[DCFusor]

Some issues arise in charging my Volt on my completely off the grid solar system. They amount to "what do you do when the sun goes behind a cloud", more or less. While my home batteries are a bit larger than the ones in the Volt (24kwh for the home), they are not designed for more than C/10 or C/15 levels of drain, and, being lead acid, don't have the round-trip efficiency you'd like to have - especially for cycles more than a few minutes deep (they are pretty good for short cycling). And the more you cycle these, the shorter their calendar lifetime. So of course, that's to be avoided to the extent possible. The 240v EVSE effectively draws about 3.5kw, which is faster than C/10 for my home batteries, and when babying them - less is better.

I have both the 120v and 240v chargers, both Voltec. While the 120v one will let you switch between current draws, it's on the coarse side, and you have to unplug the car to make it respond to the switch. Not an exercise in user-friendliness. At least you can put it on a switched circuit and turn it on and off, leaving it plugged in. Here the charge confirm chirp won't bother anyone, and you can turn off the "theft alert" or whatever that is for the power going off making the car act like a car alarm, so that's a start. Of course, the bigger Voltec doesn't even have a rate control, and as designed, isn't switchable (but of course...when there's an EE in the house).

To make a long story short, while this is less than what I intend to wind up with, I've got a hack that works for the moment. My solar electronics include some Xantrex (now bought by Schneider) charge controllers and inverters, which have a programmable Aux output that can be set to switch on various conditions. I am using one of the outputs to control a few large solid state relays that switch power to the two charger cords I have, and it's working well. You can't quite easily get perfection this way - for example, you can't turn on the relays via one trigger source, but off by another (but you can if you consider driving both of the SS relay inputs with different aux outputs...wired AND logic), but even just using one simple criterion and some timing the Xantrex stuff provides, I've got some thing that is now real usable - and I don't have to get up every few minutes to throw a switch or unplug/switch/plug the car.

I simply set the Aux output to go true when battery voltage was high, up near the "absorb" voltage for my system, and cut off when it got below the "float" voltage. Each trigger on or off has a time delay associated with it to add what amounts to hysteresis to the system so it doesn't "hunt". I'm using a 1 min delay (battery has to be high for over a minute to trip on) and a two minute delay for off - this is to ensure we don't waste too much time just rebooting the EVSE all the time, it takes some seconds to start charging after power is applied.

A picture of this as it sits would give an electrician/code inspector the willes, so I'll abstain till I make it look pretty. For now, all I did was bolt the overrated (480V 40A) relays down to a large aluminum panel (heatsink) and proceed to wire it up, using nice crimp+solder lugs for everything, and a phone type wire over to the solar stuff to drive the optoisolators. I'm switching both halves of the 240, and put in another relay for a separate 120v auto-switched circuit.

In testing with both sun and generator power, using the EVSE in a case where it draws more than either can put out results in a short EVSE on time - 2-4 minutes, then a few minutes for the battery voltage to recover, then a recycle of the process, very slick and neat for a "bang bang" control loop. While you're cycling the house battery a little doing that - it's right at the top of things, the electrolyte that got charged/discharged is still right there and hasn't diffused away yet - over all, if anything, kind of good for a lead acid setup as this sort of microcycle cuts down sulfation. Of course, this cycling means it might take longer to charge the car, but it also means the car can't run down my home batteries if I'm not paying enough attention, and that's more annoying yet.

Either circuit can of course also have other diversion loads wired to it, as I'm putting in switched outlets as well as the hardwired stuff to the car chargers. I have this interesting water distillation unit, hmmm...

I have noticed before that if you unplug and replug a fully charged Volt, it'll go back into some kind of "top off" mode for awhile again. With this cycling, that happens a few more cycles after the car is reporting a full charge, but not many. I doubt it hurts anything, as this mode does draw less power - more like 1kw off the 240 charger when it's happening, and it doesn't go on forever. If anything, I'm squeezing in a little extra for the range games we all play.

I expect not to need this as much this summer, when the new 2.4kw worth of panels get off the ground and up into the sky, but weather hasn't permitted that yet. I am still going to make a fully adaptive charger with the J1772 cable I bought and an EzWebLynx web server/controller I also got, so more control will be possible.

