Heater Control Advice / General Science and Electronics / Forums

Forums

4hv.org :: Forums :: General Science and Electronics

« Previous topic | Next topic »

Heater Control Advice

1 2 next

Move Thread

LAN_403

thedatastream

Fri Jul 24 2009, 11:54AM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Judging by this thread (pan and tilt)

there are a number of people here who know stuff about control loops, PID and the like.
As a follow on from this thread (thermistor questions)

I want to ask for advice regarding a solution for a project I'm working on.

I'm designing a homebrew temperature controller for my brother who does lots of homebrewing (beer, mead, etc.) to help him keep his yeast at the right temperature. I'm using an Arduino

to PWM control 5 individual heaters with feedback via thermistors (or possibly LM35s) being sensed by the analog inputs. There's also a multiplexed LED display for showing the temperature and some buttons to set temperature.

Set up of heater and fluid is like this

My question regards the method of controlling the heater power to achieve a stable temperature. I can only heat and not cool and I don't want any overshoot as too high a temperature will start to kill off the yeast. Whilst I'm going to put a "safety feature" in that cuts the heater power if the temperature rises too much I don't think this would be suitable for controlling within a degree of the set point.

My options are (I think):

> on/off controller with hysterisis
> PID controller
> Custom

On/off control
Normal thermostat style, turn heater on max power until temperature reached.
This is my favoured method due to it's simplicity
+ very simple to implement
- risk of overshoot
- how to deal with thermal capacitiy of heating element (is this an issue)? turn off before set point?

PID
PID functions for the Arduino do exist (

but these use floats and probably take a while to compute. On the other hand, my temperature is changing so slowly that shouldn't be a problem! As long as I multiplex my LED display from an timer interrupt it could take as long as it needs to compute temperature.
+ Once tuned should work well with no overshoot
- Difficult to implement and tune

Custom
Since the parameters of the system are reasonably well defined (specific heat capacity of the fluid, heater power) I could come up with a control scheme based on these parameters.
+ could be made fairly simple
- using system in non-ideal conditions wouldn't work properly

Your advice and experience is welcomed
Thanks
James

Nik

Fri Jul 24 2009, 07:04PM

Registered Member #53 Joined: Thu Feb 09 2006, 04:31AM
Location: Ontario, Canada
Posts: 638

With a bit of work you could adapt a regular digital house thermostat to do this. Just extend its own temperature sensor into the jug and use the heating-enable signal to close a relay that turns on your heater. You can adjust the "lag/delay" in most units and if you do a trial run with just water you can determine if you need to set the thermostat down by a degree or 2. It will be 10x faster then building your own unit and fairly easy to operate.

Parts: Fancy digital/programmable thermostat
24v XFMR
24v coil 120v contact relay
2 wires to move the thermistor/temp sensor from inside the thermostat to inside the jug.
Box to put it in so its not a mess

MinorityCarrier

Fri Jul 24 2009, 09:22PM

Registered Member #2123 Joined: Sat May 16 2009, 03:10AM
Location: Bend, Oregon
Posts: 312

PID is actually a very good method for heating a liquid. There are many thermocouple type digital PID controllers out there, Omron is one mfg that comes to mind, who has a very large selection of economical temp controller types. These usually control to within 1 deg. C. Many PID temp controllers can be 'trained' to their load to minimize temperature overshoot.

If you do not need tight tolerances, then more relaxed designs can be used, a mechanical thermostat (large hysterisis) or a thermistor/comparitor IC-driven relay, where hysterisis can be determined.

