|
asdf32 posted:Or what's the arduino's built in DAC again?
|
#
?
Jan 17, 2015 06:32
|
|
|
|
| # ? Jun 4, 2024 22:44 |
|
|
Stealth Like posted:I have a question about a fully-differential amplifier and an ADC input if anyone could help. Yep that's what you do. Captain Cool posted:As far as I can tell, most Arduinos don't have one. The Arduino Due has two pins with a 12-bit DAC; otherwise, the analogWrite() function outputs a PWM waveform at a default 500hz. Oh bummer.
|
|
#
?
Jan 18, 2015 01:12
|
|
|
asdf32 posted:Oh bummer. The name of the analogWrite() function is designed to get your hopes up I think.
|
|
#
?
Jan 18, 2015 03:58
|
|
|
I'm using EAGLE to design a PCB with a big giant matrix of connections. I could whip up a spreadsheet of connections very easily, but hand connecting them in EAGLE would be tedious to the max. Is there a way to automate import of bulk connections? I don't even care how clean the schematic looks at this point, I just want to tell EAGLE what's up so I can get to the layout tool and be done.
|
|
#
?
Jan 18, 2015 06:29
|
|
|
asdf32 posted:A 16-bit serial DAC controlling a proper current source (easy to make in many different ways depending on details) sounds good for your application. An arduono should be able to control that at 2-4 kHz easily. Given the precision you're going for a current source would be ideal. RE:LEDs To add to this here is one of the simpler ways to have proper controllable current regulation for an LED. The TL431 (low voltage version shown) is like a combo voltage reference/op-amp in 3 pins. Here it nominally regulates the 6.25K resistor to 1.25V (20ma) with the 12.5k/37.5k network boosting or reducing that linearly from 0ma when the CTRL pin is 5V to 25ma when the CTRL pin is 0V. This scales if more voltage is available by allowing several LEDs in series. The 431 is convinient because it's super cheap, small (SOT-23), includes the reference and doesn't require an actual power rail, though the downside is that it works by shunting excess current around the LED. Thus it's inefficient even for a linear circuit. But good selection of resistors can keep this to a few milliamps at full power (the LTV431 only needs 100uA for itself). So the CTRL pin can come straight from a DAC which could feed many copies of this circuit to control more LED's. 
asdf32 fucked around with this message at 22:38 on Jan 18, 2015 |
|
#
?
Jan 18, 2015 22:18
|
|
|
Stabby McDamage posted:I'm using EAGLE to design a PCB with a big giant matrix of connections. I could whip up a spreadsheet of connections very easily, but hand connecting them in EAGLE would be tedious to the max. Is there a way to automate import of bulk connections? I'm not sure what you mean by matrix, but would using a BUS (a multi-wire net) instead of multiple single NETs help? It should be possible to script but it's not something I've ever looked into.
|
|
#
?
Jan 19, 2015 04:17
|
|
|
Chalupa Joe posted:I'm not sure what you mean by matrix, but would using a BUS (a multi-wire net) instead of multiple single NETs help? I played with buses, but I'd still have to run a net into each bus, so the total amount of manual clicking would be the same, right? By matrix, I basically mean it's something like a crosspoint switch layout. So if it's 128x128, I'd have to run 16384 nets.
|
|
#
?
Jan 19, 2015 07:37
|
|
|
Eagle commands are fairly simple, so you should be able to cook up a Python script or something to generate the necessary commands. http://web.mit.edu/xavid/arch/i386_rhel4/help/24.htm Add components programmatically, then use the WIRE command to connect them in a grid. It's not as easy as copying a net spreadsheet over, I know, but it's not Altium.
|
|
#
?
Jan 19, 2015 07:59
|
|
|
Stabby McDamage posted:I played with buses, but I'd still have to run a net into each bus, so the total amount of manual clicking would be the same, right? I'm a little confused. How many pins are connecting to how many pins? You can't hVe more nets than pins. I have zero eagle experience but have extensive knowledge of the view draw 2005 file format (unfortunately). For text charts, I've used a spreadsheet to generate the proper file format data (just text) and pasted it into the schematic page file. I'd crack open an eagle file in notepad and see what you're dealing with.
|
|
#
?