 

[solar_dave]

Does the control side of the J1772 spec allow for changes in the charger by altering the voltage on the J1772 control line? Is that how the 120 unit changes its draw? Is it stepped? i.e. 4 volts means 8amps, 6 volts means 10amps .... And is it at all variable and dynamic?

 

[bonaire]

As the price of home-battery sets using LiFePO4 comes down, you could have higher-cycle counts off of those type of cells with less electrolyte worries and faster response times. They're more expensive but could probably take more of the abuse over the Leads. I hear various off-grid setups are now considering LiFePO4 (mostly Chinese-origin over higher-cost A123s).

 

[sparks]

Dude, you are having WAY too much fun.

Where did you get the relays? Are they salvage? Military salvage can often provide real heavy-duty, hardened components that will never wear out (even if you're hit by a small mortar round).

Also, do you switch both halves of the 240V with different arms of the same relay, or are they different relays? Is there any reason to be concerned about the possiblility of one-half making contact while the other half fails to? Just one of the things I would obsess about if I was doing the project!

I love home-brew stuff, as I spent my childhood (we're talking born 1954 here, aaarrrgh) building ham radio gear from junk. My family couldn't afford the price tag of anything approaching real ham gear. I did eventually get a low performance Heathkit general coverage receiver (GR-64) and spent all my time building outboard enhancements, including a preselector (to improve the scandalously poor image rejection), BFO, Q-multiplier, voltage regulator, new power supply (when the factory supply blew), T-R switch for break-in keying, etc. All tubes! Got fried by the B+ supply a few times too.

BTW, Zero-Hedge is now in the doghouse. Yesterday they ran a scathing post on the 5-week Volt production halt, referring to Volt as the exploding car, spontaneous combustion in dealer lots, and all that rot. So now I have an incontestable example of Zero Hedge either not fact-checking, or worse, stooping to falsehoods to do hit pieces for political purposes, just like the other media clowns they condemn. I seem to be checking their site a lot less frequently now.

 

[DCFusor]

Well, ZH certainly has some issues, don't they -

The real one is they are uber-perma-bears - and a great place to find the bad news first. They've predicted 128 of the last 1 collapses, after all. Closet gold bugs tend to be like that. I never did get that religion. Why would you want to be the only "rich" guy when everyone else is starving and rioting? (It would seem to me that in that case, riches would be food and safety, not some pretty rock)

To want to be the king of a very degraded hill, by wanting the hill itself to fail, seems a little sick to me, for one thing. Another is - well, you still can't eat the stuff, defend your home with it and so on. I think having a lot makes you a target for others of that religion, actually. At any rate, since they hunger for bad news, and hate GM (while cheering on all other bailouts as they think they make gold more valuable, when actually all they do is make dollars less so), they'd pick up on that one - but they have a real short attention span and a few have actually responded well to my owning a Volt as long as I couch it in "well, I'm more ready for Armageddon than you are, as I have gold and a Volt (and a farm, and a few other interesting things), not just the one thing". A few of them get it.

The SAE j1772 spec has the EVSE controlling what the car is allowed to draw via the pulse width of a 1khz pulse it supplies to the car. A variable pulse width is trivially easy to do...yet GM has a fixed width in the 240 charger, and a choice of only two in the 120v one, and for whatever reason - you can't change the pulse width on the 120v one while the car is plugged in (even though the spec says you can and the open evse project has tried this with success on Leafs). This is how I plan to control my homebrew one when I get it built - but I needed a fast solution while that and other projects fight for priority on my list, so I did this.

In turn, the car tells the EVSE it's there by loading this pilot signal down a little. When it's ready for charge, it loads down the signal more - and then the EVSE switches on the AC (in normal operation). When you pull the plug, the load on the pilot signal goes away first, giving the EVSE time to drop the AC power before there can be a spark at the terminals (or you get shocked) - fairly slick for a one-wire analog protocol.

The relays were a surplus deal from Marlin P Jones, at $9 each (low for this kind of thing). They only switch on or off at zero crossings, so are more benign than mechanical contact ones. I do note that in the spec, the mechanical relays don't ever switch under power drain except in fault conditions - they thought ahead there. There is this negotiation dance that goes on with the pilot signal, and the car tells the EVSE when to switch on the AC to it. The connector design is such that certain contacts break first or last as the plug is moved, which handles the unplugging case under power too - if the EVSE stuff is quick enough, it drops the relays before you can get the plug out (this is under power, and so they'll spark for this case).

