|
Thats why I am so glad transistors are pretty cheap. I can order some on a whim, test and if it will not work not a big deal.
|
#
?
Feb 13, 2013 00:43
|
|
|
|
| # ? May 9, 2024 04:59 |
|
|
Kilersquirrel posted:So I guess that's a no on sources for reasonably-priced clamp-on inductor probes, then. Should I be able to just take a wide enough ferrite bead, do a couple turns of solid-core insulated wire around it, and then pass the spark plug wire through the eye for a functional induction probe? If all you want is to sense the pulse, simple turns of wire are all you need. Nothing else complicated. Wire, optoisolator, comparator (schmitt trigger) and you've got a pulse sensor. I did this when I wanted a "is this thing on" input for a microcontroller. I just kept adding turns around my thing-to-be-sensed until I got reliable triggering.
|
|
#
?
Feb 13, 2013 00:47
|
|
|
I know the LA talk was on the last page, but I'll throw in this: http://www.seeedstudio.com/depot/open-workbench-logic-sniffer-p-612.html?cPath=174 It's a real logic analyzer (captures to high speed memory) instead of being dependent on USB latency/throughput. It can capture up to 200Mhz signals on all 16 channels, is totally open source, and jawi has been busting his rear end working on the software for it on the dangerous prototypes forum. It's really quite nice now. I've never had a problem with buffer length. Just use larger # of channels than needed (16 or 32, gives longer RLE length than 8 channels) and use the triggering properly. (jawi is currently working on implementing all the crazy advanced triggers in the current 'demon' core too, which is basically anything a real benchtop $$$$ agilent LA can trigger on).
|
|
#
?
Feb 13, 2013 02:33
|
|
|
Rescue Toaster posted:I know the LA talk was on the last page, but I'll throw in this: Are you JawnV6's coworker? If not, you should go back and read his post (that started the LA discussion) very closely. 
|
|
#
?
Feb 13, 2013 02:44
|
|
|
Delta-Wye posted:The problem with "how do I select parts for guitar pedals" is that small details in the transistors can color the sound significantly, complicating choosing one compared to most applications. Hell, a stack of the same make and model can produce different sounds across units. That latter point is important, though - unless you get a transistor for a particular purpose (high current, high beta, etc) manufacturing variations can swamp difference between models. While its not wrong to pick exact NOS transistors for a project, it's also not wrong, 95% of the time, to just use 2N3904/6 or BC550/560s for everything (assuming you get the pinout right, they're different), and its not wrong to just try random transistors of the right polarity/pinout until you find one you like. Of course, that's assuming you want "a good sound" rather than "THE right sound". The classic TB303, originals even more than clones, had a reputation for some models sounding great and others garbage; it might have been finicky and superstitious users but it might also have been huge manufacturing variations in transistors that all had the same part number stamped on the package. If a circuit calls for a germanium type, though, do try it with a germanium transistor. The big difference is Ge's have lower Vbe and lower beta than Si, and that can make a big difference. A lot of NOS germaniums however were originally used for switches (like the 2N404 originally used in CDC computers), which illustrates how relatively interchangeable small-signal transistors can be. There's also cases where high beta is important, like 2N5172, but that's unusual. I have an old Tek 454 scope that had a messed up brightness control that turned out to be a failed transistor, unmarked and in an old style package not used any more. After figuring out the pinout and polarity I just started sticking transistors in the socket, which immediately fixed the problem, then tried a few others until I found one that worked best (in terms of control range).
|
|
#
?
Feb 13, 2013 02:59
|
|
|
Base Emitter posted:words about transistors How much does beta vary from unit to unit these days? The art of electronics 2e mentions the problem and says that values can frequently vary from nominal by a factor of 2(!) - has the situation improved at all in the intervening 20-some years?
|
|
#
?
Feb 13, 2013 03:10
|
|
|
Otto Skorzeny posted:How much does beta vary from unit to unit these days? The art of electronics 2e mentions the problem and says that values can frequently vary from nominal by a factor of 2(!) - has the situation improved at all in the intervening 20-some years? I honestly don't know... I'm a hobbyist rather than an industry guy and most of the sources I rely on are that old at least. It wouldn't surprise me if it hasn't changed that much, most of the innovations in the last couple decades seem to have been in MOSFETs.
|
|
#
?