If it is practical, put a teflon-encapsulated stir-bar magnet in the vessel and use an external magnetic stirrer to constantly stir the liquid. You can minimize temperature overshoot and thermal stratification of the liquid that way.

mikeselectricstuff

Fri Jul 24 2009, 10:21PM

Registered Member #311 Joined: Sun Mar 12 2006, 08:28PM
Location:
Posts: 253

There is no need for floats for PID as the values will not span a huge range. Integer maths with binary fractions stored in longs (e.g. 8:24 bits) will easily give enough resolution.

aonomus

Fri Jul 24 2009, 10:33PM

Registered Member #1497 Joined: Thu May 22 2008, 05:24AM
Location: Toronto, Ontario, Canada
Posts: 801

Adding the stirrer will even distribution of heating into the vessel, and prevent the heater (if at maximum power) from overheating a localized area and killing the yeast.

Steve Conner

Sat Jul 25 2009, 08:07AM

Registered Member #30 Joined: Fri Feb 03 2006, 10:52AM
Location: Glasgow, Scotland
Posts: 6706

mikeselectricstuff wrote ...

There is no need for floats for PID as the values will not span a huge range. Integer maths with binary fractions stored in longs (e.g. 8:24 bits) will easily give enough resolution.

This is true, I've done a PID controller on a PIC using just 8x8 multiplies and 16-bit state variables. The only problem is that you need to think harder when designing it, than if you just mindlessly used floats everywhere.

Computing power is now so cheap that it often makes economic sense to waste power in this way, because it gets the development done quicker and makes the code more maintainable.

thedatastream

Sat Jul 25 2009, 09:21PM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Hey guys, thanks for all the replies.

First off, I already have all the hardware built, I just need the software writing to drive it. Thanks for all the alternate ideas though. I'm not going to leave my Arduino board in there, rather I'm going to replace it with an ATmega168 (or maybe an 88) in an IC socket. Should save a few bob! Plus most of the parts I used were junk box or free issue so it didn't cost that much really.

Shortly after posting the start to this thread, I knocked up a simple on/off controller with the heater in free air on my desk and a 0.25 degC hysterisis. It overshot by 6 degC on the first warmup from room temperature but wasn't too bad after that; I need to put it into water to see what it does then.

If it doesn't work, I'm going to code my own PID functions rather than use a library as this will be the best way to learn!

Also, we have some thermal simulation software at work, so I might press that into action to see if it can be made to model a heater suspended in fluid. It would be interesting to see the temperature differential present in the water, how much convection there was and if a magnetic stirrer would be worth it.

Also I don't know if the brewing my brother does creates sediment (I suspect the answer would be yes) so that would be a definite no vote for a stirrer.

Happy weekend
James

Dennis Rogers

Sun Jul 26 2009, 02:30AM

Registered Member #1837 Joined: Tue Dec 02 2008, 02:20PM
Location: NYC
Posts: 65

Maybe you should put a regular thermostat in series with your high tech solution as a back up. That way you have best of both worlds. Redundancy is next to Godliness.

thedatastream

Sun Jul 26 2009, 07:47AM

Registered Member #505 Joined: Sun Nov 19 2006, 06:42PM
Location: Yorkshire!
Posts: 329

Dennis Rogers wrote ...

Maybe you should put a regular thermostat in series with your high tech solution as a back up. That way you have best of both worlds. Redundancy is next to Godliness.

Hi Dennis,

There's going to be a safety feature in the software (is that an oxymoron?) that cuts the heater power if the control loop allows it to overshoot

Thanks
JAmes

tobias

Mon Jul 27 2009, 01:02AM

Registered Member #1956 Joined: Wed Feb 04 2009, 01:22PM
Location: Jersey City
Posts: 172

James

You have a nice project coming up!

My advice: You should use PID. Get four pots on arduino ADs and name them "SetPoint", "Kp", "Kd" and "Ki". Then write a simple code using the AD's values plus the temperature sensing of the LM35 to calculate the PWM.

If you are woried about not getting enough amplitude of the LM35 (1V) you can put a opamp between the that IC and the controling for a 5x gain or so...

Then google "Zieglerâ€“Nichols" and use that method to tune your system. It's a fun thing to do!! I did it before for a heating device too but I used a "opamp only" design. Yes, I used a lot of opamps! haha
Good luck out there!

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob