4 Channel Thermostat using PID loops

Guys,

Over the past few months I've been working with DT on a project which I will now share with the PBP community. Darrel put together 99% of the code and routines (some of which have already been published on his website and here on the forum). My contribution was constructing the test equipment and running the real life trials and then tweaking the values to get the best accuracy for the equipment used.

Background

I keep reptiles, well snakes to be exact. These need to be maintained in a controlled environment, with a steady heat source at one end of their enclosure, which needs to be thermostatically controlled.

http://www.picbasic.co.uk/forum/atta...9&d=1266672273

Whilst there are already commercially available pulse proportional thermostats on the market, these are single units. I really wanted to use one unit that had 4 separate pulse proportional thermostats which would cut down on lots of wiring etc.

The original idea was to have 4 pots to set the target temperature for each vivarium with the temperature displayed on an LCD, however for ease of testing we opted for using hyper terminal and a simple RS232 connection.

Development

I have an EasyPIC5 development board, and to make life simple purchased a few of their add on modules such as the 1307 based real time clock, pot board, and several connection boards to connect the solderless breadboard to.

http://www.picbasic.co.uk/forum/atta...8&d=1266671836

To test the code and develop the project as we went along I constructed a box from 18mm melamine board, approx 400mm x 400mm x 200mm (LWH) and installed a 50w powerplate (a flat panel ceramic heater) to the roof inside the box. A 240v 4amp solid state relay was used as the means to turn the heater on and off, especially as the opto isolation could be driven directly from the PIC.

DS18B20's were used for the sensors, and one was connected via lengths of wire to the EasyPIC5 board. The sensor was placed inside the box which was placed on a flat surface. The SSR connected to the mains and it's input connected to the EasyPIC board too.

Darrel produced the 1st version of the PID code to drive just one channel. I then compiled and ran it to see how it performed. Over the course of several evenings I managed to get the values tweaked enough to maintain a stable temperature that would drift no more than 0.5 a degree in the course of the evening.

Armed with the feedback, DT then went on to work on multiple channel PID loops, and after a few weeks of frantic PM's between him and Henrik came back with some further code for me to test and try. The result was that after some further tweaking at my end, we managed to have 4 independent channels all running at the same time and in the test rig, maintaining a constant temperature.

Future Development

The plan was to now develop the fancy options such as using the clock module to program a night time drop in temperature if required, menu buttons, and possibly a means of monitoring / setting via software on a computer, even to storing temperature data in an eprom for use with the PC software. However this meant porting the code to a different chip and so Darrel suggested the 18F4550.

We got as far as having testing out the USB coms, and getting the PID loops running, however due to personal reasons we've had to halt development of this side of the project, at least for the present time.

I'm now going to concentrate on developing a prototype PCB based on the 16F877A code, which has achieved the main goal, which was 4 independent pulse proportional thermostats. I'll use this prototype in the two vivs that house my two Royal pythons, which was my original intentions. However if anyone would like to contribute and take this project further, possibly to its full potential then please feel free to drop me an e-mail or PM. I've attached the code and all the includes for use in MCS (PBP 2.60 is required).

In the meantime I would like thank Darrel for all his hard work and contribution to the project. It's guys like him that make this forum a pleasure to visit. Whilst I may not entirely understand the complicated maths that goes on in a project such as this, I've learned a lot by way of his examples. - Darrel, Thank you !

You're welcome Malcolm, and thank you too.

And I'll add a "Thank You" to HenrikOlssen, who wrote the Multi-Channel PID filter that is at the core of the program.
It allows any number of PID loops to be running concurrently, and it works really well.

Cheers,

Well, I didn't continue the development with the 16F877A... I ported the code to an 18F4580 which has a lot more code space instead :)

I've added the clock module and used Mel's example code for RTC and with her help and assistance with other forum members I'm now able to release a fully working beta version of this project.

Now need to do some further testing and then build the prototype PCB

Please note the header at the top of the main code. It details the history, development and contributors. If I've missed off anyone in the credits please accept my apologies

I've just drawn up the schematic in Eagle (yet un-proven as I've yet to prototype this project). I've attached the image here for reference.