Feb 13, 2013 03:19
|
|
|
Delta-Wye posted:Are you JawnV6's coworker? If not, you should go back and read his post (that started the LA discussion) very closely. Well... I guess just put me down as agreeing with his co-worker then.
|
|
#
?
Feb 13, 2013 04:12
|
|
|
babyeatingpsychopath posted:If all you want is to sense the pulse, simple turns of wire are all you need. Nothing else complicated. Wire, optoisolator, comparator (schmitt trigger) and you've got a pulse sensor. Thanks, I asked this earlier in the thread and had gotten the impression that I really needed a proper inductance probe to pick things up. Turns of wire it is!
|
|
#
?
Feb 13, 2013 04:20
|
|
|
Otto Skorzeny posted:How much does beta vary from unit to unit these days? The art of electronics 2e mentions the problem and says that values can frequently vary from nominal by a factor of 2(!) - has the situation improved at all in the intervening 20-some years?
|
|
#
?
Feb 13, 2013 04:46
|
|
|
Speaking of Magic smoke: I am working on a project using ATTINY45, It is quite a simple affair (controllable room lighting) but whenever you pressed the mode-button or adjusted the brightness-pot, the application went crazy. Sometimes the micro outright crashed. The Pot was working better than the button, but the range was somewhat skewed. I re-wrote the software for it roughly 3 times, bread-boarded it, serial-debugged it - shelved it in frustration and repeat. The bug that made the toggle button go crazy remained elusive... Until yesterday, when I pulled it all apart. Turns out that as I was working with two voltages on the board, whenever I pulled a pin high I ended up using the 12V line instead of the 3.3V line intended for the ATTINY. Ouch. It survived fine, but there was lots of hair pulling and debugging on the software side for what finally turned out to be a hardware issue due to me being careless. So if your software is going crazy, it might as well be the hardware. BTW, still seeking info on that 3.5mm plug for Hakko Irons...
|
|
#
?
Feb 13, 2013 06:24
|
|
|
I'm studying logic design right now and I really want to build some stuff on these perf-boards I have laying around that would implement what I'm learning. Does anyone have any good ideas for using a bunch of logic gates (+ any other components) to do something neat? I would love to build a parity generator on one board, have it send via an infrared LED the signal to another board (with an infrared detector), and have a corresponding parity checker on that board which would light an LED anytime it detected an error. Right now the remaining issue as I go over it in my head is generating a byte (probably with switches) and having the boards be able to synchronize so that the serial communication via light would translate into the gate system properly (via a serial-to-parallel mux). I'm not sure how I could synchronize two wireless boards easily.
|
|
#
?
Feb 14, 2013 04:27
|
|
|
Kire posted:I'm studying logic design right now and I really want to build some stuff on these perf-boards I have laying around that would implement what I'm learning. Does anyone have any good ideas for using a bunch of logic gates (+ any other components) to do something neat? Why the infrared and not just a cable? Added coolness factor?
|
|
#
?
Feb 14, 2013 04:32
|
|
|
It's been a while since signals theory class but we had a digial audio demo board in class that used relatively simple logic to implement a single-channel digital audio stream. IIRC the block diagram showed it XORing in the clock, and presumably it used a PLL (like the CD4046) to recover the clock from that and then xor it back. Fake-e: Looking at it now there was a receiver clock, and a module that we don't know the specifics of to handle timing and synchronization of the shift registers, latches, D/A and so on.
|
|
#
?
Feb 14, 2013 09:03
|
|
|
Otto Skorzeny posted:I looked through the spec and didn't see anything about the front end wrt. what logic levels it will work at; might want to check if you're looking at anything that isn't common for hobbyists. Incidentally, are you looking for a pure LA (as your coworkers' suggestion seems to be), or something that is useful for analyzing analog signals as well? peepsalot posted:Or you can get a 3 clones for $10. http://www.aliexpress.com/item/Free.../689100550.html Kire posted:I would love to build a parity generator on one board, have it send via an infrared LED the signal to another board (with an infrared detector), and have a corresponding parity checker on that board which would light an LED anytime it detected an error. Right now the remaining issue as I go over it in my head is generating a byte (probably with switches) and having the boards be able to synchronize so that the serial communication via light would translate into the gate system properly (via a serial-to-parallel mux). I'm not sure how I could synchronize two wireless boards easily.
|
|
#
?
