This might be a weird question but wouldn't it be technically feasible to setup a microcontroller/PI or other low-cost computing hardware with GPIOs to communicate with a PCIe device through it's header? A video card for example.
I've got an old video card and I'm curious how hard it would be to get in into a running state in a stand-alone setup. Not trivial of course. But I think it would make an interesting project.
... Really fucking hard.
You realize you'd have to reverse engineer the GPU drivers enough, to figure out how to init the card, and then use it past that, right?
Maybe if you used a Raspberry Pi or something, that runs Linux, and wrote some sort of layer to replace all the PCIe stuff, with GPIO driving instead, but... there's shitloads of timing concerns, I'm not sure how you'd do all timings correctly etc. with that setup.
[QUOTE=Aathma;49976152]This might be a weird question but wouldn't it be technically feasible to setup a microcontroller/PI or other low-cost computing hardware with GPIOs to communicate with a PCIe device through it's header? A video card for example.
I've got an old video card and I'm curious how hard it would be to get in into a running state in a stand-alone setup. Not trivial of course. But I think it would make an interesting project.[/QUOTE]
You'll never really manage it with GPIO, at the very least you'd need a PCIe interface chip or a processor with it built in, PCI on the other hand could be done with a fast processor but it isn't the sort of thing you'd usually do.
As to interfacing with a GPU you're going to have your work cut out, it'd be easier and quicker to make your own GPU (low resoulution VGA output is actually really simple).
it would be easier to just write your own software renderer
and doesn't raspberry PI has it's own GPU?
The intention was never really to use it as a practical solution for graphics. I'm just curious if you get get it running in some sort of low level state.
Some neat info here: [url]http://www.hardwaresecrets.com/everything-you-need-to-know-about-the-pci-express/[/url]
You could work backwards from the free driver stack on Linux (in-kernel code and drivers shipping in Mesa).
The question is, how many years do you have?
Plugged in a battery charger today and it started making fizzing sounds, after prying the case open and a visual inspection I found this diode cracked in half with burn marks and measuring it showed low resistance in both directions so that's obviously dead
[t]http://i.imgur.com/KE0aZhj.jpg[/t]
After some inspection with a magnifying glass it's a 1N5819 schottky diode, how big is the chance that this part is the only thing failed? I've never really repaired anything before and usually I'll see people talking about capacitors failing but the 3 that are on here all seem fine, this diode is the only part that appears to be damaged with the naked eye
[editline]21st March 2016[/editline]
I'll probably attempt to draw a proper schematic in the morning:
[t]http://i.imgur.com/xKYKbIq.jpg[/t]
What's some good software to draw schematics?
There's a tiny chance the wall wart I used had the wrong polarity but the diode's datasheet says the Peak Repetitive Reverse Voltage is 40V so the 12V shouldn't matter right? And if that was the case wouldn't the cap have blown up?
[QUOTE=Aathma;49976152]This might be a weird question but wouldn't it be technically feasible to setup a microcontroller/PI or other low-cost computing hardware with GPIOs to communicate with a PCIe device through it's header? A video card for example.
I've got an old video card and I'm curious how hard it would be to get in into a running state in a stand-alone setup. Not trivial of course. But I think it would make an interesting project.[/QUOTE]
If it's a card that Linux already has a driver for, you should be able to implement a bottom-layer PCI driver that emulates an actual PCI controller and wiggles pins in the correct way, and the existing driver stack will be able to sit on top of that. All the PCI-based drivers should then function as normal, because the actual PCI interface is abstracted.
If you want to implement PCI[B]-E[/B] that sounds pretty hard, because it's a very fast bus. I don't know details, it might be possible to clock it off of the GPIO and slow it waaaay down. Go and surf through the spec.
A good starting point would be [URL="http://free-electrons.com/doc/books/ldd3.pdf"]the Linux Device Driver book[/URL]; chapter 12 looks relevant, but you'll want to read the first 11 too if you're new to kernel and driver development.
Technically it doesn't sound impossible, but it will not be very fast, and will be a pretty significant software effort. However, it shouldn't require any reverse engineering.
RIP desoldering gun heating element.
2015-2016
Now to feel pleb and use the pen sucker for a month.
