|
movax posted:Interesting project! I'd imagine the ESR of the caps would be of interest to you, perhaps followed by the ESL (smaller as package shrinks obviously). The ESR and ESL are the series resistance and series inductance with the rated capacitance. What you need is the parallel resistance, which is not usually spec'd (unfortunately). You'd have to do some research into the dielectric material of your specific capacitors to know what the parallel resistance is, or you could try to measure it. A simple way would be to charge the capacitor to some voltage, wait, measure the voltage, and compute the RC time constant. I imagine it's likely in the 10's of gigaohms for most ceramics. You could ask the manufacture. As to the pull ups and pull downs, that will depend on the leakage current of your input pins. The leakage current is the maximum amount of current that will flow into a "high-impedance" input. Your pull resistor should be sized such that it can still provide a valid logic level if the maximum leakage current is flowing. Now, if you have a good enough DMM, you could try to measure your specific part's leakage and adjust accordingly - just don't expect it to be the same across different dies and at different temperatures. For really low power applications, some people will specifically screen parts for very low leakage.
|
#
?
Feb 9, 2012 06:33
|
|
|
|
| # ? Jun 3, 2024 15:02 |
|
|
Quick, dumb question that might get answered in this thread: I'm looking to make a built-in GPS in my car. I've secured the factory flip-up monitor, and the plan is to gut that, and insert something else. Problem is, the factory screen is 6.5" widescreen format, and none of the big GPS brands make anything above 5.0". What are the odds of finding a new screen with the same resolution (and touch screen to boot), but just bigger, and retrofitting that? Is there a connection standard, or is it all proprietary?
|
|
#
?
Feb 9, 2012 15:17
|
|
|
bolind posted:What are the odds of finding a new screen with the same resolution (and touch screen to boot), but just bigger, and retrofitting that? Is there a connection standard, or is it all proprietary? If I'm understanding you correctly, you're asking if you can swap an LCD screen into a GPS? If so, that's probably not (easily) feasible. While many GPS units have some kind of Linux onboard, the LCDs don't use a standard display interface. Instead, they generally use some sort of parallel interface directly to the LCD (which probably has it's own controller, too). A more convoluted solution would be to see if any work has been done with custom firmware for certain GPS units. Alternatively, consider looking for some cheap chinese phone/tablet/pda running Android, with GPS. But that's a stretch.
|
|
#
?
Feb 9, 2012 15:51
|
|
|
Thanks for the ideas everyone!movax posted:Interesting project! I'd imagine the ESR of the caps would be of interest to you, perhaps followed by the ESL (smaller as package shrinks obviously). Seems like ESR wouldn't be of much help because I'm not doing high-frequency work. I've usually seen ESR come into play when you're either 1) dumping/drawing a ton of current very quickly and trying to figure out ESR losses or 2) building a very touchy analog filter and the ESR effect comes into the calculations. I could be mistaken! It's amazing how many little details get glossed over in class. SnoPuppy posted:The ESR and ESL are the series resistance and series inductance with the rated capacitance. What you need is the parallel resistance, which is not usually spec'd (unfortunately).  Assuming that the drain is reasonably fast so I can measure it.I will be using passives I have laying around for this project so being able to characterize them is good enough; as long as I can identify WHY I'm getting more current drain than ideal, I'll be okay. Shrugging and going "I don't know" isn't acceptable to me. This is for a GPS tracker; I am hoping to get a long period of inactivity until it's moved, and then have it come to life and start tracking. At this point, none of these uAs here or there matter a damned bit, because the radios are going to be drawing significantly more power. The professor was thinking we could just plug it into USB and have it idle that way, but I think this would be a slicker solution if I can get the battery life reasonable. Being that it is above and beyond expectations, a well quantified near-success would be an alright result. However, a well-qualified success is even better! SnoPuppy posted:As to the pull ups and pull downs, that will depend on the leakage current of your input pins. The leakage current is the maximum amount of current that will flow into a "high-impedance" input. Your pull resistor should be sized such that it can still provide a valid logic level if the maximum leakage current is flowing. I've been reading some promising stuff about profiling battery life, and there is a pretty strong correlation about realized battery life and current draw. By drawing tiny amounts of power, I should get more than the sticker value out of the battery (see "Battery capacity measurement and analysis using lithium coin cell battery" by Sung Park et al). I'm really curious to see if I can actualize this effect to get a year or maybe even two of shelf life.
|
|
#
?
