|
longview posted:I have a small audio circuit that's powered by a step-up switching converter, I want to reduce the RF noise coupled to the output (a few mV). It depends on how large your transients will be - the capacitance after the inductor/ferrite will have to supply the charge. As far as stability is concerned, Q=(1/R) * sqrt(L/C). Try to minimize Q, but definitely keep it < ~0.7. Adding series resistance to the inductor can help, at the expense of power loss. Sometimes thats the only way though, if you have a large inductor and don't have room for a large amount of capacitance. Adding an LDO will help clean up the low frequency noise too, if you can tolerate the power loss. edit: If you don't want to use a full on LDO, and can tolerate ~0.65V drop in your rail, you can use a single NPN transistor, a resistor, and a capacitor to make a very low noise voltage source. Make an RC filter from your noisy power rail to the base of the transistor. Also connect the collector to the noisy power rail. The clean output is the emitter. Down side is the voltage drop, and the clean output is no longer regulated so it will drift a bit more with temperature. SnoPuppy fucked around with this message at 15:56 on Jun 13, 2012 |
#
?
Jun 13, 2012 14:40
|
|
|
|
| # ? Jun 11, 2024 07:28 |
|
|
SnoPuppy posted:That sounds like ringing due to the cable inductance. Can you take a scope screen shot and post it? I have never done a scope shot, but I'll give it a go today. The power rail has a little noise, but doesn't seem much different than with a short cable. On the capacitor, I usually hook it up correctly, but thought you guys were suggesting shorting it for some reason I didn't understand. 'shorting the ground pin' meant connecting the ground pin on the pandaboard's serial port directly to the ground rail on the far circuit via a short test lead. I'll try the power\ground swap today and report back.
|
|
#
?
Jun 13, 2012 20:54
|
|
|
SnoPuppy posted:It depends on how large your transients will be - the capacitance after the inductor/ferrite will have to supply the charge. Wow that definitely sounds like more trouble than I want, I'll skip the series inductors and add LDO regulators like the LT1963/LT3015 with the right capacitors. I measured the noise level on the output and we're only talking 5-6mV of 360kHz RF, noise in the audio band was less than 10µV so it's hardly a big problem. E: I'll also put in a capacitor in the feedback loop to limit the bandwidth to about 100 kHz, currently it extends to about 800 kHz so there may be some coupling through the input. longview fucked around with this message at 16:15 on Jun 14, 2012 |
|
#
?
Jun 14, 2012 13:32
|
|
|
longview posted:Wow that definitely sounds like more trouble than I want, I'll skip the series inductors and add LDO regulators like the LT1963/LT3015 with the right capacitors. It's actually not that bad. A 1uH inductor (ESR >= 20 mOhm) and a 100 uF cap (ESR >= 100 mOhm) should work for 500 mA transients and provide ~27 dB attenuation at 360 kHz. Or 10 uH @ 250 mOhm ESR and 200 uF @ 50 mOhm ESR would give you ~45 dB loss. To be honest, I usually just use spice to see if I'm getting too much ringing.
|
|
#
?
Jun 14, 2012 16:23
|
|
|
SnoPuppy posted:It's actually not that bad. I've already ordered the regulators but maybe I'll put in a LC filter between the SMPS and the LDOs to improve the high frequency rejection, then it'll still have a properly regulated supply and the RF rejection. Thanks!
|
|
#
?
Jun 14, 2012 18:23
|
|
|
Back to fundamentals time. I was wrapping my head around the concept of high voltage\lower amps to reduce transmission power loss(which I finally 'got') but I'm confused on another point. given a simple textbook circuit... 10v battery --> 100ohm resistor --> lightbulb --> 100ohm resistor --> negative terminal of battery, the book will always give you the resistance of the lightbulb and ask you for the the amps, of give you the amps and have you compute the resistance of the bulb... Same thing, if you replace the bulb with a pot, you can change the current flowing through the circuit at any time. What I don't get, is when you replace the bulb with a device that pulls more amps. The voltage of the battery can't change, and a wire with 100ohms of resistance is going to continue to have 100ohms of resistance, but MORE amps can be pulled out of the battery. I'm missing something simple about ohm's law or something:(
|
|
#
?
Jun 17, 2012 18:49
|
|
|
bobua posted:I'm missing something simple about ohm's law or something:( P=I2*R <-> P=I*E Since the voltage/resistance stays constant, as power increases, current increases.
|
|
#
?
