|
lollynoob posted:Hello people smarter than I am, I have a question about three-state logic chips. Say I had a few registers with three-state outputs (in my case, 74hc574's) connected to a bus. What I want to do is have the output enable pins connected to some sort of decoder, so that only one of the outputs will be enabled at any given time. What I'm wondering, though, is if there will be some in-between state as one register is disabling and another is enabling, where the two will short together and damage each other. If I had an oscilloscope I'd test it myself, but I am a hobbyist with no money so that's out of the question for now. Thanks! There's no way that a pair of 74-series ICs can source or sink enough current, especially in such a short time, to cause any damage to each other. I mean if that was an issue then all of the computers from the 70s/early 80s with loads of 3-state 74-series ICs wouldn't have worked. BattleMaster fucked around with this message at 21:07 on Aug 12, 2009 |
#
?
Aug 12, 2009 21:05
|
|
|
|
| # ? May 22, 2024 16:53 |
|
|
lollynoob posted:Hello people smarter than I am, I have a question about three-state logic chips. Say I had a few registers with three-state outputs (in my case, 74hc574's) connected to a bus. What I want to do is have the output enable pins connected to some sort of decoder, so that only one of the outputs will be enabled at any given time. What I'm wondering, though, is if there will be some in-between state as one register is disabling and another is enabling, where the two will short together and damage each other. If I had an oscilloscope I'd test it myself, but I am a hobbyist with no money so that's out of the question for now. Thanks! Unless you insert successive delays in-between each output enable pin, you will definitely have transitional states where multiple outputs are on simultaneously. With that said, this probably isn't a big deal. The maximum amount of shoot-through time you'd be looking at would be the rise time from when the output enable goes high until when the tristate activates. This is probably somewhere between 10ns-100ns. Since the shoot-through time is so short, it's unlikely to actually damage any of your ICs. If you repeatedly toggled the output enables (at maybe 10MHz / 50% duty cycle), you could definitely blow up your chips. But if you're talking about an event that happens once every 1kHz or so then you've got nothing to worry about. One thing to be aware of though - although it won't damage your ICs, it _will_ dump current spikes onto your ground plane every time it happens. You'll want to be sure and keep your digital ground separated from your analog ground all the way back to the source. edit: Battlemaster 
|
|
#
?
Aug 13, 2009 01:08
|
|
|
Your answer was better anyways. lollynoob didn't mention anything about it, but if the bus master is a CPU there should definitely be delays between bus accesses. If it's a microcontroller with firmware-controlled signals then just make sure everything is put into an inactive state for a moment between accesses. If you don't have control over it you shouldn't sweat it though. Edit: drat did they switch the position of Creative Convention on the forum list? It seems like I've been having trouble finding this subforum every day for the last week 
BattleMaster fucked around with this message at 02:07 on Aug 13, 2009 |
|
#
?
Aug 13, 2009 02:02
|
|
|
Well for some background information, I'm planning on using these chips in a simple 8-bit processor, sort of like what this guy did (but nowhere near as complex), to try to teach myself digital logic and electronics in general. I've already built a circuit that uses input from a computer's parallel port to test static RAMs, so I'm not completely clueless about these things, but I definitely don't know as much as I'd like to. Back on the subject of the three-state ICs though, I don't like the idea of the chips shorting at all, especially if it could cause damage at higher frequencies (I'm hoping to eventually run this processor at around 4MHz or so). You mentioned inserting a delay; is there a way to do this other than having the registers sit idle (with none enabled) for a full clock cycle? Also more importantly, is it even sane to think about building such a thing without an oscilloscope? I've never used one so I don't exactly know what it would help with, but I've heard they're very useful.
|
|
#
?
Aug 13, 2009 02:16
|
|
|
Not only would waiting an entire clock cycle with no bus access be the safest way to avoid bus contention, but probably the easiest. What you're building isn't exactly going to be a speed demon anyways, so I don't think you're going to notice it. Even the Z80 and 6502 played it safe and waited a couple of clock cycles between accesses.
|
|
#
?
Aug 13, 2009 02:27
|
|
|
Waiting a clock cycle is definitely doable; thanks for the help.
|
|
#
?
