How to rig up an off grid AC system

[sharpener]

Is it as simple as buying a pure sine waive inverter, connecting it to a battery to produce an "artificial grid" for the sunny boy to synchronise to, then plug in the DC panels to the sunny boy to supply power and then fit a battery charger to the battery to charge it from the "artificial grid"

Moxi

A small Victron MultiPlus inverter https://essandsolarsolutions.co.uk/mult ... 0v-ve-bus/ will do all of the above straight out of the box.

Mine can control the output of the legacy Steca inverters using frequency control, don't see why this would not work equally well with your SB.

[Moxi]

Thanks all for the input, I do need to go and do some reading up when I get quiet times and I think I need to sketch out what I want to achieve so that I don't take any wrong turns as I start to install things.

I must say I do like the look of the Victron stuff and I own a standard smart battery charger from them which "just works" without fuss or calamity, the remote interface is flawless too, so as time goes on I will probably be replacing stuff with that brand.

Moxi

[Saladin]

... I think I need to sketch out what I want to achieve so that I don't take any wrong turns as I start to install things.

I have a mantra "If the first one you make is the best you're lying to yourself"
I usually have a decent build by the Mk-IV

I must say I do like the look of the Victron stuff and I own a standard smart battery charger from them which "just works" without fuss or calamity, the remote interface is flawless too, so as time goes on I will probably be replacing stuff with that brand.

My experience with blue boxes differs. I find them a mite fragile under duress, I always got the sense it was off-grid made by gridders: marketing and office workers not dyed-in-the-wool field tested to fidelity. I never could get one of their battery chargers to charge a lead acid to specific gravity 1.275. They husta always quit at 1.26. That's actually pretty standard, Morningstar were the only product I ever found to properly charge a lead acid battery. That all changed with li-ion because you can pretty much charge them with a CV PSU. Still Morningstar are the only ones to implement proportional low temperature charger derating which is a much neglected cell longevity feature for lithium. Cells aren't logical. A 0°C charger cutout isn't good enough it needs to ramp down progressively from 15°C

The amount of times I hear things like the battery is charged because the light on the charger says so...

Take all that with a pinch of salt. I run a merciless proving ground. I won't sign off on a build unless I can't break it. I've killed many Victrons retired more on performance inadequacy, killed or retired all Mastervolts I've tried, toasted a coupla Morningstars (death by boot with live solar and battery isolation) but I've often ressurrected those, deranged a minor proportion of Studers (mostly my own idiocy), Sterlings...actually work and reasonably reliably but they're populated with C-rate semis more often than not, killed or binned nearly every off-brand yolk. Come to think of it I've never killed an SMA, the best I could do was blow the MOVs in a SB-HF with a nearby lightning strike...haven't used a sunny island.

In terms of sine waveyness, Studer are head and shoulders above the rest. I think the Multiplus drives only 4 FETs...

Swiss dudes know their shyt:



Xtender 3.5kVA

[sharpener]

What is the point of having two banks of 36 FETs if none of them are bolted to any kind of heat sink?

The fan noise from my Multi is noticeable as I write (bc the intervening door is open), but at least it gives me comfort that the FET/heat sink arrangement is getting cold air blown over it! I agree the engineering of it could be better but at least it exists.

[Tinbum]

Come to think of it I've never killed an SMA, the best I could do was blow the MOVs in a SB-HF with a nearby lightning strike...haven't used a sunny island.

The problem is a lot of stuff is done to a price. It didn't used to be the case but many rely on their past reputations that are no longer deserved. JCB being a typical case. I used to run their old stuff that you could leave for months unused and would be fine next time you came to it. Bought a brand new machine in 2010 and the chrome on the hydraulic cylinders was rusty in one month. The replacements then went rusty despite being specially chromed.

I've been running sunny islands and do fige them loads of punishment and not had a single problem. Think they are 13 years old now.

[Saladin]

What is the point of having two banks of 36 FETs if none of them are bolted to any kind of heat sink?

The heat sink is thermally and electrically conductively coupled to the FETs via the power bus. It's the Metal Shroud.