To prevent half of the 240 from being switched and the other half not (insofar as that can be done assuming no relay failures), I just hooked the inputs of the two that handle the 240 in series - they have to either both go, or neither - the inputs are optoisolated and floating so you can do things like that. Since they are optoisolator inputs that can take 3v to 32v, and I have 12v signal, they get 6v apiece - no problems there. And they handle only changing state on zero crossings themselves. The Solid state relays are single pole, so I had to use two for the 240.

Really, the only thing bad that would happen if I only switched one side is that the GFIC stuff in the EVSE would go nuts and refuse to work without both being there - fine, but I figured on doing it right, since I bought these honkin relays in bulk when they were on sale for 1/5th price or thereabouts.

Ah, it turns out MPJ is still selling these: http://www.mpja.com/prodinfo.asp?number=17157+RL But the price went up a bit. Still, compared to other vendors, this is a steal for something this huge. You really only need 240v and 15 amps for this job. After all, that's what the ones in the Voltec unit are rated at (mechanical relays too). I thought that was cutting it kind of fine - but since they rarely operate under current draw, I guess it's OK, at least they work. If they were switching "hot" they wouldn't last very long there. The car is pretty gentle on the EVSE - it ramps the current up and down gracefully. IF DigiKey or another mainstream new parts vendor stocked these, they'd probably be around $50-60 each.

I'll be testing a homebrew thing fairly soon, certainly have it done and installed before summer is out. I plan to put it on my home network so I can see and control it from the feet-up position as desired.

I got a couple of these things - http://www.ezweblynx.com/product_info.php?cPath=Store&products_id=EZWEBLYNX_5V
To be miniature web servers - they rock! They've implemented some extensions to HTML that allow you do control some I/O pins that can be a/d converters, counters, plain I/O, and PWM (funny that they would have that!) from the network, and even run a little state machine kind of thing out there in the server - which is about the size of my thumb past the last knuckle. I'm testing and programming one now. I may use discrete logic to make the PWM signal the standard needs, and just control that from the uP, making it more crash-proof if that seems like an issue. So far the thing has run 3 days on my network with a couple other machines dedicated to banging on it though - seems good.

So, one of these could do battery monitoring if you didn't have the nice Xantrex stuff to do it for you, for example, and do fancier logic if you think you needed that. To this engineer, the bang-bang control is inelegant, so I want proportional control, only going full on or full off when the occasion really warrants - so I don't have to brute-force switch the AC and make all the down stream computers reboot all the time (wastes time, for one thing). On the other hand, for all I know making them reboot all the time makes them less likely to die some odd way from an error in their internal ram, you never know if they don't let you see the design.

But the very simple thing I already did - works like a charm even if it is ugly to both eyes and mind. So it's probably going to stay there for awhile - once I like the homebrew better and it's shown to be reliable, I'll have a Voltec for sale...

 

[Henry_FL]

If you want a elegant solution, make a Xanbus monitor that can the programmaticly adjust the J1772 signal for available current.

Xanbus is open doc now as is the J1772, so all you need is a little time

I very familiar with the Xantrex stuff, I worked with them for 5 years and was the principal designer of the XW-60

While I think it would be cool to have a J1772 station that could have monitoring abilities, I think I could count on on hand how many units a product like this would sell. I'm sure UL would have allot to say also about having the J1772 variable and want all sorts of fail safes.

 

[DCFusor]

Well, not so interested in making a bigtime product just now out of this - as you say, the volume would be pretty low, but those who need it, really need it - nice niche, that. The kind of thing where you can sell a few at a decent price, make your NRE back, and be happy. With the ridiculous overpricing of everyone else's EVSE, there's a little room for a custom I think. Been through UL before - what a bunch of know nothings. I'd never have passed the existing Voltec if I were them. Why would they care about variable? It's fixed at max now - that's worst case, you can only go down from there. Less power, less heat, less strain on relays and connectors, all that. At any rate, UL stamp is nice, but not actually required all the time. "It depends". There are European bodies with standards that are a lot nicer and more well spelled out for things like that to make insurance companies all happy.