Aug 13, 2009 02:47
|
|
|
Welp I've pretty much finished my work on the wireless power crap. I can't really post the final report because I don't own the IP rights to it, but the bottom line is basically that the theory is sound and accurate, but getting decent efficiency requires extremely low loss coil inductors. So unless you have the resources to make an inductor out of silver plated litz wire, your efficiency is going to be pretty crappy. I was actually surprised at how close I was able to get to theoretical yields: So around half at mid-range distances. I know now I could have gotten better results if I did some things differently, but I didn't really have the resources necessary. I have trouble thinking of good applications for it, except in medical stuff, which is already being done. I think my supervisor wants to implement it where one coil is actually a pcb trace. Good luck with that, I won't be around to do it. ANIME AKBAR fucked around with this message at 19:43 on Aug 13, 2009 |
|
#
?
Aug 13, 2009 19:41
|
|
|
What were the simulations?
|
|
#
?
Aug 13, 2009 21:29
|
|
|
LTSPICE. Red is with an ideal voltage source driving it, green is with the driver circuitry I used.
|
|
#
?
Aug 14, 2009 00:08
|
|
|
lollynoob posted:Well for some background information, I'm planning on using these chips in a simple 8-bit processor, sort of like what this guy did (but nowhere near as complex), to try to teach myself digital logic and electronics in general. I've already built a circuit that uses input from a computer's parallel port to test static RAMs, so I'm not completely clueless about these things, but I definitely don't know as much as I'd like to. Back on the subject of the three-state ICs though, I don't like the idea of the chips shorting at all, especially if it could cause damage at higher frequencies (I'm hoping to eventually run this processor at around 4MHz or so). You mentioned inserting a delay; is there a way to do this other than having the registers sit idle (with none enabled) for a full clock cycle? An oscilloscope is not required, but it can be pretty dang helpful for tracking down mistakes. If you're looking to buy your own scope, however, it might not be worth the money. The scopes which are best for digital work often cost well into the thousands. Analog scopes are much cheaper, but are necessarily what you need. I hear there are decent USB scopes these days which use your PC for most of the processing and interface. I've never used one myself but I've seen them hit the sub-$200 price range. If you're talking about speeds on the order of maybe 5MHz, a simple 10MHz scope will be OK. One more note about that tri-state business - The easiest way to guarantee that there will never be two items enabled simultaneously is to use a separate enable line for each device on your bus. You then use each enable as a "strobe", meaning that you enable it, complete your operation, and then disable it again. Typically you would include weak pull-up resistors (maybe 200k or so) on your data bus so that it has a default value when the tri-states are all inactive.
|
|
#
?
Aug 14, 2009 00:11
|
|
|
lollynoob posted:Well for some background information, I'm planning on using these chips in a simple 8-bit processor, sort of like what this guy did (but nowhere near as complex), to try to teach myself digital logic and electronics in general Also, congratulations on moving into the electronics world! it's a great place to be. Your project sounds awesome  A simple way to 'prove' your design before you go through all the effort of wiring it up is to simulate it first. SPICE simulations are great for analog electronics but really lousy for complex digital circuits. You might consider downloading the Xilinx Webpack and messing around with it a little bit. It is an FPGA/CPLD development environment that Xilinx gives away (at least, the Webpack version). I would try to find Webpack version 9, for reasons explained below: You can enter digital circuits using a fair straight-forward schematic editor, or you can also type them in as gates in a Verilog netlist. Once you've made a schematic and/or Verilog version, you can simulate the entire design with a pretty fast simulator. Versions 9 and below all come with a graphical testbench editor that lets you design test signals for your design. I think Version 11 (the current version) has ditched the graphical testbench tool, but Version 9 has it for sure. Even if you never use an FPGA, you can still use the schematic editor and simulator for some pretty bitchin' simulations. Poopernickel fucked around with this message at 00:23 on Aug 14, 2009 |
|
#
?