Jan 19, 2015 15:32
|
|
|
asdf32 posted:RE:LEDs It would still be way simpler just to use a more-modern processor with sufficient timer peripherals. A quick search showed a ton of parts from STM with 12+ 12-bit timers, along with higher resolution ones as well.
|
|
#
?
Jan 19, 2015 16:29
|
|
|
Slanderer posted:It would still be way simpler just to use a more-modern processor with sufficient timer peripherals. A quick search showed a ton of parts from STM with 12+ 12-bit timers, along with higher resolution ones as well. Granted I've only spent a day with an Arduino but in that day I went from not knowing what a "shield" was or how it interfaces with the Arduino to having a decent project running. It's pretty good stuff. I'd say there is quite a bit of value sticking with it. Especially if that's what you're familiar with. Just setting up an IDE and getting hello world on a new platform can be a task. Though anyway, the current source shown (or variant) would still probably be my choice for driving the LED's regardless of how I got the PWM/DAC output. Especially where this is some type of visual experiment.
|
|
#
?
Jan 19, 2015 16:46
|
|
|
asdf32 posted:RE:LEDs Interesting. The reason I've been trying to stick with PWM over direct current adjustment for my LEDs is because most LEDs have non-linear regions near 0mA that are trickier to account for with current regulation than PWM. The measurements I've taken from previous PWM'ed equipment to assess linearity has been as accurate as my lab's photometers, which is why I'd prefer to stay with that, but I may have to whip that up and give it a whirl and test it out, especially since 12-bit DACs that are I2C compatible are under $1.25 on digikey. Slanderer posted:It would still be way simpler just to use a more-modern processor with sufficient timer peripherals. A quick search showed a ton of parts from STM with 12+ 12-bit timers, along with higher resolution ones as well. In the long run, I'll end up transitioning from 8-bit AVRs to ARMs, but I really don't have the time to get the hang of the tool-chain and re-write my existing code until after I graduate.
|
|
#
?
Jan 19, 2015 22:42
|
|
|
poeticoddity posted:Interesting. The reason I've been trying to stick with PWM over direct current adjustment for my LEDs is because most LEDs have non-linear regions near 0mA that are trickier to account for with current regulation than PWM. The measurements I've taken from previous PWM'ed equipment to assess linearity has been as accurate as my lab's photometers, which is why I'd prefer to stay with that, but I may have to whip that up and give it a whirl and test it out, especially since 12-bit DACs that are I2C compatible are under $1.25 on digikey. Oh, cool. That's information I didn't have. I've never used LED's beyond debug indicators. Depending on how far you want to take this the current regulation still adds value in the PWM context by allowing more controlled matching between LED's. The LMV 431 has bandwidth over 1MHZ so you could PWM the whole circuit shown and remove the adjustment network. Though I have a better circuit for doing both. I'll diagram in when I get home.
|
|
#
?
Jan 19, 2015 23:16
|
|
|
With the addition of a mosfet this circuit has a number of advantages and is also better suited for PWM control. First, it's a proper series regulator, it only draws as much current as the load. Second, you can squeeze in more LED's because it only needs about 2V overhead (1.25V for the sense resistor, and ~0.75V extra for power supply variations etc). Thus for a 12V circuit you can squeeze in a full 10V+ of LED (counting their maximum potential drop). Current will be spot on all the time within ~1%. This is significantly better than a resistor limiter where the trade-off is the ratio between the forward drop variation and the nominal resistor voltage. So for example if you have a string of LED's with 1V of potential variation and a nominal resistor voltage drop of 4V you'll have up to 25% variation in current over time, temperature and circuit instances. You can waste more volts on the resistor to bring that error down, but trade-off ends up being pretty poor all around. I threw this together quick, it uses the regular TL431 with a 2.5V reference, not the 1.24V variant I prefer. It also shows some random LED's with very high voltage drops, hence there are only two. It's shown being driven from a 5V 5kHz PWM which works pretty well as long as the Vgs of the mosfet is less than ~3.5V which is easy to find. Running off 3.3V is possible, but you'd really want a low voltage mosfet (or a BJT) or you'd want to add an additional buffer of some kind. 
asdf32 fucked around with this message at 02:34 on Jan 20, 2015 |
|
#
?
