[BullCreek]

Hello all,

After finding this site about a month ago, I'm definitely itching to build a CNC machine, but before I can do that, I need power in my new shop. I'm installing solar panels, and while initially, I'll probably just statically mount them on poles, the dual axis trackers you can buy for big bucks actually do make quite alot more power by semi-intelligently tracking the sun. I'm thinking they would be pretty easy to build yourself and control using CNC type servo/stepper motors and software.

My newbie question regards planning for the future with the wiring. I'm running underground conduit from the PV combiner box and PV poles back to the shop in the next couple of days, and would like to include enough wire to control the two axis motors on each pole at some point in the future when I have time/funds/understanding to add them. In addition to the wiring for the PVs, what might be needed going to each pole? Would running just one additional #10 hot wire for 48v power to the motors, and then maybe two cat5 wires (one for each axis) for control be sufficient?

I don't have the advantage of knowing alot about how the control circuitry for servos or steppers works - but I'm hoping it isn't more than 8 wires and the currents involved in control are small enough that the 8 #24 wires in cat5 would work. Would this work?

[ViperTX]

Each servo would have 3 wires for power and direction control and you'll need 4 wires for each servo encoder (A, B, +5, and GND). You'll also need 4 wires for the cell which will determine where the sun is at.

[BullCreek]

>Each servo would have 3 wires for power and direction control

The 72v PV array and 48v battery bank share a common negative - so I figured I wouldn't need an extra wire for -, just for + from the batteries. What voltage and amperage is the third "direction control" wire?

>and you'll need 4 wires for each servo encoder (A, B, +5, and GND).

The encoder is what tells the servo where it is right? I posted this in servos instead of steppers because from what I've read, servos are generally more powerful and while the vertical axis shouldn't take alot of power to move, tilting the array up and down along the horizontal axis might. Do you think cat5 would work for this over 125'? If the current is low enough maybe the voltage drop wouldn't be a problem.

Relatedly, since steppers seem to be cheaper and from what I gather don't involve an encoder, how many wires for them since one of those might work for twisting the vertical axis? In either case, I guess I should probably also include limit switches on both axis just in case - which probably also takes 1 wire back to the controller no?

>You'll also need 4 wires for the cell which will determine where the sun is at

Rather than use a sensor, I was just thinking I'd use software to drive each unit in parallel based on my latitude, date, and time of day. The software might also have a home function, and maybe a function to tilt the array to maximum for knocking off snow in the winter. From what I have read, the sensors tend to cause more problems than anything else on the commercial units.

Sorry to be such a newb - your help is greatly appreciated.

P.S. I used to live in Austin. I sure miss Gumbos and Rudys.

[Geof]

When you are sizing servos, wire, etc., you should consider that you only need small motors running through a large gear reduction to drive your motion. By typical CNC standards your axis motion is snail-like; you want to track less than 180 degrees in 12 hours or so. Also you do not need a rapid return because you have all night to get back to the dawn position. Even your tilt for dumping snow does not need to be fast because snow does not accumulate instantly without warning.

Have you thought about the mechanical details of the drive? I suggest you do not do something that is small and close to the axis of rotation because you have to worry about high torque loads. Design something that allows you to put the motor drive at the periphery of a half circle comparable in size to the dimension of you PV array.

[BullCreek]

Geof,

Very good points about speed - and using gear reduction rather than powerful motors. Modest steppers might work for both cases given sufficient gearing. I haven't given alot of thought to mechanics other than looking at what the existing solutions do (www.wattsun.com for example). Seems like a worm drive for the vertical axis, and a stainless acme screw or some such for the horizontal might do the trick.

This is unrelated, but since the movement on each of the 4 poles will be identical - I wonder if there are any controllers out there beefy enough to run say 4 steppers per controller? That would allow me to just have two controllers rather than eight.

I guess it's time to order a few cheap stepper motor off eBay, pull out my old BASIC Stamp protoboard, 100' of cat5, and see what I can or can't make work. I guess if the cat5 won't work, I can always pull larger wire although I'm wondering if the voltage drop on such a long run will cause any problems.

