Hacker News Reader

Comment by remipch | Hacker News Reader

Thank you very much for your kind and detailed reply!

Indeed, the closed-loop control was the initial idea which convinced me that it would be possible to build the mechanical parts by hand with common tools. In other words, the software "smartness" compensates for the mechanical "ugliness".

Another initial idea was to do multi-panels (several orientable panels) with a single camera looking at the target. Indeed, it's not easy, so I finally went back and decided to finish and release something with a single panel.

Nevertheless, I have some ideas to do multi-panels with a few more cameras. I would like to work on them in the near future.

Thank you for all the references, I will spend time to explore them.

There is also a company that uses vacuum to adjust the mirror shape, I'll try to find it and post it here.

I wanted to emphasize the inherent risk because my project is not a finished product, but a work-in-progress/proof-of-concept.

View on HN

FYI, this is the "vacuum adjustable focus mirror" I mentioned:

https://lm.solar/order/4-square-rigid-aluminum-composite-mir...

Yeah exactly!

I like your cable-drive concept by the way. Did you describe it in more details anywhere? Heliogen were also developing a cable-drive system for their commercial heliostats, but I don't know if they are still working on it.

How are you getting the right mirror orientation for each mirror (aka canting)? Custom spacers for each mirror?

One trick for closed-loop control with many heliostats/panels is to have a few cameras surrounding the receiver. When they look back at the mirrors, they will see the circumsolar radiation (how the sky gets brighter as you get closer to the sun). By comparing the brightness of the sky at different cameras, you can estimate which cameras is "closest" to seeing the real sun, and get an estimate for the real position of the sun.

> I like your cable-drive concept by the way. Did you describe it in more details anywhere?

I didn't take the time to describe the cable-bot concept in detail.

I modeled almost all the mechanical parts with OpenSCAD, but I struggled to model the cable itself.

In the "mechanics" README [0] you can click on any image to view it in an online 3D viewer.

The following note in the same README tries to explain how the cable is used :

    Each cable is actually wrapped around the motor axis, then passed through the pulley and tied to a fixed ring in the corner of the panel.

English is not my first language, is this sentence clear enough ?

You can see these elements in the main 3D viewer [1]

> How are you getting the right mirror orientation for each mirror (aka canting)?

I use one bolt that pulls the mirror holder in the center and 3 bolts that push it in the corners.

By screwing or unscrewing the corner bolts you can precisely orient each mirror independently.

The "panel_board_exploded" view tries to show this [2]

> One trick for closed-loop control with many heliostats/panels is to have a few cameras surrounding the receiver.

Super clever, thanks for the explanation!

I think it might be tricky to calibrate.

[0] https://github.com/remipch/solar_concentrator/blob/master/me...

[1] https://remipch.github.io/solar_concentrator/view_3d.html?mo...

[2] https://remipch.github.io/solar_concentrator/view_3d.html?mo...

> Each cable is actually wrapped around the motor axis, then passed through the pulley and tied to a fixed ring in the corner of the panel.

Thanks, when seeing the video again now it makes sense! I didn't catch the counterweight the first time I saw it. Nice! In the Heliogen concept I mentioned previously they got around having to use a counterweight by attaching the other side of the cable to another part of the panel, such that the cable length stays approximately constant. Then they used a spring to compensate for the small changes in cable length that are inherent to the geometry.

> I use one bolt that pulls the mirror holder in the center and 3 bolts that push it in the corners. By screwing or unscrewing the corner bolts you can precisely orient each mirror independently.

Nice! Even in large heliostat fields it is often done in a similar way. It becomes quite labor intensive when you have thousands of heliostats in a field, with 10+ segments each, so there are ongoing efforts to find ways to do it automatically or to get around the need for doing it in the first place.

I wonder if you could set up 3 multi panels with a known pattern of red green or blue gels on each, then filter the camera image by color to identify which of the panels is pointed where so you can adjust them individually?

With the low-cost camera used here, the small red/green/blue pattern would be indistinguishable because the whole spot is saturated.

It might be possible with an expensive camera.

Reply on news.ycombinator.com