PID Water Bath Temperature Controller Kit

[dantheman88988]

Hi All,

I'm new to this forum, I've spend several years working in process control in the Oil/Gas industry here in OZ.

I've developed a kit which is simply:
-PID + SSR enclosure
-Pt100 (not thermocouple)
-400mm stainless thermocouple & grommet
-Water Bath
-Heater
-USB temperature recorder

What this kit does is control fermentation temperature better than I believe can be achieved. The kit is installed inside a running fridge, tested with an output step response, results sent to me and I reply the correct PID parameters for your fridge and setup.

The fridge gives the environment a slowly decaying temperature floor, in effect a starting point. The water/water heating gives a high gain/highly controllable (T/D for those who know) linear system. Tuning of the D term is what makes this so highly advantageous over on/off controllers.

It's a relatively simple but elegant solution to a problem. Let me know if anyone is interested!

 

[mklojay]

I do work in the environmental sector for a lot of oil/gas clients...if I can remember correctly a PID runs about $1,000?

 

[Cambone]

What this kit does is control fermentation temperature better than I believe can be achieved.

There is so much going on here, I just don't know where to start.

 

[passedpawn]

A PID for our purposes goes for about $40. The ones we buy have autotune processes that should optimize control system coefficients automatically in situ. At least my Auber did, and seems to be perfectly critically damped.

I can't tell what's going on with the OP, but if he wants to offer more details of his system I'd love to look. Or are you selling it?

 

[dantheman88988]

I wont get into how PID actually works, there's plenty of other material for that. Further PID is not $1000? It's not even really a physical thing, it's a software process in a stand-alone controller, a PLC or a DCS. Even an arduino can do it.

One of the main problems with temperature control is the requirement of the presence of a D term in the PID equation. Typically D term's are avoided in most processes or rather they are engineered out by use of feed-forward control, cascade control etc (all way too complex for a simple homebrew setup). The D term amplifies noise and can lead to even more instability.

In temperature control they cant be avoided unless large overshoots are acceptable. Effectively you rely on very large temperature gradients to drive energy into a system, and then allow it dissipate. The problem can be if the driving force (ie heater) is at 90oC, the brew is at 24oC and the controller is then off the temperature will continue to go up. D control counteracts this by predicting that overshoot will occur.

However then, once a D term is introduced, you need to have a controllable system. A controllable system is defined in terms of the state space matrix that governs the system and whether the rank equals the number of poles (again not going into this). In basic terms this means that the output of the PID controller has to start to act, then act, then settle in a proportionate time, otherwise the PID controller has no meaning and ON/OFF is your only option.

I've tuned PID to control 100km 10" gas pipelines, gas fuel systems, chemical systems and these principles are well understood. Brewing is no different, in effect harder actually because such tight setpoint control is required.

 

[jfrizzell]

"better than I believe can be achieved."? So you're doing the impossible?

 

[mklojay]

Further PID is not $1000? It's not even really a physical thing, it's a software process in a stand-alone controller, a PLC or a DCS. Even an arduino can do it.

I thought you were refererring to THIS used for detection of volatile organics in an atmposhpere. I guess the photoionization detector is the software inside the handheld device?

 

[passedpawn]

I wont get into how PID actually works, there's plenty of other material for that. Further PID is not $1000? It's not even really a physical thing, it's a software process in a stand-alone controller, a PLC or a DCS. Even an arduino can do it.

One of the main problems with temperature control is the requirement of the presence of a D term in the PID equation. Typically D term's are avoided in most processes or rather they are engineered out by use of feed-forward control, cascade control etc (all way too complex for a simple homebrew setup). The D term amplifies noise and can lead to even more instability.

In temperature control they cant be avoided unless large overshoots are acceptable. Effectively you rely on very large temperature gradients to drive energy into a system, and then allow it dissipate. The problem can be if the driving force (ie heater) is at 90oC, the brew is at 24oC and the controller is then off the temperature will continue to go up. D control counteracts this by predicting that overshoot will occur.

However then, once a D term is introduced, you need to have a controllable system. A controllable system is defined in terms of the state space matrix that governs the system and whether the rank equals the number of poles (again not going into this). In basic terms this means that the output of the PID controller has to start to act, then act, then settle in a proportionate time, otherwise the PID controller has no meaning and ON/OFF is your only option.

I've tuned PID to control 100km 10" gas pipelines, gas fuel systems, chemical systems and these principles are well understood. Brewing is no different, in effect harder actually because such tight setpoint control is required.

Like I said, the PIDs available cheap to the homebrewer have an "autotune" process that can be run once to set these PID terms. Through an iterative process, with water and kegs and all, they automatically determine the ideal terms for the control system to be perfectly dampened. I could look at mine and see what the terms are, but anyway I get to temp quickly without overshoot (this wasn't the case before I tuned it!).

 

[dantheman88988]

Hi passedpawn, yeah autotune makes some assumptions though like the system is largely linear, which a fridge on a thermostat, and a water bath heater on a PID generally isnt. But yes this will work and along with a bit of fine tuning will work.

The key that I've found is the fact that you run the fridge at a temperature floor, minimise the water bath water volume and use the largest heater you can. Then you use an RTD which has less lag than a thermocouple and minimum noise filtering on the PID the system becomes very controllable.

I guess what I'm trying to see if there is interest in a kit that makes this viable for non technically minded people. The step response tuning (which is better than autotune) can be done by way of an email of the .xls file to me.

 

[dantheman88988]

Ok I've now completed the trials of the system, I'll post the temperature logs etc up on a blog once I figure out how to do it.

For those who are interested:

- I used a 23litre corny Keg as a fermenter. It sits inside a normal 30lt plastic fermenter with the gap filled with water. The important point is to minimise the water volume and maximise the height of the outer skin. This enables easier PID tuning and more effective heating of the internal column.

- A 500W aquarium heater sits in the water bath. The position of the RTD in the water bath is also quite important.

- A corny keg being stainless allows excellent heat transfer from the outside water bath. A plastic fermenter minimises heat loss to the environment (ie fridge).

- The whole assembly sits inside a fridge controlled to around 10oC. Taking cyclic defrost into account, any constant temperature can be achieved from 14oC and above.

- Temperature control was to within +/-0.1oC. Quite remarkable. This took many hours of tuning and particularly getting the D term right. It was quite interesting comparing the auto-tune with my manual tune parameters as they were vastly different.

- Total Cost:
Fridge - Already Have
30lt Plastic Fermenter as water bath - Already Have
3 - wire RTD + PID - $50
Components to Build Kit - $40
Aquarium Heater - $20
23lt New Keg - $140
Extra Corny Fermenter Lid - $20
9/16" Drill Bit for airlock - $20
Airlock + grommet - Already have

Thinking of building a DIY kit plus instructions and the all important tuning parameters.