Embrace the Weirdness

No, that's OK, you can renege on your promise

Tue, 17 Nov 2009 02:21:17 GMT

I won't mind. Not at all. No, don't do it. Noooooooooooooooooooooooooooooooooooooooo.

Adventures in thru-hole soldering

Tue, 17 Nov 2009 01:47:17 GMT

I need to get a smaller iron tip. And a table lamp. And one of those magnifying glasses. If you do much soldering you start to want one of those.

This is an assembled kit that follows the arduino standard. It's 100% thru-hole components. It has two less LEDs than the "official" arduino, and it doesn't do USB at all. But that's no problem - I have a USB to RS232 dongle laying about, and it works just fine with it.

It's somewhat cheaper than the USB version, but the nice feature I like about it is that you can program it while it's hooked up to an external power supply. The other style, you have to program it, detach it from USB and then hook it up to your external circuits.

And now I have a backup (or main) in case of another magic smoke event.

Hmm, charge pumps aren't that bad

Mon, 16 Nov 2009 19:52:01 GMT

http://www.edn.com/contents/images/321805f1.pdf

I actually understand this schematic, and why everything is where it is. The basic principle is to put capacitors initially in parallel and then switch them in series. It's slightly tacky - it actually inverts the waveform, because those are PNP BJTs, but I can understand why he's using PNP's. Oh yes. He doesn't care - because he's dealing with a 50/50 duty cycle. For me, inverting a PWM pin means that the max value becomes zero and zero becomes the max value. Annoying but manageable.

I think the RC values would have to be retuned as well. Not sure what frequency he's tuned this for.

I dunno. The pMOSFETs are still more straightforward.

I should add Colorado Springs in there

Mon, 16 Nov 2009 18:04:09 GMT

From Colorado Springs to Boulder to Littleton to Granby to Longmont to (now) Fort Collins, welcome to Colorado.

Watch your back.

Side effects may include cross-dressing...

Mon, 16 Nov 2009 17:04:56 GMT

http://en.wikipedia.org/wiki/Pramipexole#Adverse_effects

Twisted transistor

Sun, 15 Nov 2009 02:43:48 GMT

Take a walk on the high side

Sun, 15 Nov 2009 02:39:07 GMT

Grr.

nMOSFETs are really really annoying to use on the high side. Charge pumps, push-pull arrays, level lifters, and none of those things are trivial. Just on charge pumps alone, try it - do a Google search. People are patenting shit on that stuff, if they're not writing academic papers on it. There are people out there and that's all they do. Someone wakes up in the morning and says, I need to build a better charge pump. Must build a charge pump for my thesis advisor.

pMOSFETs aren't quite as annoying, but not straightforward either. To get it to switch, you still need something that will sink to ground and you probably need a pull-up resistor to keep the gate stable. But still, it's a circuit I can visualize in my head, and a schematic that does not look hilarious.

One of the design goals of this thing, is to keep the number of integrated components to a minimum. The only real black box in this whole thing should be the Atmega chip - and everything else should be widely available. I don't want to put another black box in the design, like a charge pump chip.

You know, people publish these theoretical schematics, and you see a switch there, and there's some handwaving, and presto, this is how it works, and then you go and try to build something that actually does work, and nobody, and I mean nobody says something like "Oh, and if you're gonna switch something on the high side, you might want to use a pMOSFET and control the gate with a BJT and a pull-up resistor".

Something that would've been really helpful to know.

Meep.

Sat, 14 Nov 2009 19:51:59 GMT

http://theodoramichaels.com/articles/meep.php

Offishuls lie to us. Film at 11.

Sat, 14 Nov 2009 00:51:30 GMT

http://www.examiner.com/x-23316-Madison-Independent-Examiner~y2009m11d10-Ft-Hood-Official-story-full-of-holes

Yeah, I think it's working

Fri, 13 Nov 2009 22:50:39 GMT

With a 1.6kOhm resistor, that's a whopping load of .20 W, that is, when the panel voltage is set to 18V (which is the peak, it probably won't get there much). But at least I can twiddle the PW, of the PWM pin on the arduino and get varying voltages across the "load". Widen the PW, get more voltage, shorten it, get less.

That's pretty much all a buck converter is supposed to do. What turns it into a charge controller is reading the voltage and temp sensors and dynamically changing the PWM pin on the fly. But that's just software. Ones and zeros.

So now I have several paths to take. I can focus on turning the schematic into a PCB, and get a real 12V lead-acid battery. I've done the current capacity calculations for the PCB design.

