Electrical Engineering V2

[MetricMoose]

I'd also make sure you get a bigass heatsink/fan for it. I mean, I'm [I]terrible[/I] at waking up in the mornings and even I think simultaneously setting your room on fire and blinding yourself seems a little excessive.

[DrLuke]

Maybe a better idea is to just buy a dimmable daylight bulb and try to drive the dimmer/build your own.

[Cakebatyr]

Perhaps an airhorn might be better?

[DrLuke]

[QUOTE=Cakebatyr;45731115]Perhaps an airhorn might be better?[/QUOTE]

The point isn't to wake you up, but to wake you up in a less stressful manner. By being exposed to a lot of light you trick your brain into thinking it sees sunlight, thus waking up the body more efficiently.

[ben1066]

I actually got something done with an FPGA for once. Has anyone here got experience with the Lattice parts, how is the ice40 software? Or should I skip those and stick with Altera or Xilinx? I am really using an FPGA for the sake of using an FPGA here so low cost would be good which certainly seems to be an area in which lattice wins.

[Subby]

I have a 100w led module, if you look into it, it leaves a dark spot in your vision for quite a while.

[Fourier]

[QUOTE=DrDevil;45728837]Just keep in mind that this WILL blind you if you directly look at it. You should wear appropriate eye protection when you work with it, and design your PWM controller to fail into the off-state at all times.[/QUOTE]

I was thinking about adding a diffusion, but when I think about it, if I add diffusor too close whole thing can catch a fire :v:.

[editline]19th August 2014[/editline]

[QUOTE=benjgvps;45730891]I'd also make sure you get a bigass heatsink/fan for it. I mean, I'm [I]terrible[/I] at waking up in the mornings and even I think simultaneously setting your room on fire and blinding yourself seems a little excessive.[/QUOTE]

I have some heatsink that is quite good at cooling (it was used for q9650 which is 95w max).

[editline]19th August 2014[/editline]

[QUOTE=Subby;45734834]I have a 100w led module, if you look into it, it leaves a dark spot in your vision for quite a while.[/QUOTE]

Thanks for warning, I though it would be something like this anyway.

[editline]19th August 2014[/editline]

[QUOTE=DrDevil;45731462]The point isn't to wake you up, but to wake you up in a less stressful manner. By being exposed to a lot of light you trick your brain into thinking it sees sunlight, thus waking up the body more efficiently.[/QUOTE]

Yep, exactly. School (math) is coming closer and I don't want to wake up all groggy every morning, especially in winter when I am depressed fuck.

[pentium]

I would like to point out from personal experience with those 100W LED's that they require a large heatsink or they burn out.

[DrLuke]

According to wikipedia, LEDs have an efficiency of 0.15 to 0.4. So in the best caste, you're generating 200W of heat for 100W of light (If I understand it correctly)

[Zero-Point]

If 100W is the total power draw, then it can only generate up to 100W of heat, can't it?

[DrLuke]

[QUOTE=Zero-Point;45739490]If 100W is the total power draw, then it can only generate up to 100W of heat, can't it?[/QUOTE]

Ah, I thought 100W was the total light output for some reason. Well, in that case you still generate 60W of heat in the best case.

[r0b0tsquid]

Can anyone think of a reason why my AVR's ADC is suddenly giving all zeroes for every reading? Whether it's connected to an external pin or to the internal bandgap reference. It was working fine yesterday, same code, and I haven't touched anything. Problem with the chip?

[editline]blah[/editline]

ATmega324PA on this board
[t]http://wrenstech.files.wordpress.com/2014/08/20140817_163140.jpg[/t]

[DrLuke]

[QUOTE=r0b0tsquid;45740509]Can anyone think of a reason why my AVR's ADC is suddenly giving all zeroes for every reading? Whether it's connected to an external pin or to the internal bandgap reference. It was working fine yesterday, same code, and I haven't touched anything. Problem with the chip?

[editline]blah[/editline]

ATmega324PA on this board
[t]http://wrenstech.files.wordpress.com/2014/08/20140817_163140.jpg[/t][/QUOTE]

Might not be initialized properly, AVCC maybe not be powered, pin is being driven active low would be my first guesses.