What can I say the inverter doesn't get hot. The overkill FETfield is the thermal design and a reliability feature. I wonder if they hand pick them for matching RDSon...it's not easy drive that many synchronously. The fans do run when it's doing some heavy lifting...not much. I run the house on twins; 7kVA combined. It's pretty rare that they'd start to sweat.
The XTH has probably double that and iirc the heat sink is also the PCB trace which is a solid aluminium plate in places

I think the Multiplus drives 12 FETs; four banks of three parallel. I watched a promo vid where I recalled (perhaps inaccurately) they claimed 4 a long time ago. I had a look on tinternet for some pics with the wrapper off; Looks like 6 above, 6 below because it would have to be a numerator of 4.

It didn't used to be the case but many rely on their past reputations that are no longer deserved.

Yurp. Couldn't agree more. BMW, Bosch, Mercedes...gone to the dawgs...
Most of my good gear is circa 2000 - 2010.
I've no interest in Apps or smartphones for that matter nor lightweight less rugged boxes I was gonna bolt to a wall anyway. It does amuse me that the Xtender is a 30yo design and it's still the most efficient on the market. All they've been doing is updating firmware since the 90s. They also changed the name from a "multimode power supply" to "hybrid inverter" to stay trendy. Impressively capable boxes.

Another thing I've observed is that the 90% efficiency is consistent across the load range. Most competitors are a lot less than the datasheet figures with low & intermediate load or high temps.

[sharpener]

What is the point of having two banks of 36 FETs if none of them are bolted to any kind of heat sink?

The heat sink is thermally and electrically conductively coupled to the FETs via the power bus. It's the Metal Shroud.

Doesn't look like it to me and I have done a fair bit of electronics thermal management design for radar power supplies and the like. The TO 220 and similar plastic packages have the semiconductor chip bonded directly to the metal tab, and are designed for the heat to be conducted away by bolting the tab to a heat sink which is not the case here. It appears that they are cooled by air flow aided perhaps by the ducting effect of the shroud, this would give a much higher thermal resistance junction to ambient and hence the chips run much hotter or you need many more of them.

Total mystery to me why they would have done it like that. Seems to be a failing in the design process. In my Victron the FETs (can't remember how many) are bolted to a plain heat sink plate but then the fan is positioned badly in respect to it. Also the fan speed management was by PWM and very buzzy but they finally brought out a mod kit last year which has made the noise a lot less intrusive.

[Saladin]

The MOSFET metal heat sink tab is the drain which is soldered to a PCB trace you'll need a 500°C tip temperature on the solder iron to budge. Hence electrically and conductively coupled to copper conductor with a massive aluminium plate on the bus and two huge copper cables on the plate.
I swap the cable bolts from stainless to brass and sub-hang a temperature sensor for a stand-alone fan speed controller for external case-mount fans when I build them into confined spaces. The case fans I've set to turn on at 35°C, they run intermittently at low duty cycle. The Studer fans are even lower duty cycle.



I also have a temp. probe on the "exhaust outlet" of my twinset in my powerplant monitored and alarmed by a SImarine Pico. Unless they're pushing >5kVA they don't spin and the temp stays in the 30°Cs or below most of the time.

The proof of the pudding is in the eating.

[Marcus]

It does actually make sense to me: by sharing the load between a large mumber of devices you reduce the heat dissipation to the point where a heatsink becomes unnecessary.

I can't tell if there's another 2 x 36 fets underneath or not, but if not that's still18 fets per corner of the H-bridge. Assuming a 48v(?)battery and 3.6kw, that's 70-80amps to/from the battery, so each fet would see about 4A at <50% duty: low enough to not need a heatsink.

The advantages are very low heat dissipation due to a very low combined Rdson, very compact design with short high current runs as you don't need to make room for a big heatsink, and of course high efficiency because you're not generating much waste heat.

You would need robust gate drive circuitry though to switch them all on and off cleanly.

[Saladin]

Hi Marcus.

That one is a 24Volter(3.5kVA @ ~93% eff.) but the 48volter (4kVA @ ~95% eff.) is the same PCB and layout I run both and they AC couple together, albeit at 26volt and 60volt because LFP and I can, possibly just a different traffo and firmware flash between versions.
There's nowt underneath 'cept pcb traces and pin headers to the logic board that's stacked behind.

Technically it's 4 x 18 FETs..{tomato emoji}