|
BattleMaster posted:
Why would anyone stock individually wrapped transistors in the first place... Fun fact: about 1% of those bags were empty. Another 1% had two in them!
|
#
?
Sep 29, 2011 18:59
|
|
Why would anyone do that?
|
|
| # ? May 18, 2024 09:56 |
|
|
What are my options if I want to demodulate an AM signal with a 6-8 GHz carrier? Is there a way to make an envelope detector at that high frequency?
|
|
#
?
Sep 29, 2011 21:33
|
|
|
ANIME AKBAR posted:Did they pack each IC in a three foot tube, with all of the tubes in a triangular cardboard box? Because that's how arrow always ships stuff to me, and it's loving annoying. I had to order some EEPROMs for work, ordered 10 AT25256B (little baby TSSOPs), showed up in 4 foot long tube. Vendor? Arrow. 
|
|
#
?
Sep 29, 2011 21:54
|
|
|
BattleMaster posted:
It seems like many suppliers like have a fetish for overkill when it come to packaging. I just don't under stand how I get a 12"*6" package for a 10 pin header. (From the US to the UK, free P&P, farnell).
|
|
#
?
Sep 29, 2011 22:04
|
|
|
Krenzo posted:What are my options if I want to demodulate an AM signal with a 6-8 GHz carrier? Is there a way to make an envelope detector at that high frequency? You'd probably need to use an I&Q product detector. I suspect this is one of those things that's far cheaper to buy than to make yourself.
|
|
#
?
Sep 29, 2011 22:23
|
|
|
Krenzo posted:What are my options if I want to demodulate an AM signal with a 6-8 GHz carrier? Is there a way to make an envelope detector at that high frequency? How wide bandwidth is the AM signal? What is the power level? Do you need to block out interfering signals, or is your signal going to be the strongest thing in the area? Recovering AM is actually really easy. You just rectify it and then low pass filter - a diode and a cap would work, if your input power level is high enough. Unfortunately, you'll probably get 12-15 dB loss using a passive diode demodulator. You'll also need a good RF diode - probably GaAs, but you might be able to get a Si one to work. If you use active components, your options are much greater. You could do a super het down conversion to get it to some low frequency IF, then do all your filtering and demodulation at your IF. However that would require substantially more microwave engineering, and an expensive VCO if you really need 2 GHz of tuning.
|
|
#
?
Sep 29, 2011 23:08
|
|
|
So some of you might recall the design of the H-Bridge I had earlier in the thread...turns out that's no-where near what we need. I put a current clamp on the motor and measured Imax of ~200A, and Iave of about 60A. So allow me to explain my problem specification and if anyone can come up with a light, effective solution...I'll buy you a beer. Essentially there is a bushed DC motor (from a cordless drill), that draws at the above specs. Its to be direction controllable for 200ms pulses, max with 10-15 seconds cool off before firing again. We have two control lines one goes high for one direction, the other will go high to go in reverse direction. When a signal line is not used it goes high-z. There is no requirement, or control for features such as active-braking etc. My question is, what is the lightest, quickest, most reliable, cost effective way to do this? Due to the high inrush, I'm keen to stay away from mechanical relays. I've found some SSRs, but we would require 4 of them (as far as i can figure) to get the motor to be able to go both directions. Any ideas? PLEASE!
|
|
#
?
Sep 30, 2011 02:04
|
|
|
MisterSparkle posted:So some of you might recall the design of the H-Bridge I had earlier in the thread...turns out that's no-where near what we need. I put a current clamp on the motor and measured Imax of ~200A, and Iave of about 60A. MOSFETs, lots of MOSFETs. An H-Bridge is still fine, but you will need to put several FETs in parallel to achieve your current requirements. When using an h-bridge at high currents, you must ensure there is suffiecient time between switching directions that both halves of the gate are not on together, or a short between power and ground can result and the mosfets can explode! Ensure you have sufficient heatsinking, and large, low resistance paths between the power source, mosfets and motor. I built an h-bridge to drive a wheelchair motor from a pair of car batteries. It had 8 fets per section, so a total of 24, and large aluminium heatsinks. The first couple turned into molten goo because of a minute overlap between directions. Caused no issues with a bench power supply, but on battery (with several thousand potential amps) it was too much, and the release of much magical smoke occured.
|
|
#
?