Jan 20, 2015 02:31
|
|
|
ante posted:Eagle commands are fairly simple, so you should be able to cook up a Python script or something to generate the necessary commands. Duh, I should have thought of that. I was thinking I'd have to put all my logic and loops into that godawful language, but I could just script creation of simple wire commands in a good langauge. I need to read more, but hopefully there's a way to say "net pin X on thing Y to pin Z on thing Q"...the WIRE command I'm looking at seems to just do coordinates, and it would suck to have to do a bunch of coordinate math just to hook this to that. asdf32 posted:I'm a little confused. How many pins are connecting to how many pins? You can't hVe more nets than pins. Sorry, I shouldn't have said nets. I mean I'd need to do 16384 clicks. As in, if I were doing the 4x4 below, I'd need to do 16 clicks (actually 32, since each switch is two clicks):  EDIT: My new favorite quote from an EAGLE tutorial: "WIRE should be used for drawing things that aren't wires(!!)" Stabby McDamage fucked around with this message at 06:17 on Jan 20, 2015 |
|
#
?
Jan 20, 2015 06:13
|
|
|
Yeah, Eagle language sucks balls. To what the dude above said about check what the files itself looks like: It's just XML, but you'd have to create several different groups of objects in serveral different places, and it would unlink the schematic and board, so the board netlist wouldn't populate. Yes, when you unlink Eagle schematics and PCBs, there's no way of linking them up again  The coordinate math shouldn't be that hard, though. Just place one LED over the other, then you need to repeat this shape in wires a whole bunch: code:I'm trying to spec out some power mosfets. I need at least 90V, 20A, and preferably <20ns rise time. If I wanted to turn it on with a 5v micro, I'd need a Vgs of <3v or so to get it really on, too, right? Are there any good ones that people know about, offhand? Preferably available on AliExpress or something because shipping to Canada kills me. ante fucked around with this message at 06:27 on Jan 20, 2015 |
|
#
?
Jan 20, 2015 06:25
|
|
|
ante posted:
Umm you have a few challenges: 1) You're most likely going to need a heatsink. Only the lowest Rdson mosfets will stay below a watt at 20A and above that you quickly need heatsinking. 2) No mosfet is happy with 5V, they all prefer 10+, and the higher power higher voltage mosfets arn't often going to be tailored for low Vgs (though a search does turn up some) 3) Turn-on speed is going to be related to your gate drive too, and if it's just a 5V I/O from an arduino that's not good. High power switching mosfets usually have dedicated gate drive chips and/or isolation. Ground bounce is an issue especially with only 5V. To actually achieve that turn on speed and current, assuming you can find a fet meeting the other requirements, you very well may need dedicated gate drive circuitry. The D to G capacitance for example kills your gate drive as the D drops 90V in 20ns any C there pushes the gate back down. A search for 90V+, 100A+, <4V turn on gave this which is in the ballpark: http://www.ti.com/lit/ds/symlink/csd19531q5a.pdf But at 6mOhm at 6V Vgs heat is a real problem. Good luck. Look for fets designed for high frequency switching. I'd scrap the 5V idea pretty quickly unless something pops up. asdf32 fucked around with this message at 07:07 on Jan 20, 2015 |
|
#
?
Jan 20, 2015 07:00
|
|
|
asdf32 posted:I'd scrap the 5V idea pretty quickly unless something pops up. I could do that. Good to know, I'll redesign my parameters a bit.
|
|
#
?
Jan 20, 2015 07:06
|
|
|
asdf32 posted:With the addition of a mosfet this circuit has a number of advantages and is also better suited for PWM control. Wow. Thanks for the impressively detailed feedback.
|
|
#
?
Jan 20, 2015 23:11
|
|
|
I have recently acquired a crazy ultrasonic soldering station that is designed for soldering to glass and ceramics and hard to solder metals. http://www.bellexinternational.com/products/sunbonder/pdf/Sunbonder-USM-IV-Catalog.pdf Any cool ideas to do something with this before I try to flip it on ebay?
|
|
#
?