Xanbus - I spent hours searching for xanbus, protocol and so on on the net last week. Not one hit on the new CAN based thing (I do have an old copy of their modbus version, which is VERY different).

Do you have a link to that? Seems to me Xanbus has some serious issues with its "publish/subscribe" model in that you can't add very many things to it before the "too chatty" peripherals clog up the bus with collisions and errors, and evidently they don't know about random backoff times - I may have to split a fairly small system (on XW 4024, 5 charge controllers, one SCP) up to make it work. I can't add that last charge controller (or any other one more thing, like their gateway) without it crashing and the inverter going to 70 hz, which of course, gives the GM charger the willies - and makes my lath run faster.

All I could find on Xanbus was promo puff and some words about it being totally proprietary and closely held, along with comments by one ex engineer pointing out their failure to understand how to do good protocol, or in this case, wisely use an existing model. So if you've got a link, please pass it on! Else I'm going to have to reverse engineer mine, and that's a PITA to have to do, especially if the design was dumb, as that one guy said. I do have a can bus dev board that has a snoop mode, but...it's still a lotta work to really RE anything at all complex.

Of course, locking this into Xantrex/Schneider might be a thing that would reduce, not increase the market possibilities - they're big, but they're far from the only guys out there doing this. As far as I can tell, they're the best at everything except this silly comm protocol stuff - but that's a deal breaker if you can't make big systems with it.

 

[Henry_FL]

Xanbus is the physical canbus using modbus protocol

I know at one time they had the informational available, I see if I can get a hold of the current spec.

 

[DCFusor]

I'd be real grateful, Henry. I have the modbus stuff that shows what registers and addresses are supposed to do what, I think (haven't scanned it as close as I would to actually start writing code to it yet, but it's a lotta pages with what looks like the right stuff).

Seems reasonable that they'd not toss all that out just to go to a new physical layer in things, I'd bet that insofar as they could, they copied the same C code headers to the new stuff unless there was some really pressing reason not to.

Funny how they redefined the can bus pinouts just to make things incompatible...at least that data was easy to find online.

 

[Henry_FL]

Its all off the shelf canbus chips and hardware so the physical layer is all canbus. Xantrex just then used modbus for the higher layers, similar to how TCP know nothing about the lower layers in the OSI model. I'll message you either way when I hear back, i still know a few people in Burnaby that use this stuff.

 

[DCFusor]

Right, I'm familiar with protocol stacks (designed a few myself), and I figured they wouldn't toss all their old higher layer stuff out the window completely. One wonders how they are clogging up a 250kbit layer with a protocol that used to fit into 9.6kbaud, however. Maybe they did the modbus layer request response instead of publish/subscribe? CAN can support the former too, but if their stuff is programmed otherwise, you're kind of stuck with it.

Now, here's a product for the solar biz - fix it so that you CAN have a big Xantrex system centrally controlled and monitored, by making an interface that emits multiple CAN buses for each little cluster of the max size their stuff can handle, then combines all the info in itself...and allows central control via taking it to more of a star topology from the master controller (remember 10 base T?). Not too hard once you get the basics handled, I'd guess. I'm mean, I'm maxing their thing out well before I have enough stuff to do a good size no kidding no compromise off the grid single home...that's not good!

 

[Henry_FL]

My understanding is that the devices on the bus blurt out status, regardless of the need for the information at a fixed rate, pissing away all the bandwidth. My last solar project was with Morningstar to do networking, and that's at 1200 baud!. I did a lot of caching, bus-snooping to make sure update rates would back off if the bus utilization got to high.

The XW is a great system, I had a large 14kw pv, xw-6048 system for 5 years with MorningStar charger controllers ( My last project i was part of ) so you really don't need them to talk with each other

Having a hub/switch configuration could be a real problem, especially with the way the bootloaders/firmware updates work, hopefully Xantrex will get an update out that mitigates the larger system issues

 

[DCFusor]

Right, I don't *need* them to all talk, all that much. Nice not to have to have a battery temp sensor for each thing, though, and have net inflow-outflow, which requires input from it all - or just having a separate shunt and metering system (which is what I did for myself - but then I only know totals, not the contributions of things). It's more a "want" thing I suppose, and I'd like to make some nice logs of what really happens with the system.