Hi,

Your project is very interesting, thank you for sharing it. I am doing the same kind of work (also for BP snakes). I need to control radiant heat panels and heat pads. I use some DS18B20 as well, and right now I am at the power stage part. I have several questions:

1. You use a SSR for driving the heat channels. The ceramic heater elements are powered via main AC line right? I saw many thermostats circuits using an optotriac in combination with a triac. What is the difference between using SSR and a combination of an optotriac such as a MOC3022 and a triac? Does the SSR heat a lot?

2. What is the period of the heating cycle? In other words, for how long are you triggering the SSR to have for example 10%, 50% and 90% power?

Thank you very much.

Denis.

Hi,

interesting to see you are working on a similar project.

http://www.opto22.com/images/products/MP240D4_p_200.png

The SSR's I'm using are Opto's MP240D4 which are rated at 4 amp - 240V, and in the test I was driving a 50 watt ceramic power plate, and the relay wasn't hot at all, using Ohms law a 200w heater (typical rating for a ceramic bulb heater used in vivariums) will pull 0.8A at 240v, and that would be driven 100%, so they wouldn't need any thermo bonding to a heat sink. If you were running 1Kw heaters then I guess you would need to bond them to a large heatsink.

The reason I used them is because I had them, they can be driven directly from the PIC, and means you don't have to mess around with lots of isolators, FETs etc etc. The downside is they can be expensive $15 / £15 - although I paid around £2 each for mine a few years back from an e-bay listing (£10 inc shipping for NIB :) ). However you could use an opto isolator and triac / thyristor combo to drive the heaters

To answer your second question, there is no set % the heat is applied. The PID loop works by reading the temperature then through Henriks complicated maths routines works out how long to apply the power to maintain the temperature. Thus as it gets near the set point the pulse gets shorter and shorter. Ideally it works out the ideal pulse length to maintain that temperature, but in reality, due to the nature of ceramics, it often over shoots a degree or so for the first time, and then settles down. My test vivaruim consisted of a box 500mm x 500mm x 300mm (l x w x h ) made from 18mm wood, and I was able to maintain the temperature to withing 0.5C of the set point, which is ideal for boa's and pythons (and most tropical reptiles).

I've also added a couple of routines to a further version of the code that uses the RTC to turn on and off the two lights I have in the vivs. I'm still trying to work out the best way to prototype the board as it's proving difficult to produce a single sided PCB and my skills at producing a double sided PCB are lacking. I guess I'll start by using stripboard and see how it performs.

Hi,

Thanks for the answer. Does the SSR needs extra circuitry such as RC filters as I have seen in some of the MOC322/triac stages?

You are talking about pulses getting shorter and shorter as the temperature reaches the set point, right? Is is a periodic signal? If yes, then what would be the period of your signal? Also what would be the lenght of the shortest pulse?

For my project I figured out a period of 1 or 2 seconds, the sortest pulse would be 20ms (1s period) or 40ms (2s period), corresponding to 5% of the period. That's where my % came from. Trigerring the heat source during 20ms over 1s correponds to 5% power applied.

Denis.

Hi Denis,

I think you are missing the point of the PID routine. There is no set period that the pulse is on / off for.

Lets say you make the set point 32 deg C, and the current temperature is 25 deg C. You switch on the thermostat and the PIC reads the current data from the sensor, then using the PID routines works out that it's way below the set point and turns the output on 100%. The heater thus starts to get hot and the increasing temperature is monitored by the sensor until it is within a few degrees of the set point at which time the maths in the PID routine calculates the percentage period to turn off the heater. The frequency at which the PID routine runs it's cycle is around half a second. Eventually as the temperature reaches the set point the PID should be just turning on the heater for the shortest of pulses, and if it is over the set point then the output remains off until the heater cools down and the temperature falls below the set point, at which the pulses increase to maintain the temperature.