Feb 14, 2013 19:36
|
|
|
JawnV6 posted:I'm expecting some signals to hit over 24MHz. I'm used to having to minimize traces and trigger properly, and that's without having the kind of hardware control I can get on a FPGA. Not too worried about trace length given all that, but that price is hard to ignore. If some of the signals will go anywhere near 24MHz then you should definitely get a LA with local storage. This isn't like an analog to digital converter where you can take 2X the frequency and say 'good enough' due to the nyquist theorem. You want a logic analyzer that is much faster than the signals of interest so you can actually say with some confidence what the timing looks like and where the transitions actually are. EDIT: Sorry it's obvious you've used a LA at work, so clearly you know this. I guess I'll leave it here for anyone else looking for LA's though. Definitely consider the OLS if you're comfortable with triggering. (And the RLE actually works very well if you keep it set to 16-24 active channels, instead of 8) Unless somehow the two channels of a Logic 16 would be enough? Remember that a Logic 16 is still dependent on usb traffic throughput, though. So the '100Mhz' on 2 channels depends on an uninterrupted 20MB/s over the usb link. Rescue Toaster fucked around with this message at 22:29 on Feb 14, 2013 |
|
#
?
Feb 14, 2013 22:26
|
|
|
Rescue Toaster posted:Definitely consider the OLS if you're comfortable with triggering. (And the RLE actually works very well if you keep it set to 16-24 active channels, instead of 8) Unless somehow the two channels of a Logic 16 would be enough? Remember that a Logic 16 is still dependent on usb traffic throughput, though. So the '100Mhz' on 2 channels depends on an uninterrupted 20MB/s over the usb link. OK I'm confused here. Can you elaborate on why RLE would work better with more channels active? Do you just mean that it will have a better compression ratio overall, assuming you are making use of all those channels? If I only care about 4 channels, I wouldn't be able to somehow log more samples by enabling more channels and RLE right?
|
|
#
?
Feb 15, 2013 01:54
|
|
|
peepsalot posted:OK I'm confused here. Can you elaborate on why RLE would work better with more channels active? Do you just mean that it will have a better compression ratio overall, assuming you are making use of all those channels? Ok so normally your memory is split up based on channel width, so if you use 8 channels you have 4x more storage space as you would at 32 bits, right? The RLE length is stored in the same number of bits as width in terms of # of channels. So if you have 8 active channels and enable RLE each byte used for RLE length can represent up to 255 clock cycles, but if you have 16 active channels and enable RLE, each two bytes used represents up to 65535 clock cycles. And obviously at 24 active channels it represents millions of clock cycles. So if your edge transitions tend to be farther apart than 255 clock cycles, you probably want to enable more channels even though the extra sample bits are 'wasted' you get a lot more max RLE length.
|
|
#
?
Feb 15, 2013 02:27
|
|
|
Parallel Paraplegic posted:Why the infrared and not just a cable? Added coolness factor? I was hoping to be able to pull the boards further and further apart and watch the error rate climb via the parity-check-output going to an LED. Or put different materials in between (plastics, cellophane) and see how that affects the transmission's error rate. JawnV6 posted:Might be too complex for hand-wired, but if all you need is the same byte at the same time on both boards, make a LFSR and synchronize turning them on. Although from your description I'm not following why you'd need to have the source byte replicated on the second board from a sideband. What's an LFSR? A switch to turn both boards on at exactly the same time? I don't think I need to replicate the byte through a sideband, since it's what's being transmitted. Rather I thought I might need to synchronize the clocks if I have a ~1kHz uC generating bytes (like the ASCII table or something) to send. But I guess that doesn't make sense, the reception of the light pulses should be all the synchronicity it needs... One problem I see would be if I start the system up, and the receiver gets out-of-sync by getting an errant infrared flash from something (not sure if that is common or not) so when the sender sends a byte, the receiver misses the first bit and gets out-of-sync which would mess up the serial-to-parallel multiplexer between the infrared receiver and the parity checking logic. Kire fucked around with this message at 19:24 on Feb 16, 2013 |
|
#
?
