|
sixide posted:I wouldn't recommend BNC for anything besides low frequency stuff, VHF and below. Your typical commercial BNC is an absolute piece of junk. I've had 50-ohm BNC connectors that were 75 ohms from immediately behind the interface all the way to the cable. A $50 BNC might be "good" to 8 GHz. Meanwhile, a $1 SMA from a fly-by-night factory in China will reach 12.4 GHz without issue. It's astounding how variable the quality is, and what you often have to pay for mediocrity. The BNC/TNC connectors we use for a lot of coax terminations in aviation can run *hundreds* of dollars each depending on the particular application, and most of them honestly aren't any better or worse than the $5 Radio Shack connectors I used to buy for CB antennas, aside from better strain relief. Of course, every last goddamn one of them also needs its own special $1200 snowflake tool to make a proper crimp. I'm starting to think I'm in the wrong end of this business.
|
#
?
Dec 6, 2013 05:17
|
|
|
|
| # ? May 8, 2024 13:24 |
|
|
The fact that there's a 75 Ohm BNC in the first place is also pretty terrible, because only the spergiest of engineers can tell the difference looking at them. I've found tons of 50 ohm adapters used with 75 ohm video cables at work. The bayonet connection is pretty nice for things like oscilloscope probes though, and I never liked the combination of thick cables and SMA connectors, but there's always N connectors for that. SMA patch cables with semi rigid coax is wonderful for connecting up racks and inside various equipment though.
|
|
#
?
Dec 6, 2013 16:52
|
|
|
Acid Reflux posted:I'm starting to think I'm in the wrong end of this business. From what I can tell, the best business to be in is agency approvals.
|
|
#
?
Dec 6, 2013 17:12
|
|
|
You can also get SMA style connectors with big N-Style Wheels for hand fastening. We got some V-Kables like this, though of course you are still supposed to use that torque wrench.
|
|
#
?
Dec 6, 2013 18:14
|
|
|
Someone at my local hackspace wants me to design a board for him. It looks fairly straightforward, it will control 18 channels of servomotors, and I expect it to take maybe three days of work, to design, test, and prepare for mass manufacturing. He has a site that sells robotics kits that he'd like to sell it on. What are fair compensation terms for me? Well, for both of us, I guess.
|
|
#
?
Dec 7, 2013 22:55
|
|
|
ante posted:Someone at my local hackspace wants me to design a board for him. It looks fairly straightforward, it will control 18 channels of servomotors, and I expect it to take maybe three days of work, to design, test, and prepare for mass manufacturing. He has a site that sells robotics kits that he'd like to sell it on. What are fair compensation terms for me? Well, for both of us, I guess. If you have no control or duty in the sales channel/promotion you should be looking to be made whole on your time to design and test the board and then seek a percentage of sales. The negotiation enterer in the part about "what percentage" vs. what your initial compensation should be. Without knowing production costs, sales price, potential market, and how confident you are this person can sell them it's very difficult to say what that calculation should be. This gets into things like "who owns the design?" You could design it and allow hime to sell it under license, in which case your initial compensation would (should) be smaller, or you could do this as work for hire and he owns the board when it's done.
|
|
#
?
Dec 7, 2013 23:30
|
|
|
I should add some more info. I'm unemployed right now, so my time isn't really worth that much, I'd be doing this more as a portfolio thing (hey look, employers, I can do stuff!) I think he's aiming at the $15 price point, and in runs of 100, I can probably do $8 for manufacturing, or a little less. As for that second question, I have to talk to him more to see what he wants, but I was thinking about that. It seems the whole point of owning the design would be to eventually do something else commercially with it, but I'm not really into that. So long as I was able to take pictures of the final project and blag about it, I wouldn't feel the need to release the designs or retain control or whatever.
|
|
#
?
Dec 8, 2013 01:36
|
|
|
Is there a concise description anywhere (perhaps in a book my company can buy) of the licensing requirements to sell a product with bluetooth? My eyes have glazed over trying to determine whether we have to pony up to join the SIG in all cases (but then again I am a dumb).
|
|
#
?
