Ok, where to begin...
Its pretty obvious which is the original above... it was grown in place, quickly and with little thought. It simply had to work and there were certain attributes it needed to possess to make it happen. Heating in a van isn't quite as simple as one might think thanks to the crazy differentials, "leaky" surfaces, and outright actual leaks letting cold in from the outside.
Air flow has to be managed, and with nowhere to put ducts and the like, it has to be done using the spaces and channels that everything else in the van create. What this translates to is keeping the drafts mixed with the warmer air... A challenge that I had no idea would be so tricky.
I've seen differentials as high as 40 degrees between the floor and ceiling during my earliest attempts at heating this thing. Given that this occurs over a height of just shy of 4 feet, it was very uncomfortable to not keep it mixed.
One of the keys to this was a buck converter to run the fans at the minimum speed required to work against the drafts at all times, but most notably at night. Until the insulation was sufficient, any heat generated would head straight up for the metal in the roof, deposit it there, and creep back down the walls, fighting my every attempt. Adding the insulation helped, but I found that the cold air would find its way up out of the lower parts of the outer walls. Anywhere there was a crack, it would ooze in like there was a spring there, a steady supply of cold that was impossible to eliminate. I used tape to seal the joints between the plastic parts of the van, only to find it would just creep up to the next highest "vent" and spill over there.
So fan "idling" was used to combat it. The actual amount of air kept moving was rather low to reduce losses, and instead, boundaries were formed, creating invisible "seals" that would keep the cold air out and away from the center of the living space. It was pretty elaborate at first, with channels created under the bed specifically to allow a draft to form where it didn't matter as much; behind my head all the way at the back of the van.
The result was the recycling of already cooled air that was going to be cooled regardless, but slightly warmer than the air it was being made to fight off. It worked, but the controller saw so many revisions that the scary mess in the pic above was created in the process of sorting all this out.
Then there was the issue eventually of needing to actually add small amounts of heat to the drafted stream. The only viable power level that could be used was 100W to do this, by a switching system that put the heater elements in a series-parallel configuration. This needed to be automatic however, as I desperately needed a thermostat to control the temps. They can swing wildly, and when morning comes, the sun hitting the side of the van will drive you right out if there's nothing to cut the heaters off.
There was a major problem however with the switching. I was on the grid still at the time, and if the thermostat released the heaters to go back to idle mode, the inductance in the wiring would create a sufficient kick to cause serious arcing at the relay contacts. This led to a flashover condition when the relay would let go near the peak of the sine. The plasma combined with the way the relay had to be wired would cause a dead short, and trip the breaker at the source. I'm still a little baffled by this, as the source breakers were 15A (and 20 at one point), but the front gate breaker in the van is only 10A. More often than not, the van's breaker would remain untouched, while the source breaker went, as well as the GFI I put immediately following the breaker. There was never a ground fault, but the resultant spike was sufficient to take out everything except the breaker that was supposed to catch it. Its just not as sensitive I suppose. I digress. The flashover was dealt with as well as could be done with what I could throw at it. The orange caps (only one really visible above) on either side of the center relay quenched the arc sufficiently to prevent popping the breakers anymore, but every time it disconnected on that peak, it did more damage to the contacts. By the time I removed it, I could barely see the contacts thru the casing due to all the deposition on the sides. It made it through the torture last year, but my bet is it would have failed before this coming one was out.
There was also a high/low mode so that it could run from battery if the need ever arised in an emergency situation. In addition, during the warmer cold nights, it could switch 0/100W under the control of the thermostat rather than 100/800 (450 on battery).
The fan speeds changed depending on what was flowing through the elements to optimize heat generation. Ceramic heating elements exhibit both negative and positive temperature coefficient during operation. From room temp to operating temp, they start out slightly low and ramp up to full power, showing a negative coefficient. If they go past their design temp, they start consuming less power, showing the expected positive coefficient. This was baffling at first, but once I realized what was happening, controlling fan speed was the key to optimizing operation.
So the buck converter had to be modulated as well. It was interlocked as much as possible with a system like this so that the fans had to follow the elements activity.
When you add all of the above to the mix, the rat nest isn't quite as surprising anymore. But something had to be done about it, because it scared me honestly, even though out did it's job once the contacts were quenched.
The replacement controller has all of the same capabilities, but does a couple things a little differently, namely the switching.
In the new design, the idle/full changeover is dealt with by 2 relays instead of one. This also allowed me a better layout for the grid/battery mode. The inductance is reduced because it no longer has additional wiring to run out to the changeover relay box at the back of the van. It's handled right there in the controller and just gets a control signal from the back.
I did manage to find a condition where it would fairly reliably pop the source breaker, but for normal operation, there isn't even significant arcing.
The thermostat was modified so that the fan on/auto switch changes the controller operation from 0/idle to 0/full (or idle/full for anti-draft, changed at the controller). This mode cannot be changed while the heaters are actually generating heat without risking flashover. As long as the heat is turned off first, there is no problem whatsoever.
The heaters are all centralized now in a mixing box that produces a very uniform temperature air stream regardless of mode. The bed no longer has an "underneath" to channel anything, so there was no point in trying to use air currents against themselves since it would have to travel directly over me to get to them. The ceiling fan is now solely responsible for draft elimination, even though I am likely taking a slight efficiency hit by doing so. I already know it works, so that aspect is dealt with.
On battery, of course, for normal operation, this is all but moot. The primary heat source is a catalytic propane heater that as of the tail end of last season I was experimenting with using water to increase control and efficiency. It worked fairly well, but I need to do some more development on it. The controller was also supposed to get a switch to select propane/electric mode. This will kill the 6 fans that are used for electric mode to further save power (and disable the elements in the process), and control the single fan that pulls air across the water containers to regulate temperature inside the van.
One other tidbit... There is a "heat on battery" alarm comprised of a piezo buzzer and a switch that mutes it. The switch is set up so that if it's muted, and you plug in to grid, the alarm goes off again to remind you to unmute it before the system goes back to battery. The idea here is if it's plugged in and running on electric and the grid supply dies for any reason, the changeover won't automagically kill the batteries. The condition is therefore left up to me to decide if it needs to continue on battery or change modes.
That's about all for now, I'll throw some pics up soon showing the details of the new layout.
Steve
