|
Pizza Dude posted:Thanks! I'm working from this schematic: You can tie pin 6 of the PT2399 chip to ground through a 10k-30k potentiometer/resistor to remove the photoresistor from the equation and see if it's causing the issue or if it's something else.
|
#
¿
May 2, 2011 19:26
|
|
|
|
| # ¿ May 22, 2024 00:27 |
|
|
Out of curiosity, how many of you are members of hackerspaces? DIY electronics are a big part of most of those places. I'm thinking about starting a thread on them, though I'm not sure which forum would be best for it.
|
|
#
¿
May 12, 2011 14:00
|
|
|
PDP-1 posted:I live in Stupidtown, USA and there aren't enough people around here interested to support anything like that but if you do make a thread please post a link here - I'd enjoy living vicariously through you guys until I can move out of this shithole. Started a thread! http://forums.somethingawful.com/showthread.php?threadid=3412940
|
|
#
¿
May 18, 2011 18:31
|
|
|
Yeah, I design schematics all the time, usually with large pin count parts, and every pin is labeled. Bigger parts also typically have the part name on them, though not necessarily the long name you would use to order them.
|
|
#
¿
Jun 6, 2011 13:03
|
|
|
We actually have a design standard at work where schematics should not be drawn with 4 way intersections because it's too easy to confuse them with one line passing over another on printouts. Here's examples of a part and nets on a typical schematic of mine:
|
|
#
¿
Jun 6, 2011 18:01
|
|
|
http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=626-1116-ND http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail&name=626-1311-ND Those look like what you want, or should get you close enough that you can find one based on the datasheets. "CONN D-SUB 7W2" is the search term you want to use, then click the "D-sub (xxx items)" link and choose 7 for the number of positions. Hillridge fucked around with this message at 14:12 on Jun 22, 2011 |
|
#
¿
Jun 22, 2011 14:05
|
|
|
This is a long shot, but does anyone know what the HDMI spec calls for in regards to unshielded twisted pair? I'll explain. A type A (most common) HDMI cable has 19 pins. Pins 1-12 are 4 sets of shielded twisted pairs with a drain wire in each set. Pins 13-19 consist of single conductors. I've dissected a couple cables, and in one pins 15&16 are a twisted pair, and in another pins 17&18 are a twisted pair. Which is correct according to the standard? It's important because I am using an HDMI cable for non-HDMI purposes and need to have a couple signals on the twisted pair.
|
|
#
¿
Jul 12, 2011 16:19
|
|
|
SnoPuppy posted:I'm 95% certain that there is no requirements on the other signals because they are intended to be low speed control signals (I2C, hot plug, a power line, and a few others I can't remember). That's the weird thing, this is a totally different twisted pair type, used just on these two wires. See Unit B (P5) in the below image. I'm also trying to find a copy of the HDMI 1.4 spec because I want to use an A to D cable, which has a different pinout, and uses 5 shielded twisted pairs. 
|
|
#
¿
Jul 12, 2011 20:59
|
|
|
I got a few ATMega32u2 Micropendous boards dumped on me to mess around with. I have Flip for loading Hex over the usb interface, and I have a copy of AVR Studio 5 installed, now how do I output a hex file from AVR studio 5 to load with Flip?