I can try to fake out a 12V battery by taking a bunch of rechargeable AA's and wiring them in series. Except that NiCD and NiMH have different charging characteristics than a lead acid battery. So I'd have to write the controller software twice and there's no guarantee that much of it would be reusable either.

I dunno. I need to think about it.

Cross your fingers and hope we don't have a burnout

Fri, 13 Nov 2009 18:14:52 GMT

Those components on the left side, they're too big to fit on a breadboard. So I punched holes in a cardboard box and have a bunch of wires with alligator clips connecting them all up on the other side. I call it a redneck breadboard. Yee. Ha.

I suppose if the arduino would pump out a higher PWM frequency, I could reduce that honking inductor down to something more manageable. I've got the frequency as high as it will go, otherwise that inductor would be even larger than it is. It's 100uH (remember we're measuring things in microTons, so this is a huge amount), but it's a high current component - rated to 14A or so, although if you notice in the datasheets, they qualify that with a gale force cooling system blowing on it. But still, at 50% capacity, that's 7A which is nothing to sneeze at.

That 4th TO220 (the black square thingy way over on the left) on the breadboard is a 300V,30A power MOSFET. That in turn (you might have to squint) is connected to a PWM pin on the arduino. That's the switch part of what makes a switched mode power converter.

Everything on the redneck breadboard is downstream from the MOSFET, and is a standard buck converter.

I don't really have a proper load to test this out with, and I need to be very careful, as although the components can handle high current, those wires cannot. At this point though, I just wanted to get it all wired together and see if the magic smoke would stay inside everything. So far the magic seems to be holding.

It's good to prototype before committing to a PCB. I think I'll redo the schematic and put the MOSFET on the ground side of the buck converter, as it makes it easier to wire the PWM pin in.

Another reason to use something like an Arduino

Thu, 12 Nov 2009 22:15:13 GMT

The consequences of failure are cheaper. I miswired the arduino to the high-output LED on the left and created a short a few days ago. I didn't see the magic smoke, but I did start smelling something plastic and hot. Those little boards are designed quite well, because the only thing fried was the atmega328 chip. Simple - pop it out, order another one and pop it in.

If I did this to that FPGA, something like this would really really hurt. And I knew sooner or later I'd see the magic smoke.

But this is nice, I can dial the "panel" power up and down (I'm using a lab DC power supply to simulate the panel), the arduino keeps ticking, and the LED on the left changes colors at 17V, 16V and 14V.

While I was waiting for the replacement atmega, I got bored and ripped some more stuff out of the DSL modem and put two of its LEDs just past the 9V and 12V linear regulators, so I have some idea of what's on when, if there's no arduino around.

I'm getting pretty good at picking resistors now. Quick, you need to drop 7V and limit current to 20mA, what resistor do you need?

Another nice thing about the arduino, is it's a completely open design, and you can source these things from multiple places. They're even sold in kit form, and in RS-232 form as well.

Um, er

Thu, 12 Nov 2009 19:23:29 GMT

http://gizmodo.com/5402152/surfing-the-road-on-an-iphone+controlled-oldsmobile

Just because you can do something, doesn't necessarily mean that you, um, should.

You know, with a little bit of mechanical knowledge, they could've hooked into the throttle with a servo. Would've required less power and they could've kept the seats in the car. I think they could've built something that controlled the steering wheel without destroying it.

To measure the speed, they could've hooked into the OBD. Jesus Christ this could've been done way more elegantly. Yeah, it's only a $300 car, but still why be more destructive than you have to be?

Great, they're using wifi. You do know the security model of wifi is swiss cheese...

It's harder than it looks

Thu, 12 Nov 2009 04:20:44 GMT

http://dontjudgemyhair.com

It requires discipline of a certain sort, just to look and accept.

Philip Glass - Pruit Igoe

Wed, 11 Nov 2009 02:21:03 GMT

Anyone want to take bets

Wed, 11 Nov 2009 00:47:15 GMT

On how long Nidal has to live, now that's he's awake and can talk? Something tells me he won't be giving press interviews, that's for sure.

Did you catch that?

Mon, 09 Nov 2009 21:29:22 GMT

http://www.nytimes.com/2009/11/09/opinion/09krugman.html

At this point Newt Gingrich is what passes for a sober, reasonable elder statesman of the G.O.P. And he has no authority: Republican voters ignored his call to support a relatively moderate, electable candidate in New York’s special Congressional election.

Newt Gingrich is now considered reasonable. Repeat that again in your head. And that he's being ignored because he's too leftist. Newt Gingrich is too leftist for the Republican Party. Repeat that again in your head. I think I understand why people are resorting to McNuggetini's to cope with what's going on.