Sep 30, 2011 03:17
|
|
|
MisterSparkle posted:My question is, what is the lightest, quickest, most reliable, cost effective way to do this? Due to the high inrush, I'm keen to stay away from mechanical relays. I've found some SSRs, but we would require 4 of them (as far as i can figure) to get the motor to be able to go both directions. However, only N channel FETs will get meet your specs. So you either need a fancier H bridge with high side drivers, or a different switching topology for handling both directions. The reversal of polarity is sometimes handled with a normal relay or just a switch, while the on/off control is with the mosfet. If you interlock things correctly, you can ensure that the relay/switch is never toggled under load, thus making is much easier to spec. edit: here is a document describing in detail what I'm talking about : http://www.zilog.com/docs/z8encorexp/appnotes/AN0255.pdf They do the speed control with one big mosfet, and use a two pole relay for direction control. edit2: but they seem to neglect putting a catch/flyback diode on the motor..? Weird. ANIME AKBAR fucked around with this message at 13:59 on Sep 30, 2011 |
|
#
?
Sep 30, 2011 05:03
|
|
|
movax posted:I had to order some EEPROMs for work, ordered 10 AT25256B (little baby TSSOPs), showed up in 4 foot long tube. Vendor? Arrow. On the other hand, once I got an order from newark and several of the ICs came in awesome plastic, clamshell hinged, padded antistatic cases. No idea why, the parts weren't special. I still use the cases for storing my ESD sensitive transistors.
|
|
#
?
Sep 30, 2011 05:07
|
|
|
SnoPuppy posted:... I don't want to do anything complicated. I have an UWB transmitter right now that transmits a 250 ps pulse to the receiver. It gets distorted during transmission and reception by my antenna and filters. I found a paper where they used an 8 GHz carrier wave and modulated their 250 ps pulse with the carrier. Then on the receiver, they used IQ demodulation (don't know the correct name) to get back the 250 ps pulse without any distortion like I'm having. I would like to accomplish the same thing without much effort as this is not necessary for my project. If I can create a simple envelope detector on the receiver, then I would be willing to go this route. Can anyone recommend some GaAs diode model numbers that are readily available such as on digikey?
|
|
#
?
Sep 30, 2011 06:15
|
|
|
Krenzo posted:I don't want to do anything complicated. I have an UWB transmitter right now that transmits a 250 ps pulse to the receiver. It gets distorted during transmission and reception by my antenna and filters. I found a paper where they used an 8 GHz carrier wave and modulated their 250 ps pulse with the carrier. Then on the receiver, they used IQ demodulation (don't know the correct name) to get back the 250 ps pulse without any distortion like I'm having. I would like to accomplish the same thing without much effort as this is not necessary for my project. If I can create a simple envelope detector on the receiver, then I would be willing to go this route. With a simple diode detector, you'll need a lot more preamplification, and probably a good zero-biased detector. I don't know of any part numbers off hand, since I don't work in the GHz range myself.
|
|
#
?
Sep 30, 2011 12:57
|
|
|
ANIME AKBAR posted:An IQ demodulator really isn't too complicated, though. You need a couple mixers, a 90 degree hybrid coupler (microstrips at your frequency), and a decent reference oscillator (that might be the tougher issue...). It's major advantages are less conversion loss and distortion, and it's probably easier to get a very wide bandwidth. Could you link the paper? I'm kind of curious to see the details of their receiver chain. Here's the paper: http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=4511506 They don't really go into much detail on it, and they do the combining of the I and Q in software which I can't do because I'm not using ADCs.
|
|
#
?