Feb 16, 2013 19:21
|
|
|
Kire posted:I was hoping to be able to pull the boards further and further apart and watch the error rate climb via the parity-check-output going to an LED. Or put different materials in between (plastics, cellophane) and see how that affects the transmission's error rate. The fact that your physical layer is light doesn't really effect the problem. Basically you're just trying to implement a serial bus. There are lots of existing ways to do this, USART's for example, or many variations of clocked serial busses such as SPI. USARTS avoid transmitting the clock but must have the data-rate previously agreed upon and require accurate clocks on each side of circuit. Clocked serial busses are simpler in that they need only one master clock and thus the receiving circuitry can be simpler. This is why I2C and SPI are so popular as interfaces between cheap IC's. LFSR is a type of shift register. Not sure what the LF part of it has to do with this problem but a shift register would be a very simple way to implement a clocked serial bus. asdf32 fucked around with this message at 19:33 on Feb 16, 2013 |
|
#
?
Feb 16, 2013 19:31
|
|
|
Ok, I need some help here. I am currently looking for a relatively cheap way to make a circuit of some sort that I can record at least 4 minutes of sound on, and then have it play back on a loop endlessly while still being pretty compact. Am I just being crazy is or is this doable? Pretty much I have been finding chips meant for greeting cards that hold about 4 seconds.
|
|
#
?
Feb 19, 2013 01:30
|
|
|
Stravinsky posted:Ok, I need some help here. I am currently looking for a relatively cheap way to make a circuit of some sort that I can record at least 4 minutes of sound on, and then have it play back on a loop endlessly while still being pretty compact. I've seen little kids play microphone toy things that you can get on the cheaps, you could tear one of those apart and look at what makes it tick, if not outright use the circuitry inside it and hack it to your purposes.
|
|
#
?
Feb 19, 2013 01:57
|
|
|
What's relatively cheap and compact, and do you want to "record live" or play back a prerecorded track? The main thing is 4 minutes of audio requires a respectable amount of memory (for a single cheap chip). Sparkfun has an Arduino shield that's got MP3 playback and a memory card interface but its $40, and you need an Arduino, some programming, a memory card, and an MP3 file. https://www.sparkfun.com/products/10628 Of course once you have the basic platform you can hook up all kinds of extras, triggers, etc.
|
|
#
?
Feb 19, 2013 03:37
|
|
|
http://www.digikey.com/product-search/en/integrated-circuits-ics/interface-voice-record-and-playback/2556441 Should be lots of chips that fit the bill in there. Like, second link I clicked on was a $3.50 chip that can store 64 seconds of high quality audio.
|
|
#
?
Feb 19, 2013 04:08
|
|
|
Ideally, I am going to be stuffing these things with small speakers into unassuming objects (hollowed bricks, mcdonalds cups and other detritus of a city). I want to keep this as cheap as possible, as I want to build at least ten of these things. If I can keep it under a hundred that would be rad as hell. Currently the ISD17240 is looking alright, but if anyone has any other good ideas please shoot me some suggestions.
|
|
#
?
Feb 19, 2013 04:59
|
|
|
When hooking up an external osc to an ARM uC, what are some rules to follow? I gather that the crystal should be as close to the actual chip as possible, but I'm using a QFP64 breakout board and can't really mount anything super close to the CPU on the breakout board without some shenanigans. Is this likely to be a huge problem? I know next to nothing about crystals  Once I'm finished prototyping my thing I'll be in a better position to lay out my own custom PCB, but for now I'm using what I have.
|
|
#
?
Feb 19, 2013 05:32
|
|
|
How fast are you going?
|
|
#
?