And if Tea Party Republicans do win big next year, what has already happened in California could happen at the national level. In California, the G.O.P. has essentially shrunk down to a rump party with no interest in actually governing — but that rump remains big enough to prevent anyone else from dealing with the state’s fiscal crisis. If this happens to America as a whole, as it all too easily could, the country could become effectively ungovernable in the midst of an ongoing economic disaster.

In fact, I'd bet on it. Although I don't think that's such a bad thing. What I've seen, if they're just merely useless, that's better than being worse than useless. In any case, I wouldn't make any plans that depended on government policy, that's for sure...

It is government policy to not have a middle class anymore

Mon, 09 Nov 2009 19:44:57 GMT

http://online.wsj.com/article/SB10001424052748703932904574511794170939688.html

Something you'll never hear from any politician, oh no. I'm sure if you asked them they'd say they're for Mom and Apple Pie too.

However, actions speak louder than words.

Now that's what I call cynical

Mon, 09 Nov 2009 19:23:44 GMT

http://prorev.com/2009/11/hidden-reason-for-marijuana-prohibition.html

One of the most important reasons for marijuana prohibition is that, without it, it would be hard to justify the drug war. Not on moral or medical grounds, but based on weight and popularity. While pot is a drug less harmful than vodka or cigarettes, it is also is the bulkiest and most used illegal substance and thus the easiest for cops to find.

And I wonder, which god would that be?

Mon, 09 Nov 2009 19:11:04 GMT

http://www.cnbc.com/id/33783316

Can't turn away from horrible disaster. Can't stop looking. Can't unsee.

Mon, 09 Nov 2009 06:36:18 GMT

And now, neither can you.

Update: Well, at least Georgia should know her cocktails - she's a veteran AA member...

You know, this pretty much sums up everything that is wrong with open source software

Sun, 08 Nov 2009 19:25:20 GMT

http://openofficemouse.com/pr110609.html

I suppose it will come with its own HOWTO? And you might actually need to buy some consulting time to have someone either a) train you or b) configure it?

Guard it, but with secret instructions not to guard it well

Sun, 08 Nov 2009 04:19:06 GMT

http://www.nytimes.com/2009/11/07/us/07truck.html

Brilliant. People will beg to be vaccinated. Oh yes, spank me with that mercury. Slap me with that squalene. Oh yes, yes, yes!!!

Y'know I can't help but think of Rory Sutherland

Sun, 08 Nov 2009 03:41:37 GMT

And that TEDtalk I posted a while back, when I read stories like this one: http://www.google.com/hostednews/ap/article/ALeqM5jemnWB5veRnM9evI2vqkMks9HwkwD9BPM93G0

Maybe they should restrict it to only the upper classes, and guard it, but not guard it well? Now that's what I call marketing. The rebranding of the swine flu vaccine.

In my mind, someone's finally doing some lateral thinking. That's the way you sell it. That's what our entire society is built on, you know. The art of persuasion and the selective drowning out of irrelevancy.

I'm hungry, I'll think I'll get me some diamond shreddies...

Phase 1 complete

Sat, 07 Nov 2009 20:55:37 GMT

What are you looking at? Ok. This is the first leg of the charge controller in prototype form. I've already got the schematic all done (I've even got Gerber files for it, woohoo). But I'm not ordering a PCB until I've seen it all work on a breadboard first.

So you've got a solar panel that does 18V at peak and a 12V battery that does - you figure it out. The solar panel will output whatever it feels like, depending on the time of day and how overcast the sky is. The battery will be somewhat more reliable, as long as it has a charge. I want to circuit to prefer the panel when it's running and only defer to the battery when the panel is crapping out.

The arduino spec says it can take 7-12V and it can take a max of 16V but they don't recommend it since the onboard linear regulator isn't heatsinked. Or to put it another way, the magic smoke might leave the board if you push it too far.

So, I've got the battery behind a 7809 and the panel behind a 7812. When the panel's going the 7812 will output 12V and the battery's regulator will output 9V all the time. Thanks JM for the suggestion of putting the +9V and +12V outputs behind two diodes, and then wiring those in parallel with respect to the common ground rail on the output side.

So why are there three regulators instead of two? Well I don't exactly have a 12V battery yet to drive that side of the circuit, so I'm using a 12V wall wart I scavenged from a junked DSL modem I don't use anymore. It's a constant current power source, which means you have to regulate it, so it's behind another 12V linear regulator of its own.

And then there's a bunch of filter capacitors sprinkled all about. They don't do much except look good.

What are those resistors doing? Ah those are high resistance voltage dividers, there to feed into ADC pins. One to measure the state of the battery and the other to measure the state of the panel.