Sep 30, 2011 16:56
|
|
|
Krenzo posted:I don't want to do anything complicated. I have an UWB transmitter right now that transmits a 250 ps pulse to the receiver. It gets distorted during transmission and reception by my antenna and filters. I found a paper where they used an 8 GHz carrier wave and modulated their 250 ps pulse with the carrier. Then on the receiver, they used IQ demodulation (don't know the correct name) to get back the 250 ps pulse without any distortion like I'm having. I would like to accomplish the same thing without much effort as this is not necessary for my project. If I can create a simple envelope detector on the receiver, then I would be willing to go this route. At 8 GHz, very few things are not complicated. What is your transmitter architecture? Or is that still TBD? Keep in mind that a 20 mil long trace (the size of an 0402 pad) will have ~10 deg phase shift at 8 GHz, so your layout will be critically important. Ideally, you will have some type of RF simulation tool like AWR or ADS to help you design and validate your circuit. Also keep in mind that your 250 ps pulse is going to have around 4 GHz of bandwidth (assuming some type of pulse filtering), and only be 2 cycles of your 8 GHz carrier. This might be too weak to capture with an envelope follower, if you don't have any amplification in the front end. Just the free space path loss at 8 GHz through 10 meters will be ~70 dB. It sounds like you're already capturing a signal today - what is the amplitude you're seeing? As far as parts go, you might try the BAT15 series from Infineon - I'm pretty sure those are available on digikey/mouser. If you decide to take the IQ mixer approach, you will have more of a design challenge. Not to say that's bad though. First, there aren't too many companies that make good mixers in the 8 GHz range because this isn't a commercial frequency band. You might have to go with something by Hittite - I took a quick look at RFMD and Skyworks and didn't see anything. You could do a discrete mixer design, but that will only add complexity. Second, trying to do a direct conversion to base band will be difficult because it's extremely unlikely that your transmitter and receiver will be tuned to exactly the same frequency and phase. Even if you have a +/- 1 ppm LO, that's still potentially 8 KHz of frequency drift, causing your RX signal will be modulated with the frequency difference. If you're dealing with a true AM signal (not DSBSC), you can lock to the incoming carrier wave, which can make things easier. Another option is to use a super heterodyne approach and bring your signal to a lower intermediate frequency. This will make the design easier, since instead of 8 GHz you can work with hundreds to tens of MHz.
|
|
#
?
Oct 1, 2011 02:58
|
|
|
SnoPuppy posted:At 8 GHz, very few things are not complicated. My current transmitter is: pulse trigger (>10 ns) -> pulse sharpener (to 1 ns) -> pulse sharpener (to 250 ps) -> 3-4 GHz filter -> 0-6 GHz amp -> antenna. I plan to put a mini circuits mixer that works at 8 GHz in there and feed the pulse into the mixer for modulation. SnoPuppy posted:Keep in mind that a 20 mil long trace (the size of an 0402 pad) will have ~10 deg phase shift at 8 GHz, so your layout will be critically important. Ideally, you will have some type of RF simulation tool like AWR or ADS to help you design and validate your circuit. Yeah, I'm not interested in putting that much work into this. SnoPuppy posted:Also keep in mind that your 250 ps pulse is going to have around 4 GHz of bandwidth (assuming some type of pulse filtering), and only be 2 cycles of your 8 GHz carrier. The paper I cited used a 300 ps pulse, and I can easily change my pulse to whatever works best. SnoPuppy posted:This might be too weak to capture with an envelope follower, if you don't have any amplification in the front end. Just the free space path loss at 8 GHz through 10 meters will be ~70 dB. I plan on putting amplification on the front end, and my current receiver has two amplifiers in it right now to get it to a level I can comfortably work with. The paper said they could manage a distance of 5 meters which is what I would be happy with. I really just want the orientation of the antenna/polarization to not change the shape of my signal which complicates things.
|
|
#
?
Oct 1, 2011 07:17
|
|
|
what arduino starter kit is recommended for someone vaguely experienced with programming concepts and circuit building? really interested in getting back into this sort of stuff, after building some guitar pedals a few years ago.
|
|
#
?
