project idea - configerable Buck Boost DCDC for single solar panel?

[max_torque]

With the cost of energy rising around the world, looking on the usual outlets, one can find cheap solar panels.  The cheapest ones are often second hand, or only available in small quantities, which means different panel specs and powers / voltages etc

Does anyone think there would be a market for a small (~300W) DC/DC with a buck boost architecture that could be directly attached to each panel and act as a MPPT controller for just that panel, and drive current out to a common higher voltage DC bus?  ie the panels are all wired in parallel to one DC bus?

This would help keep the cost of the wiring down too, as the higher current segment of the system would be short.

The device could run off the panel, and have some form of setup to tell it what the MPPT voltage was for the particular panel (or it could attempt to learn that itself?). Output DC volts of around 400 Vdc might be sensible to keep currently low and to provide a very easy second step to symthesise an AC wave form from that DC if necessary?


Basically, similar to micro-inveters but staying in the DC domain?

[M0HZH]

Sounds like an interesting idea, can't believe something similar doesn't exist already. I've been toying with a CN3722 based module recently for a similar purpose.

[ejeffrey]

Solaredge does almost this.  They use per panel "power optimizers" which are DC-DC converters (either buck or buck-boost, I'm not sure) that are then wired in a series string configuration feeding a large AC inverter.  This allows per module MPPT.  The individual converters only have to handle relatively low voltage but you get the advantage of reduced current in the wiring of using a high string voltage.

[JohnG]

There exist such systems, and they are usually called "DC Optimizer" systems. There are many variations on them, and there are companies doing this commercially.

John

[thm_w]

Looks like the solaredge units are in the 400-600W range: https://www.solaredge.com/sites/default/files/se-p-series-module-add-on-power-optimizer-datasheet-row.pdf

Simple diagram here: https://unboundsolar.com/blog/micro-inverters-vs-string-inverters
Maybe a good middle ground between full on micro inverters?

[max_torque]

Coms over power would allow for module monitoring, diagnostics and setup as well.

I think a DC link at around 400V  or 500v(once you are over 50v, you may as well be high enough to leverage low current efficiencys) is sensible which would allow a direct sythesis of a 240VAC mains waveform pretty easily.

So the DC DC would probably need to be a boost only architecture, with something like a 5x step up ratio.  What sort of physical size do people think this entails? (say 400W max ie 1 A @ 400V)

Any suggestion on dc dc architecture and converter controller ic etc?

(can't see that it needs to be of an isolating type architecture, ie all panels can share a common ground)

[penfold]

[...]
So the DC DC would probably need to be a boost only architecture, with something like a 5x step up ratio.  What sort of physical size do people think this entails? (say 400W max ie 1 A @ 400V)

Any suggestion on dc dc architecture and converter controller ic etc?
[...]

One point is that if the DC is going to be transformed to a mains AC at any point, it may well need a layer of isolation somewhere. I forget whether the insulation of cables, connectors, and panels actually allows them to be coupled to mains circuits... but if the unit is targeting second-hand panels, you may need to make some guarantees about its safety which the second-handedness of the panels may not meet without testing or re-certification.

Since cost is one of the main factors and you may in that case want to be a bit 'light' with EMC filtering, it is worth considering a bit of the system-wide EMC with converter choices. In particular with systems of many power converters with similar switching frequencies, beat frequencies can be a common problem. The beat-frequencies themselves don't necesarily exist (just a linear sub of sines with close frequencies) but the total sum of all noise will have a periodic variation. Maybe spread-spectrum controllers will get you out of some possible issues there, the alternative would be synchronization, but I couldn't hazard a guess in the differences of cost between spread-spectrum or implementing a synchronization element to the comms protocol... or whether you'd even have a problem in the first place TBH.

A flyback might be a better alternative to a boost converter, or perhaps a coupled inductor boost converter. In comparison to a standard boost, those two alternatives allow for more favorable voltage stress on the switch and don't leave you quite as heavily constrained between step-up ratio, efficiency, size etc. The advantage of the flyback is of course that you can get the isolation if desired. The boost however might be the better solution in terms of cost, but it is worth seeing what effects the parasitic capacitance on the switch-inductor-diode node is going to have on efficiency.

[max_torque]

Interesting points on the EMC / Hetrodyne effects!

I guess if the DC DC controller is a current based controller then depending on its control strategy, with different pannels receiving different solar insolations, that may automatically result in different frequencies and duty cycles for the boost converter?

Having isolation at the point of generation is useful, from both a safety and a redundancy aspect, so i will need to consider the best converter architecture to include this. I guess like all converters, the cost will be dominated by the smoothing capacitance, switching elements (losses) and the inductor / transformer itself, like wise the package space as well.

[penfold]

[...]
I guess if the DC DC controller is a current based controller then depending on its control strategy, with different pannels receiving different solar insolations, that may automatically result in different frequencies and duty cycles for the boost converter?
[...]

That is definately an option, if the control is variable frequency you will get a natural variation that'll probably stop any problems. EMC and noise stuff can get a bit accedemic at the best of times, lots of things that'll help and hinder in various cases, beats are just a specific thing that can be a problem, and worth being aware of choices that'll make eliminating it more or less easy.

I'm glad you mentioned current mode control, because, yeah, that's a bit important. I guess you may know that as you sweep from high to low current or high to low voltage from the PV source, the incrimental impedance changes a lot, and you'll need to work over quite a wide range if you're working with unknown panels. That can be a big problem for converter stability (Middlebrook criterion) and it is a big enough variation that trying to neutralise the probelm with input (as you might with a conventional voltage source having some impedance variation) isn't really practical or cost effective, so better to handing the stability problem functionally and ooptmise the filter for the EMC purposes.

For the single point that the individual converters feed, there will be some stability concerns there too. The response of each converter will be dependent on its PV source and the output voltage; the output voltage (input voltage to the inverter) is defined by the contributions of all other sources and how much power is being sucked out by the inverter. That's a bit more than I can come up with a solution in my head right now, needs a diagram, I'll check back later when I have a bit more time.