Hi Arpi,

You seem to have a fairly flat period for a few minutes at around 350F, was this by design or as a result of PV racing ahead of SV and therefore the element being largely off to try and bring them back together?

My friend has managed to solder in an extension cable into the PXG, so the dataloader is back working. The graphing is still not going, it may be something to do with the RS232-USB convertor I have so going to borrow a friends laptop with a native 232 port to try.

I obviously haven't used your roaster, but I'd be willing to bet you could get a much more 'accurate' product by manually tuning it yourself. Alot of it can be done running the profile with the drum empty to get a feel of how things are acting and how much individual changes in PID settings affect things.

Cheers, Chris

Hi Gyro.

These were my ramp settings for that particular roast

SV-x TMxr
360F 4 mins
370F 2 mins
400F 3 mins
430F 10 mins



I wanted to enter the 400Fs slow because of the thermal inertia. The resulting roast did taste OK. It will take 4 days to mature but I can say now that it is not bad.

One of the problems I have is that I am experimenting I like to try things my way and see the results. What one person may find as good or bad may be the other way for another. With this type of roast I was trying to emphasize the caramelization stage. I think I did ok.

Cheers

Hi again,

Its good to experiment and see what you like etc. The point I am trying to make is that if you can accurately tune the PID, then it will do what you tell it to do better and more accurately. That way when you want to experiment more with subtle changes, it will actually achieve those changes for you and not be in a state of 'catch up' or 'overshoot' if you know what I mean.

I've presently just got mine programmed with a very simple curve and am attempting to get the PID to follow it as closely as possible. Once I have confidence the machine will do what I tell it to in terms of following the profile, then I can really experiment with faith that the profiles I want will be what is achieved.

Once I've got it nailed down, I'll post my settings. They will out of necessity be different to yours, but they may be of some use.

Another thing I have tried and am liking at the moment is limiting the minimum power to a set value. At the moment I've got it set so that the minimum power output is 25%. That way if I do get a little overshoot, heat is still going into the roaster which will in turn limit the eventual thermal momentum on the way down. This shouldn't really be necessary if its perfectly tuned, but its a easy way to limit the damage to the roast while you are figuring it out. I'm fairly certain my biggest problem lies in the integration settings still.

Cheers, Chris

Hi Gyro.

I haven't tried my self, but maybe the Fuzzy control (same PID as before but with less overshoot) would work better. The other option I may try is PID + self tuning (self readjusts parameters on the fly).

I've seen people that work only with soak parameters (staircase approach) but I am not sure what is best, if only soak, only ramp, or a combination.

The behavior of the roaster at the beginning/middle (beans absorb heat) of the roast is very different than at the end (beans release heat). Maybe one of the control methods mentioned above would work better.

When I did my auto tuning, I selected 'normal type.' But the 'low type' would avoid overshooting. This is good for the end of the roast but not for the beginning/middle (slower). This is another thing to try.

*pg 44 of the manual. There are seven different PID sets (palette) that can be stored. Then using the Digital input (push button), I can flip over the palette. That means that after 1C the beans are exothermic and the previous settings don't work (because the beans were endothermic before 1C). But if I flip the PID settings with the switch at 1C, I should get in theory something good

Let me know what you find

Cheers

To the best of my knowledge, the beans aren't exothermic for long (or at least, not dramatically so). I get an exothermic spike through 200C at 1st crack that pushes the PV ahead of the SV by a few degrees but thats about it.

I've got a few roasts planned for tomorrow, so another chance to try and narrow done the tuning a little more.

I'm very happy with the consistency of my BT probe, its much better than I had thought it would be. I was concerned with its stability but it hasn't presented any problems, yet...

Hi.

I readjusted the PID settings. My previous PID settings were done with an empty drum at 300F. They were:

p=7.3
i=172
d=33.1

My new PID settings were done (auto tune - normal mode) with a normal load at 360F. Huge difference in terms of numbers. My new numbers are:

p=22.5
i=499
d=96.0

While running the program for auto tune, the PID stopped at 362F but the T went on climbing till 395.6F. The thermal inertia beans/drum took it well into the 1C but without going into 2C. Then it when back down. The program did this a couple of times. The program lasted for maybe 30 mins. I am roasting now some Yemen beans and I hope that these changes will play a big difference in terms of following a profile.

Cheers

Nope. No good. Very low respond at low temp and the setting don't anticipate the thermal inertia at the end.

The first settings were much better to work with and I was happy with them So I'll go back.

Well, now I know je je

Cheers

Gyro,

I recently bumped into your article here on HB regarding the PID system install in your Toper.

I have been thinking about doing the same thing with my Cafemino and wondered how well the PID system is working for you.

Have the roasts improved? How long did it take to figure out how to modulate the PID to get good control of your roasts?

Have you experienced an increase in heating element life? And, if so how much.

If you where to do this again, is there anything that you would do differently? And why?

Many thanks.

CJ

I have the exact same setup as you and looking to do the same thing. Would you be willing to share the part numbers you used and any wiring diagrams? Thanks in advance!

Mike

Hi guys, sorry haven't been checking this thread. I managed to get it to profile a roast fairly well with 250g batch size, but anything over that it wasn't doing a very good job and rather than spend a LONG time trying to get it to work, I actually found it pretty easy to manually control the power by setting a percentage, ie I generally go 100% till 150C, 90% at 180C back off to 65% approaching 1st crack and then just ride it through to the end... thats with 1kg load and read from my bean temp probe which is in the revolving bean mass.

As for wiring diagrams etc, I have no recollection sorry, unless I have posted anything previously... I just pretty much pulled stuff out and replaced it as I thought appropriate. Too busy with little kids these days.

So yes it give more control etc but I think to profile a full size batch automatically you would really need to do the ET for profile control. I did BT and above 250g it was just too oscillatory. But I like it for the level of manual control I have now.

To be honest though, even though its a very old school contactor set up etc, it did a decent job before I hacked into it and I would be tempted to leave it alone. Even though its only ON or OFF as far as the element goes at present, there is a whole lot of thermal mass in the element itself, it still glows orange for a long time while the power is off to it, I think that smoothes out the reasonably long on/off contactor cycles. And you can still manually 'bump' the cycle by upping or downing the preset temp while roasting... thats what I used to do anyway.

A fun little project though, but the Fuji PID I used was a bit too expensive really.

Cheers, Chris