Oct 1, 2011 08:25
|
|
|
You guys are going to absolutely murder me, since I'm pretty sure that this will just use a simple circuit, but here goes: http://volpinprops.blogspot.com/2009/02/bioshock-little-sister-adam-syringe.html So, I was planning on just taking an LED Flashlight and gutting it for parts, and hooking it up to a simple 9V, and ideas on a good type of switch mechanism for this? I was considering making it toggle by squeezing the handle. Any idea where to get a frosted cap, whatever that is?
|
|
#
?
Oct 1, 2011 14:55
|
|
|
MisterSparkle posted:So some of you might recall the design of the H-Bridge I had earlier in the thread...turns out that's no-where near what we need. I put a current clamp on the motor and measured Imax of ~200A, and Iave of about 60A. IGBTs, heavy-duty ones like are used in industrial VFDs. You'll only need two bricks (two IGBTs each) to make an H-bridge. Just make sure your driver circuit is timing things right, shoot-through is no fun.
|
|
#
?
Oct 1, 2011 15:23
|
|
|
KaiserBen posted:IGBTs, heavy-duty ones like are used in industrial VFDs. You'll only need two bricks (two IGBTs each) to make an H-bridge. Just make sure your driver circuit is timing things right, shoot-through is no fun. IGBTs would be a bad idea at 12V. They would immediately eat up 25% of the delivered power since they're bipolar devices. And industrial bricks are probably far too big for what he wants to do.
|
|
#
?
Oct 1, 2011 15:38
|
|
|
ANIME AKBAR posted:IGBTs would be a bad idea at 12V. They would immediately eat up 25% of the delivered power since they're bipolar devices. And industrial bricks are probably far too big for what he wants to do. D'oh, I thought he was asking about higher voltage DC, more like 72-96V. The industrial ones come in all sizes, starting around matchbox sized for 25A versions. 100A versions are ~1x1.5x4", but pretty heavy. If you're pulling 60A average current, you won't get very small, period; you have to have somewhere to land the wires.
|
|
#
?
Oct 1, 2011 16:25
|
|
|
MisterSparkle posted:So some of you might recall the design of the H-Bridge I had earlier in the thread...turns out that's no-where near what we need. I put a current clamp on the motor and measured Imax of ~200A, and Iave of about 60A. There are plenty of MOSFETs that are rated to 200A continuous, let alone 200A pulsed. Here's one from IRF: http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=IRLB3813PBF-ND If you can use surface mount parts, you're going to have an easier time finding something to handle the current. The problem isn't in the FET itself, it's in the package. You're still going to have to heat sink the FETs though, and you're going to want to switch them fast to keep them out of the linear region. You should also be driving them with a Vgs of like 10V to make sure that the Rds-on is as low as it can be (you might be able to get away with a 5V driver given your duty cycle, but 10 will be better).
|
|
#
?
Oct 1, 2011 17:20
|
|
|
SnoPuppy posted:There are plenty of MOSFETs that are rated to 200A continuous, let alone 200A pulsed. But for 60A continuous it's definitely plausible. But you will have to be careful with thermal management. KaiserBen posted:If you're pulling 60A average current, you won't get very small, period; you have to have somewhere to land the wires. ANIME AKBAR fucked around with this message at 19:19 on Oct 1, 2011 |
|
#
?
Oct 1, 2011 19:16
|
|
|
MisterSparkle posted:Essentially there is a bushed DC motor (from a cordless drill), that draws at the above specs. Its to be direction controllable for 200ms pulses, max with 10-15 seconds cool off before firing again. So.. if I'm understanding you correctly there will be a 200ms pulse inevitably followed by a 10-15 second cooldown? In other words - a 2% duty cycle? If that's the case then a single beefy MOSFET with low on-resistance should be sufficient, even without a heatsink. Also, if you don't need instant switching of directions, I'm fairly sure there are relays designed to switch motor directions.. I suppose you could use one of those to switch directions inbetween pulses, and then turn power on/off with the MOSFET. But don't take my word for it - I'm usually satisfied when my stuff doesn't instantly disintegrate. Edit: Apparently it's called a changeover relay Astrolite fucked around with this message at 20:13 on Oct 1, 2011 |
|
#
?