[QUOTE=Goz3rr;49978882]After some inspection with a magnifying glass it's a 1N5819 schottky diode, how big is the chance that this part is the only thing failed? I've never really repaired anything before and usually I'll see people talking about capacitors failing but the 3 that are on here all seem fine, this diode is the only part that appears to be damaged with the naked eye[/quote]
Diodes tend to fail open, it being burned up likely indicates there is another problem.
[QUOTE=Goz3rr;49978882]What's some good software to draw schematics?
There's a tiny chance the wall wart I used had the wrong polarity but the diode's datasheet says the Peak Repetitive Reverse Voltage is 40V so the 12V shouldn't matter right? And if that was the case wouldn't the cap have blown up?[/QUOTE]
[url]http://www.digikey.co.uk/schemeit/[/url] is free and doesn't look like ass.
A reverse polarity protection diode is often designed to short out, unfortunately a lot of designers are too incompetent
to put a fuse before it so it ends up burning up, you probably didn't have it on long enough to damage the capacitor, then again it isn't always
possible to tell a bad cap by looking at it or measuring capacitance, ESR or Dissipation is the only true test.
[QUOTE=Chryseus;49979809]you probably didn't have it on long enough to damage the capacitor, then again it isn't always
possible to tell a bad cap by looking at it or measuring capacitance, ESR or Dissipation is the only true test.[/QUOTE]
It was on for a good 30 seconds I'd say, before I noticed the smell because initially the status LED came on so it appeared to be charging, which probably rules out the wrong polarity?
My cheapo multimeter can't test capacitance, however the resistance appears to be 600 Ohms but I'm not sure about that because when I was trying to measure the resistor that's labeled 1K it was showing completely different values, I suppose that might have to do with the transistor? The label reads 25V 100µF, could I just put 12V on it for a short while and then measure the voltage?
[QUOTE=Chryseus;49979809]Diodes tend to fail open, it being burned up likely indicates there is another problem.
[url]http://www.digikey.co.uk/schemeit/[/url] is free and doesn't look like ass.
A reverse polarity protection diode is often designed to short out, unfortunately a lot of designers are too incompetent
to put a fuse before it so it ends up burning up, you probably didn't have it on long enough to damage the capacitor, then again it isn't always
possible to tell a bad cap by looking at it or measuring capacitance, ESR or Dissipation is the only true test.[/QUOTE]
KiCad is not too bad nor is fritzing also. fritzing is super simple and basic but not too bad for quick ones.
[QUOTE=Goz3rr;49982058]It was on for a good 30 seconds I'd say, before I noticed the smell because initially the status LED came on so it appeared to be charging, which probably rules out the wrong polarity?
My cheapo multimeter can't test capacitance, however the resistance appears to be 600 Ohms but I'm not sure about that because when I was trying to measure the resistor that's labeled 1K it was showing completely different values, I suppose that might have to do with the transistor? The label reads 25V 100µF, could I just put 12V on it for a short while and then measure the voltage?[/QUOTE]
You're right that doesn't sound like wrong polarity.
Easiest way to test a suspect cap is to remove it from the circuit and measure the resistance, it should climb fairly quickly to OL, this won't exactly tell you if it's healthy but that doesn't matter in your case since you're looking for a short somewhere.
I certainly would pop out that transistor and test it, you should start to reverse engineer the circuit as well get a clearer picture of what it's doing.
A few months ago, I bought one of these [url=http://www.aliexpress.com/item/3V-6V-0-56inch-Four-Digital-Clock-Kit-DIY-Electronic-Clock-Kit/32599961150.html?ws_ab_test=searchweb201556_2,searchweb201602_4_10036_10035_10034_507_10020_10001_10002_10017_10010_10005_10011_10006_10003_10021_10004_10022_10009_10008_10018_10019,searchweb201603_2&btsid=ca1f815f-6674-4f59-a786-c499f204407a]DIY Digital Clock Kits[/url] for myself.
I assembled it, everything worked perfectly, but after a week powering it on, I realize that the time is not precise, it's like the clock has been counting seconds/minutes wrong. So, for example, instead of it displaying 12:00 when it is midday, it would display something like 11:40. And when I got it to work I adjusted the time to match the one in my computer/cellphone. And in the product page it said "precise second calculation".