This method doesn't use any paused pauses etc as you describe, and is by far the safest method as the actual temperature is controlling the feedback loop and it's almost impossible for the drive to lock hard on and over heat the vivarium. If the DS sensor develops a fault or becomes disconnected the code makes the pin low so there is more risk of the vivaruim cooling than overheating, which most reptiles can tolerate better. Having said that it would be a simple matter to have a buzzer or some other audible warning if the temperature exceeded 5C above or 15 degree C below the set point for example.

The SSR requires no additional circuitry, connect one input pin to GND the other to the port pin, then I connect the neutral wire from the mains to one of the output pins, and then the neutral from the heater to the other output pin - the data sheet can be found here http://www.opto22.com/documents/0859...data_sheet.pdf

Hope that helps

Hi Malcolm,
I don't think Denis is missing anything. What you are doing is some form of PWM, probably at a reasonably low frequency. 100% means the output is on all the time 0% means it's off all the time. At 50% it's on half of the time and off for the other half.

The question Denis is asking is how long in terms of milliseconds, seconds, hours or days is one period of this PWM signal.

On slow responding systems or "loads" like your heater for example one period can be long, like several seconds or even longer while on fast responding systems like regulating the current thru a motor or whatever the period needs to be MUCH shorter.

A 50% duty cycle or output means the output is on 50% of the time but in the first case, with the heater, it means it's on for 5 seconds and off for 5 seconds while in the later case with the motor it means it's on 5us and off for 5us. Still 50% in both cases but in the first the period is 10 seconds, in the later the period is 10us.


Really it's all a matter of how much "jitter" you can live with on the resulting response from whatever it is you're driving. There's usually no need to use 20kHz PWM if regulating the temperature in an oven where applying "100% power" for one minute will result in a temperature rise of 0.5° - as an example.

Sorry for the long winded response.

/Henrik.

Henrik,

I think we were / are describing the same thing, IE a low frequency PWM signal. In that the frequency of the cycle is approx half a second, with a varying pules width between 100% and 0% depending on the closeness the current temperature is to the set point.

I thought Denis was simply running a loop with a set mark / space ratio for a given cycle period. If I thought wrong then I apologize

Well a few months have passed and I'm able to get back to this project.

I used Eagle to design a PCB, but every time I ran the build it required a double layer PCB and my DIY PCB skills resulted in several coasters ! I've just found one of many stripboard design applications and have come up with the attached. Hopefully it won't need modding too much, and will either update the image when completed and tested, or confirm that I've got it right first time (be a miricle :) )

I'll also take some pictures of the unit as it's put together and will add them to this thread as I progress.

Note that before anyone jumps in and states that I need to break tracks around the relays, due to the lack of being able to design a custom component, I had to compromise, the real relay doesn't have all the pins on the package as shown

Well due to work and other personal commitments I've been too busy to turn the drawing into a practical prototype... until today.

I called into Maplin and purchased some stripboard and a suitable box this morning and then spent an enjoyable afternoon building the board. I've not fully tested it yet, bit on powering up the EasyPIC board and taking a 5v supply and using a test LED, the pins to the LCD flashed the same way as they did on the development board (and there was no smoke :) )

Wife has plans to get me to sort out the garden tomorrow... but hopefully I'll find time to build the PSU board, and LCD and test this further before hooking up the mains to the relays. For convenience I'll probably take the sensors straight in through the back panel rather than rig up some connectors... we'll see.

It's taken me six months since DT (and Henrik) assisted me with developing the code to get this far. With out their help it would of taken me a lot longer. I'm determined to see this through to the end - I feel I owe it to you guys to prove the project works in the real world :)

Hi Malcom,

A truly nice project Sir. May I ask if you've developed menu and display subsystems and would you have display examples you might share? Just curious to see how your user interface works.

Cheerful regards, Mike

Hi Mike,

The menu's as they stand are fairly basic. If the option button is pressed the main menu is displayed.