|
|
#
¿
Jul 19, 2011 20:39
|
|
|
I'm just getting around to messing with the AVR family of parts, particularly an ATMEGA32U2 on a micropendous board. What's the easiest way to develop code for these? I know how to write C, but I'd prefer to not reinvent the wheel, and would love something that has built in features like delay loops and button debouncing, and easy ways to setup I/O (something like PORTB1=OUT rather than writing bits to registers). Would I need to do something like LUFAduino to get that, or is there another option? Hillridge fucked around with this message at 17:16 on Jul 27, 2011 |
|
#
¿
Jul 27, 2011 17:11
|
|
|
I know how to do them, I'm just feeling lazy 
|
|
#
¿
Jul 27, 2011 17:35
|
|
|
yup, just found that I could get a 1 second* delay by doing: #include <util/delay.h> and _delay_ms(1000); *delay is horribly inaccurate and 1 second is actually like 2/3 second
|
|
#
¿
Jul 27, 2011 20:20
|
|
|
As soon as I found the issue, I knew someone would have posted a reply addressing it. F_CPU is indeed set to the default, not what I need it to be. I'm compiling with AVR Studio 5 rather than a makefile, can I redefine F_CPU in my main .c file? I also get this warning: #warning "Compiler optimizations disabled; functions from <util/delay.h> won't work as designed" Where can I enable this? Thanks
|
|
#
¿
Jul 27, 2011 20:51
|
|
|
Dear Altera, Please explain to me why a FPGA programmer is a 112MB installer that then downloads and installs a 750MB program. I'm not talking about your full Quartus II suite (which is over 2GB), just your Stand-Alone Programmer. Please clarify the need for this much bullshit. Sincerely, Hillridge
|
|
#
¿
Aug 5, 2011 15:22
|
|
|
I really need to learn PCB layouts. I have a guy at work whose entire job is to do that, and I'm swamped with schematic design, so I never get a chance.movax posted:You rang? That's my current task at work. Using Altera, Cyclone IV GX specifically, x4 Hard IP, but I've done Xilinx as well with a Spartan-6. I'm working with a Cyclone IV right now having a weird issue I'm hoping you can shed some light on. I program the C4 from a micro using PS mode. When the bitload is finished conf_done is supposed to get released and be pulled high by an external 10k resistor. It does, but immediately afterward, the C4 goes off into lala land and acts like it was never programmed. Here's the weird bit - if I TOUCH the conf_done net while the programming finishes and it goes high, it works just fine. I also got it to work by soldering an 18pf cap from conf_done to ground. What's up with this output? Hillridge fucked around with this message at 23:46 on Aug 14, 2011 |
|
#
¿
Aug 14, 2011 14:57
|
|
|
movax posted:Slanderer: any luck with your Cyclone IV? Assignments->Device->Device & Pin Options->Dual-Purpose Pins perhaps. If you hook a scope up to CONF_DONE, what does it look like after configuration is done? That was me actually. We have an outside consultant doing the actual FPGA dev, so I'm waiting to hear back from him on how the pin is setup. I can't learn anything from scoping it because applying a scope fixes the problem. Hillridge fucked around with this message at 16:48 on Aug 17, 2011 |
|
#
¿
Aug 17, 2011 16:40
|
|
|
movax posted:Adding capacitance fixing the problem makes me think it's a hardware issue with the nets connected to CONF_DONE (you probably can't share the schematic). Then again, maybe that pin is being set to something else post-programming as well. It's a pretty simple net, and something I've used before without issue. The conf_done net goes between the micro and the FPGA with a 10k PU to 3.3V. it also goes to the ~OE line on a couple buffer chips to disable them once the FPGA is programmed. I have a 0 ohm on this line, so I'll try disconnecting it and seeing if the issue goes away. I'm also going to double check that this line requires a 3.3V PU and not a 2.5V.
|
|
#
¿
Aug 18, 2011 16:46
|
|
|
kwantam posted:Also try adding 1 to 10 pF to ground on the line, since that's what your probe is anyhow. This isn't a solution, but it'll give you some evidence as to the source of the problem. Oh it's definitely something to do with capacitance. I've already "fixed" the problem with an 18pf cap to ground. A well placed finger also works. I also managed to make it work without a cap (or finger) by removing the ~OE inputs from the net. Basically the conf_done line was connected to an input on my micro and to the ~OE pins on two sn74lvc1g125 buffer chips. Removing the buffers also fixes it. Next step is to try a harder pullup. Edit: 5k pullup makes it worse! Digging up the number for my FPGA rep. Hillridge fucked around with this message at 18:31 on Aug 19, 2011 |
|
#
¿
Aug 19, 2011 16:47
|
|
|
I finally figured out my FPGA issue. It was, as I suspected, something dumb I did. After conf_done goes high to signal that the device received its programming, the FPGA requires 2 more clock pulses before it initializes. By controlling my buffers with conf_done, I was shutting off the clock line before these last 2 pulses went in. The slight amount of capacitance delayed the buffer shut off just long enough to make it work. Unparagoned posted:I want to move into SMD PCBs and was wondering how important is soldermask. Would it be possible to solder SMD stuff as a novice on a naked board without solder mask? I build SMD PCBs all the time and I've never used a soldermask. If you designed a product and plan to run big batches of them it might be worth it, but not for prototypes.