Feb 9, 2012 22:16
|
|
|
Delta-Wye posted:I've been reading some promising stuff about profiling battery life, and there is a pretty strong correlation about realized battery life and current draw. By drawing tiny amounts of power, I should get more than the sticker value out of the battery (see "Battery capacity measurement and analysis using lithium coin cell battery" by Sung Park et al). I'm really curious to see if I can actualize this effect to get a year or maybe even two of shelf life. It really depends on the individual chemistry and internal structure of the particular battery you're using. I mean, they make batteries specifically for extremely long lifetimes for these exact sorts of situations, with hilariously small self-discharge rates. But batteries not designed with that in mind might be problematic, and there'll probably be a lot of variation between cells from different manufacturers.
|
|
#
?
Feb 9, 2012 22:23
|
|
|
Anyone used the Murata Okami line of converters? They're cheap and would actually be cheaper for my project than the SC4525C based converters (since I have to buy all the external components for that, whereas here I only need caps).
|
|
#
?
Feb 10, 2012 05:54
|
|
|
I haven't used the Okami line, but I have quite a bit of trust in the Murata name from using their brick style converters. I would assume it does what they say it does.
|
|
#
?
Feb 10, 2012 06:02
|
|
|
DC-DC converter modules are getting pretty insane these days. I recently bought a few of these guys and they're just awesome. I just wish that they made them with higher max input voltages. For that I'm still sticking with modules like this. Murata stuff is probably fine, but I wished they gave example plots of EMI. What I'd really like to see are modules that give negative outputs, but there aren't any in anywhere close to that size. Unless they have split tracking outputs, which I don't want. I'm still stuck using charge pumps for negative voltage generation.
|
|
#
?
Feb 10, 2012 13:51
|
|
|
What does split tracking mean? I'm having a hard time googling that.
|
|
#
?
Feb 10, 2012 16:10
|
|
|
It means there are two equal outputs of opposite polarity.
|
|
#
?
Feb 11, 2012 00:54
|
|
|
Tracking usually means both poles are controlled by a single reference. This can be bad when you have unequal loads.
|
|
#
?
Feb 11, 2012 02:18
|
|
|
I'm finally jumping back into electronics, since it's something I've always been fascinated by and was one of my favorite classes. I just bought an Arduino and I'm having fun with it. Although it was weird to see them just sitting there in Radio Shack when they've been shrinking that section more and more. When I do get around to soldering, is this still great if I can find one on eBay? Otherwise my budget is around $100, since I realized the reason I suck at soldering is probably from buying $10 pen style ones. EDIT: Wow I asked about things a year ago too...I'm actually throwing myself into this time though.
|
|
#
?
Feb 11, 2012 14:58
|
|
|
We have a dual setup similar to this one and it hasn't had any problems. 80W instead of 60W & it has digital temp. e: also get some flux (flux pens are great) and some tip tinner. taqueso fucked around with this message at 18:33 on Feb 11, 2012 |
|
#
?
Feb 11, 2012 16:08
|
|
|
Fooley posted:Otherwise my budget is around $100, since I realized the reason I suck at soldering is probably from buying $10 pen style ones. I wouldn't say that; I do all my soldering with a 15W Radio Shack iron. Smaller SMD stuff is kind of annoying but doable. Get some flux and practice some more.
|
|
#
?
Feb 11, 2012 18:44
|
|
|
Fooley posted:I'm finally jumping back into electronics, since it's something I've always been fascinated by and was one of my favorite classes. I just bought an Arduino and I'm having fun with it. Although it was weird to see them just sitting there in Radio Shack when they've been shrinking that section more and more. Tips for soldering: 1. Always have a wet pad of some sort available. Clean the tip of the soldering iron OFTEN. (every joint ideally..) 2. Flux. Use it. To much just means you need to use some alcohol to clean up. to little means dry, nasty, joints. I have a tub of paste flux I usually dip wires and components in. 3. Heat the pad/wire, not the solder. 4. Wait for the solder to flow into the joint before removing the iron. 5. Don't move the wires until the joint is cool. That should cover it. :-) I"m glad you're having fun with the arduino. They're awesome digital glue.
|
|
#
?
Feb 11, 2012 19:42
|
|
|
Make sure you get a soldering iron that has many different tips available for purchase. Those hobby irons usually only have a few tips available. This is the station I use: Weller WESD51.
|
|
#
?