Dec 9, 2013 16:27
|
|
|
I came across something in an academic paper I wanted to ask about : "The voltage drop across a diode-tied transistor [they just mean a diode made from a transistor] under 2nA bias is much different than one drawing 2 micro-A of current. This voltage difference is typically 100mV for every decade of current difference." I can't find any info on further studies or explanations about this change in Vth across a diode based on bias current. I am really interested in learning more about how the current used to create a bias voltage changes the actual voltage needed, since I've never heard of that before. I don't know what to google, can anyone give me some more info or point me somewhere?
|
|
#
?
Dec 9, 2013 20:41
|
|
|
reading posted:I came across something in an academic paper I wanted to ask about : I don't have any nice curves of diodes at currents that small, but that sounds like small signal diode behavior. The first diode model anyone learns is just the ideal diode model, which says that the current is 0 below the forward voltage. Then we get to a linear model, which has the forward voltage increasing with current, but still with a minimum forward voltage (and a max current as well). In reality, a diode will still conduct below it's nominal forward voltage (which is really just the forward voltage for a particular current), but the I-V curve is nonlinear there. You can look into diode modelling, but the Wikipedia page may be less than helpful.
|
|
#
?
Dec 9, 2013 21:12
|
|
|
Slanderer posted:I don't have any nice curves of diodes at currents that small, but that sounds like small signal diode behavior. Here is a picture of small signal diode behavior:  EDIT: VVV is this just a nomenclature issue or am I missing something more important? Delta-Wye fucked around with this message at 03:56 on Dec 10, 2013 |
|
#
?
Dec 10, 2013 00:08
|
|
|
Diodes don't have a threshold voltage...  It's just a simple exponential function, with no obvious trip voltage, as opposed to MOS structures.
|
|
#
?
Dec 10, 2013 03:32
|
|
|
ANIME AKBAR posted:Diodes don't have a threshold voltage... I thought they had a threshold voltage that is defined as where that curve becomes "vertical." Usually 0.7 or 0.3. I know those are the voltage drop but I thought of it as the same thing- the voltage needs to reach that point before the diode conducts fully.
|
|
#
?
Dec 10, 2013 21:40
|
|
|
So I want to goof around with some guitar effects, they seem easy and cheap to mess with. I'm having a hard time finding how traditional components affect audio frequencies though. Anyone know a good place to start? I see quite a few diy stompbox type forums, but it seems rather shallow with knowledge. I'm assuming a oscilloscope is rather mandatory later on. I just reread the OP, it has some nice links. Most seem to be build-a-kits though. I don't want a kit I want to learn how they work. I will peruse those links in the meantime, maybe i'm just an idiot an missing something fundamental. darkhand fucked around with this message at 23:02 on Dec 10, 2013 |
|
#
?
Dec 10, 2013 22:46
|
|
|
reading posted:I came across something in an academic paper I wanted to ask about : The current through a diode is an exponential function of the voltage across it, or viewed the other way around the voltage across a diode is a logarithmic function of the current through it. The Shockley diode equation is a pretty common model that shows this dependence. It sounds like the folks who wrote your paper have a tiny current source that they are driving into a diode to produce a voltage that is proportional to the log of the current (100mV for every decade of current). Since 2nA and 2uA are orders of magnitude apart the resulting voltage change would be easy to observe. It may be easier/cheaper for them to use this trick than to to have a meter that can measure nanoamps directly. reading posted:I thought they had a threshold voltage that is defined as where that curve becomes "vertical." Usually 0.7 or 0.3. I know those are the voltage drop but I thought of it as the same thing- the voltage needs to reach that point before the diode conducts fully. The idea that diodes have a threshold voltage comes from some common approximations that engineers use to make working with diodes easier. Doing math with exponential functions leads to a lot of inconveniently unsolvable transcendental equations, so it's much easier to just approximate the device as turning on at some point and maybe having a bit of series resistance. The simplified model is often good enough and since it's just lines the math is a lot easier. Unfortunately the design approximation "diodes only conduct above 0.7V bias" gets used so often that folks tend to forget that it's just an approximation and not reality. In the chart below current is vertical, voltage horizontal, the black line is a real diode curve and the red line is an approximate threshold model with a series resistance. You can see that if the current is pretty big (top of the chart) the simple approximation is not terrible. Unfortunately you're operating at nanoamps, so you're way down near the bottom of the chart where the models don't agree at all and you'd really need to use something like the Shockley equation linked above. 
|
|
#
?