|
|
#
¿
Aug 29, 2011 16:33
|
|
|
What's the hobbyist PCB fab of choice these days? I'm primarily interested in low cost options. It's 2 sided, and I don't need any frills, but would appreciate a reasonable turn around time.
|
|
#
¿
Aug 31, 2011 21:09
|
|
|
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
|
|
|
Maybe you guys can help me with something I've been trying to puzzle out the last couple days. I want to generate a high (2100-2600V) DC voltage with very low current needs. Rather than feed a transformer with a regular AC signal, I want to monitor the output and give the transformer a 5V pulse whenever it dips below the set value (somewhere in that 2100-2600VDC range). To illustrate how low the current draw is, I only need 4-6 pulses a second to maintain my voltage. Currently this is setup with a custom coil style transformer that takes a 10V pulse and steps it up to about 1150V, where it then passes through a Greinacher voltage doubler to end up around 2300VDC. I want to pulse it with 5V and get away from the custom transformer because the guy who runs the small company that makes it is an rear end in a top hat. I don't know the exact specs on the current transformer. The problem I am having is finding any off the shelf coils designed for this sort of thing. I don't know if it's weak google-fu or a really niche use or what. So 3 questions: 1. What should I be looking for? Pulse transformer? High voltage coil? 2. Since I'm just blipping the transformer with pulses until it fills up a cap to 2300VDC, is there any reason why I can't use a circuit that if run with a steady 5VAC signal would produce a higher than desired max voltage, and then stop pulsing before I get that high? 3. Do you think a CCFL transformer would work in this scenario? They are cheap, common, and designed to output in the 1kV+ range, but on the other hand they are designed for steady 40kHz signals, not pulses.
|
|
#
¿
Feb 3, 2014 22:40
|
|
|
Related: The best knock-off I ever heard of was a circuit breaker that when opened up was revealed to just be a thick wire shunting line to load. 
|
|
#
¿
Feb 3, 2014 22:43
|
|
|
Otto Skorzeny posted:Got a set of pilot run boards from our board-stuffing contractor today. A bunch of the (through-hole) fuses were open because of failed solder joints. Turns out they trimmed the leads before soldering rather than after. How does a company that's been in business for twenty years gently caress that up? Trimming leads before soldering is pretty standard, as clipping them afterward can break solder joints and wave soldering doesn't work so well with big leads hanging off. They still should have caught the problem in inspection though.
|
|
#
¿
Feb 12, 2014 19:38
|
|
|
I'm in need of a 5VDC to ~48VDC boost converter to create an very low current bias voltage (we're talking uA here). I need it to be as noise free as possible too. Has anyone done something similar or have a recommended driver IC/circuit for this?
|
|
#
¿
Mar 26, 2014 21:24
|
|
|
movax posted:Buy a module and follow everything the app note(s)/datasheet suggest. Linear Tech or TI are good choices. In the case of uA level currents, you can probably find an IC that only needs an external inductor and possibly a diode. I've been reading this appnote: http://cds.linear.com/docs/en/application-note/an92f.pdf page 5 has a circuit that does pretty much what I want, though the performance is listed at 500uA. It's a pretty good read since it's very similar to what I'm doing (replace APD with PIPs alpha detector).