I asked a few people about this issue, and they said that the problem is that there's no "compensation" for calculating time. One of them even said that temperature can interfere in a certain component in the board and delay or advance time.
So, here's the question: is there a way I can fix this issue? Do I need to change the source code? How would I do that?
[QUOTE=RockyTV;49997823]A few months ago, I bought one of these [url=http://www.aliexpress.com/item/3V-6V-0-56inch-Four-Digital-Clock-Kit-DIY-Electronic-Clock-Kit/32599961150.html?ws_ab_test=searchweb201556_2,searchweb201602_4_10036_10035_10034_507_10020_10001_10002_10017_10010_10005_10011_10006_10003_10021_10004_10022_10009_10008_10018_10019,searchweb201603_2&btsid=ca1f815f-6674-4f59-a786-c499f204407a]DIY Digital Clock Kits[/url] for myself.
I assembled it, everything worked perfectly, but after a week powering it on, I realize that the time is not precise, it's like the clock has been counting seconds/minutes wrong. So, for example, instead of it displaying 12:00 when it is midday, it would display something like 11:40. And when I got it to work I adjusted the time to match the one in my computer/cellphone. And in the product page it said "precise second calculation".
I asked a few people about this issue, and they said that the problem is that there's no "compensation" for calculating time. One of them even said that temperature can interfere in a certain component in the board and delay or advance time.
So, here's the question: is there a way I can fix this issue? Do I need to change the source code? How would I do that?[/QUOTE]
You could look at getting a crystal that can replace the current one in a chamber. It is not really worth it for a kit like that. If you want a precise clock, look at RTC.
You could get something like [url]https://www.adafruit.com/products/264[/url] and with a [url]https://www.adafruit.com/products/881[/url] with an arduino or simalar would get you a better clock.
Most of the parts should be on ebay.
[url]http://www.ebay.com/itm/2PCS-DS3231-AT24C32-IIC-precision-Real-time-clock-module-memory-module-Arduino-/401051829692?hash=item5d608d45bc:g:tToAAMXQaZxSE0o7[/url]
That is 1.50$ for two modules
[url]http://www.ebay.com/itm/New-Red-1-2-4-Digit-7-Seven-Segment-LED-Display-IIC-for-Arduino-/262081977689?hash=item3d054d9d59:g:yFsAAOSw4HVWFJa0[/url]
It is 10.62 but has everything needed to drive them,
[url]http://www.ebay.com/itm/2pcs-0-36-inch-4-digit-led-display-7-seg-segment-Common-cathode-Red-NEW-/171907471306?hash=item28067bd7ca:g:uuoAAOSwBLlVOM61[/url]
Could be used with a shift register and be cheaper. $0.99 +
[url]http://www.ebay.com/itm/20Pcs-SN74HC595N-74HC595-8-Bit-Shift-Register-DIP-16-IC-/130806676456?hash=item1e74af7be8:g:uycAAOSwgQ9Vgoa2[/url]
Which are 20 for 1.53
And lastly an arduino, such as
[url]http://www.ebay.com/itm/USB-CH340G-Nano-V3-0-16M-5V-ATmega328P-Micro-Controller-Board-For-Arduino-/262123424219?hash=item3d07c609db:g:3rQAAOSwniRWNtSi[/url]
which is about 2.50$
In total it could be the same price plus you get the fun of putting it together.
So I'm in the market for a proper bench-top power supply with constant current capability. There's an awful lot of cheap ones on eBay, so are there any you guys would recommend? What about these two?
[url=http://www.ebay.com/itm/Adjustable-Power-Supply-30V-5A-110V-Precision-Variable-DC-Digital-Lab-With-Clip/181921919389?_trksid=p2045573.c100034.m2102&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D35958%26meid%3Db6a94acee6234b1197fab6fe90ca6c8e%26pid%3D100034%26rk%3D4%26rkt%3D8%26sd%3D281980217030]$60 one.[/url]
[url=http://www.ebay.com/itm/361089176129]$72 one.[/url]
When I asked around, I was informed that those are completely shit on the inside.