Dec 10, 2013 23:46
|
|
|
darkhand posted:So I want to goof around with some guitar effects, they seem easy and cheap to mess with. I'm having a hard time finding how traditional components affect audio frequencies though. Anyone know a good place to start? I see quite a few diy stompbox type forums, but it seems rather shallow with knowledge. I'm assuming a oscilloscope is rather mandatory later on. I'm more familiar with general electronics than with guitar effects, but from what I've seen, the whole point is that those boxes run a guitar signal through a network of components in an attempt to 'shape' and change the input signal to a desired new shape at the output. As the shape of your sinusoid is associated with the timbre of your instrument, you change its sound. This can be anything from simple changes (a 'high pass', 'low pass', or bandpass filter, to get rid of low pitches/high pitches/intermediate pitches), to using digital components (like diodes and transistors) to try and 'clip' off parts of the signal (making it sound more electronic), to oscillators (like certain types of op-amp circuits) that add vibrato onto your signal. These circuits can be combined and chained in different ways to make a more complex effect. If you had an idea on what you wanted to tackle first, you could look up a specific type of effect (or ask here).
|
|
#
?
Dec 11, 2013 02:08
|
|
|
I could be way mistaken, but here are my assumption about how some of the stuff works: A low voltage travels through the audio cable into the guitar pickups, and the strings modulate (idk the right word?) the magnetic field creating a frequency of voltage on the line. Then it travels to the amp and the transistors amplify the signal to drive the speakers at those frequencies. Small pots and resisters can lower the voltage so it lowers the volume. Transistors can can raise the voltage so they can make it louder, or create distortion. My hang up was, why or how do capacitors create filters? Reading your post I got to googling 'high pass' / 'low pass' filters and I got some promising links. I think those were the phrases I was needing, thanks a lot
|
|
#
?
Dec 11, 2013 04:03
|
|
|
Is this you? https://www.youtube.com/watch?v=4d4rdat3HdA&t=70s But seriously, no, you have the general concept of how an amplifier works (low power signal is amplified with transistors into a high power signal to drive the speakers) but you're cloudy on the details. Resistors reduce current, not voltage. If you want to reduce voltage at one point in a circuit you need an arrangement of resistors called a voltage divider. Transistors don't raise the voltage; they work like the handle on a faucet, letting a small motion (input signal from the pickups) precisely control a large flow (output signal to the speakers). The higher voltage and current come from somewhere else, usually the wall. Capacitors create noise filters by "absorbing" the spikes of higher power in an otherwise clean signal, then releasing the stored energy if the signal drops. That tends to even out the highs and lows to a middle-ground. The amount of energy the capacitor can absorb is directly related to how long of a spike it can smooth out, and therefore the frequency of the signal it will clean up.
|
|
#
?
Dec 11, 2013 04:19
|
|
|
Yup, the thing to realize with filters is that capacitors 'look' different in the circuit depending on what signal frequency they see. Let's take the following circuit: At DC/very low frequencies, capacitors 'look' like open circuits. So any of this signal would completely ignore the capacitor in the circuit and 'pass' at full strength. At higher frequencies, though, capacitors begin to 'look' like short circuits. If you were instead to pass a much higher frequency signal through here, it would get to the capacitor and be largely absorbed. That capacitor 'looks' like a short to ground (i.e., zero volts, or no signal). This is your basic low pass filter. The values of R & C chosen, in combination, determine at which frequency your filter changes from 'frequency passed' to 'frequency blocked'. Since an audio signal is a combination of both high and low frequency components, a filter like this can help separate out undesired portions and get rid of them. (Fun aside: If you switch the positions of the resistor and the capacitor, the circuit now passes high frequency and blocks low frequency. It turns into a high pass filter.) Now there is not a lot that you can do with just an LPF or an HPF. Your two options are 'no change in signal' and 'signal reduction'. However, they gain a lot of usefulness when used with other circuits, like amplifiers. When connected up together, one part of the circuit does something special (like amplify), while the RC portion helps select what frequencies these actions occur at.
|
|
#
?