Aug 14, 2009 00:19
|
|
|
Poopernickel posted:Versions 9 and below all come with a graphical testbench editor that lets you design test signals for your design. I think Version 11 (the current version) has ditched the graphical testbench tool, but Version 9 has it for sure. What did they replace it with? Seems odd they would lose functionality.
|
|
#
?
Aug 14, 2009 01:10
|
|
|
Actually, they retired it completely in favor of HDL-based testbenches. Now they have a little wizard thingie that generates HDL testbenches but you still have to write in all of the state changes by hand (or code them procedurally).
|
|
#
?
Aug 14, 2009 02:30
|
|
|
ANIME AKBAR posted:LTSPICE. Red is with an ideal voltage source driving it, green is with the driver circuitry I used. Very interesting. I have so many questions. What coils did you end up using? I assume you used a resistive load on your secondary tank, was that set up to provide the constant nominal output power as the secondary voltage changed (I assume it changed with distance?) Mmmmmmmm, silver litz wire....
|
|
#
?
Aug 14, 2009 03:00
|
|
|
One coil was quarter inch copper tubing, the other was 18awg enamel wire. One thing that blew my mind was that a layer of oxidation on the bare copper tubing cut my efficiency about in half (I actually measured the coil's resistance while sanding its surface and it would steadily get better). I actually obtained some mediocre litz wire, but by then I didn't want to remake the whole experiment so I just went with the bare copper. I didn't have any power regulation or anything. I just picked a resistive value to match the impedance of the coupled tanks for maximum efficiency.
|
|
#
?
Aug 14, 2009 04:03
|
|
|
Did you end up going with the class E or whatever? Frequency? Pancake primary coil? The efficiency at small distances is pretty impressive. Had my final year project inspection today. I think the inspectors were suitably impressed.
|
|
#
?
Aug 14, 2009 06:31
|
|
|
I used something similar to a class E, yeah, but I can't really go into specifics since that's one of the more valuable parts of the report. Suffice to say I got to the point where most of the losses are in the resonant element, so the driving circuitry isn't the limiting factor anymore.
|
|
#
?
Aug 14, 2009 14:02
|
|
|
I have one of these on the way: http://www.spykeeworld.com/ It was on Woot! earlier this week for less than half price, and I couldn't resist. I'm looking forward to hacking it, as there seems to be a pretty good community of people doing so. It has a 200Mhz ARM9-based Marvell System-on-Chip, Wifi, a camera, audio in and out, and is programmable in C# or C++. Ideas that I've had so far include: - Adding a USB Missile Launcher (http://www.thinkgeek.com/geektoys/warfare/8a0f/) - Articulation of the "head" which is where the camera is. Out of the box, you can only position it by hand. - An articulated arm with a gripper, which I may build out of LEGO using their new linear actuators, or I might just go for something from Lynxmotion or similar. The idea of remotely controlling this thing and actually being to physically interact with objects really appeals to me. - Expansion of the battery capacity/possibly conversion to lithium polymer batteries; the stock battery life is rated at only an hour, although supposedly it can automatically find its charging station when it gets low. - Give it a face: http://www.nkcelectronics.com/color-lcd-128x128-nokia-kn128128.html  Anyone else have one of these? I'll post pictures of the guts as soon as I get it!
|
|
#
?
Aug 14, 2009 15:11
|
|
|
That's pretty awesome. Other than the things you mentioned, what kind of I/O does that SOC have?
|
|
#
?
Aug 19, 2009 03:32
|
|
|
It appears to be a USB host, although I'm not quite sure of its capabilities in this area. The "head" which contains the camera and a motion sensor (and, I think, the microphone) connects to the motherboard via a mini-USB connector, but that doesn't necessarily mean it's a USB connection. Looking at the data sheet for the processor, it has some GPIO, serial, and JTAG. Other hackers have explored using the JTAG interface to flash the firmware. I'm not sure which ports are in use and which aren't, but I'm hoping I'll be able to get some insight into that by examining the motherboard. It just got delivered yesterday and I haven't cracked it open yet.
|
|
#
?