[r0b0tsquid]

[QUOTE=DrDevil;45740833]Might not be initialized properly, AVCC maybe not be powered, pin is being driven active low would be my first guesses.[/QUOTE]

uint16_t get_battery_voltage()
{
    ADMUX = 0x5e;   // AREF = VCC, input = 1.1v bandgap reference
    ADCSRA = 0x87;  // ADC enabled, don't start conversion yet, auto trigger off, no interrupt, ck/128 = 62.5 kHz
    _delay_ms(30);
    uint32_t result = 0;
    for (int i = 16; i; --i)
    {
        ADCSRA |= 0x40;
        while (ADCSRA & 0x40);
        result += (uint16_t)(ADCH << 8) | ADCL;
        bpmc_sendstring(itoa((uint16_t)(ADCH << 8) | ADCL));
    }
    return 4613734 / result;   // = 1024 (number of ADC steps) * 1.1 (bandgap voltage) * 16 (number of readings) * 256 (fxp88 fixed point)
}

It's measuring the internal bandgap reference, so no to the final one (that did cross my mind! I checked and the DDR bits are zero).

I'll have a look at AREF :)

I've checked the datasheet and my initialization seems to be okay? Like I said, this was working fine yesterday!
The bpmc_sendstring() call is just for debugging - both that and the itoa() function work fine.

[editline]blah[/editline]

AVCC is at the same voltage as VCC and AREF (4.2v). That means it's probably an issue with my software. Back to the data sheet :v:

[ben1066]

Decided I wanted to try a different display technology, this time VFDs. Ordered 5 IV 21 displays, really cute 8 character VFD tubes that were designed for calculators as far as I can tell. Should be able to do something neat, they don't need very high voltages either which is a bonus. After this I am kinda left with just nixies so I'm going to have to look out for some pretty nixie tubes.

EDIT: Add 4 ILC2-1/8s too. Been looking for something in a small DIP form factor for a while, I may of finally found it.

[r0b0tsquid]

Hnnngh. If I change the ADC's prescaler frequency in one of the functions that uses the ADC, it reads a different value (25 or 37 usually, depending on the frequency) instead of 0 in [B]every single other function[/B] that uses the ADC, even though these functions all provide their own value for ADCSRA and so the prescaler.

Also, when I make a minimum example program which does nothing but initialize and read the ADC, it gets stuck forever waiting for the ADSC bit to clear - i.e. the conversion never finishes.

What is happening?:(

Have I zapped my chip?

[editline]20th August 2014[/editline]

It's started working. [B]Help. [/B]​:v:

[ben1066]

So I decided to look at the MachXO2 FPGAs and well...
[video=youtube;3-NeNCn-p80]http://www.youtube.com/watch?v=3-NeNCn-p80[/video]

[r0b0tsquid]

Okay problem seems to have gone away, I am incredibly nervous.

However, it seems to be more or less working! I stuck a 1602 display and a shift register onto the logic header so that I can have a nice readout until the screens arrive :)

Measuring a 32-ish volt output from my power supply:

[IMG]http://i.imgur.com/XlnP2pk.png[/IMG]

The auto-ranging works nicely!

There is, however, a problem:

[IMG]http://i.imgur.com/27M9ydx.png[/IMG]

Measured voltage vs actual voltage. I'm assuming this has something to do with MOSFET leakage currents in the attenuator...

(it clicks up a range at ~3.8v)

[editline]20th August 2014[/editline]

Nothing to do with the attenuators - I've fixed it!

It was to do with the output impedance of the internal bandgap reference which it uses to measure the battery voltage every time it takes a voltage reading. If you switch to the bandgap reference and then immediately try to measure it, it gives you a false reading - if you have switched to it from a higher voltage, the reading will be higher than the true value of the reference. This means the calculated value for the battery voltage will be lower, and the voltage reading will tail off towards the top of the range. Fixed! The meter now agrees within about 0.5% with my EX330 between 0 and 32v (the biggest range I can test).

Now I can move on to measuring things like current, resistance etc.

I'm also going to switch on a high-frequency PWM pin while I'm measuring from the ADC, to inject a bit of noise and hopefully increase the resolution I get with oversampling and decimation. At the moment all of the readings are coming back pretty much the same, so I'm only getting 10 bits of precision really - if I can add 1 LSB of random dither then I should be able to get 12 bits of precision.

[LoneWolf_Recon]

In my experience I've either carefully put the ADC pin into a quick Hi-Z state and drain the pin capacitance (Which I've felt is a culprit with ADC Reference-Change inaccuracies) with an internal pull-down or waited a short bit (typically a millisecond at max).