Oct 1, 2011 20:09
|
|
|
Basic Beater posted:So, I was planning on just taking an LED Flashlight and gutting it for parts, and hooking it up to a simple 9V, and ideas on a good type of switch mechanism for this? I was considering making it toggle by squeezing the handle. Every time I've seen the word frosted I've taken it to mean either diffused LEDs, or you imitate diffusion by sandpapering the glass down on the LED. Both work just fine. http://www.instructables.com/id/Music-LED-Light-Box/step3/Frosting-and-Diffusing/ This project is a solid example of the use of various types of frosting.
|
|
#
?
Oct 2, 2011 03:10
|
|
|
I really have to get back to that fuel gauge thing, but I have another question. I was looking at solar controllers on eBay which seem more than adequate for my uses for the most part. They have two terminals for the panel/s, two terminals for the battery and two for the power feed to whatever runs off the battery. the purpose of the feed is so it can have a low voltage cutout, which is fair enough. It looks like the maximum draw through that is 2A, which to me is drat pathetic. Is there any reason I can't just put a high current relay there to run my 12 stuff? I would think the voltage for the cutout would be measured at the charging terminals. Would this be correct? Otherwise I don't want to be nerfing the cutout by using a relay which would only be providing switching current. The whole exercise here is really convoluted. I pretty much want a half decent solar charger for my VW camper because no matter what it has always suffered from a mysterious parasitic draw which defies detection. I have one of those lighter based panels which isn't up to the task, even though it provides enough power to operate the AM CB radio (barely) or the interior lights. So I want to take it up a notch, but on a budget. This parasitic loss problem has defeated me for about a decade. The multimeter only shows a tiny bit of current flowing (probably to the stereo) with the engine off. I fitted a battery isolator to stop the mystery drain with enough current to kill the battery in 3 days. But it'd be nicer to force the thing to keep it charged, and to flip over to the aux battery when wanted.
|
|
#
?
Oct 2, 2011 06:59
|
|
|
ANIME AKBAR posted:And what do you mean "wires?" I'd hope a motor that draws 60A has busbar tabs, not wires. Why would you use busbars for only 60A? That sounds like an enormous PITA, especially given that it's only #6 wire for 60A continuous, much smaller if we're talking very low duty cycle. Hell, my welder draws 100A on a 40% duty cycle on #6. Most AC motors still use wire up into the 1000A range, and many beyond that. Once you get beyond 6x 4/0 per phase, busbars start to make sense, but so does higher voltage, and higher voltage is usually easier. DC usually uses wires into the 500A range, sometimes higher. Once you hit 5kA or so though, busbars become quite common.
|
|
#
?
Oct 2, 2011 16:11
|
|
|
Hi guys, I was hoping someone could help me identify what kind of capacitors I've got here: http://longview.be/uploader/uploads/20111002074520PM2011-10-02%2021.30.27.jpg, apologies for the messy wiring, I think the yellow ones are a sort of Polyester, they're Philips branded, my real question is about the metallic ones, I've got a bunch and they range from a few nF up to about 20µF. I'm using them for IC decoupling in that circuit (OPA2134+LME49600 headphone amp with a speaker protect circuit at the top controlling the relay), are they good or bad for that (I did have oscillations without them at first), I'll replace the 2134s with higher speed LME49720s (55 MHz) at a later point, will they still be acceptable?
|
|
#
?
Oct 2, 2011 20:52
|
|
|
longview posted:Hi guys, I was hoping someone could help me identify what kind of capacitors I've got here: quote:I'm using them for IC decoupling in that circuit (OPA2134+LME49600 headphone amp with a speaker protect circuit at the top controlling the relay), are they good or bad for that (I did have oscillations without them at first), I'll replace the 2134s with higher speed LME49720s (55 MHz) at a later point, will they still be acceptable?
|
|
#
?