Code:

---


LCDOUT $FE,2,"Main Menu"
lcdout $FE, $C0, "Select Option"

IF Option = 1 THEN lcdout $FE, $D4, "Set Time and Date  "
if option = 2 then lcdout $FE, $D4, "Set Night Period  "
if option = 3 then lcdout $FE, $D4, "Set Normal Temps  "
if option = 4 then lcdout $FE, $D4, "Set Lighting Period"
if option = 5 then lcdout $FE, $D4, "Run                "

---

Depending what state you leave the option button set to the LCD display reads

Code:

---


Main Menu
Select Option

Set Time and Date

---

The normal temp sub menu simply displays the A/D from the four pots.

The sub menu for the lighting period simply shows the time which gets changed by the +/- buttons so you can set the on time and then repeated for the off time (for each viv).

If I get chance I'll take some photo's of each menu and post them up at a later stage.

Guys,

Normally when strip boarding I build as I build up the board from the schematic as I go along, and often test each section, ie get the LCD working, then the sensors, then the switches, then the RTC etc. However this time I used an application to design the board and having followed it found that the board contains an error somewhere as the LCD doesn't work. I'm no longer able to edit the post to remove the image so..

DO NOT USE THAT IMAGE

Sounds like one more reason not to use simulators and such.

I always thought, WOW!!!
Those folks that write simulators must really be good. Able to anticipate all variables and all....
Yeah right....

Hi Malcom,

Thank you for the brief description of the user interface. It sounds very easy and very intuitive.

A couple more questions, please? (1) Do you "lock out" the pots' to prevent someone from accidentally bumping them and changing the settings? (2) Could you use a single pot' to adjust the four zones, one zone at a time?

Cheerful regards, Mike

Hi Mike,

In the current format I use 4 pots, one for each channel, and having selected the manual temp setting option from the menu can set all 4 channels to whatever temps I need which are picked up when exiting out of the menu and back to the live screen. You could re-write the code to use one pot and cycle through each channel if you wished, however I preferred to use 4 pots.

No the pots are not locked as such, but you do need to select the menu option to adjust them. Once they have been set and the menu returns to the live screen you can twiddle the knobs as much as you like and it won't change the values - this does prevent accidental changing of the set temperatures.

I'm still working on taking the project from the development board to a working prototype, and once I've done that and had it running for a while I'll update the project with the final release of the code and some project notes.

Well I've just completed the two boards in the MKII prototype, and was pleased to find that when I plugged the mains lead in there was no bang, no smoke but the thing actually powered up and displayed temperatures on all 4 inputs with the LEDs flashing away indicating the outputs were also running.

All that's needed now is to make up the sensor leads properly, fit the correct 3 core cable for the supply and hook up the output board to the controller via 7 way cable - then install it and monitor how it performs...

I'll post up a few more photo's as the installation progresses.

http://www.picbasic.co.uk/forum/atta...0&d=1279223804

Well, seven months after the initial post, and with the welcome assistance from DT to develop the code, the hardware has finally been completed and tested. All that remains is to remove the old commercial ON/OFF stats and hook the heaters and light up to this new unit.

I might look at having some PCB's made up as I've been asked if I could build a unit by other reptile keepers... and who knows, there may even be a market for such a device....

http://www.picbasic.co.uk/forum/atta...2&d=1279817438

Nice! Wish you good luck with the new bussiness field!

Ioannis

Hi Malc long time no speaky hehehe....

Love the project! very nice bit of kit you guys have developed there!
Been updating my bits and bobs so that I can start developing some bits for my small reef tank etc and cant wait as along with the usual temp reader, im also looking at adding water level detectors (going to try sonic sensors as got a few laying about) and when funds allow, will be having a go at using the larger powered LED's to make a new light unit where i can control dawn to dusk etc (thing with LED's are the large amounts of them you need for penetration/cover on a reef tank)

If I can get anything that remotely resembles the quality bit of kit you created i'll be a very happy chappy :)

Mark.

Hi mate,

I'm sure if you could encapsulate the DS18B20's in some form of heat shrink boot you could use this project to control heaters in up to 4 tanks, and with the DS1307 and a bit of adaptation program seasons with shortening / lengthening days to switch in those lights.

The version of the code I'm now working on is so far advanced than that listed here, but what is listed can be used as a base if you wish.

