[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

DC-> AC Power Switching

Original poster: Davetracer@xxxxxxx


My old battle ground....  Might want to reply to Dave directly ;-))

There's been a lot of discussion of some heavy duty power switching here on the list, particularly between AC and DC supplies, and, well, I am just out of my depth on a science project my son has suggested. Can I present it and just ask for comments? It's quite obvious that the EE stuff I learned in 1976 has either faded from my brain or been overtaken ... I'm quite aware that this may be a stupid question, and am willing to take the heat for that, but it might also prove to be a fun question. If it can't work, I'd love to know *why* for the knowledge of it. I am in the oh-so-fun position of being The Dad Who Does Not Know The Answer, and since I have a scope for my computer stuff and build computers, he thinks I know All about electricity, which I just plain don't.

I'm going to ask (beg, bribe, plead) the moderator to allow this note on the TC list in the hopes of finding someone who can tell me if his idea might work.

His idea, which I find interesting, is to store power. Charge the batteries during the night, when the electrical networks around the country are pretty unloaded, then release it during the day, especially at peak (4 PM, according to the California ISO people anyway). He's done some pretty interesting research that shows that areas like California have the generating capacity but their timing is wrong; if they could store some power at night, they'd do okay during the day. (Yes, I'm aware this is in thousands of megawatts.)

He also found that in Colorado, we have done that, by pumping water from one lake upstream to another at night, and letting it flow down during the day. This explains a lot of what I've seen on Guanella Pass.

One reason TO do this thing is the politics at his school. They're having a "green" year and the science projects are supposed to be about, well, ecological stuff, solar power, whatever. Personally I'd prefer to blow something up or bounce a laser off the Moon or something fun, but you know how it goes.

He proposes using "cheap tech" to solve this -- lead acid batteries. Are lead-acid batteries efficient enough to even bother? This is an answer I am having a curiously hard time finding an answer to ... what percentage of energy going in comes back out? I don't propose to abuse the batteries (deep drain them or chill them).

Here's the idea. Take 10 lead acid batteries (basically car batteries). Connect them through a relay network (note, relays just for safety, semiconductors don't bother me) so they can be in one of two modes: Either being charged at 12V each with a trickle charger, or connected in series to form (up to) 120 VDC. [Guess I could charge the group at 120VDC but I've been told that is bad.]

(Yes, ideally, the chargers should be set to a precise voltage in the 12.x region and so forth, taper the charging, and so forth. That'll be for the government version).

Inverters are expensive and the transformers and sheer iron are terribly expensive. Again, "cheap tech". He came up with the idea of patterning a sine wave, from 0 to 120 volts and on down to 0, by connecting an increasing number of batteries in series with medium speed, computer driven, semiconductor switches (probably transistors). I mean, at 60 times per sec, I don't think relays are going to keep up, but probably someone will mention how wrong I am (grin).

Thus at (+)0, nothing is connected; at the first point in the sine wave, as it approaches +12, the first battery connects, as it approaches +24, the second battery connects, and so forth, up to 120V, then back down. Yes, I know a true AC sine wave goes up past 120V, but this is more of a "does this approach stand a chance" versus "nailing it on the dot" question. We'll definitely be outputting sinewave data much faster than Mr. Nyquist requires; 60 Hz is not a big challenge to even a slow 8 bit microprocessor.

This will give us modulated 0 to plus 120 VDC. You can see where I'm going, right? Now one idea is to simply switch poles so we're now working with (-120) to 0 VDC. If that proves impossible, maybe going for another 10 batteries, to form a negative 120 rail. Note: I don't propose to buy 20 car batteries for a science project no matter how nice my son is about it. We'll use small 12V lead-acid ones.

So, we're at an idea of running a -120 volt rail and a +120 volt rail, and using semiconductors to switch the voltage to match a normal AC sine wave. (As you can see, this is a clumsy way of doing modulation with an additive voltage ladder). Without an inductor or a capacitor to smooth things down, it's going to stair-step. Adding those is possible. Adding a 6V or even 3V battery to get very smooth is also possible.

I do see that going from 120V sources and doing PWM would also work, but I have deep concerns about matching the current and voltage needs, especially as they change with a load, "on the fly" with PWM.

I believe I can handle the DAC stuff to measure the current and match phase to an outside AC source (although of course I'd have to chat with the power company about that; probably just showing it would work would be sufficient).

Now I could probably model this system with relays and run it at 10 hz. But I am not comfortable using transistors at this voltage and power level [not to mention buying all this stuff!] without asking someone if this has a prayer of working. Believe me, I have made my share of expensive smoke clouds just with computer stuff! I did a search on the Net for this sort of stuff and found a grad student over in Taiwan made AC with PWM, and it sounds like he worked his tail off. I found nothing else!

I don't know much past what'll happen when the 60 hz cycle is working and the 'scope says it matches. Bring it out to an outlet? Isolation transformer? To be a good project it should generate a good current. "The Battery Page" has taught us a lot. I'd like to be able to steadily generate AC at 20A and peak to 40A or so, because there are going to be spikes (see also: inductive loads, turning something on, etc). The spikes worry me a lot, because I don't want evils like ringing and undershoot to hit the transistors and kill them. And I am totally out of my depth answering a question like, "What impedance do I match to", because matching that is what prevents said ringing. I clearly need to talk to someone who knows power engineering; hence this note.

For example, if an inductive load like a vacuum cleaner powers up, there is going to be an initial heavy draw (if I remember EE correctly, theoretically infinite at start). I have dreams of expensive power transistors letting the smoke that makes them work out at that point. When it powers off, I believe there's going to be a voltage spike as an inductor turns off. (See also: smoke coming out). Things like this worry me. An isolation transformer is looking really much better.

The only thing I've seen online that can handle around 40A are some power transistors. I haven't seen anything in SCR land that can. The other thing about SCR's that bothers me is the heat they generate handling even small loads.

Anyway, there you go. With the "green" teachings his school has [sigh] , if this project can be made to work, he'd probably do pretty well on the science fair side. If it's a nitwit idea, mea culpa -- he is 16. If it needs tweaking but is possible, that would be nice too. I wish I knew more on this but I just plain don't, so I'm asking some people who would know.

    Many thanks,

    Dave Small

p.s. Feel free to email me direct at <mailto:davetracer@xxxxxxx>davetracer@xxxxxxx to avoid cluttering up the mailing list, probably unless you think your answer is also relevant to something like running a DC powered Tesla Coil ...

p.p.s. I think trying to do 3-phase is probably not a good idea. *grin*

p.p.p.s. "1.21 Gigawatts! Tom, what am I going to do!!??" -- Back To The Future