I picked up the Gopher CPS-3205 because of that, it's simplistic but, the output is decently stable, it's small, the internals are good (I opened it up and checked), and I've been happy with it.
It's hard to tell anything for sure with those lower end power supplies being cloned by so many manufacturers, some can be good others not so good, I hear the UNI-T one is pretty shit though so I'd avoid that.
I'd go for the 60 bux one.
[QUOTE=nikomo;50008913]Gopher CPS-3205[/QUOTE]
Ehhh I would not recommend a switching power supply, but then it depends on what work you're doing exactly.
Also outputs on the rear :suicide:
[QUOTE=Zero-Point;50008083]So I'm in the market for a proper bench-top power supply with constant current capability. There's an awful lot of cheap ones on eBay, so are there any you guys would recommend? What about these two?
[url=http://www.ebay.com/itm/Adjustable-Power-Supply-30V-5A-110V-Precision-Variable-DC-Digital-Lab-With-Clip/181921919389?_trksid=p2045573.c100034.m2102&_trkparms=aid%3D222007%26algo%3DSIC.MBE%26ao%3D1%26asc%3D35958%26meid%3Db6a94acee6234b1197fab6fe90ca6c8e%26pid%3D100034%26rk%3D4%26rkt%3D8%26sd%3D281980217030]$60 one.[/url]
[url=http://www.ebay.com/itm/361089176129]$72 one.[/url][/QUOTE]
Here's one guy's test of the [URL="http://dangerousprototypes.com/forum/viewtopic.php?f=2&t=5705&p=53417#p53417"]PS-305D[/URL], but I couldn't find any other reviews. I suppose try putting some sort of snubber diode to safely shunt the -9v spike on startup if you plan on getting it.
[QUOTE=LoneWolf_Recon;50009468]Here's one guy's test of the [URL="http://dangerousprototypes.com/forum/viewtopic.php?f=2&t=5705&p=53417#p53417"]PS-305D[/URL], but I couldn't find any other reviews. I suppose try putting some sort of snubber diode to safely shunt the -9v spike on startup if you plan on getting it.[/QUOTE]
A spike with no load isn't terribly unusual, it's only really a risk to unprotected CMOS circuits, even then the total energy is low and most CMOS gates can deal with up to 20V.
Seems to be a common issue with many linear supplies, but then you can't expect amazing performance on a budget.
I'm a super beginner with this stuff but I'm doing a project that I need a small basic amplifier for. I pulled [URL="http://i.imgur.com/DjJijybh.jpg"]this[/URL] out of a set of desktop speakers and it has the power requirements I need. However I cut the wires off the power terminals to prepare them to solder my own wires in and I forgot which of them were positive and negative. Naturally I do not want to reverse the polarity otherwise some of those capacitors might burn out.
I tried testing it with a multimeter and marked them myself but I want to make sure it's right. [URL="http://i.imgur.com/ShflIOO.jpg"]Here's[/URL] the underside.
Edit
Figured it out. I was hoping my multimeter would tell me but on continuity it kept giving me positive numbers regardless of what probe was where so I hooked it up. Nothing burned up, just didn't work. Flipped them around and now it's good.
[QUOTE=LoneWolf_Recon;50009468]Here's one guy's test of the [URL="http://dangerousprototypes.com/forum/viewtopic.php?f=2&t=5705&p=53417#p53417"]PS-305D[/URL], but I couldn't find any other reviews. I suppose try putting some sort of snubber diode to safely shunt the -9v spike on startup if you plan on getting it.[/QUOTE]
It seems like it's not bad, but eeeeehhhh I've heard of that particular manufacturer of the one he's tinkering with... And anything that looks remotely like it have brand-names I've never heard of plastered all over them, if at all.
How about [url=http://www.ebay.com/itm/DC-Adjustable-Switching-Power-Supply-30V-3A-Output-Digital-Display-w-Low-Ripple-/281981343795?hash=item41a765d833:g:I-EAAOSw5dNWmr04]this one?[/url] Costs a bit more, but if it's any better quality at all I wouldn't mind it, I would just like to find something that costs no more than 100 American FunBux.
