PLC?

Yup.

Not unless it's adaptive -- a plain jane PID algorithm isn't adaptive, but can be tuned to be more or less robust to changes in the characteristics of the system.

The ones that I know of aren't adaptive -- they're autotune, which means they'll stimulate the system and come up with a tuning on operator command, but they won't do it all the time.

I don't think things have changed much -- you don't see much hoopla about adaptive control anymore, and the definitive texts that I have read include cautions about when adaptive control won't work (like, blind).

The more I think about this the more I think you want the roll-up curtain to roll down over a tapered hole, so that when you're closed you don't get as much area change as when you're open -- because a given area change will be much more effective when it's cold and the curtains are closed up, and because once the curtains are mostly open a bit more isn't going to change much.

Just a thought, but get it working first.

Yabbut what's the difference between "High tunnel" and plain ordinary polytunnel greenhouses?

Expiring minds want to know :-)

Mark Rand RTFM

A very cheap contol system would be 2 cheap snap acting home thermostats - 1 mounted high and 1 mounted low, although you might have to monkey with the location. You can still get units that come with an available proportional band adjustment, not much but usually 2-6 deg.F. Set 1 for cooling and 1 for heating. They boot up from power failure every time the power returns.

Good luck Steve

Tim Wescott fired this volley in news:GOidnQJOys75J4PWnZ2dnUVZ snipped-for-privacy@web-ster.com:

I agree, except that many commercial PIDs now offer it as a user-selectable option. But you need to understand the difficulties it may impose.

LLoyd

If you're looking at the structure the difference is blurred. Few greenhouses use natural ventilation - large exhaust fans. most high tunnel have huge natural vent opening.

The main difference is application. Greenhouse grow plant for resale in some sort of medium normally on shelves. High tunnel grows fruit and vegetables in the soil.

Karl

It's been a while since I messed with my Basic Stamp but IIRC they used 8 bit integer math. My Basic Stamp 1 had only about 16 bytes ram, I'm not sure that would do a PID control, maybe. I think the BS2 has about twice that amount of ram. I'm not to great at working with integer math, especially 8 bit.

RogerN

If outdoor temp is higher than your minimum, if the temp ever got high enough to open the curtain it'd stay open until outdoor temp drops.

If outdoor temp is lower than your min but there's enough sun to heat the space to above max with curtain closed, then the system would cycle, alternately opening and closing.

If outdoor temp is lower than min and there's not enough sun to reach max temp, then the curtain would stay closed.

If you only partially move the curtain and then go to a delay timer for a while to see if that move drops temp below max, might it take unacceptably long to fully open on a high temp condition? The time constant will be quite different on calm days vs windy days.

A PID controller with a temperature sensor would provide better temperature stability, but at the expense of battery life because the curtain would be making small moves very often. You'd need curtain position feedback too.

Specs are on their website. You'll need drive transistors to operate relays. Someone suggested TIP120, a good suggestion. Holler if you want some help with that.

You can't know that because it will change significantly with cloud cover, sun angle, wind, etc.

Ouch. That's pretty restrictive. You'd need at least 16 bits to do PID in this sort of thing, and mo is better.

'splain why?

The timing is closer to hours than milliseconds. A real time interrupt (external 555?) for the long delay?

The AtoD function can be external hardware, or an external mux if an internal AtoD is there. One bit to select from two sources, two bits for 4.

Motor control is one bit for direction, one for Go. An eight bit count for run time to position the vent.

Too much bloatware in the world today.

What we need are more Assembly Programmers!

Are you volunteering to assist?

Well, maybe...

If I got a "development system" for christmas?

:)

I was wondering if he could use a string pot on the curtain for position feedback? I've never eBayed for them but I don't think they are very cheap new.

RogerN

I would think a simple timing loop would do it. Run the device and see how long it takes to go from one end to the other. Divide that by 256 for max resolution, or more likely by 16 - plenty of resolution for a vent.

I would like to see limit switches at each end just for safety sake. That would take up two bits of input per vent. But if the device has comfortable "over-run" areas, it might not be needed.

It ain't hard, it just ain't cheap. I used a double reduction Sumitomo cycloid drive of 3400:1 to drive the continuous filter I invented. Web movement maximum was 60"/hour, min was close to zero. I'd mark the web, go have a cup of coffee, come back and stick a ruler on it to make sure it was working.

Pete Keillor

I talked this project over with my crop consultant today. He said I should apply for a research grant and thinks I'd probably win. In that case, this device needs better EE than I can do. A ways down the road, but would you like to help design phase two?

Karl

Sure Karl,

You get some grant money for research and development, I'm your geek.

It would be fun. I haven't done a control project in ages.

I have a little CMOS Z80 controller board worked up that should run on about 25 milliamps (not counting off board I/O interfaces, of course).

Personally, I'd go that way and just write in Z80 Assembly, or Tiny C, or as fast as this thing would run maybe even BASIC. Once it's up and running, anybody could then tinker with the software quite easily.

It would be a little overkill, but if I'm writing code, I'd prefer to do it on something I already know.

Fewer surprises.

And, like Roger pointed out, the little controllers are, well, little(!)

Richard