Dec 11, 2013 06:01
|
|
|
Thank you, that was very helpful!
|
|
#
?
Dec 11, 2013 06:04
|
|
|
OK that is definitely helping me out. Thanks! Pretty much yeah! I had no interest in this stuff as a kid, I am so late to the party 
darkhand fucked around with this message at 06:12 on Dec 11, 2013 |
|
#
?
Dec 11, 2013 06:09
|
|
|
Does anyone have an updated link for the free electronics textbook from the OP? I've recently been getting more interested in electronic circuits, stemming from an interested in digital logic design coming from a CS major. I've switched to being an ECE major now, and I figure I have a lot to learn and it would only help to get ahead in material so it'll make the classes easier later on. (Have to take a year of general engineering before I can fully switch to ECE, and another ECE friend I do have tells me like half the class fails circuits 1, so I'm trying to avoid becoming one of that statistic if possible.)
|
|
#
?
Dec 11, 2013 14:52
|
|
|
darkhand posted:OK that is definitely helping me out. Thanks! This article is one of the better introductions to the subject: http://www.generalguitargadgets.com/richardo/distortion/ I'd recommend reading that document and then finding a simpler distortion pedal and try to recognize some building blocks - here's the preamp, the clipping section, the tone stack, etc etc etc.
|
|
#
?
Dec 11, 2013 22:35
|
|
|
ante posted:Someone at my local hackspace wants me to design a board for him. It looks fairly straightforward, it will control 18 channels of servomotors, and I expect it to take maybe three days of work, to design, test, and prepare for mass manufacturing. He has a site that sells robotics kits that he'd like to sell it on. What are fair compensation terms for me? Well, for both of us, I guess. This might be my own general slowness talking, but I'd be pretty surprised if you could design, fab, and test a board and prepare it for mass manufacturing in 3 days. Depending on what the circuit looks like, I'd probably budget a day or two in schematic/component selection, two days in layout, and day of getting everything set up BOM-wise/filewise. Then you have to support getting the board fabbed, and you probably have to populate the first few articles yourself. Plus there's the testing. All in all, that sounds more like 6-7 days of labor spread out over a couple of weeks as opposed to three consecutive days. I'd say if you have nothing else going on, charge him for parts and donate your time (maybe get some points in the total sales of the board). If you're looking to charge professional prices, I'd personally charge something on the order of $400/day. You might lie somewhere in the middle of that. edit: This is all assuming that the actual circuit is already designed, or that it's trivial (such as an array of MOSFETs and a few connectors). Poopernickel fucked around with this message at 11:14 on Dec 12, 2013 |
|
#
?
Dec 12, 2013 11:05
|
|
|
The board is really trivial, so I figured 8-10 hours to bang together a schematic, board, and send it off. Wait a few weeks for it to arrive, solder on the ~4-7ish components, test it, and repeat if necessary (but really hoping that's not necessary). That's a good perspective though, thanks for that. This is the kind of thing that donating my time(and getting a cut of the profits) would be a reasonable investment, with word getting around that I am smrt and can work good sometimes.
|
|
#
?
Dec 12, 2013 11:21
|
|
|
ante posted:The board is really trivial, so I figured 8-10 hours to bang together a schematic, board, and send it off. Wait a few weeks for it to arrive, solder on the ~4-7ish components, test it, and repeat if necessary (but really hoping that's not necessary). I don't know man, I'd consider myself a pretty experienced EE and I'd budget at least a week or two of engineering effort, especially since it has 18 servomotors on it. Coupled with the fact you're going to be selling this as a product, possible SW dev time...I'm not trying to dissuade you from taking it on (you absolutely should!), just budget the appropriate amount of time. Feel free to use us for reviewing / checking stuff though, I'm curious to see what it is!
|
|
#
?
Dec 12, 2013 18:59
|
|
|
This applies to just about every profession or hobby. Take how long you think it will take and then triple it. 3 days might be okay if you get absolutely everything right from the get-go, but the second anything goes wrong you'll probably need to spend a fair amount of time debugging.
|
|
#
?