Jun 17, 2012 19:28
|
|
|
bobua posted:Back to fundamentals time. The main problem with your understanding is the notion of "Pulling" more amps. It isn't really pulling them, it is just allowing them to pass through. If there was nothing where the lightbulb was except a straight wire, the circuit would have much higher amperage because the resistance would be lower (and amperage is inversely proportional to resistance: I=V/R). I struggled with these concepts at first too, because it seems like there are too many variables moving around at once. It is REALLY IMPORTANT to get this if you're pursuing an electrical engineering degree so keep working problems until you have a really strong understanding. In most basic cases, one part of V=IR will be constant, one will be your independent variable, and one will be your dependent variable. In your case, V is constant, I is the dependent variable and the reistance of the lightbulb is your independent variable. Your function is going to be I = 10V / (100ohms + Lightbulb_Ohms +100ohms)
|
|
#
?
Jun 17, 2012 19:42
|
|
|
bobua posted:What I don't get, is when you replace the bulb with a device that pulls more amps. The voltage of the battery can't change, and a wire with 100ohms of resistance is going to continue to have 100ohms of resistance, but MORE amps can be pulled out of the battery. I'm missing something simple about ohm's law or something:( With a purely resistive load, the maximum current in this circuit would be 50mA with a dead short (10V/200R) Let's pretend there's a 1A constant current source replacing the light bulb. This might seem to not make sense, but in reality your "load" is now another "source" (these distinctions might be more of a hindrance than a boon). Your source voltage is still fixed, the loop current is fixed, and the resistances are fixed. What does this mean? Basically, your CCS "input" must be 100V (1A*100R) lower than the positive output of the voltage source. Your CCS "output" must be 100V (1A*100R) higher than the negative output of the voltage source. If you ground the negative terminal of the voltage source, this means +10V, -90V, +100V, 0V at the four nodes clockwise around the loop (starting at voltage source positive). Likewise if you ground the "output" of the CCS, the four nodes are -90V, -190V, 0V, -100V. This is probably not realizable, but in ideal-element circuits it is a good exercise.
|
|
#
?
Jun 17, 2012 20:32
|
|
|
I'm not dumb, you guys are obviously just terrible at explaining things  Let me ask it this way... if I have a 12v battery, and 4 ohms of resistance in the wires, will I have an absolute maximum current of 3amps in the circuit, no matter what I add to it?(assuming I never remove any of that 4ohms of resistance?
|
|
#
?
Jun 17, 2012 22:48
|
|
|
bobua posted:I'm not dumb, you guys are obviously just terrible at explaining things Assuming an ideal battery and no other energy sources, yes.
|
|
#
?
Jun 17, 2012 23:05
|
|
|
And assuming you're connecting everything in series. If you put another load in parallel with the resistance, the resistor will still draw 3 amps, and the other load will draw (12v/R) amps of current.
|
|
#
?
Jun 18, 2012 03:44
|
|
|
I have a little fm transmitter for my ipod, and recently it began losing power unless the power cord was bent at a very specific angle where it enters the device. I figured "Oh, there must be a loose connection, I can totally fix that. I bet." and opened it up. When I did, I saw that the solder points for the power seem to be completely intact. So, any ideas as to what's wrong? I can't imagine it's the other end of the wire (the 12v plug) because the intermittent power loss was completely dependent on this end of the wire being bent in a specific direction. I've stripped the cable sheathing about an inch from the end to see if either wire was damaged (And in the process, removed what appears to be shielding, so I've probably messed things up pretty decently anyways), but the wire seems ok. Can solder points look and feel fine, but not be functional for whatever reason?
|
|
#
?
Jun 19, 2012 00:39
|
|
|
Slugworth posted:Can solder points look and feel fine, but not be functional for whatever reason? Another possible issue is the power jack for the transmitter has bad solder joints. Find out if the intermittent behavior depends only on the wire bending, or on force/torque being exerted on the cable jack due to said bending.
|
|
#
?
Jun 19, 2012 13:58
|
|
|
What kind of fidelity can you expect to get with the photo paper toner transfer method of etching PCBs? I'm thinking about trying to save money, plus maybe learn some stuff, by avoiding ordering PCBs from europe. But they make a bit use of surface mount ICs with relatively close pins.
|
|
#
?
Jun 19, 2012 15:58
|
|
|
Lets Play Arson posted:What kind of fidelity can you expect to get with the photo paper toner transfer method of etching PCBs? I'm thinking about trying to save money, plus maybe learn some stuff, by avoiding ordering PCBs from europe. But they make a bit use of surface mount ICs with relatively close pins. I can't remember exactly what pitch I was able to get down to, but it was pretty fine. I only judged it to be a waste of my time due to the effort needed to drill the vias, the lack of soldermask, and the difficulties of soldering some SMT parts without a tinned board (tinning by hand often puts down too much solder, and SMT parts dont lie flat for hand soldering).
|
|
#
?