Aug 19, 2009 13:13
|
|
|
Has anyone here used USART on an atmega16? I'm trying to write an integer to the UDR, but no matter what I do the TxD pin is high constantly (I'm just looking at it with an oscilloscope). I know I'm entering the interrupt that writes to UDR, since I'm writing the same integer to another port, where it shows up just fine. Do I need to have some kind of external load on the pins for USART to work properly? Currently the only thing I have hooked up to the pins is my scope. I'd post my code, but my comp is dead at the moment. Anyway, the USART initialization was copied straight from the atmega datasheet..
|
|
#
?
Aug 21, 2009 11:05
|
|
|
Hey folks, it's been an age since I last flipped through the entirety of the thread, but I checked over the last few pages (as well as the OP) and couldn't find any mention of this, so here goes. I don't know how many of you read Hack a Day, but it's been a go-to blog for me since I started taking a closer look at getting into things like microcontrollers. Recently, they posted a link to a relatively new video blog by an Aussie in the EE business, and it's turned out to be a wonderful time sink for anyone who wants to learn a few things about electronics, and get some perspective on what it's like to be in the industry. The subject of the blog runs the gamut from product reviews and rants, to industry stories and job interview tips (most of which are spot-on, though a couple I'd take with a grain of salt). Anyway, it's an entertaining and informative presentation for anyone interested in the subject (which should be everyone in this thread). In an early episode, he calls out a poor power design decision on an HP calculator, which actually elicited feedback from one of the calculator's design engineers; he mentions this in a later video. The blog is called the EEVblog, hosted by David Jones: http://www.alternatezone.com/eevblog/ Perhaps this can get added to the OP sometime?
|
|
#
?
Aug 22, 2009 22:13
|
|
|
Astrolite posted:Has anyone here used USART on an atmega16? I'm trying to write an integer to the UDR, but no matter what I do the TxD pin is high constantly (I'm just looking at it with an oscilloscope). I know I'm entering the interrupt that writes to UDR, since I'm writing the same integer to another port, where it shows up just fine. make sure the transmitter is activated and the baud rate is set. The sample code for initializing and using the USART in the datasheet should work fine. The pin doesn't need any pullup or pulldown resistors.
|
|
#
?
Aug 22, 2009 23:24
|
|
|
I may be teaching a one night a week program geared at middle to high school age kids (I can suggest an appropriate age group) interested in electronics, robotics, etc. I pretty much have free reign to come up with the course material, and I'd like to do something micro controller based, probably a PIC. I'm still roughing out exactly what I'd like to include, but here's a brief list: • Inexpensive hardware that the kids can take home to work on between classes and keep after the program ends. • Simple enough that this age group can get right into development without having to endure too much of a lecture beyond the basics. • The option for them to be able to expand on the hardware with ideas of their own. • Teaching soldering would be great, maybe tie that in with the previous point. • Hardware that is good enough to do "cool" projects. PIC makes an entry level board that has a small prototyping area, 4 LEDs, a potentiometer (for A2D), and a push button. It also has a header for additional IO. The whole thing is based on a PIC 16F690. It meets just about all the requirements, but I'm not sure how well it will hold interest. I've also considered getting one of the larger PIC development systems to keep in the class for people who want to do more. I'm open to any other suggestions as well, so if any of you know of a good fit, please share.
|
|
#
?
Aug 26, 2009 21:21
|
|
|
Does anyone know of something like this that isn't such a ridiculous price? The important features are that it is in a DIP (SIP is cool too) package, runs at 3.3 volts and has a UART interface.Hillridge posted:I'm open to any other suggestions as well, so if any of you know of a good fit, please share. Not suggestions, but a couple of questions. What hardware do you plan on getting to program the PICs with? Do you intend to teach the students assembly or are you planning on using a higher-level language?
|
|
#
?
Aug 28, 2009 07:56
|
|
|
BattleMaster posted:Does anyone know of something like this that isn't such a ridiculous price? The important features are that it is in a DIP (SIP is cool too) package, runs at 3.3 volts and has a UART interface. I did a project using a module like this last year. The long and short of it is: No, there's no way to get a quality bluetooth module that interfaces via UART without shelling out some bucks. Even if you price out the discrete components, you're not saving a *lot* of money. The exception: Sometimes Chinese knock-off modules like this show up on ebay. I can't attest to their quality; We eventually broke down and purchased the sparkfun module. Another thing to note: These things are really particular about their power supply. The recommended power the datasheet mentions isn't just a recommendation-- any ripple over spec, no matter how tiny, and you'll start losing connections and other odd behavior.
|
|
#
?