|
|
#
¿
Mar 26, 2014 22:17
|
|
|
Can anyone recommend a TFT driver IC? I'm looking for something that can handle the parallel A/D lines for a 3.5" or 4.3" LCD display and interface to a micro with something simple like SPI or I2C. I've seen mention of a RA8875 chip, but that's it so far. The last time I had to do this we used an FPGA, but that was a bit of a clusterfuck. I'd like off the shelf from a reliable manufacturer if I can get it. FTDI's FT800 looks pretty promising too. Hillridge fucked around with this message at 22:42 on Jun 19, 2014 |
|
#
¿
Jun 19, 2014 22:35
|
|
|
I want to run a 6V pump off of a system that can be powered by either a 12VDC supply (but needs to handle 15V+ in case it's on a poorly regulated one/ running car) or a 2 cell li-ion pack (8.4V max down to close to 6V at full discharge). The pump draws well under an amp while running. Anyone know of a good buck/boost converter that will meet these needs (or a better solution)? Hillridge fucked around with this message at 13:35 on Aug 11, 2014 |
|
#
¿
Aug 11, 2014 13:30
|
|
|
movax posted:You could find an eval board for a buck-boost switcher IC from Linear Tech or similar, but I bet there's an eBay special that can do the job as well (search for buck-boost). If you're willing to forgo running at the low end, you could probably live with just a buck. I've changed tactics a bit and am looking at using a pretty beefy 5V supply (bucked off the battery) as the source to boost to 6V. I need a 5V rail anyway, and lots more parts are dedicated to just boosting than buck/boost. I'm still trying to find a good boost chip though. Designing using a motor as a load is a pain in the rear end since the Iout is either 0, a steady 100mA, or a start up surge.
|
|
#
¿
Aug 11, 2014 17:45
|
|
|
I'm working out the details on this one right now: http://www.analog.com/static/imported-files/data_sheets/ADP1612_1613.pdf The compensation network calcs are a bit confusing, but I think doing them based on the typical steady state current is probably the best bet. It has soft start, which will help. I planned on an output cap as well, though I don't think I can go too high or it will affect stability.
|
|
#
¿
Aug 11, 2014 18:21
|
|
|
movax posted:That's a current-mode boost converter, so it should actually be simpler for you to control and compensate, which is nice. I mostly use LTC parts myself, not super familiar with the AD offerings, do they have any calculation spreadsheets or similar that can help you out? Looks like a very simple Type I compensation network there, but at your power level you should be fine. Boost converters have an inherent right-half plane zero which can make maintaining sufficient phase margin difficult, so the compensation network boosts ( It's not too bad, but after a decade of mostly digital embedded systems work, things like right-half plane zeroes, poles, phase margin, and other transfer function jargon are a bit more fuzzy than they were when the ink was still wet on the diploma. 
|
|
#
¿
Aug 11, 2014 19:34
|
|
) gain around that frequency.
|
|
| # ¿ May 22, 2024 00:27 |
|
|
movax posted:Hah - definitely understandable, but it just makes you a more dangerous (in a good way) design engineer if you can jump over into the analog sandbox, play around for a bit and then hop right back over to digital Thanks, I may take you up on that offer! For layout, we had dedicated PCB guys at my last job (giant company). At my current job (teeny tiny company), I get to do it all! The biggest downside is that lack of experience makes everything take 10x as long, and I second guess everything. Also, there's no one else to blame layout mistakes on . The upside is that I'm getting experience, and stupid mistakes that used to happen due to communication lapses should (hopefully) be less. Stuff like when the old PCB guy (started when they still taped boards) would give power nets fat pipes - until they passed through a ferrite or jumper and got a new net name, then they became 5mil traces. For the amount of boards I make at this job (few), and the complexity (probably no more than 4 layer), Eagle has been working great. Schematic and PCB are constantly linked, so schematic changes are instantly reflected on the PCB. Parts are relatively straightforward to build, and tons are available pre-made (you do need to check them though).
|
|
#
¿
Aug 12, 2014 17:29
|
|
Always good to understand power more than just a block of 3V3 power comes in for you to use at all temperatures, loads, etc! I would recommend finding a similar part from LTC and check out their layout guide as well, within reason. Integrated switch makes it easier, but you still have a few hot paths that can radiate like crazy. Feel free to post what you end up with here if you want a second set of eyes.