Feb 12, 2012 00:03
|
|
|
The WESD51 is fantastic, possibly my favorite soldering station. Hakko also makes excellent stations. For someone getting started on a budget, I'd recommend Aoyue. I have a few gripes with mine (stand isn't that great, display is Celsius only) but it solders amazingly well for $20 more than a radio shack special. DethMarine21 posted:I wouldn't say that; I do all my soldering with a 15W Radio Shack iron. Smaller SMD stuff is kind of annoying but doable. Get some flux and practice some more. The problem with cheap irons isn't that they can't do a good job, it's that they wear out very quickly (especially cheap tips) and they don't have the muscle you need when it comes time to solder something with a lot of thermal mass. sixide fucked around with this message at 03:29 on Feb 12, 2012 |
|
#
?
Feb 12, 2012 03:25
|
|
|
Okay I have an embedded system question question: For a project I'm working on, I need to log a large amount of data very fast. Basically I have to do a quick series of reads from an external ADC (anywhere between 16 and 200 bytes of data) within between 10us-60us. I then have the same amount of time to store that data somewhere, before I need to do another burst of samples. Eventually I want to be able to do several thousand of these read-store operations continuously without stopping. So far I've been using the internal SRAM of my MCU (AT32UC3B0512), but it doesn't have enough for longer sequences (I'm using 80KB of its available 96KB. I think I'll need as much as 256KB). So I'm looking for some kind of external media I can write the data to. My MCU doesn't have a built in external memory interface controller, so I'm stuck with things that interface via SPI. The closest thing I've found is actually using an SD card, but it only allows write operations in blocks of 512 bytes, and I definitely don't have time to write that much (my max SPI speed is 30MHz). Is there anything out there that can meet my needs? I'd rather not have to switch to another MCU with a parallel memory interface controller, but that's the only option I can see at this point.
|
|
#
?
Feb 12, 2012 03:40
|
|
|
ANIME AKBAR posted:Okay I have an embedded system question question: We were able to continuously store audio onto an SD card with at atmega (or possible xmega), along with several other sensors. What kind of sampling rate are you looking at? I don't see why, with ADC interrupts that are correctly written, you shouldn't be able to continuously store the data in between samples...
|
|
#
?
Feb 12, 2012 04:08
|
|
|
Slanderer posted:We were able to continuously store audio onto an SD card with at atmega (or possible xmega), along with several other sensors. What kind of sampling rate are you looking at? I don't see why, with ADC interrupts that are correctly written, you shouldn't be able to continuously store the data in between samples... I'm already sampling at the absolute fastest speed I could possibly get (1.6us for a pair of simultaneous 14 bit samples) without losing consistency in the timing of the samples. My MCU only has one SPI port, so I can't use it to store the data externally until all the sampling is over with (and if I were able to sample in less time, then I would want to use that time to take more samples, not store the samples). Even if I did have another SPI port it would be tricky because I would have to use my peripheral DMA for both the read and write operations, and they might conflict.
|
|
#
?
Feb 12, 2012 05:01
|
|
|
ANIME AKBAR posted:I'm already sampling at the absolute fastest speed I could possibly get (1.6us for a pair of simultaneous 14 bit samples) without losing consistency in the timing of the samples. My MCU only has one SPI port, so I can't use it to store the data externally until all the sampling is over with (and if I were able to sample in less time, then I would want to use that time to take more samples, not store the samples). Even if I did have another SPI port it would be tricky because I would have to use my peripheral DMA for both the read and write operations, and they might conflict. Only 1 SPI port? Only 1 DMA channel? It may be time to find a different uC.
|
|
#
?
Feb 12, 2012 05:07
|
|
|
ANIME AKBAR posted:I'm already sampling at the absolute fastest speed I could possibly get (1.6us for a pair of simultaneous 14 bit samples) without losing consistency in the timing of the samples. My MCU only has one SPI port, so I can't use it to store the data externally until all the sampling is over with (and if I were able to sample in less time, then I would want to use that time to take more samples, not store the samples). Even if I did have another SPI port it would be tricky because I would have to use my peripheral DMA for both the read and write operations, and they might conflict. Oh shoot, I missed that you were using an external ADC. In that case, not sure how to help. I mean, if you had the ability to directly store the incoming bytes from the ADC with DMA into a a couple of buffers, then you might be in luck, but it looks like the limitations of your processor are currently hindering all possible strategies. I never cared for the AVR32s...
|
|
#
?