Aug 28, 2009 12:20
|
|
|
BattleMaster posted:Does anyone know of something like this that isn't such a ridiculous price? The important features are that it is in a DIP (SIP is cool too) package, runs at 3.3 volts and has a UART interface.
|
|
#
?
Aug 29, 2009 00:13
|
|
|
So I'm continuing my exploration into the high voltage/high power side of things, and I'm planning at making a run at a DC supply for a tesla coil we'll be making. Needs to go from a 120V line to 300V/150A max out. Also its output voltage needs to be controllable via serial interface, and it has to have good transient response. Ultimately we want to use the supply to modulate audio into the coil (whereas most singing coils use pulse density modulation on their coil drivers to make audio). So I'm planning on going AC line to bridge rectified 300VDC to three-phase buck supply out. But switching supplies at high voltage are far more complicated than low voltage ones. The buck transistors have to be N-channel, meaning each gate driver needs an isolated supply voltage of its own. Fast transient response will require high switching frequency, so we'll have to use FETs instead of IGBTs. I want to implement active current limiting on each phase (it will certainly die without it), and I'll be doing voltage feedback and PWM control with an AVR. There are a lot of fun challenges in this; I just hope I don't fail any classes due to being distracted.
|
|
#
?
Aug 29, 2009 00:24
|
|
|
ANIME AKBAR posted:but that's not a bad price at all, really. and having a UART interface is pretty unique; bizarre, actually. I doubt you'd be able to find anything else like it. There are loads of Bluetooth modules with UART interfaces, they are just much more expensive than ones with USB interfaces for some reason. And it's a bad price because it would cost several times as much as the rest of my project combined. Edit: That Sparkfun one is still the cheapest I've seen so I think it's just a case of people trying to gouge hobbyists who don't know better. BattleMaster fucked around with this message at 01:03 on Aug 29, 2009 |
|
#
?
Aug 29, 2009 00:58
|
|
|
That is a ripoff. National operates a pretty good sampling program. Order some samples of the LMX9838, it's an excellent chip. You can operate it in a transparent serial mode without ever sending it a command string (unless you want to change the default bluetooth password), and you can set the UART rate with hardware pullups. Tricky package though.
|
|
#
?
Aug 30, 2009 01:33
|
|
|
ANIME AKBAR posted:So I'm continuing my exploration into the high voltage/high power side of things, and I'm planning at making a run at a DC supply for a tesla coil we'll be making. Needs to go from a 120V line to 300V/150A max out. Also its output voltage needs to be controllable via serial interface, and it has to have good transient response. Ultimately we want to use the supply to modulate audio into the coil (whereas most singing coils use pulse density modulation on their coil drivers to make audio). Can you tell me any more about the proposed setup? You're planning to achieve the tesla coil voltage control through a variable DC supply? Any reason you can't do this with the primary driver circuit control?
|
|
#
?
Aug 30, 2009 01:38
|
|
|
catbread.jpg posted:Can you tell me any more about the proposed setup?
|
|
#
?
Aug 30, 2009 02:03
|
|
|
Very interesting. I wonder about the bandwidth requirements of the supply compared to the tesla coil resonant frequency and the number of cycles required to energise the tank to the point where it sparks. I don't see why a variable driver duty cycle is necessarily inefficient, of course it would be if you were using a rigid resonant switching technique (and would make no difference if you were already hard-switching), you'd just need a more intelligent switching technique. My tesla coil knowledge is limited, what is relation between input voltage and the tank behaviour? DRSSTCs are generally switched in pulses, are they set up to provide one spark per pulse?
|
|
#
?