Might have a look through the code as I'm rather rusty lol. But hoping to write most of it myself to dust off the cobwebs and it's also a good way to fully understand what's going on etc. But might end up still asking a few q's etc which is why this forum is cool as so many pol willing to help etc.

Well I received 6 PCB's that I had made by PCBWing in Hong Kong, which I must say offer a fantastic quick service even on the "economy" option. I chose the 7 day turn-around, and the boards were received around 12 days after placing the order.

Here's the 1st board, not quite completed, but you can see the quality of the PCB. I'm hoping to get the first "production" unit complete in the next couple of weeks, funds permitting. I've already had people on the reptile forum dropping me PM's with offers to purchase it... - Just need to resolve the coms issue in the firmware first !

http://micro-heli.co.uk/pcb4.jpg

Excelent Work! Congratulations!

Great post! It must be cool to have a history showing your project going from start to finish. Just curious, did you end up using the Triacs and Optoisolators for cost, size or both? I have a HV control project and I am considering the difference between the two. Thanks.

Hi,

Yes I opted for the isolators and triacs for both cost and size.

Has the incPID_mc.pbp file been modified?
I can see several commented lines.
Has Henrik posted this file somewhere on the forum?

As far as I'm aware there was no modifications since the original one posted at the start of the thread. Here's the version I have used in all subsequent builds (change the TXT ext to pbp)

Thought I should update this thread with some more images of the finished MKII prototype which has been running now for just over a year now without any issues.

http://micro-heli.co.uk/MK2%20back.jpg

From the back, and

http://micro-heli.co.uk/MK2%20front.jpg

From the front.

The RS232 connector on the board is obsolete, and is connected to a panel mounted socket on the rear. I was thinking of adding one of those new 232 > USB TTL cables to make connectivity easier, but to be honest I've only ever needed to adjust the time from GMT to BST and back, which was easy from the unit rather than from the PC, so haven't bothered.

To many people like me, the result is almost a work of art (grin). I'd love to see some detail on how you mounted the LCD to the front panel...

Happy Holidays.

Cheerful regards, Mike

Over the years I've received several requests regarding using this code for a commercial product. Whilst I had lots of request there were two deciding factors why it was not pursued. The first was the the fact that others had contributed to the development of the code, namely Henrik and Darrel, and, at the time of development it was for my own private use with no financial gain. It may well be that they would have no issue if the unit was made and sold as a hobby rather than business, but I respect that without formal signed agreements it would be the wrong thing to do. The post here details the development and the code was (is) available as is for others to learn and develop their own PBP projects. If parts of the code are used, then please respect the condition in the header and include all contributors in your code.

The second reason is that any item sold in the EU needs to me certain standards, especially electronics that switch mains potential. The risk in supplying a unit which is not CE marked to someone could result in the supplier / developer being sued or facing a custodial sentence should it be proved that the device was the cause of any loss to person or property. To get the device through the testing, documentation and due-diligence procedure would of cost near £10K, which even if I had the approval from the other parties involved in the code development, was ruled out as I didn't have that sort of money, and given the specialty of the project, there simply isn't enough of a market to recover that sort of cost, let alone make a small profit.

I have also been approached for the development files and schematics etc. These are now archived, and other than what's featured here, will not be posting up any further files, or updates that I have made.

In closing, I would like to re-iterate my thanks to Darrel and Henrik for their help in developing the code for this project, which insistently has been running now for over three years, 27/7, 365 days a year with not so much as a hick up. I've not even had to replace the battery in the RTC !

Malcolm

May I ask who makes the case and where you purchased it please?

The case was purchased from Mapln Electronics (UK). http://www.maplin.co.uk/instrument-a...losures-265883

Sorry for being such a noob I'm good with soldering but that's about it. The question I have is I noticed that several of the pins on the pic are not connected to anything I am assuming that they are used to program the pic, can someone tell me how to connect them or point me in the right direction.

Thanks much

I don't use them for programming the PIC, they are just unused ports. The PIC is programmed via the EasyPIC development board shown at the start of the thread