Posted Tue May 18, 2004, 11:59pm Subject: Bean Temperature Control On a PID'd Air Roaster
I posted this on alt.coffee but then thought that it might be of some interest here as well.
Recently it was observed by Jim Schulman and afterwards by others, that it is advisable to keep the supply air temperature in a hot air roaster rising as smoothly as possible for best results in the cup. But there are large excursions in temperature if a PID'd hot air roaster is controlled by the bean temperature.
There are previous threads on alt.coffee and maybe here as well discussing this in detail. But briefly, when controlling the roast with a PID controller by monitoring the bean temperature and using the feedback for PID control, large swings in the supply air temperature can be observed and these swings adversely affect the quality of the resulting coffee. On the other hand, when the roast is controlled by the heated supply air which then indirectly controls the bean temperature, the rise in temperature of both the supply air itself and the bean temperature is much smoother and the resulting cup is superior.
The problem lies in the fact that for roasting coffee, the primary interest is the temperature of the beans and since I changed over to running my hot-air roasts by PID-controlling the heated supply air, it's been hit-or-miss on some of the bean temperature profiles since I was no longer controlling the temperature of the beans directly but instead, controlling the supply air. It's been very difficult to set up a specific desired bean temperature profile indirectly by controlling the hot air at the inlet.
But recently I hit on a way to get a nearly perfect duplication, roast after roast, of a desired bean temperature profile translated, for the PID, into supply air parameters. It was so simple and was staring me in face all along but I didn't pick up on it until a few days ago. Then "eureka"; I knew how it could be done.
First I run one "calibration" roast using the PID to control the bean temperature ("the old-fashioned way"). Of course the supply air temperature goes all over the place as explained above and in previous threads. After the calibration roast I analyze the supply air temperature curve generated by that roast on a minute-by-minute basis and program the PID parameters for the supply air temperature from the results of the analysis. It works like a charm!
I've uploaded a graphic image of two roasts that I did yesterday comparing the two methods of PID control, as an example. You can see them on alt.binaries.coffee (Someone posted a Rugrats movie as well but it's buried in there somewhere) or you can see it here:
From these charts, the swings in inlet air when controlling by bean temperature and subsequent smoothness when controlling by supply air is quite clear. Actually the swings in this example are not as bad as many I've seen. After analyzing the resulting supply air curve in the top graph, I generated a new PID ramp/soak sequence and for the next roast, the one on the bottom, you can see that the PID is controlling the heated supply air using the results of the analysis. The two bean temperature curves are, for all intents and purposes, identical. No more guessing and I can now program any bean temperature profile that I desire and still control by supply air for an evenly rising profile.
Well - I had removed the "click here" graphic from my web page because of the "crashing silence" I received after the above post from May 19th. But, because of an off-list email from you, it's back up. You can, once again, click on the "click here" above and see the graphic with a brief explanation. I'm attaching a photo of my PID assembly controlling a Sirocco air roaster but I also use it for a Poppery1 with no problem.
coffeehabit Said:
Is there someone that can post details on how they PIDd the roaster? What PID controller? What input? What output? and from where to buy the stuff?
The PID is a Fuji PXR4. This PID is a bit "overkill". You can do this a lot cheaper but I wanted to be able to see the program vs. actual that this PID affords.
Input: I use a K-type thermocouple but other types can be used on this PID.
Output: I output to an SSR (a solid state relay). Other types of outputs are available as well.
I don't have written instructions, step-by-step, of how I put this together, but I'd be glad to give general guidance and answer any questions that come up if anyone is interested.
Yes - it has an RS485 output that I have interfaced to an RS485/RS232 converter and connected to the com port of my laptop. The PID comes with free software that will run in Windows through which you can set up most of the PID parameters and upload data. You can just make out the laptop in the corner of the photo. It will also upload data in a text format which can then be converted by Excel into a spreadsheet. That's what you see in the "click here" charts in my first post.
jim_schulman Senior Member Joined: 19 Dec 2001 Posts: 3,772 Location: Chicago Expertise: I live coffee
Posted Fri Jun 4, 2004, 7:09pm Subject: Re: Bean Temperature Control On a PID'd Air Roaster
Oops, missed this post. Here's the mini-faq:
Which roasters can be PID'ed? Only airroasters that are overpowered, i.e. that roast too fast on their own. The net effect of the PID is to switch the heater on and off very rapidly, thus slowing the roast down to the rate one wants. The effect of heat changes on temperatures in a drum roaster are too slow for simple PID control, and their automation requires more sophisticated computations. I would recommend the FR or poppers.
How much does it cost? About $175 minimum , more if you want a computer interface. A lot less if you are an EE and can build your own interfaces. The upside is that the PID is not part of the roaster and can be transferred.
What's a PID? A fancy thermostat that reads the temperature one wants to control, compares it to a setpoint, works out the error, and adjusts the heat accordingly. Unlike a conventional thermostat it can 1. tweak the heat to different levels rather than turn it full off or on; 2. Compensate appropriately by considering how the temperature is moving over time. It does this by adding integral (I) and derivative (D) components to normal proportional control (P). Google for more details, there's plenty of .edu lectures on the web.