Jun 19, 2012 16:03
|
|
|
Lets Play Arson posted:What kind of fidelity can you expect to get with the photo paper toner transfer method of etching PCBs? I'm thinking about trying to save money, plus maybe learn some stuff, by avoiding ordering PCBs from europe. But they make a bit use of surface mount ICs with relatively close pins. Here's a test board I did about a year ago:  pcbtest by bradluyster, on Flickr .5mm pitch is possible only if you go back with magnification and cut out teeny slivers of leftover copper. Even then, it's iffy and labor intensive. In these days of cheap Chinese PCBs, you're not going to save any money, especially if you legally dispose of your copper-infused etchent. And with the cheap chinese PCBs, you get solder mask, silk screen, the whole 9 yards.
|
|
#
?
Jun 19, 2012 16:22
|
|
|
Slanderer posted:(tinning by hand often puts down too much solder, and SMT parts dont lie flat for hand soldering). You can buy liquid tin that you immerse the board in for several minutes, and it coats the copper very nicely: https://www.alliedelec.com/search/productdetail.aspx?SKU=70125554 I never tried the toner transfer method, but I did use the photo developer method. It was easy to do, but it didn't turn out very good for SMDs. It's difficult to get the exposure just right and then dissolve the right amount of ink for very precise work.
|
|
#
?
Jun 19, 2012 20:11
|
|
|
Thanks for the replies. Kinda kills it when they do most of the leg work for you, but then it kinda kills it when you're paying extra for the privilege of doing the leg work yourself. Hobbies, huh?
|
|
#
?
Jun 19, 2012 20:14
|
|
|
Zuph posted:Here's a test board I did about a year ago: I'm fairly certain I got 5mm pitch to work just fine--except that when it came time to solder the part, the lack of soldermask meant preventing bridges was harder than usual. I also rain into trouble with a few long, thin traces on the board that were broken in the middle. But, still, drilling and filling vias was the hardest and most annoying part. While the top and bottom layers ended up lining up well enough, putting everything through the drill press was a huge timesink. I know they have weird rivet-via things that are easier than using tinned bus wire, but it's still a hassle.
|
|
#
?
Jun 19, 2012 21:19
|
|
|
I haven't touched hardware design in ... about 12 or 18 months, so I thought I would ask for some advice touching up an old design. What's the cleanest way to automatically detect and switch between two power supplies, let's say USB and battery power? Switches or jumpers are out. I also don't want to do something cheesy like diode or-ing the supplies together... but then again I might if no other solutions pop up. Fake edit: I have fuzzy recollections of a part from LT specifically for this job, but I thought I would ask anyway.
|
|
#
?
Jun 19, 2012 21:42
|
|
|
My Rhythmic Crotch posted:I haven't touched hardware design in ... about 12 or 18 months, so I thought I would ask for some advice touching up an old design. Look for "ideal diode mosfet controller" or "power supply or-ing controller". I just made a board that uses a couple LM5050-1s to select between two sources. Voltages can be up to 24V in that project. The LM5050 might not be the best choice for a lower-voltage setup. I know I was looking at a few options that could do what I wanted with just 1 IC, but not at 24V. Pretty much all of these ICs have sample circuit diagrams for USB/battery switching. I'd bet that it is one of the most common use cases. taqueso fucked around with this message at 21:57 on Jun 19, 2012 |
|
#
?
Jun 19, 2012 21:49
|
|
|
My Rhythmic Crotch posted:Fake edit: I have fuzzy recollections of a part from LT specifically for this job, but I thought I would ask anyway. http://parametric.linear.com/powerpath_controllers_*_ideal_diodes
|
|
#
?
Jun 20, 2012 00:05
|
|
|
Why aren't capacitor ESR meters built into multimeters? I just improvised one using an audio generator to locate a broken capacitor in a crossover filter, measuring with a capacitor tester/diode tester/running a test signal through the filter into an AC voltmeter it read fine, but with the ESR meter it developed significantly higher voltage than a known-good one.
|
|
#
?
Jun 20, 2012 12:30
|
|
|
I realize this might be starting a religious war, but can somebody recommend me a good family of microcontrollers to learn? I've got a fair amount of electronics experience, but I'd like to move some of my designs from FPGA-based systems to smaller microcontroller-based systems. I've worked with "soft" microcontrollers inside of FPGA logic, but I've never done a standalone uC design. My main criteria are: 1) Easy to design circuits around. 2) Easy to obtain, although unit price isn't much of a concern for me. 3) Good design tools (ideally free ones) which work on Linux or Mac OS X. The obvious choices are PIC, AVR, and MSP430 but I've had a hard time narrowing down my choice further. Or should I just start on 32-bit and stay there? Poopernickel fucked around with this message at 08:56 on Jun 21, 2012 |
|
#
?