Jan 20, 2015 23:21
|
|
|
Probably a stupid question, but: I got this button. It says it accepts 12V DC power. Does this mean I can plug in + and GND from a DC power source directly to the leads without any resistors? Thanks
|
|
#
?
Jan 21, 2015 00:02
|
|
|
The review of the green version says it has a resistor. Also, just saying "12V" would lead me to believe it has a resistor, because a) there are no single LEDs with 12V forward voltage, and b) they don't give enough information to figure out what resistor to use, and c) they don't discuss it anywhere and it would be a really common problem.
taqueso fucked around with this message at 00:29 on Jan 21, 2015 |
|
#
?
Jan 21, 2015 00:27
|
|
|
taqueso posted:The review of the green version says it has a resistor. Also, just saying "12V" would lead me to believe it has a resistor, because a) there are no single LEDs with 12V forward voltage, and b) they don't give enough information to figure out what resistor to use, and c) because they don't discuss it anywhere and it would be a really common problem. That makes sense. Thanks!
|
|
#
?
Jan 21, 2015 00:30
|
|
|
asdf32 posted:With the addition of a mosfet this circuit has a number of advantages and is also better suited for PWM control. I have a few stupid questions about this circuit: 1. The output current is determined by the value of the current sense resistor (in conjunction with the reference voltage of the TL431), so I could use a lower value resistor and use this circuit to control a high power LED, assuming I spec a FET with low Rds and sense resistor that can dissipate the heat? 2. I won't release the magic smoke from the TL431 by cranking up the current, as long as I stay within the part's 36V cathode voltage max, correct? All the thermal dissipation is done through the transistor and resistor? 3. The efficiency of this circuit will be better than a linear regulator (power is switched, rather than being burned off as heat), but not as good as with a carefully designed smps (since this circuit still burns off some power in the current sense resistor and FET)? Basically, what I'm looking for is a way to do constant current regulation for high power LEDs without wasting poo poo-tons of power using linear regulators or having to spec out inductors and deal with the layout requirements most switch mode regulator ICs need. This sounds too good to be true.
|
|
#
?
Jan 21, 2015 02:59
|
|
|
So I'm trying to find an easy way to trigger remote flashes that only have an optical slave unit but no actual trigger connection, and I have some probably-dumb LED-related questions. I realize that the datasheet for any LED I purchase from a decent reseller will have decent information, but in this case, what I have is a handful of random "assorted LEDs" and I honestly have no idea what the general range of specs are. My questions: with bargain basement LEDs, can I get a really short-duration pulse from them (like under a millisecond)? Sorta related, how hard can I exceed their rated current if I'm only doing it for half a millisecond? I don't mind blowing a few up to find out but I don't even have a ballpark idea about this.
|
|
#
?
Jan 21, 2015 03:20
|
|
|
nobody- posted:I have a few stupid questions about this circuit: 1) Right, though the resistor wattage may demand larger resistors. 2) Yep, it's just controlling the FET. The FET is doing all the heavy lifting. 3) No, here you're mistaken. This is still all linear. It's a current controlled series linear regulator topology and no better than other low voltage current regulator options (like the LM317). Efficiency can be decent because the circuit only needs 1.25V to operate. In the earlier example you could realistically drive 10V of LED's from 12V. That would be 80%+ efficiency. But this goes down at lower voltages and won't be great with a single LED. Lots of inductor based integrated switching LED current controllers are out there so it's definitely an option. (The key component of a true switching supply is the inductor) poeticoddity posted:Wow. Thanks for the impressively detailed feedback. I've been fleshing out a linear capacitive/inductive discharge circuit lately which is a higher power relative of this circuit. So I've been evaluating and simulating various current controlled TL431 based circuits for a couple weeks. Actually I used that as a test bed for a couple LTC431 models but I still can't find one that works in LTSpice. The TL431 is also worth advertising. I discovered it relatively late but looking back it would have had a place on basically every board I've ever done in various roles as a reference/comparator/opamp. asdf32 fucked around with this message at 04:03 on Jan 21, 2015 |
|
#
?