What's a roaster PID? For roasting, the PID controller will need an add-on called ramp/soak control. This is a method for programming a time and temperature trajectory, say starting at room temperature, going to 350 in 2 minutes, to 450 in another 4, then to 500 in another 6 minutes. Each of these slopes is called a "segment," and many PID controllers come with 4, 8 or 16 of them.
How does one put together a PID controller for a roaster? You will need an appropriate PID controller, the Fuji PXR-3 sold by TTI global is excellent and cheap at $129 plus costs for any optional add ons. You will also need a 20 amp, 220V solid state relay activated by 12 volts DC with a heat sink (these can be switched on/off in a time proportioned 1 second cycle that the PID controller will calculate automatically - the 1 second cycle is fast enough so the heat flow to the roaster does not change. You will also need 2 thermocouples and one thermocouple reader. Action electronics and a few other places sell these cheap. Finally you will need a box, some electrical fixtures, and some knowledge of basic elecric wiring. The circuit instructions will come with the PID. If you have no knowledge of basic wiring, either get that knowledge or do not attempt this.
How does one program a PID? RTFM. The math one used to need to get the P, I, and D constants right are no longer necessary, since these units are self tuning. One gotcha -- these units use a moving average filter on the temperature input to filter out noise in industrial environements, set the time constant down to a haf second for roasting.
How does one set up a roast profile for a PIDed roaster? This was the gist if Bob's original post. If you don't know what the best bean profile should be, I recommend one manually profiled roast. In this roast use the front panel of the PID to manuall move the temperature up. Try to roast as quickly as possible below the first crack while keeping the beans at an even color appropriate to that bean temperature, and finishing the roast, start of first crack to end of roast, in 3 to 6 minutes, depending on your preferred balance of acidity and sweetness (fast for more acidity, slow for more sweetness). Here's the aprropriate bean colors: Go fast to 250F, lower temperatures don't do anything. Don't go above 300F until the beans are no longer green but have turned yellow, you can do this by going slow between 250 and 300F. The beans should stay very even and go from yellow to a light milk chocolate/cinnamon color from 300F to 380F. At 380F, the beans will get ugly, some going dark, some staying light, some being piebald. Go into the first crack slowly until they even up again, then speed up for the preferred roast finish.
Posted Sat Jun 5, 2004, 12:31am Subject: Re: Bean Temperature Control On a PID'd Air Roaster
jim_schulman Said:
"How does one set up a roast profile for a PIDed roaster? This was the gist if Bob's original post. If you don't know what the best bean profile should be, I recommend one manually profiled roast....."
This is not only a good way to set up a PID profile, but also an excellent general explanation of how to approach hot-air roasting, especially if one can control the profile using a Variac or something similar (on-off switching, etc.).
But to connect some dots and apply what Jim has described to the original premise of this post, If one can data-log and chart both the supply-air and bean temperatures simultaneously, using 2 TCs with this manually-profiled roast, one can then look at the resulting chart of the supply-air temperature profile obtained indirectly from the manually-controlled bean temperature.
Then, on a minute-by-minute basis, one can program the PID profile to match, as closely as possible, the supply-air graph obtained by the manually-controlled roast. Afterwards, one can run the PID program thus obtained and try a roast with the supply-air profile now controlling the PID. The bean temperature profile resulting indirectly from this roast should look fairly similar to the original manual roast. If not, a few tweaks should get even closer the next time and the resulting cup should be even tastier than the one produced by the original manually-controlled roast. This method seems better, to me, than hit-or-miss attempts to dial in on an acceptable profile when using a PID to control a roast using the supply-air temperature. That was the observation I was trying to convey by raising this subject.
But I also realize that some of you have not yet PID'd a hot-air roaster, so this method is kind of like jumping the gun, a bit...
Symbols: = New Posts since your last visit = No New Posts since last visit = Newest post
Forum Rules: No profanity, illegal acts or personal attacks will be tolerated in these discussion boards. No commercial posting of any nature will be tolerated; only private sales by private individuals, in the "Buy and Sell" forum. No cross posting allowed - do not post your topic to more than one forum, nor repost a topic to the same forum. Who Can Read The Forum? Anyone can read posts in these discussion boards. Who Can Post New Topics? Any registered CoffeeGeek member can post new topics. Who Can Post Replies? Any registered CoffeeGeek member can post replies. Can Photos be posted? Anyone can post photos in their new topics or replies. Who can change or delete posts? Any CoffeeGeek member can edit their own posts. Only moderators can delete posts. Probationary Period: If you are a new signup for CoffeeGeek, you cannot promote, endorse, criticise or otherwise post an unsolicited endorsement for any company, product or service in your first five postings.
![boby: PID.jpg](javascript:ImagePopup("/images/9524/PID.jpg", 704, 528, "boby: PID.jpg");)
= No New Posts since last visit 