Thanks for the feedback. --J

[Evodyne]

BullCreek,

Hi! You can really do it all at the PV panels. All you need is to have DC to run the motors. Take a peek at Red Rock Energy: you can get sun trackers there that will run DC motors (not servos, not steppers, just plain-jane permanant magnet motors) to move the panel until pointed at the sun. If you don't like what he has, Google "sun trackers" or something similar: there are several other sites that sell units or plans. Also, because the dohicky needs to move so slow, you can get by with small motors with a good amount of gear reduction. Depending on your mount type (polar, azimuth-elevation, etc.) you might be able to use the linear dish positioners made for the old, big satellite dishes. They are weatherproof and strong and fairly cheap. SO...local sun tracker(s), simple DC motors, NO drives, NO computers, etc. Keep it simple! Please keep us informed as to how this all works for you.

Sincerely,

Evodyne

[ViperTX]

*chuckle*...you actually only need 2 wires to each motor....for tilt I would make it mechanical summer and winter.

yes you can also use just plain old DC motors, a simple H-bridge driver and a split sensor...that was my school project back in '76.

[Evodyne]

*chuckle*...you actually only need 2 wires to each motor....for tilt I would make it mechanical summer and winter.

True. But (and I'm really thinking this through) your committing to a polar mount to get this simplicity, right? I think so. But it's early and I've not had my coffee.

yes you can also use just plain old DC motors, a simple H-bridge driver and a split sensor...that was my school project back in '76.

Wow Viper, your dating yourself. Did you have to fend off the dinosaurs on your way to school? Listen to me: I'm not so young myself. He he he...

BullCreek-while your looking into this stuff, look at peak power trackers. A peak power tracker is an electronic device that sits between the solar panels and batteries. It monitors the array output and varies the load that the array sees to ensure maximum power transfer.

Lance

[BullCreek]

Thanks for the replies to all. I've learned alot just by reading more since posting.

Regarding MPPT - yes, I am doing that already using an Outback MX60 (which also has the advantage of allowing you to run your array at a higher voltage than your battery bank - 72 vs 48 in my case). Regarding the sensors from Redrok - I had come across them a while back from the otherpower links page but forgotten about them. I suspect their sensor is similar to the commercial ones and may suffer similar reliability problems but for $39 fully assembled, it sure can't hurt to try one or two on for a season.

I hear you on keep it simple - NO computer - but I'm going to have a dedicated low power PC logging wind and weather data as well as output from the Outback equipment so computer control seemed sensible - but maybe not. I suppose if I did proceed, there would probably be lots of kinks to work out - one being that the drivers probably don't have an idle mode - and suck up a fair amount of power unless disconnected - plus all the mechanical issues of actually building the tracker. Using one of the sensors to run a dish actuator on just one axis would certainly be cheaper and simpler and probably get me 90% of the effeciency gain I'm after.

Thanks again for your wisdom to all, --J

[j m]

Hi all,
BullCreek, Just out of curiosity what benefit you are going to get from using servos or steppers?
How are you planning to sense where the sun is, are you going to design some thing your self or buy something. If you are going to do it your self I think the idea of using the green LED’s is a really good one, they are very cheep and are surprising sensitive to sun light. I have seen The red rock site before and it is well worth a look, there is a nice picture of a model sterling engine being driven from a solar collector.

[Mariss Freimanis]

A dumb idea: How about a 6" by 6" piece of vertical sheetmetal oriented north to south on your array. Have two photo transistors, one the east side, the other west side of this barrier. If a shadow is cast on the east one, the motor would move the whole thing west until the shadow dissapears. The photo transistors would be sensed differentially, meaning only the difference between them would be detected. This would cancel the effects of clouds/sun since both detector outputs would decrease/increase the same amount.

Mariss

[ViperTX]

Mariss...that's exactly what I did on my school project....just use the error signal to turn on the motor....it's works beautifully.

[Geof]

Simple is good but not too simple. Your first posts mentioned two axis tracking on four collectors with home capability and extreme tilt for snow clearance; now you are simplifying down to a single axis using a sun tracking sensor. Try to make sure you do not rule out the possibility of upgrading to the fully capable system some time in the future. You can use linear actuators for both axes; a compound linkage can give you more than 180 degrees of motion on your panning axis and the first linear actuator for tilt can be a turnbuckle that you adjust a few times a year. A sensor for sun tracking, a bit of circuitry and you have the simple starting version. Just make sure your tilt axis has enough motion for the full vertical tilt for snow clearing or to put the collector facing down for hail protection even though you will not use this at first.