Jan 21, 2015 03:58
|
|
|
I have a sainsmart solid state relay that I am trying to use to control the state of a LED lamp that runs at 9V, 500mA. If I power the SSR with +5v and GND and apply +3.3V to the input the indicator for the relay turns on and so does the LED lamp (relay load). However, when I remove the 3.3V input, the LED lamp stays on while the SSR indicator LED turns off. I'm not sure how to fix this. After some searching, I found this article: https://www.ia.omron.com/support/faq/answer/18/faq02155/index.html ...but I don't fully understand what to do to fix the issue or what resistance resistors I would use. Relay: http://www.sainsmart.com/8-channel-5v-solid-state-relay-module-board-omron-ssr-4-pic-arm-avr-dsp-arduino.html Mr. Bubbles fucked around with this message at 00:59 on Jan 22, 2015 |
|
#
?
Jan 22, 2015 00:48
|
|
|
Mr. Bubbles posted:I have a sainsmart solid state relay that I am trying to use to control the state of a LED lamp that runs at 9V, 500mA. If I power the SSR with +5v and GND and apply +3.3V to the input the indicator for the relay turns on and so does the LED lamp (relay load). However, when I remove the 3.3V input, the LED lamp stays on while the SSR indicator LED turns off. I'm not sure how to fix this. After some searching, I found this article: https://www.ia.omron.com/support/faq/answer/18/faq02155/index.html After further reading, it looks like I have a SSR that is designed to run AC loads, and is not compatible with DC loads. I guess I'll have to get a different relay or use a transistor. Could anyone recommend me a solid state relay (or transistor) that I'd be able to use to drive such an LED lamp from a 3.3v (RPi) input? Thanks! Mr. Bubbles fucked around with this message at 01:59 on Jan 22, 2015 |
|
#
?
Jan 22, 2015 01:11
|
|
|
Mr. Bubbles posted:After further reading, it looks like I have a SSR that is designed to run AC loads, and is not compatible with DC loads. I guess I'll have to get a different relay or use a transistor. Could anyone recommend me a solid state relay (or transistor) that I'd be able to use to drive such an LED lamp from a 3.3v (RPi) input? It must be an SCR or triac based SSR which inherently latch. There are other forms of SSR which would work with DC too. Buy any 5-10 amp+ fet in a TO-220 package and it will work. Smaller ones would work too if you double check the on-resistance and keep the power well below a watt.
|
|
#
?
Jan 22, 2015 02:33
|
|
|
I'm trying to build a giant seven-segment display out of lengths of LED strip. Something like this: Two quick questions: 1. Is there an IC or other cheap thing that will let me switch several relatively short lengths of LED strip? The only thing I've used in the past are IRF530 MOSFETs direct to an Arduino, but that seems overkill, since each segment will just be around 1' long. 2. Does anyone know of a good material to act as a mask in front of the display? I'm thinking of something like smoked glass that is dark but will let most of the LED light through clearly. See the thing I'm pointing to here with the chunk bit out of it: 
|
|
#
?
Jan 23, 2015 06:21
|
|
|
Stabby McDamage posted:I'm trying to build a giant seven-segment display out of lengths of LED strip. Something like this: 1. Voltage/current requirements? A ULN2003 Darlington driver might be enough for a buffer. One channel per strip. 2. You can get smoked acrylic in all sorts of colours
|
|
#
?
Jan 23, 2015 06:40
|
|
|
ante posted:1. Voltage/current requirements? A ULN2003 Darlington driver might be enough for a buffer. One channel per strip. On the general subject of load switching there is a category of parts that are drop-in 3 pin replacements for fets but are self protected against voltage, current and temperature. Examples: NCV8401A (SOT-23, 1.3A) AUIPS1031 (TO-220 and surface mount, 10A) They're confusingly spread between a couple categories on digi-key which makes them harder to find than they should be. But for basic load switching they're clearly better, not expensive, and footprint compatible with millions of regular fets if it doesn't work out.
|
|
#
?