Jun 21, 2012 08:52
|
|
|
Poopernickel posted:I realize this might be starting a religious war, but can somebody recommend me a good family of microcontrollers to learn? I've got a fair amount of electronics experience, but I'd like to move some of my designs from FPGA-based systems to smaller microcontroller-based systems. I've worked with "soft" microcontrollers inside of FPGA logic, but I've never done a standalone uC design. All of those would be fine. I would posit that MSP430 has a steeper learning curve than PIC or AVR, but if you are starting at FPGA soft cores it shouldn't be overly outrageous. Any idea what peripherals you may need?
|
|
#
?
Jun 21, 2012 09:08
|
|
|
What sort of projects do you want to use them with; do you need a lot of computing power in them? There's a lot of fancy looking AVR and ARM things out there that someone else will have to talk about, personally I use PIC ones for no good reason other it's simple enough and it's what my school taught me. Getting started with making something is very simple since you just need any clone PICKit 3, a micro of your choice and a (free) copy of MPLAB X and the HI-TECH C compiler. The latest MPLAB X is based on Eclipse I think, it works pretty well, and no hidden pay-for features except some compiler optimizations. The HI-TECH C compiler for low-end devices is pretty decent, they charge money to enable certain optimizations but that's usually not critical. Also the documentation on chip-specific things is often poor in the C-documentation so I often have to keep the datasheet open for registers to set, then cross-reference that with the .h file to see what names they gave it. It's not really a big deal once you've gotten started though. They've also fallen slightly into the Arduino-trap of implementing a bunch of bit-banging functions as part of their API. It's probably useful for when you just need to make something work but it doesn't really warn you that you're doing SPI in software instead of using the controller, for example. Microchip parts are also easy to find, both on ebay for buying bulk cheapos and reputable suppliers if you need really advanced ones or better QC. Their parts-selector is also very good.
|
|
#
?
Jun 21, 2012 09:08
|
|
|
longview posted:The latest MPLAB X is based on Eclipse I think, it works pretty well, and no hidden pay-for features except some compiler optimizations. The HI-TECH C compiler for low-end devices is pretty decent, they charge money to enable certain optimizations but that's usually not critical. Netbeans, for whatever it's worth. I've used Eclipse for some TI DSP chips and it was a... negative experience. I've actually been able to use most of my development software totally crossplatform (OSX, Windows, and Linux) but have been hampered by hardware support on non-windows platforms for my programmers.
|
|
#
?
Jun 21, 2012 09:39
|
|
|
longview posted:PIC stuff Nb. that if you use Microchip's compiler instead of the Hi-Tech one a lot more of the register names match. It also doesn't try to hide the fact that you really only have one ISR the way Hi-Tech does, and iirc has some handy bit manipulation helper things.
|
|
#
?
Jun 21, 2012 13:12
|
|
|
I'm a PIC fan-boy, but if you're OK with the slightly-lower learning curve (steeper learning curves mean easier  ), the MSP430 has the lowest cost-of-entry thanks to the LaunchPad. PICKit3s are $50 I think, but the LaunchPad is $4.30. I do so much with PICs I bought a Real ICE and a LVDS driver-kit (~$650 all together I think) to make my life easier.MPLAB X is starting to grow on me a bit; I stayed true to MPLAB IDE for the longest time, but it's OK I guess. Looks like Microchip released "new" compilers, or renamed their old ones? You can now get XC8/XC16/XC32.
|
|
#
?
Jun 21, 2012 16:48
|
|
|
Current electronics status: meeting with the analog guy from Cypress in about an hour to see if we can get ENOB > 17 bits from the ADC on our PSOC 5 project without disabling Every Freaking Thing PS movax what is your av from? I've always wondered.
|
|
#
?
Jun 21, 2012 16:52
|
|
|
Otto Skorzeny posted:Current electronics status: meeting with the analog guy from Cypress in about an hour to see if we can get ENOB > 17 bits from the ADC on our PSOC 5 project without disabling Every Freaking Thing It's from the Algis Budrys book Who?  Pulp as gently caress (1958) but awesome. I got the super cleaned-up version for my avatar from a friend, forget which country's edition it came from.
|
|
#
?
