming that I could pretty easily use 20 gauge wire for the microswitches and just run the DC side at 24v or so and not have any problem. Just test whether the relay turns on or not by putting however much your max run is or measuring the voltage at the end. That said, its easy to find an optoisolated SSR with a BIG input voltage range and almost no current draw. As an example this one has a 3 - 32V input range. Note that you can find big current SSRs and you should consider more than just static current rating (e.g. cooling, duty cycle, in rush current) when deciding how you want to turn on mains voltages.
|
|
#
?
Jan 30, 2020 17:53
|
|
|
Blotto Skorzany posted:I would caution you that state machines are not synonymous with non-blocking/asynchronous execution. You can write a 'blocking' FSM pretty easily, and in fact the simplest way to bang one out in C is going to block. The way I tend to write microcontroller firmware for stuff that doesn't like complicated ISR's and can't support something like FreeRTOS is with each "task" I need to do being its own separate little state machine function, and each of them is called from the main system loop. Each one does one step of whatever it's doing (note that's not necessarily one state transition, just one thing, where thing is whatever unit of work makes sense for your application) and returns, and then the next one goes, etc. You still have to make sure none of the steps block but once you get into this "one thing at a time" mindset it's not that hard to manage. Usually I also have some other function for each task that says if it has work to do or not, so I can skip it if it doesn't. If you want to get a bit more elaborate you could also put the processor to sleep if none of your tasks have work, though you'd have to be careful that you don't get an interrupt between checking for work and sleeping.
|
|
#
?
Jan 30, 2020 17:57
|
|
|
|
| # ? May 9, 2024 08:57 |
|
|
Hypnolobster posted:Followup, if I'm running the DC control wire a significant difference (like 50 feet), do I need to think about voltage drop/load on the DC side of the SSR? It's hard to find a spec for that, but I'm assuming that I could pretty easily use 20 gauge wire for the microswitches and just run the DC side at 24v or so and not have any problem. You should be fine -- voltage drop over distance is proportional to current, and SSRs eat near zero current. The one thing to do if you're going to put 12V over thin wire at distance is to either put a resistor in the way (to limit current) or a fuse (to *really* limit current). That way if somehow your long wire shorts out to DC negative, you don't melt your wire or blow your 12V supply.
|
|
#
?
Jan 30, 2020 18:43
|
|