[ben1066]

Am I missing something with these 0.5ppm TCXOs [url]http://www.farnell.com/datasheets/1804154.pdf[/url] , they seem to cost an order of magnitude less than others that at least on the surface seem very similar. The only reasoning I can think of is because they are used a lot for GPS.

[Ddrl46]

[QUOTE=ben1066;45748740]Am I missing something with these 0.5ppm TCXOs [url]http://www.farnell.com/datasheets/1804154.pdf[/url] , they seem to cost an order of magnitude less than others that at least on the surface seem very similar. The only reasoning I can think of is because they are used a lot for GPS.[/QUOTE]

Probably because they are only available in wacky frequencies?

[ben1066]

[QUOTE=ddrl46;45748792]Probably because they are only available in wacky frequencies?[/QUOTE]

Fair enough, that doesn't really pose a problem when you just want to keep accurate time on an FPGA though so all is good. Was just expecting there to be something hidden in the datasheet that meant they weren't all they seemed.

[pentium]

I should really finish designing this new Teletype interface but fuck, I really don't want to redraw eight pages of schematics.

ddrl, I need your eyes.

[r0b0tsquid]

[QUOTE=LoneWolf_Recon;45747842]In my experience I've either carefully put the ADC pin into a quick Hi-Z state and drain the pin capacitance (Which I've felt is a culprit with ADC Reference-Change inaccuracies) with an internal pull-down or waited a short bit (typically a millisecond at max).[/QUOTE]

I'm not changing the reference (AREF), I'm connecting the bandgap reference to the ADC input (set MUX to 0x1e) and reading that known voltage to find out the value of AREF (which is the same as the battery voltage) :)

But yeah more or less the same thing, I just wait for ~250us for the internal capacitance to discharge.

[pentium]

[QUOTE=pentium;45738665]I would like to point out from personal experience with those 100W LED's that they require a large heatsink or they burn out.[/QUOTE]

Case in point:

[IMG]http://i11.photobucket.com/albums/a166/ballsandy/CGS_0852.jpg[/IMG]
[IMG]http://i11.photobucket.com/albums/a166/ballsandy/CGS_0853.jpg[/IMG]

When running at 30v 1A you get something like five minutes without a fan before the heatsink is too hot to be efficient and the LED would burn out.

[Fourier]

[QUOTE=pentium;45758692]Case in point:

[IMG]http://i11.photobucket.com/albums/a166/ballsandy/CGS_0852.jpg[/IMG]
[IMG]http://i11.photobucket.com/albums/a166/ballsandy/CGS_0853.jpg[/IMG]

When running at 30v 1A you get something like five minutes without a fan before the heatsink is too hot to be efficient and the LED would burn out.[/QUOTE]

Somehow impressing that it lasts 5 minutes :D

Cool, I will have fan and I will run it only for 15 minutes, I guess I am on safe side :D

---------------

Wait a second, you said 30V 1A.. this is 30W.. holy hell, now that is problematic..

[r0b0tsquid]

[QUOTE=Fourier;45759202]Somehow impressing that it lasts 5 minutes :D

Cool, I will have fan and I will run it only for 15 minutes, I guess I am on safe side :D

---------------

Wait a second, you said 30V 1A.. this is 30W.. holy hell, now that is problematic..[/QUOTE]

100 watts is a lot.

[Fourier]

[QUOTE=r0b0tsquid;45759240]100 watts is a lot.[/QUOTE]

Yeah, I am quite getting sense of it. Still, it is doable. He said "without a fan". Most CPUs in a PC would die without a fan (where required), so the fan is crucial.

[Metallics]

[QUOTE=Fourier;45759306]Yeah, I am quite getting sense of it. Still, it is doable. He said "without a fan". Most CPUs in a PC would die without a fan (where required), so the fan is crucial.[/QUOTE]

A modern cpu draws about 70W... and presumably you've seen the size of the heatsink/fan combos you get to run them cool....

[gamerpaddy]

[QUOTE=gamerpaddy;45762059]Hah, looks good. made one too (few weeks ago)
2x100W + 4x Constant Current sources + 2x 24V to 36V stepup 150W each & big lenses for 90° light cone

i built a working flashlight, but after some test runs one stepup died (shunt resistor...) and one lead acid battery.
[T]http://s7.directupload.net/images/140822/9kp8k6mz.jpg[/T][/QUOTE]

they are pretty cheap. 5.50€ per led, 3€ per stepup or 5€ for stepup with constant current
8€ per lens and heatsink from junkyard