Dec 12, 2013 19:39
|
|
|
reading posted:I came across something in an academic paper I wanted to ask about : The Shockley diode equation as mentioned by the previous posters predicts a 60mV increase in diode voltage for every decade increase in diode current at room temperature. The reason why the author said 100mV instead of 60mV could be due to the details of the physics of the MOS transistor below threshold, assuming that the diode-connected transistor you were reading about was a MOS transistor. The physics of how a MOS transistor turns on below threshold is very similar to how a diode current turns on with voltage except that often, normal bulk MOS transistors don't turn on as sharply as pn junction diodes. This has to do with the silicon underneath the gate seeing only some fraction of the full gate potential. Since the silicon underneath the gate, which forms one side of the p-n junction in the MOS connected transistor, only sees a fraction of the gate voltage, you need to apply more gate voltage to turn the diode-connected MOS transistor on. As an aside, this slower transistor turn-on is engineered out in the new tri-gate or FinFET transistors, or at least it should be ideally. silence_kit fucked around with this message at 20:24 on Dec 16, 2013 |
|
#
?
Dec 16, 2013 08:49
|
|
|
Delta-Wye posted:If I had to totally guess, a diode failed in the rectifier and is now shorted. Then can you help me figure this board out? I'm not super strong in electronics. This thing doesn't have any rectifier that I recognize, and it looks to me like it doesn't have enough big component diodes to do full wave rectification, let alone having those diodes attached together.   Or is the problem that big burn mark inside the LCD display?  Still, I'd love to be able to confirm what the problem is, even if I can't fix it. The only other slight damage is some smoke near the big resistors in the board cutouts and a little underneath zener diode Z3.
kid sinister fucked around with this message at 22:44 on Dec 16, 2013 |
|
#
?
Dec 16, 2013 22:39
|
|
|
kid sinister posted:Then can you help me figure this board out? I'm not super strong in electronics. This thing doesn't have any rectifier that I recognize, and it looks to me like it doesn't have enough big component diodes to do full wave rectification, let alone having those diodes attached together. It's not the burn mark. That's not actually a LCD display. That is a Vacuum Fluorescent Display or VFD. A VFD works just like any other vacuum tube. It has an filament that generates electrons, a grid to accelerate them, and a target grid. The target grid is covered in a phosphor that glows when it's hit by the electron beam. Of course, you need a vacuum for it to function. The better the vacuum the better the performance, longevity etc. When you pump the air out of the panel and seal it some is still left in or stuck on the various surfaces inside it. To get rid of this residual gas they leave some reactive metal in it and then use an induction heater to blast it onto the inside of the glass. It's called a getter. You can see one being fired in a different vacuum tube. It's the orange glowing thing at around 4:26. The angles they've chosen make it kind of difficult to see, but the glass does start to turn darker behind it. Another one, but much worse video quality.
|
|
#
?
Dec 18, 2013 06:47
|
|
|
I also have a diode question. I was talking to another EE here about RF switching diodes and he mentioned that applying RF without DC on the diode can short the oxide layer. I've never heard of this before. Can anyone point me to some discussion on this?
|
|
#
?
Dec 18, 2013 17:21
|
|
|
So it's basically near impossible to get samples from Atmel, huh?  Every other company has sent me some free poo poo to try out but Atmel is pretty stingy. Not like I'm resentful having to pay $2 for a chip or something, but it's neat to try stuff out gratis 
|
|
#
?
Dec 18, 2013 17:24
|
|
|
Martytoof posted:So it's basically near impossible to get samples from Atmel, huh? I'm not sure if Atmel started being like that before or after hobbyist stuff took off, but yeah, I can't really blame them. Although honestly the reason that PSoCs were one of the first things I used was because Cypress gave tons of samples (in DIP packages). I also got some Linux industrial controller dev kits from them too, back when their sample system was terrible (and thus allowed me to get $400 kits). The RF ADCs that I got from TI were still the craziest sample ever, though. I think they were like $800 each when I got them.
|
|
#
?