Sometime in the future install the linear actuator on the tilt; now you don't have to go out in cold weather to clear snow. Later on build some absolute linear position encoders to attach to the actuators so you don't need to worry about a bird pooping on your sensors. You do not need fine resolution; an 8 bit grey code encoder will give you a resolution of plus or minus 1 degree on the panning axis. Now all you need to develop is the control software to do your absolute position tracking based on date and time of day. The linear to angular motion conversion in your linkages introduces some non-linearity in the angular position but that is tolerable. You don't need a large capacity motor controller because you can have the PC adjust the eight axes individually. The only thing you have to include from the beginning is enough speed in your actuators so you can cycle through them like this.

If you do develop a system with evolving complexity it would be interesting to document the increasing collection efficiency. At the end you would be able to do a proper cost benefit analysis to see if the full automated system is truly justified.

[synthetiklone]

I hope Im not jumping in far too late, but the title interested me. I have designed and built a solar tracker before. I am not much help anymore, as the whole design was sold including manuals etc years ago. The thing I would like to mention to you here is that you dont require a complicated PC or microprocessor based system to track the sun. From memory, all I used was op-amps, a quad optic sensor array, some simple circuitry for homing. It was all powered at the panel, by the panel, and required no other maintenance. It used linear drives, no exposed moving parts. The design was very simple compared to system required to monitor a quadrature encoder. The positional feedback was from the sun! The system didnt really need to know where it was all the time. hope this helps anyone.

[NC Cams]

There's a way to simplify the servo drive system even more.

Do a web search for NE544 servo driver IC. It was used to take a PWM'd signal and turn it into motion in whatever direction need be for MANY, MANY R/C servos. WE drove R/C motors at ungodly high currents via fet drives that the NE544 actuated. Bidirectional drive with the IC is a snap.

EDIT
http://www.datasheetcatalog.com/data.../4/NE544.shtml

END EDIT
WIth the logic level high and low side fet drivers available today, in concert with the level fets that are now sold, you should be able to run a drive from a 12v car battery. Simply add some surplus car actuator motors (seats, windows, etc) and the appropriate gear reduction to a dual channel driver card and you should be able to point the collector almost anywhere.

It was one of the finer, easier servo IC's that was ever concieved. It is should still be available from by Ideal Semiconductor - they bought the die from Signetics/Phillips (the original designer and developer) and sell it as well as other "legacy"/"orphan technology" chips.

Contact them at:
www.idealsemiconductor.com/

The specification sheet from Ideal/Signetics shows the internal logic of the IC (downloadable from the net). If you study the internals closedly, you can figure out how a discreet component servo driver could be formulated.

Clearly, you don't need a PC to make this project work....

[BullCreek]

Thanks for the additional input. I just finished getting one set of panels up on a single axis manual system but I still want to go two axis at some point once I get a bigger mill with either DRO or CNC.

I can see the advantages and simplicity of using DC motors and simple electronic sensors to track the sun - but also like the additional features offered by a PC controlled system (primarily wind protect and possible hail protect). I already have the low power PC data logging performance for the system - as well as wind speed via an attached anemometer. I prototyped a two axis sytem using steppers - using a home made worm gear I made on my lathe with an acme tap. The gear turned out nice, but unfortunately, some of the other mechanicals didn't quite fit/work right, and I concluded the accurracy needed to make them were beyond my abilities with my current small 3-in-1 manual mill/lathe.

As a mostly software guy - I must say I have learned alot from reading this site over the last few months - and I have an incredible amount of respect for you guys that seem to be able to build anything. Hopefully I'll get there someday!

[fkaCarel]

I like the worm gear. If the pitch angle of the worm is < 3 degree it is considered self-braking, so there is no energy required to hold the system. Complete worm gears can be bought at reasonable prices. Off course "reasonable" is different for everyone. The idea of Geof to use 8 bit Gray code for absolute positioning is also good.

I want to throw into this the length of cable between the collector and the PC. Maybe even at this point, it would be a point to consider to "embed and command" the position. It would reduce the cabling to power and a serial line. Long cabling, for me that would be above 3-5 meter, can require "tuning" for the signals and still have hick-ups through interference and pickup. With a wisely designed communication protocol you can still address individual units.

[NC Cams]

If you use the NE544 or other similar devices, you can easily make a minimal parts count remote servo amp. All you would have to feed it would be a simple to generate analog voltage - easily done with a D-A IC and the parallel port. Or from foto cells and an op amp remotely.

The issues of RF 'tween the amp and the PC would be minimized via a slowly fluctuating analog signal and a 3 wire cable from the amp to the D-A/BOB on the PC - power for the servo drive could come from a solar cell and a batter.