[QUOTE=Chryseus;50008960]Ehhh I would not recommend a switching power supply, but then it depends on what work you're doing exactly.[/QUOTE]
It's 30mVpp in CV mode, it's fine for most stuff you typically do, when you're looking for a cheap supply.
Today I found out what happens when you use your building's earth as the counterpoise for your HF antenna.
It's called an easy way to test every GFCI breaker and outlet in the house.
[QUOTE=nikomo;50011856]It's 30mVpp in CV mode, it's fine for most stuff you typically do, when you're looking for a cheap supply.[/QUOTE]
It's not so much the amplitude that causes problems but more the frequency distribution which makes it more problematic and difficult to remove than say 100Hz / 120Hz mains ripple.
Of course for many thing this doesn't matter and as a beginner with simple circuits you'll probably never enounter an issue (certainly not with a decent supply like that), however there is always the odd
unexpected case that leaves you stumped for hours on end (trust me I've been there), so I always try use a linear supply when experimenting.
The reason switching power supplies work fine with many products is simply decoupling, on a PCB you can put capacitors exteremely close to a chip which will deal with any such noise
issues, on a breadboard however things are much more difficult and that's where the benefits of a linear supply come in, there have been numerous cases where I've had to stick my beadboard
in a tin can to try keep out noise from nearby switching supplies which goes to show how sensitive some circuits can be.
[QUOTE=papkee;50012672]Today I found out what happens when you use your building's earth as the counterpoise for your HF antenna.
It's called an easy way to test every GFCI breaker and outlet in the house.[/QUOTE]
Sounds like fun.
[QUOTE=Chryseus;50012926]It's not so much the amplitude that causes problems but more the frequency distribution which makes it more problematic and difficult to remove than say 100Hz / 120Hz mains ripple.
Of course for many thing this doesn't matter and as a beginner with simple circuits you'll probably never enounter an issue (certainly not with a decent supply like that), however there is always the odd
unexpected case that leaves you stumped for hours on end (trust me I've been there), so I always try use a linear supply when experimenting.
The reason switching power supplies work fine with many products is simply decoupling, on a PCB you can put capacitors exteremely close to a chip which will deal with any such noise
issues, on a breadboard however things are much more difficult and that's where the benefits of a linear supply come in, there have been numerous cases where I've had to stick my beadboard
in a tin can to try keep out noise from nearby switching supplies which goes to show how sensitive some circuits can be.[/QUOTE]
I checked the output under different circumstances with a scope at school, and I couldn't see any significant noise on the FFT.
It's really well filtered, but it's obviously not as good as a linear regulator.
If I wanted to build my own USB soundcard interface that could also detect when a key is pressed, what chip/chips should I be using?
Since the button in question is a push-to-talk key would I be better off putting it in series with the microphone line (open circuit on the switch->grounded input sort of thing) or have it as another USB device?
[QUOTE=Cakebatyr;50023146]If I wanted to build my own USB soundcard interface that could also detect when a key is pressed, what chip/chips should I be using?
Since the button in question is a push-to-talk key would I be better off putting it in series with the microphone line (open circuit on the switch->grounded input sort of thing) or have it as another USB device?[/QUOTE]
Maybe try a [url]https://www.pjrc.com/teensy/[/url] They can act as usb devices.
What's your setup with the USB soundcard? A little more info would be useful.
It sounds like you're thinking of putting a switch in series with an analogue audio line which is generally not great, since switches are very noisy and you don't want that getting into the signal, and any filtering would be awkward since you don't wanna touch the frequency content of your audio signal if possible.
[editline]29th March 2016[/editline]
There are plenty of microcontrollers on the market that support USB, so finding one of those shouldn't be an issue, PIC and AVR are pretty popular. If you're looking to send the audio signal over USB (still not sure what you mean by 'usb soundcard interface' - does it interface with a usb soundcard? or is it a usb device that handles audio?) you're gonna want to find a microcontroller that has a decent ADC, or find a separate ADC chip for audio stuff. 24 bits @ 44 to 48 KHz is pretty standard when it comes to digital audio.
Sorry, you need to Log In to post a reply to this thread.
Electrical Engineering V3