Feb 12, 2012 05:07
|
|
|
There are several DMA channels I could use, with configurable priority, but I'm pretty sure only one can access the SRAM at a time. The datasheet is quite vague on exactly how arbitration works, so I'm not really comfortable leaving the possibility of collision, since my application is extremely sensitive to timing errors.Delta-Wye posted:It may be time to find a different uC. Slanderer posted:I never cared for the AVR32s... Also I could switch over to the AT32UC3CXXXX device family, which has two SPI ports, an external memory controller, and a peripheral event system which is also neat. However, I'd rather not unless it's absolutely necessary. And I've never used a true SRAM chip (with a parallel bus and all the flow control that comes with it). It looks somewhat complicated, and takes a ton of I/Os and will complicate my layout quite a bit. And I'll STILL need an SD card to store the large amount of data I'll be accumulating. edit: poo poo I just checked and only the devices in 144-pin packages have the memory controller. That is not loving happening. ANIME AKBAR fucked around with this message at 05:34 on Feb 12, 2012 |
|
#
?
Feb 12, 2012 05:24
|
|
|
Have you thought about using an FPGA? I don't think you can beat the speed+price+memory all in one package.
|
|
#
?
Feb 12, 2012 05:45
|
|
|
Krenzo posted:Have you thought about using an FPGA? I don't think you can beat the speed+price+memory all in one package. I ruled out FPGAs a while back because the role it would fill is actually a relatively small part of the overall project, so I needed something that would be relatively fast to develop and debug. And I've never seen a cheap FPGA with 256KB or more of RAM built in. Also, after I'm done with this project the hardware will probably passed on to some BME grad student to use in future research, and it's likely that they would want to modify the code at some point. And it would be pretty awful for them to have to try and work with an FPGA.
|
|
#
?
Feb 12, 2012 05:52
|
|
|
ANIME AKBAR posted:I ruled out FPGAs a while back because the role it would fill is actually a relatively small part of the overall project, so I needed something that would be relatively fast to develop and debug. And I've never seen a cheap FPGA with 256KB or more of RAM built in. Sounds like the bolded part isn't working out so well. I had an FPGA project way back in school that continuously sampled a very fast ADC. Had storing the data been the only requirement, we would have spent 5 days tops--learning FPGA development in 3 days and finishing the program in 2. Typical FPGA evaluation hardware these days has an obscene amount of DRAM onboard, so buffering a couple hundred kilobytes isn't an issue at all.
|
|
#
?
Feb 12, 2012 06:06
|
|
|
movax posted:Anyone used the Murata Okami line of converters? They're cheap and would actually be cheaper for my project than the SC4525C based converters (since I have to buy all the external components for that, whereas here I only need caps). What I came to ask is how do you folks keep your work table from becoming a mass of alligator clips and test probes? Here's my table about 5 minutes of a debugging session, which is right about the point things are started to get tangled.  The oscilloscope probes are wrapped up in the power alligator clips, there's PCBs being pulled to the floor by wires, the laptop often is losing USB devices as I short them... and this is a pretty small amount of wires to be dealing with compared to what people who do this for a living work with right? So what's the secret? Shorter wires? More desks? Extra helping arms of Vishnu?
|
|
#
?
Feb 12, 2012 06:10
|
|
|
ANIME AKBAR posted:There are several DMA channels I could use, with configurable priority, but I'm pretty sure only one can access the SRAM at a time. The datasheet is quite vague on exactly how arbitration works, so I'm not really comfortable leaving the possibility of collision, since my application is extremely sensitive to timing errors. If you had two SPI ports, you'd be okay. On chips I've used, you can set the DMA to fire instead of an interrupt routine. Set up your ADC, set up your SD card and get it ready to receive data, then point your DMA module from SPI0 to SPI1 and go do something else while it records data as fast as possible. CapnBry posted:What I came to ask is how do you folks keep your work table from becoming a mass of alligator clips and test probes? Only thing I've found that works is to not use it 
|
|
#
?
Feb 12, 2012 06:17
|
|
|
sixide posted:Sounds like the bolded part isn't working out so well. I had an FPGA project way back in school that continuously sampled a very fast ADC. Had storing the data been the only requirement, we would have spent 5 days tops--learning FPGA development in 3 days and finishing the program in 2.
|
|
#
?
Feb 12, 2012 06:17
|
|
|
Delta-Wye posted:If you had two SPI ports, you'd be okay. On chips I've used, you can set the DMA to fire instead of an interrupt routine. Set up your ADC, set up your SD card and get it ready to receive data, then point your DMA module from SPI0 to SPI1 and go do something else while it records data as fast as possible. My DMA can only communicate between peripheral memory and SRAM, not from peripheral to peripheral 
|
|
#
?