Oct 2, 2011 23:49
|
|
|
Interesting, I have a ton of them in a large bag. As I understand they were leftovers from when F-16 parts were made by a local airforce contractor. I also got a bag of LM101AHs (TO-99, they seem like 741s without the 22pF cap), a small bag of JFET switching transistors and a stack of boxes with 1000 1k, 1R, 2k2 etc. resistors. I don't have a calibrated LCR meter handy but my multimeter put most of the metallic ones at around 1-20µ while the smallest was 68 nF. That said I'll probably get some ceramic/tantalum caps as the datasheet for each IC recommends for a final build. Now a question about the amplifier project: I built my own DC-protection circuit but I'm wondering how to make it more effecient, the schematic: http://longview.be/uploader/uploads/20111003080248PMdcprotect.png. Basically it low pass filters the output and if the signal goes outside +-0.7V it kills the relay (-15V to the gate), the problem is it requires four comparators + 2n7000s for stereo. The RC filter bit is already used in the circuit for a DC servo, so that has to stay. Can I make this use less chips in a way that won't compromise the circuit? For those wondering, the DC servo protects against DC coupled from the input, but if one supply rail or the output stage fails, it would put the other supply across the output, frying the headphones, I've already built and tested the circuit shown.
|
|
#
?
Oct 3, 2011 21:08
|
|
|
longview posted:Interesting, I have a ton of them in a large bag. As I understand they were leftovers from when F-16 parts were made by a local airforce contractor. I also got a bag of LM101AHs (TO-99, they seem like 741s without the 22pF cap), a small bag of JFET switching transistors and a stack of boxes with 1000 1k, 1R, 2k2 etc. resistors. Why not just stick some big diodes to ground? That should clamp your voltage if it strays too far. Or do you need to allow for instantaneous voltages that might be outside the +/- 0.7V range, so long as they don't persist?
|
|
#
?
Oct 4, 2011 02:18
|
|
|
SnoPuppy posted:Why not just stick some big diodes to ground? That should clamp your voltage if it strays too far. Yes, the circuit should only care about DC voltages (i.e. frequency >1 Hz). Also keep in mind I'm using audiophile design rules here, so putting something non-linear in the circuit path is a big no-no. I'm not married to using +-0.7V as threshold voltages though, the important rule is the circuit should cut immediately if one supply rail drops (relay connected between Vcc and Vee), and in case the entire supply voltage (either side) appears on the output for more than a fraction of a second, cut it. Now, regarding opamps: am I right in assuming that if I connect a voltage follower input to ground, any current sourced or sinked to the output will NOT go through the referenced ground but rather the opposite polarity power rail? I.e. something tries to put 10mA from the positive rail to the opamp virtual ground, that current will flow to the negative rail to maintain the 0V level?
|
|
#
?
Oct 4, 2011 08:10
|
|
|
longview posted:Yes, the circuit should only care about DC voltages (i.e. frequency >1 Hz). Also keep in mind I'm using audiophile design rules here, so putting something non-linear in the circuit path is a big no-no. It's only non linear when it's on  You're quite right though - it probably wouldn't be the best for an audio application, especially if you actually need to pass transients that would exceed your turn on voltage. What you have is probably good enough, although you can buy integrated window comparators if you want. They'll make for a smaller design, but likely more expensive. longview posted:Now, regarding opamps: am I right in assuming that if I connect a voltage follower input to ground, any current sourced or sinked to the output will NOT go through the referenced ground but rather the opposite polarity power rail? Yes - inputs to modern op-amps have very little current flow on their input terminals (usually on the order of 1 nA, good ones will be in the pA range).
|
|
#
?
Oct 5, 2011 05:16
|
|
|
I want to add a hardware entropy source to a board with an FPGA & uC. I'd like to find a cheap, low part count solution. I was thinking an LC oscillator would be simpler and cheaper than any of the avalanche noise designs. The FPGA will toggle a bit (at around 100-500x the freq of the LC oscillator) that is shifted into a register on the oscillator edges. After filling the register, the data would possibly be post-processed on the FPGA. Regardless, the data is then passed to the uC for error checking and more processing. I think a Colpitts oscillator will be the simplest and cheapest. Any other ideas? Should I be looking at a different oscillator topology?
|
|
#
?