Dec 18, 2013 17:51
|
|
|
Aurium posted:It's not the burn mark. That's not actually a LCD display. That is a Vacuum Fluorescent Display or VFD. A VFD works just like any other vacuum tube. It has an filament that generates electrons, a grid to accelerate them, and a target grid. The target grid is covered in a phosphor that glows when it's hit by the electron beam. So each light-able segment is like a tiny one-color CRT? I'm actually a little familiar with repairing those from fixing old arcade cabinets. I had to fix an H-K short in a Joust monitor once. Still, that would explain the lack of a rectifier, right? They use AC instead of DC? I'm still trying to wrap my head around this board. What should I check next? kid sinister fucked around with this message at 18:33 on Dec 18, 2013 |
|
#
?
Dec 18, 2013 18:18
|
|
|
Slanderer posted:I'm not sure if Atmel started being like that before or after hobbyist stuff took off, but yeah, I can't really blame them. Although honestly the reason that PSoCs were one of the first things I used was because Cypress gave tons of samples (in DIP packages). I also got some Linux industrial controller dev kits from them too, back when their sample system was terrible (and thus allowed me to get $400 kits).  And here I thought I was being pretty abusive getting $20 analog multipliers from TI
|
|
#
?
Dec 18, 2013 18:35
|
|
|
TI is pretty ridiculous about sending things out. Plus I'm never not amused by what has to be a free $20 overnight delivery for a $2 part. The only downside about TI is that most of their samples are SOIC or other non-(general)hobbyist friendly stuff. Not like I expect them to cater to hobbyists though, and SOIC adapters are cheap It's the BGA and QFN stuff that blows
|
|
#
?
Dec 18, 2013 18:39
|
|
|
Martytoof posted:So it's basically near impossible to get samples from Atmel, huh? Unless you're with a company, it's tough. Our local sales rep can get us samples, but only because I work with a fairly well-known firm. TI gives all kinds of stuff away, and I'm amazed at the samples Samtec will ship out.
|
|
#
?
Dec 18, 2013 20:40
|
|
|
TI throws these big gala all-attendees-welcome catered parties every year at APEC. A few years ago, they were giving out 1" thick 'How to Grill' cookbooks at the door. So I think TI maybe just likes spending buckets of money on potential customers. 
|
|
#
?
Dec 18, 2013 20:53
|
|
|
Martytoof posted:So it's basically near impossible to get samples from Atmel, huh? Wow, I had no idea. I got several decent chips, but I indicated that I was working with a high school robotics team. I just put that for completeness -- I didn't think it mattered. Maybe it's actually why I got the chips? Also, I didn't go nuts -- just a few tinys, a 328, and a 644.
|
|
#
?
Dec 19, 2013 01:51
|
|
|
|
| # ? May 8, 2024 13:24 |
|
|
kid sinister posted:So each light-able segment is like a tiny one-color CRT? I'm actually a little familiar with repairing those from fixing old arcade cabinets. I had to fix an H-K short in a Joust monitor once. Quite similar yes. They use multiple filaments to produce a more uniform area electron source, instead of a point source that a CRT would use. There's also typically no cathode heater, the filament typically is the cathode. That said I looked at the board and don't see some things I expect. VFDs have a number of drive requirements. The filament is typically driven with an AC current to heat it (say 2vpp), and raised at a DC potential off ground(say 3v) , and then the anodes are driven much higher (around 20v). Usually you get this arrangement using a center tapped transformer. In the above 2vpp with a center tap connected to 3v. There's no transformer in the above. Now all this could be generated by a power module elsewhere, but I only see 2 connection points, which isn't near enough to for this, unless some of the voltages are generated on board. Even then, they must have taken some cheats. There are some zeners. You could measure them. The one one with the cutout(Z1) should be 22v, the other one(Z3) is 8.2v. These could potentially be used in part of a scheme to generate anode voltages. I have no idea what Z2 should be, but you could measure it and make sure that at least it's a different value on both sides. Or if the filament was seeing straight AC, it would be driven negative half the time, which would turn on all of the anodes, but would still prefer the ones that should be on. This would give you 60hz flicker on the rest of the elements. Another thing to look at is what's connected to E1 and E2. I wouldn't be surprised if there was a rectifier there.
|
|
#
?
Dec 19, 2013 02:48
|
|