What dust sensor do you recommend for detecting dirt on a solar panel surface?

Hello everyone,

I’m working on a project for automated cleaning of solar panels, and I’m looking for a reliable sensor to detect dirt (dust, sand, etc.) directly on the panel surface.

I’ve looked into sensors like the GP2Y1010AU0F and DSM501A, but these are designed for detecting airborne particles.
What I need is a way to detect dust accumulation on the panel itself, not in the air.

Do you have any recommendations for:

  • A sensor suitable for surface dirt detection?
  • Or an alternative technical approach to determine if a panel is dirty?

Thanks in advance for your ideas and suggestions!

Knowing nothing about the material properties of solar panels, so just a wild guess, but maybe you could test how reflective the surface is - maybe by measuring the amount IR is reflected back using a IR Tx/Rx pair.

Or maybe just clean the panels regularly regardless.

2 Likes

How does the dust particles travel?

Just as @markd833 I find it difficult to think of any sensors for this, and would recommend regular cleaning. But there might be one way of knowing actually, a reference PV panel that is cleaned every day, week or so often deemed necessary.

1 Like

It also occurred to me that - given the lack of information from the OP - that the surface of the panel may need to be scanned rather than just a simple point check. That would involve 2 axis mechanical stuff, motors, electrics etc per panel. Would all that extra equipment take more power from the solar panels and kinda defeat the purpose.

I've no off-grid knowledge - other than watching too many episodes of Homestead Rescue - but others here may have first hand experience of this sort of thing.

Or maybe you are wanting to carry out laboratory testing and then the scenario is completely different.....

1 Like

Yes, tests now and then it will be in real situations.

Yes, I also thought about using photoresistors to check the reflection of the panels once illuminated, but I can't yet put this device in place.

If the panel is dirty (does not get the full sun) - you’ll produce less

So What about using the output of the panel as the sensor ? If the panel delivers less than expected for some amount of days, then it’s clean up time.

You might trigger a unnecessary clean up if it’s rainy for days but overall might work.

1 Like

It wouldn't be difficult to make a "sensor" with a visible light source combined with light sensor measuring light through a flat transparent "reference" glass or measuring scattering on the same side of that glass. It would give you good estimate how dirty that reference glass is.
The problem is that also rain, fog, condensed humidity and frost changes the glass transparency, not just dust.

Just use a couple or LDRs. They will get as dirty as the panels. So their output is directly related to amount of dirt.

1 Like

@arieloss
You might think about having your reference LDR arranged so that you could protect it from contamination and just uncover it when you wanted to make a measurement. That plus an LDR that was positioned to get the same contamination as the panel would make for a pretty reliable system.

2 Likes

Solar production forecasts are available. Google a bit - you might be able to find one that's reliable enough over several days to tell you that your panels need a clean.

Can't be a new problem either - what have other folks done?

I only clean my panel when it has been snow covered for more than a day.

1 Like

And how do you use them to estimate the dirt accumulation caused effect of your panel production?

Maybe use a particle counter as a proxy?

That is measuring very small particle in the air (0.3, 0.5, 1.0, 2.5, 5.0, and 10 microns.)

This is more for pollution detection.

I was thinking that you could compare the forecast production with actuals (by talking to the MPPT or whatever controllers), keep records about how the two compare and when the actuals consistently lag the forecast, call for a clean.

I was suggesting a concept more than a specific model of equipment.

I understand your point, but forecast doesn't count clouds, (mis)orientation of panels, shadows from trees etc.
Those not counted variables have more effect than dust.
Ok, orientation of panels and fixed shadows could be calculated to match the forecast. But local weather with clouds would be more difficult.
With very clever algorithm it would be possible though I expect.

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.