Jan 23, 2015 15:52
|
|
|
I just smoked a 10 kohm resistor and I don't get why. I was trying to see what it takes to run a little DC motor so I set up a voltage divider with 3 of the 10k resistors. I had my Uno R3 into the wall and ran from the 5V pin. The motor didn't run on the first step, so 5/3V, and then 3.33V did it but quickly smoked the first resistor in the setup. What could have happened?
|
|
#
?
Jan 24, 2015 02:14
|
|
|
The Royal Scrub posted:I just smoked a 10 kohm resistor and I don't get why. I was trying to see what it takes to run a little DC motor so I set up a voltage divider with 3 of the 10k resistors. I had my Uno R3 into the wall and ran from the 5V pin. The motor didn't run on the first step, so 5/3V, and then 3.33V did it but quickly smoked the first resistor in the setup. What could have happened? Can you draw a diagram of how you had it hooked up? That shouldn't happen. Even if you dumped a full 5V across a 10k resistor, that's only 2.5 milliwatts dissipated as heat.
|
|
#
?
Jan 24, 2015 02:27
|
|
|
How big of a motor? When did it fry, when you applied the voltage or after you took it away? Flyback diode across the inductive load? Back emf?
|
|
#
?
Jan 24, 2015 02:58
|
|
|
Do any of you do embedded stuff with cellular GSM modules? If so, where do you get GPRS data service? Most of the plans out there specify Android phones - is that really an issue, or just what they'll support?
|
|
#
?
Jan 24, 2015 02:59
|
|
|
The Royal Scrub posted:I just smoked a 10 kohm resistor and I don't get why. I was trying to see what it takes to run a little DC motor so I set up a voltage divider with 3 of the 10k resistors. I had my Uno R3 into the wall and ran from the 5V pin. The motor didn't run on the first step, so 5/3V, and then 3.33V did it but quickly smoked the first resistor in the setup. What could have happened? It wasn't 5V or it wasn't a 10k resistor.
|
|
#
?
Jan 24, 2015 03:50
|
|
|
It would be pretty neat if you could get nationwide GPRS data only for $2 a month or something.
|
|
#
?
Jan 24, 2015 04:00
|
|
|
I'm having a bit of a hard time with my table saw wiring, my ignorance in the matter is becoming very apparent. 120V AC single-phase motor. It just died the other day, and after testing I discovered the relay (Klixon 4CR-54-782) was shorted. It sits behind a manual reset thermal motor protector. I'm not really sure how to tell if that is good or not, at least partly because the schematics don't really make sense to me. Going by the very helpful schematics, mine wasn't hooked up properly, yet it was working. Here is a drawing of the whole circuit as it is now- link. I could be reading them wrong, I can't really tell which one should apply, seemingly none because the line power is coming in on pin 2, not 1. 1 and 3 went to the relay. What am I not understanding there? I could use some guidance on what relay would be a suitable replacement and the best place to buy, preferably one that isn't $50+ (I only paid $75 for the whole tablesaw). Again, an oddity is that the original parts list and diagrams for this table saw show a different relay, one with only 3 terminals. The part isn't available anymore and the few places that have the whole parts lists thought it a great idea to just completely remove any references to model numbers if they are out of production. I was looking at replacing it with a solid state relay, but with it being a completely different type of relay and needing a low voltage feed to activate it, I thought it easier to just replace it with the same type. They certainly seem cheaper though, just not sure what else I'd need to do. Hopefully that made some sense. Time to start reading up more...
|
|
#
?
Jan 24, 2015 04:51
|
|
|
|
| # ? Jun 4, 2024 22:44 |
|
|
JawnV6 posted:How big of a motor? When did it fry, when you applied the voltage or after you took it away? Flyback diode across the inductive load? Back emf? It's essentially this https://www.sparkfun.com/products/retired/9608. I didn't test the resistor myself before using it, but it was a 10k resistor. Setup was simple as it gets, 3 resistors in series on a breadboard, 5V/Ground on the Arduino, and putting my motor leads in parallel to the first resistor, first and second resistor, etc. Nothing screwy about it. Fried the resistor when I applied voltage though. Even if it was a 100 ohm resistor, that's still only 250mW, which should be fine for even the cheapest resistors no?
|
|
#
?
Jan 24, 2015 04:58
|
|