Aug 30, 2009 02:25
|
|
|
catbread.jpg posted:Very interesting. I wonder about the bandwidth requirements of the supply compared to the tesla coil resonant frequency and the number of cycles required to energise the tank to the point where it sparks. quote:I don't see why a variable driver duty cycle is necessarily inefficient quote:My tesla coil knowledge is limited, what is relation between input voltage and the tank behaviour? DRSSTCs are generally switched in pulses, are they set up to provide one spark per pulse? Here's a couple pages for comparison: Singing coils using pulse density modulation, with videos Singing tesla coil using duty cycle modulation, short video at the bottom I want the power and spark length of the former with the audio quality of the latter.
|
|
#
?
Aug 30, 2009 03:06
|
|
|
catbread.jpg posted:That is a ripoff. Thanks for the suggestion, it looks just about perfect. I wanted a UART interface for simplicity but I think I could also put the audio codec interface to work. I am playing around with a wireless speaker idea a friend had and I am basing it around a dsPIC33 microcontroller which has a compatible interface in addition to a UART. I need to figure out how to deal with that wacky-rear end package before I get one, though.
|
|
#
?
Aug 30, 2009 05:08
|
|
|
BattleMaster posted:Not suggestions, but a couple of questions. What hardware do you plan on getting to program the PICs with? Do you intend to teach the students assembly or are you planning on using a higher-level language? I was looking at this group of Starter Kits: http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en023805 They all come with USB programmers, and the prices are 25% lower than marked due to the academic discount. Now that I've looked again, I think the PIC18F4xK20 Starter Kit might not be such a bad idea. http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1406&dDocName=en535806 With the discount it's only $75, which I think is a reasonable extra cost considering the kid will get to keep it. It also isn't as restricted as the PICkit 2 Starter Kit ($37.50). I'm going to do a brief overview of assembly, and have them run through some of the simple lessons PIC includes. Since they would be able to take these home, they'll have plenty of time to experiment. After that I'm probably going to use whatever C compiler comes with the kit. Once they have the basics, I'd love to have the class evolve into the kids coming up with their own project ideas and then figuring out how to implement them.
|
|
#
?
Aug 30, 2009 14:57
|
|
|
Well if you're worried about keeping the student's attention, then it looks like that PIC18F4XK20 Starter Kit is the way to go because it has quite an impressive list of features. Without any extra hardware you could do a lot of interesting projects like alarm clocks or digital thermometers. Also for what it's worth PIC18s are easier to program with assembly than PIC16s because you don't have to change memory banks several times every time you want to use a peripheral.
|
|
#
?
Aug 30, 2009 18:15
|
|
|
If you're not already dedicated to PICs, I would definitely give the AVR butterfly a look. A ready-made platform for all kinds of project ideas. It has a light sensor, temp sensor, voltage sampler, LCD screen, joystick, RS-232, and FLASH memory all built in, and gives you access to two of the MCU's I/O ports. They cost $20 each on digikey. Last fall I was poised to do a seminar on microcontrollers (until the president of the student IEEE group found out they didn't have the funding to pay for it), and I was planning on using these guys. For what you get they're pretty incredible. I'm pretty sure Atmel intentionally sells them incredibly cheap in order to undercut competition and get people hooked on their stuff early.
|
|
#
?
Aug 30, 2009 19:54
|
|
|
Yeah, the butterfly is pretty cool. Thanks for those links, they make things a bit clearer. I was having a very interesting discussion with somebody about applying some certain high-power RF amplification techniques to tesla coil design, if I stick around university for postgrad next year I will definitely be looking into that. As far as the LMX9838 package goes, it's not too bad, the internal pads are all N/C. If you mess up the reflow you can re-tin the pads.
|
|
#
?
Aug 31, 2009 01:06
|
|
|
|
| # ? May 22, 2024 16:53 |
|
|
While I was designing a gate driver board today I may have found another, much easier way to modulate the output power. Typically the controller will use feedback from a current transformer on your primary to control the switching of your drivers (so that you always switch at zero current). It's also typical to use the current signal to put active current limiting on the controller, so that when a certain current threshold is reached it will shut off the drivers until the next pulse begins. I'm thinking that if one could dynamically control the current limit, one might be able to get a good linear control of the output. I've yet to simulate it though.
|
|
#
?
Aug 31, 2009 02:40
|
|