Feb 19, 2013 05:59
|
|
|
I don't have time to post a comprehensive reply. I'll just mention to be sure that you understand the difference between a canned oscillator, which is an active powered component and produces a clock signal, and the crystal which is a passive component that is driven by an oscillator circuit (located inside the CPU, usually). Most micros can run in either external oscillator or crystal mode. Also I would expect an ARM CPU will likely have an internal PLL to multiply the clock from say 8Mhz up to the actual core speed which is much higher. An external oscillator will be more relaxed as far as layout/wiring goes, (though some care in ground connection is needed, you could do it in your situation fine). But a simple crystal attached to the CPU will require much more careful layout (directly on a PCB) and usually has some load capacitors too, that is to say, will likely not work on a QFP breakout.
|
|
#
?
Feb 19, 2013 06:05
|
|
|
I haven't decided on a final speed yet. The chip itself can hit 120MHz but I haven't really figured out how much horsepower my application needs. The chip does run an internal PLL and in fact has an internal clock or timer source, but my understanding is that it's not precise enough to drive USB functionality. Apologies if this is really vague, but this is the first time I've tried to work with anything but a pre-built prototyping board so I'm going to be asking some really dumb questions until I get comfortable with the ideal system design  For reference, the chip I'm using is an NXP LPC1769 but I'm partly asking just to gain generic knowledge which I can apply to any of the dozen bare ARM chips I have lying in my drawer.
|
|
#
?
Feb 19, 2013 09:05
|
|
|
If the crystal frequency is relatively low (like 12MHz or less) then an inch or two of traces probably won't hurt too much.
|
|
#
?
Feb 19, 2013 14:54
|
|
|
You will likely need a specific frequency (or one of a set of choices that divide nicely) for USB, make sure you check the datasheet.
|
|
#
?
Feb 19, 2013 14:56
|
|
|
Martytoof posted:When hooking up an external osc to an ARM uC, what are some rules to follow? I gather that the crystal should be as close to the actual chip as possible, but I'm using a QFP64 breakout board and can't really mount anything super close to the CPU on the breakout board without some shenanigans. Is this likely to be a huge problem? I know next to nothing about crystals I've added the crystal to the pin part of a breakout board before, and it worked fine.
|
|
#
?
Feb 19, 2013 15:38
|
|
|
As much as I know about RF and tuned circuits and all that, I can't build an oscillator in Multisim or on a breadboard to save my effing life. I just don't get crystals yet. I mean, I do, but I don't. I really wish I could get a better hang of them to work on RF stuff more. Really need a scope or something. There's only so far a good DMM can take you.
|
|
#
?
Feb 19, 2013 15:49
|
|
|
Martytoof posted:When hooking up an external osc to an ARM uC, what are some rules to follow? I gather that the crystal should be as close to the actual chip as possible, but I'm using a QFP64 breakout board and can't really mount anything super close to the CPU on the breakout board without some shenanigans. Is this likely to be a huge problem? I know next to nothing about crystals In your case I think you may need to adjust the load capacitance values specified by the xtal to compensate slightly for the added capacitance/inductance of your longer traces, but at low speeds they are somewhat forgiving.
|
|
#
?
Feb 19, 2013 18:17
|
|
|
double-post, I am the worst mod
|
|
#
?
Feb 19, 2013 18:19
|
|
|
movax posted:double-post, I am the worst mod Clearly. Jonny 290 posted:As much as I know about RF and tuned circuits and all that, I can't build an oscillator in Multisim or on a breadboard to save my effing life. Usually my problem is getting rid of the oscillator I accidentally made  Talking about RF, is it just me or is a lot of RF/analog becoming a lost art? Seems like my parent's generation had quite a few self-taught HAM types who are comfortable with antenna design and a bunch of other things that I, with my formal education, am not comfortable with. It could just be practice (I don't do much antenna stuff, they have years and years of experience) or willingness to experiment (they are picking up analog radio stuff so they can play with the antenna, I'm doing digital 2.4 GHz so I follow the datasheet exactly), but I'm always sad when I see old electronics magazines and they gloss over details on stuff like antennas because 'duh' and I don't get it.
|
|
#
?