Jun 21, 2012 16:58
|
|
|
movax posted:Pulp as gently caress (1958) but awesome. This is frequently the case; eg. Poul Anderson's Flandry stuff from the mid-late 50s was awesome in a very fun way, and probably peaked in the mid-60s when he had an optimal mix of pulp and (rather) hard sci-fi in the stories in that series.
|
|
#
?
Jun 21, 2012 17:04
|
|
|
Otto Skorzeny posted:This is frequently the case; eg. Poul Anderson's Flandry stuff from the mid-late 50s was awesome in a very fun way, and probably peaked in the mid-60s when he had an optimal mix of pulp and (rather) hard sci-fi in the stories in that series. Aw man, those aren't available on Kindle  Guess I'll order the paperbacks after I move, they look awesome.
|
|
#
?
Jun 21, 2012 18:15
|
|
|
Delta-Wye posted:All of those would be fine. I would posit that MSP430 has a steeper learning curve than PIC or AVR, but if you are starting at FPGA soft cores it shouldn't be overly outrageous. Any idea what peripherals you may need? It's funny - I started with FPGA-based systems so I'm pretty good at making really complex electronics, but I have almost no experience with making really simple electronics (arguably an art in of itself). Do the peripheral sets differ much between different manufacturers? Anybody have any peripherals that set them apart? I'd say my main selection criteria is about half "ease of engineering" and half peripherals. I don't build the volumes where I care if a chip costs $2 or $5, and learning curve doesn't worry me much as long as there are good tutorials. EDIT: Also, which of the various families have reasonable RTOS support? I've been using a home-rolled cooperative task scheduler on my Microblaze projects, but I'd like to step up my game to something with threads and communications primitives. Poopernickel fucked around with this message at 19:49 on Jun 21, 2012 |
|
#
?
Jun 21, 2012 19:38
|
|
|
movax posted:
 Everything I know is a lie!quote:Do the peripheral sets differ much between different manufacturers? Anybody have any peripherals that set them apart? I'd say my main selection criteria is about half "ease of engineering" and half peripherals. I don't build the volumes where I care if a chip costs $2 or $5, and learning curve doesn't worry me much as long as there are good tutorials. It's mostly the same stuff everywhere. Configuration difficulty and edge-case features differ, but unless you're doing something fairly exotic everyone you mentioned provides processors in their product lines with whatever peripherals you may need. Delta-Wye fucked around with this message at 20:19 on Jun 21, 2012 |
|
#
?
Jun 21, 2012 20:16
|
|
(steeper learning curves mean easier
|
Poopernickel posted:It's funny - I started with FPGA-based systems so I'm pretty good at making really complex electronics, but I have almost no experience with making really simple electronics (arguably an art in of itself). Most of them will share the same basic stable of peripherals; ADCs, SPI, I2C, UARTs, etc. The differences come in what families each brand chooses to shove them in (product/market segmentation I guess). MSP430s even at the low-end have a nice LCD control peripheral. Some tiny PICs come with an EEPROM for "free", or an integrated thermal sensor. As for RTOS support, last I looked at FreeRTOS, they had solid support for bigger AVRs (ARM/AVR32), PICs (16-bit/32-bit) and MSP430, so you should be good there. Delta-Wye posted:
I got it from a list of common misconceptions, now I'm "that guy" and will point it out to people. Steep learning curves are good!
|
|
#
?
Jun 21, 2012 20:22
|
|
|
Otto Skorzeny posted:Current electronics status: meeting with the analog guy from Cypress in about an hour to see if we can get ENOB > 17 bits from the ADC on our PSOC 5 project without disabling Every Freaking Thing Btw as per this meeting, the answer to high precision analog problems with the PSoC 5 is "use a PSoC 3 instead because the PSoC 5 isn't actually a mature product; the SP1 for the PSoC 5 which actually works is in tape out and will be available Real Soon Now(tm)"
|
|
#
?
Jun 21, 2012 20:24
|
|
|
movax posted:I got it from a list of common misconceptions, now I'm "that guy" and will point it out to people. Steep learning curves are good! Just to also be that guy, a learning curve could be [gain vs. time] just as easily as it could be [time vs. gain]. So be that guy at your own risk 
|
|
#
?
Jun 21, 2012 21:01
|
|
|
|
| # ? Jun 11, 2024 07:28 |
|
|
Poopernickel posted:Just to also be that guy, a learning curve could be [gain vs. time] just as easily as it could be [time vs. gain]. So be that guy at your own risk B-but the dependent variable goes on y  (I'm bad at this)
|
|
#
?
Jun 21, 2012 21:06
|
|