Feb 12, 2012 06:20
|
|
|
Fooley posted:When I do get around to soldering, is this still great if I can find one on eBay? Otherwise my budget is around $100, since I realized the reason I suck at soldering is probably from buying $10 pen style ones. For the price compared to a Weller I think you can't beat it.
|
|
#
?
Feb 12, 2012 06:24
|
|
|
ANIME AKBAR posted:I can't find any FPGA under $100 and in a leaded package with at least 1Mbit of RAM. And using a dev kit isn't an option because I have size limitations. The size of the FPGA and its supporting hardware would literally take up half the board area of what I have now. Size requirements (and ridiculously low price requirements) in a research project sucks. Having a ton of massively parallel logic to throw at this would make it a cakewalk. My general experience in both the research and commercial fields indicates this is the typical solution, as well. I've seen several million gates used in a single motor controller (commercial) and something like 4 FPGA eval boards and a PC chained together for use in ultrasound image processing (academic research, naturally). It sounds like you're already aware that hardware limitations are holding you back, so my only advice is to find the right hardware before throwing every software trick available at covering up your hardware's weaknesses. Would using a pair of microcontrollers lessen the pain? It wouldn't help on the memory front, but it may help lessen the processing power limitations.
|
|
#
?
Feb 12, 2012 06:30
|
|
|
I think I'm kind of sending the wrong message. This really isn't a matter of the hardware not being as good as expected. I originally didn't think I'd be able to sample data faster than I'd be able to store it. Up until now I have been sending the data to a PC host instead of storing it locally, and that has a bandwidth of about 6Mbaud max. I never thought I'd be able to collect data faster than that interface could get rid of it, but I was able to do so, so now it's become a "problem." It's not the end of the world if I can't do it. I'll still meet my original specs, but I'll be disappointed that I won't be able to fulfill the hardware's true potential. Also right after finalizing the hardware, my adviser decided to throw in a requirement for it to store large amounts of data without being connected to an external host, which kind of screws things up...
|
|
#
?
Feb 12, 2012 06:58
|
|
|
CapnBry posted:
Wires coiled up with twist ties and power bars mounted on your desk. Also, a scope with more channels so you don't need to use a multimeter at the same time.
|
|
#
?
Feb 12, 2012 07:03
|
|
|
sixide posted:The problem with cheap irons isn't that they can't do a good job, it's that they wear out very quickly (especially cheap tips) and they don't have the muscle you need when it comes time to solder something with a lot of thermal mass. Edit: I seem to be channeling yoda this morning.
|
|
#
?
Feb 12, 2012 12:05
|
|
|
ANIME AKBAR posted:My DMA can only communicate between peripheral memory and SRAM, not from peripheral to peripheral Delta-Wye posted:It may be time to find a different uC.
|
|
#
?
Feb 12, 2012 19:43
|
|
|
Maybe take a look at the dsPIC33F series. Also the Altera DE1/2 I thought were "affordable", for academia anyways.
|
|
#
?
Feb 12, 2012 20:14
|
|
|
Delta-Wye posted:Just out of curiosity, is there anything (other than ARM) that could help me out? Mainly I need at least two UARTs, fast interrupt handling, at least a few 16 bit timers, two SPI ports, at least 50MIPs, at least 5 DMA channels, at least 12 GPIO on top of all the other peripheral I/Os... I haven't seen that on anything but ARM parts, and I didn't want to take a leap that big. movax posted:Maybe take a look at the dsPIC33F series. Also the Altera DE1/2 I thought were "affordable", for academia anyways. ANIME AKBAR fucked around with this message at 20:47 on Feb 12, 2012 |
|
#
?
Feb 12, 2012 20:44
|
|
|
Ah sorry, I was on my phone and couldn't check specs in detail. Maybe the PIC32 is the last step you could take before going ARM.
|
|
#
?
Feb 12, 2012 23:48
|
|
|
|
| # ? Jun 3, 2024 15:02 |
|
|
movax posted:Maybe the PIC32 is the last step you could take before going ARM. In any case I don't think it's worth it for me to make such a big change at this point. I think I'll just go with a slightly different device with two SPI ports (and of course a completely different pinout  ), and use one SPI port just for an SD card and use that for fast storage.
|
|
#
?
Feb 13, 2012 01:50
|
|