Oct 5, 2011 20:44
|
|
|
taqueso posted:I want to add a hardware entropy source to a board with an FPGA & uC. I'd like to find a cheap, low part count solution. I was thinking an LC oscillator would be simpler and cheaper than any of the avalanche noise designs. The FPGA will toggle a bit (at around 100-500x the freq of the LC oscillator) that is shifted into a register on the oscillator edges. After filling the register, the data would possibly be post-processed on the FPGA. Regardless, the data is then passed to the uC for error checking and more processing. What FPGA are you using? Some have built in ADCs for doing power rail diagnostics, so you could just sample the LSB of that (if it exists). Another option would be to use an LFSR to generate a PRBS data stream on one clock domain, and then sample that data stream on a different, uncorrelated clock domain. You could even use a ring oscillator to generate the clock for the LFSR, if you don't want to put down anything else on the PCB. How random does your random data need to be?
|
|
#
?
Oct 5, 2011 23:30
|
|
|
SnoPuppy posted:What FPGA are you using? Some have built in ADCs for doing power rail diagnostics, so you could just sample the LSB of that (if it exists). quote:Another option would be to use an LFSR to generate a PRBS data stream on one clock domain, and then sample that data stream on a different, uncorrelated clock domain. You could even use a ring oscillator to generate the clock for the LFSR, if you don't want to put down anything else on the PCB. Your setup seems like it would be reasonable space-wise, though. quote:How random does your random data need to be? I am not going to worry about an attacker with enough physical access to modify the PCB, other than to detect things like "the bits are always 1". I'd like to avoid using sources that are obviously controllable, like temperature or network interrupt timing. I am planning on using a good PRNG that has the seed changed/entropy added periodically, similar to arc4random, probably with a simpler/less secure algorithm. I shouldn't need a huge amount of data per second/minute.
|
|
#
?
Oct 6, 2011 00:21
|
|
|
Wait, so what's wrong with like a Zener noise source?
|
|
#
?
Oct 6, 2011 02:51
|
|
|
Slanderer posted:Wait, so what's wrong with like a Zener noise source? I think the zener would make "better" noise, but I can make the oscillator with 5 passives and a transistor. I don't know how to amplify zener noise anywhere near that cheap (or mm^2 or parts count). That doesn't mean it isn't possible, of course. Suggestions welcome e: actually was just thinking maybe I could use the ADC instead of amplifying to digital, possibly get away with 1 or 2 transistors. taqueso fucked around with this message at 03:31 on Oct 6, 2011 |
|
#
?
Oct 6, 2011 03:29
|
|
|
taqueso posted:I think the zener would make "better" noise, but I can make the oscillator with 5 passives and a transistor. I don't know how to amplify zener noise anywhere near that cheap (or mm^2 or parts count). That doesn't mean it isn't possible, of course. Suggestions welcome You could reverse bias a transistor, AC couple the base, and then use two more transistors to make a 2 stage amplifier. Boom, done in 3 transistors + some resistors for biasing and AC coupling caps. You might still want to buffer it just in case. I still think that a ring oscillator in the FPGA fabric and an LFSR wouldn't take up much FPGA logic, requires 0 external parts, and would probably be pretty decent.
|
|
#
?
Oct 6, 2011 04:06
|
|
|
What is the absolute cheapest thing I can use for potting compound? I had to remove the inch thick jello like coating from the ignition computer in my truck to diagnose a problem, and now I need something to cover all the electronics back up and protect them from moisture. 5140 Methoxy Silicone seems ideal, but it's like $20 a tube.
|
|
#
?
Oct 6, 2011 16:54
|
|
|
|
| # ? May 18, 2024 09:56 |
|
|
What's the best way to get a negative voltage supply? I needed +5 and -5 voltage sources for a comparator I'm working with. So, I used a +10V source as the +5, a +5V for ground, and the actual ground for -5V. Is this ok? It works fine now, but I'm about to put multiple devices on a single board that I would like to all use the same ground. Things seem to act up when I start connecting other things to the +5 as a ground. Is there a better way I should be getting -5V?
|
|
#
?
Oct 9, 2011 21:51
|
|