And that's what I heard too - they just blurt out status, and don't have a good backoff-when-in-conflict scheme, so the bus gets blown really quick due to retries taking even more bandwidth that only results in more collisions.

As far as I know, Xantrex has never pushed out an update that would let me for example, sync my older SW 4024 (and forget the 3524) with my new XW-4024. I guess they just expect me to buy all new stuff every cycle. GoodLuckWithThat, it ain't happenin.

 

[DCFusor]

A new first for me!

Well, in just two days of testing, this new hack works great. But today, the last (finally unbroken) solar panel for my system upgrade came in, and I got it at least lashed up...but as you can see from the picture, I have my work cut out for me indeed, unless I want to call that a "solar garden". The result is going to be a completely covered roof once the racks and scaffolding are built up - should be cooler in summer and there will also be spare power for A/C! Yeaaah - it's been a couple decades for that one.

But on to the "first" - today I went out early (after stirring it up on ZH a bit) to vote (Ron Paul), and pay my phone bill in Floyd, which is about 1k feet higher in altitude, and a lot farther on the map by car than by helicopter. Got back with about 10 mi electric range left, found the new panel, lashed up that last group and...I no longer need the hack on days like today - and I'll have full charge again by 3:15 (and house batteries full too), to go out to two other towns and do other errands (Cburg and Bburg for those who know this area). The first for me will be the first day with over 45 miles or so pure electric range - I've never had enough power to do things like that before. In fact, some of the first charge was from today, this AM from about 9:00 to about 11:00. Seems like it's wise to top the car off after sitting all night in the cold. It reports full charge and blinks the green light - but still draws full snot from the 120v cord I used this AM when the solar wasn't putting out near enough for the big guy - and it did for a couple of hours. First time I've ever seen 42 miles estimated in the driveway too - and by the time I get down the hill I live on, it was 45!

Victory! YAYYYYYYYYYYYYYYYYYYY

 

[bjhorton2005]

Haha yes you are a tinkerer! If it were me I would just do all you can to double your capacity and add inverters. I know that gets crazy expensive though. I'm about to do a simple grid-tie system through missouri wind and solar. Really pumped about it! It should make for a close to zero electric bill.

 

[solar_dave]

Haha yes you are a tinkerer! If it were me I would just do all you can to double your capacity and add inverters. I know that gets crazy expensive though. I'm about to do a simple grid-tie system through missouri wind and solar. Really pumped about it! It should make for a close to zero electric bill.

Hope you are,staying away from their wind offerings. Do a search on them at http://wind-sun.com Forums.

 

[DCFusor]

An update - this system is getting it done fine. At the current panel capacity, the main goodness of having the automation is that I can start charging the car a lot earlier in the day, and get it done quicker, while never putting my house batteries in any pain at all - they've had to get up to near max voltage (but not max charge) at least once to trip the system on. Then the system cycles on and off a few times in the first half hour, then just stays on, and a few times I've been done by noon, which helps my scheduling. Today, in a partly cloudy day, I hit full on both the car and the house by 2 pm, not too shabby. This IS a doubling or more of the system, but of course on a day that's just dark, I'm still doing no car charging, and am in hunker-down mode re the house itself, as my "job" involves having a rather large, fast, computer on 12 hours a day or thereabouts. So a couple dark days in a row still means using some sort of backup - which can now include the Volt via an inverter I put onboard - it's the most efficient gasoline->kWh converter in the stable.

I DID add an inverter and 5 more xantrex charge controllers in this mod. The new inverter can also run my fusor, my lathe and be a charger off a 120 or 240v genset. I like having that separate from my "main power" as when an inverter switches from inverting to charging (which then passes raw gen power to the output) it can crash computers, and the gen power isn't as high a quality in general. My system for this building has 3 inverters now, specializing in different tasks, and all able to backup the others in a pinch.

I am hoping to start getting all these panels into the sky next week. They will completely cover that roof and extend beyond it both top and bottom, leaving a few left over. I may put a few of the extras on the east and west walls angled out a bit to catch earlier and later sun than the roof will. The building was built oriented pure solar south on purpose...