Feb 19, 2013 18:35
|
|
|
Delta-Wye posted:Usually my problem is getting rid of the oscillator I accidentally made I'm in my final year of university for EE, and looking at the job market I'm definitely seeing people talk about how hard it is to find people who know what they're doing with RF or high frequency analog. Sadly, that's not me - I'd like to learn, but I don't know how. My university doesn't offer any serious classes in that area, we kind of talked a little bit about it in one class but nobody really understood it (I don't know if the professor did, even). I took a "Wireless Communication Systems" class, but it was all stuff about path loss, link budgeting, network layout, and so on. Not a word was breathed of antenna design or anything like it, other than looking at some data sheets for existing antennas.
|
|
#
?
Feb 19, 2013 18:59
|
|
|
I work in microwave/rf passives. Essentially everyone I work with is self-taught, usually with a mechanical engineering background. People in this field are rare, and the ones that do exist have graduate degrees and usually end up working on the coasts for a Raytheon, Boeing, Agilent, etc. Whenever I'm involved in the hiring process I have to point out to my boss(es) that you won't find a microwave engineer for this location at this pay, because they don't exist outside the building.
|
|
#
?
Feb 19, 2013 19:35
|
|
|
Speaking of radio I'd really like to buy a bladeRF unit: https://www.kickstarter.com/projects/1085541682/bladerf-usb-30-software-defined-radio But what bums me out is they have a 484 pin BGA chip on there and only just a single tiny expansion header. I assume some of which is the ADC/DAC bypass as well (bypassing the chip's internal tuner, since for instance it only goes down to 300Mhz). If I could get BOTH an awesome SDR system and a decent FPGA dev board with USB 3.0, that would be absolutely amazing. But as it is with only like 12-20 IO lines from the FPGA into that little header... it's hard to put up $400-600 for it since it can basically be used for absolutely nothing but SDR.
|
|
#
?
Feb 21, 2013 04:53
|
|
|
|
| # ? May 9, 2024 04:59 |
|
|
I'm interested in making my own capacitive discharge spot welder, for thin sheet metal, battery pack tabs, etc. I guess what I'm trying to figure out is what kind of current I need for this. I'm looking at maybe using car audio cap like this: http://www.amazon.com/Boss-CAP4C-Farad-Capacitor-Chrome/dp/B0014E2BSK advertised ratings: 4F 24V 1.95mOhm ESR I would probably only run it at 12-14V. If i could perfectly short out this capacitor at 14V/1.95mOhm would give a max current of ~7kA. Thats a lot but obviously the rest of the circuit is going to have some amount of resistance to add, I just have no idea how much. For the switching circuitry, I saw some CD welders use SCR/Thyristors, but due to the high cost of very high power SCRs, I'm wondering if a bunch of parallel MOSFET would be more cost effective. Also the fets would allow me to turn the thing off, for more precisely controlled pulses, and give the possibility of dual pulse operation. Theres this MOSFET for $2.55 on mouser http://www.mouser.com/ds/2/200/irfb7430pbf-38118.pdf 20V, 195A continuous, 1.3mOhm R(on) Say I put 10 of these MOSFET in parallel, would give only .13mOhm resistance across them, and 2kA continuous current capability. If I got my total resistance to just 7mOhm, it would limit current to the rated 2kA. At 7mOhm resistance, the RC time constant would be 28ms. I was thinking soldering mosfets directly to some bus bars for heatsinking and low resistance. Then I need some heavy gauge cabling from the cap/fets/bus to the electrodes. Maybe roughly 1meter of cable total. 1m of 5awg cable for example would be just 1mOhm resistance according to this: http://en.wikipedia.org/wiki/American_wire_gauge#Tables_of_AWG_wire_sizes But how much contact resistance should I expect on the lugs between the cable and the bus bars, electrodes. I don't really have an idea of even the order of magnitude, if it would dwarf all the other resistances I'm talking about. Any suggestions, am I on the right track here? Professional units seem to be in the thousands of dollars, I'd like to build something as nice as possible for under $200. peepsalot fucked around with this message at 22:58 on Feb 24, 2013 |
|
#
?
Feb 24, 2013 22:48
|
|